diff --git a/introduction/introduction.tex b/introduction/introduction.tex old mode 100644 new mode 100755 index eea2a57..3c77cdf --- a/introduction/introduction.tex +++ b/introduction/introduction.tex @@ -125,7 +125,8 @@ \subsection{The biology of meiotic recombination} The axial elements are attached to chromatin, containing the compacted DNA of the sister chromatids in a series of loops that radiate outwards from the core axis of the SC. In the zygotene stage, peak levels of MSH4 foci are found, which mark most DSBs and is thought to promote synapsis\cite{Oliver-Bonet2005}. -MLH1 foci, thought to specifically influence DSBs to be repaired as crossovers, begin to appear in late zygotene. +MLH1 foci, thought to specifically influence DSBs to be repaired as crossovers and not non-crossover\cite{Baker1996}, begin to appear in late zygotene\cite{Oliver-Bonet2005}. +%MLH1 foci counts have been shown to mirror those of crossovers detected through linkage studies\cite{Tease2004,Gruhn2013}. By the end of zygotene, the paired sister chromatids in each axial element complete synapsis. The axial elements progressively join together at homologous regions, assisted by transverse elements that ``zip'' the structure together, and are bound in the central core by central element proteins\cite{Yang2009}. @@ -209,7 +210,7 @@ \subsection{Timing of meiotic events} Another possibility, suggested by a study in the Icelandic population\cite{Kong2004}, is oocytes that have higher recombination rates are more likely to survive to become successful embryos. Since a higher rate of recombination is linked to a lower incidence of aneuploidy, it is possible that oocytes with lower crossover counts were more likely to be aneuploid. These aneuploid oocytes would therefore be discarded somehow, by a cell cycle checkpoint mechanism for example. -Therefore, these could not be observed in a study looking as presumably healthy, or at least viable, offspring and would giving the impression of a recombination rate increase. +These could not be observed in a study looking at presumably healthy, or at least viable, offspring and would giving the impression of a recombination rate increase. %%% other possibilities? rec events during arrest period. @@ -415,15 +416,14 @@ \subsection{Distribution within the genome} The distribution of recombination within the genome has been shown to vary considerably between individuals, sexes, and populations. At the broad scale, crossovers are not distributed equally across the genome. The recombination rate is higher towards the telomeres\cite{Broman1998,Mcvean2004,hapmap2007}. -This effect seems to be primarily driven by males, with crossover occurring more frequently in telomeric reions, while females have higher recombination rates closer to the centromeres\cite{Kong2002,Coop2008,Kong2010}. +This effect seems to be primarily driven by males, with crossover occurring more frequently in telomeric regions, while females have higher recombination rates closer to the centromeres\cite{Kong2002,Coop2008,Kong2010}. Numerous studies in humans have found that recombination tends to be depressed within gene regions overall, with recombination high just upstream and downstream of gene regions\cite{Mcvean2004,Myers2005,hapmap2007,Spencer2006,Kong2010}. -In addition, when looking further away from gene regions, around 500 kb away, recombination appears to be elevated, then falling off after a few hundred kb. - +When looking further away from gene regions, recombination appears to be elevated up to around around 500 kb away, then falls off after a few hundred kb more. Evidence for sex differences in recombination has also been suggested to occur around gene regions. \citet{Kong2010} found that recombination was lower within genes overall, and that female recombination was lower within genes, but higher in the surrounding regions. -Additionally, while females have a lower rates within genic regions, males have higher rate within exonic regions. +Additionally, while females have a lower rates within genic regions, males have higher rate within exonic regions\cite{Kong2010}. In addition, results from LD maps in humans have examined the proportion of recombination occupying various proportions of sequence. From these analyses, it is estimated that the majority of recombination, 80\%, occurs in approximately 20\% of the sequence\cite{Mcvean2004,Myers2005,hapmap2007} @@ -436,19 +436,20 @@ \subsection{Recombination under genetic control} A number of genetic factors have been identified that alter properties of recombination, a summary of which can be found in Table \ref{tab:introGenes}. In a sperm typing analysis, \citet{Jeffreys2005} found a SNP whose minor allele suppressed the ratio of crossover to gene conversion near a particular hotspot. -Additionally, this variant was found to be overtransmitted, an example of meiotic drive. +Additionally, this variant was found to be overtransmitted in the offspring. +This is an example of meiotic drive, in which forces acting during meiosis alter the expected transmission ratios for a particular locus, causing over- or under-transmission. Another study in the Icelandic population found an association with an inversion at 17q21.31 on recombination rate\cite{Stefansson2005}. The presence of this inversion is associated with an increase in crossover rate in females, but not males. % RNF212: indentified in Saccharomyces cerevisiae and Caenorhabditis elegans -RNF212 has been associated with recombination rate in an Icelandic population\cite{Kong2008}, and this has been replicated in a number of follow-up studies\cite{Chowdhury2009,Fledel-Alon2011,Reynolds2013,Kong2014}. +In addition, RNF212 has been associated with recombination rate in an Icelandic population\cite{Kong2008}, and this has been replicated in a number of follow-up studies\cite{Chowdhury2009,Fledel-Alon2011,Reynolds2013,Kong2014}. Interestingly, the linkage of two particular SNPs near this gene is associated with an increased recombination rate in males, and a corresponding decreased rate in females. RNF212 is a homologue of ZHP-3, a \textit{C.\ elegans} gene involved in crossing over and disjunction, localizing to sites of crossover, and aids in the change in chromatin structure that occurs with the disassembly of the synaptonemal complex\cite{Bhalla2008}. Mouse RNF212 was found to have similar function, facilitating synapsis, and forming crossover-stabilizing structures\cite{Reynolds2013}. %Rnf212 KO mice are sterile but achieve complete synapsis. In addtion RNF212 possibly influences the decision to repair a DSB as a crossover rather than a gene conversion\cite{Reynolds2013}. -A study by \citet{Chowdhury2009} using 2310 meioses found significant associations with variants associated with four gene regions. +A study by \citet{Chowdhury2009} using 2,310 meioses found significant associations with variants associated with four gene regions. Two were associated with female recombination rate (KIAA1462, PDZK1), and the other two with male rate (UGCG, NUB1). These genes are poorly characterized and their function, beyond potential meiotic roles, remains unknown. However, this study reinforces the suggestion that male and female recombination rate are controlled via differing genetic factors. @@ -462,13 +463,13 @@ \subsection{Recombination under genetic control} Genome-wide associations with hotspot usage were replicated in the Icelandic population with strong effects\cite{Kong2010}. In non-human studies, both RNF212, and REC8, a SC-associated protein, have also been shown to associate with recombination rate in cattle\cite{Sandor2012}. - +% \afterpage{ \begin{table}[p] \centering \small \begin{tabular}{|l|c|c|p{1.3cm}|p{1.3cm}|c|c|} \hline Gene / region & Chr. & Association & Female rate & Male rate & Study & Replication \\ \hline - RNF212 & 4 & rate & + & - & \citet{Kong2008} & \cite{Chowdhury2009,Fledel-Alon2011,Reynolds2013,Kong2014,Campbell2015} \\ +\rule{0pt}{2ex} RNF212 & 4 & rate & + & - & \citet{Kong2008} & \cite{Chowdhury2009,Fledel-Alon2011,Reynolds2013,Kong2014,Campbell2015} \\ 17q21.31 inversion & 17 & rate & + & none & \citet{Stefansson2005} & \cite{Fledel-Alon2011,Chowdhury2009,Kong2014} \\ KIAA1462 & 10 & rate & + & none & \citet{Chowdhury2009} & \\ PDZK1 & 1 & rate & + & none & \citet{Chowdhury2009} & \\ @@ -487,20 +488,14 @@ \subsection{Recombination under genetic control} \end{table} \clearpage} - -\subsection{Heritability of recombination modifiers} - +\paragraph{Heritability of recombination modifiers.} +Several lines of evidence indicates that modifiers to recombination are heritable. The 2004 deCODE study, by analyzing siblings in large families, suggested that recombination rates were broadly heritable\cite{Kong2004}. -The pedigree analysis of the Hutterize population by \citet{Coop2008} was the first to find extensive variation among individuals in terms of their hotspot overlap (the proportion of an individual's crossover that overlap with known hotspots). -Another significant finding was that hotspot overlap was heritable. -This raises the possibility that other aspects of the recombination process may also be heritable. -And supports previous data from sperm typing finding that a hotspot-proximal SNP had recombination-suppressing effects, and that this SNP is over-transmitted to progeny\cite{Jeffreys2005}. - -A follow-up study in 2011 reinforced the heritability of hotspot usage as a phenotype, and expanded the scope\cite{Fledel-Alon2011}. -This study found that both male and female recombination rates are heritable. -Additionally, associations with RNF212 and the inversion on 17q21.31 were replicated. - -% Ottolini2015: selection for maternal recombination rates. (Kong2004 suggested this as well). +The pedigree analysis of the Hutterites population by \citet{Coop2008} was the first to find extensive variation among individuals in terms of their hotspot overlap (the proportion of an individual's crossover that overlap with known hotspots). +Furthermore, it was found that the extent of hotspots overlap was also heritable. +Further work reinforced the heritability of hotspot usage as a phenotype, and supported the finding that both male and female recombination rates are heritable\cite{Fledel-Alon2011}. +In addition, selection has been shown to act to retain eggs with higher recombination rates\cite{Ottolini2015}, a phenomenon previously suggested to explain the apparent increase in recombination rate in older mothers\cite{Kong2004}. +This raises the possibility that other aspects of the recombination process may also be heritable, but further work in larger cohorts is will necessary to determine the full extent of these modifiers. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% @@ -510,7 +505,6 @@ \section{Sexual dimorphism in recombination.} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsection{Heterochiasmy} - \afterpage{ \begin{table}[p] \footnotesize %\begin{tabular}{|P{3.2cm}|c|c|p{1.6cm}|p{1.3cm}|c|p{1.7cm}|} @@ -557,7 +551,6 @@ \subsection{Heterochiasmy} \end{table} \clearpage} - The Marshfield map\cite{Broman1998}, provided some of the first genome wide evidence of recombination rate variation across the human genome, and between males and females. Particularly interesting was the finding that recombination rates are higher in the telomeres, especially in males, and that females have a 1.56-fold higher rate of recombination in the autosomes. The estimation of this ratio proved to be accurate, despite the low marker coverage and few meioses used, and has been reinforced through numerous follow-up studies\cite{Broman2000,Kong2002,Coop2008,Kong2010,Bleazard2013,Campbell2015,Bherer2016}. % (Table \ref{tab:introHeterochiasmy}). @@ -565,7 +558,7 @@ \subsection{Heterochiasmy} This example in humans provides an illustration of heterochiasmy, the unequal distribution of recombination rates between the sexes of a species. Humans are among a large majority of species with data currently available in which the female recombination rate is higher than that of the male (Table \ref{tab:introHeterochiasmy}). -Related to female-biased heterochiasmy is the Haldane-Huxley rule\cite{Haldane1922,Huxley1928}, which states that in organisms in which one sex has an absence of recombination (achiasmate) it is the heterogametic sex. +Related to heterochiasmy is the Haldane-Huxley rule\cite{Haldane1922,Huxley1928}, which states that in organisms in which one sex has an absence of recombination (achiasmate) it is the heterogametic sex. Because recombination is suppressed between the sex chromosomes in heterogames, \citet{Haldane1922} hypothesized that this suppression carried over to the autosomes as well. But the Haldane-Huxley rule does not apply to species in which neither sex is achiasmate. There are a number of known species in which heterogametes have an equal or higher recombination rate than homogametes, and several other possibilities have been proposed. @@ -579,13 +572,13 @@ \subsection{Heterochiasmy} % (recombination disrupts favorable haplotypes in the most fit individuals, therefore is selected against). This idea has been criticized however, especially because selection in an adult, diploid organism is unlikely to effect recombination rate at the haploid stage\cite{Lenormand2003}. -\citet{Lenormand2005} suggested that selection at the haploid stage may act differently in males and females and that this, and not heteromorphic sex chromosomes, could account for the heterochiasmy observed in plants. -A more recent study\cite{Otto2015} supports this idea, suggesting an additional level of parental control. +\citet{Lenormand2005} suggested that haploid selection may act differently in males and females and that this, and not heteromorphic sex chromosomes, could account for the heterochiasmy observed in plants. +A more recent study\cite{Otto2015} supports this idea, suggesting an additional level of parental control over the level of selection experienced by a given gamete. In mammals, the idea of haploid selection to effect dimorphism is an attractive one to explain heterochiasmy, although the strength of selection is markedly lower. Since oogenesis is arrested at dictyotene and does not complete until fertilization, the haploid stage is essentially absent in females. Therefore, there is a higher potential for selection in male haploids, even if only a handful of genes are actually expressed\cite{Dadoune2004}. -In addition, there is evidence for elevated female recombination rates in imprinted regions\cite{Lercher2003}, which further confounds this issue. +In addition, there is evidence for elevated female recombination rates in imprinted regions\cite{Lercher2003}, which further confounds this issue of male or female control over haploid selection. In most studied species, including humans, females have a higher rate and a longer map length than males, with a ratio ranging from just above 1, to 14 in the European tree frog\cite{Berset-Brandli2008}, and nearly 17 in Atlantic salmon\cite{Danzmann2005}. Estimates in humans, perhaps the most well studied organism, have demonstrated little variation since the earliest linkage studies, with a ratio of 1.6. @@ -601,10 +594,7 @@ \subsection{Heterochiasmy} Cytological studies of meiotic cells have provided convincing evidence that the sex differences in crossover count are driven by, or parallel differing chromatin configurations across meiotic prophase. -Cytological evidence for crossovers focuses on markers for MLH1, a protein involved in breakpoint repair that is specifically associated with crossover, and not non-crossover\cite{Baker1996}. -MLH1 foci counts mirror those of crossovers detected through linkage studies\cite{Tease2004,Gruhn2013}. -The entirely of the SC can also be visualized and measured with markers to SYCP3, a transverse element protein. - +Using this technique, the entirely of the SC can be visualized and measured with markers to SYCP3, a transverse element protein. The SC has been shown to be substantially longer in females than males, translating to a correspondingly larger DNA loop size\cite{Tease2004,Gruhn2013}. This relationship has been proposed to account for the differences in crossovers observed between males and females. Males and females have chromosomes of identical size (among the autosomes), so packaging the same amount of DNA into a smaller total SC length would naturally mean a larger loop size. @@ -628,30 +618,33 @@ \subsection{Maternal age effect on recombination rate.} %%% A study by \citet{Hussin2011} examined in recombination in 195 maternal meioses from French-Canadian pedigrees. -Here, the opposite effect was seen, with recombination rate found to decrease with maternal age, with a larger effect size, estimating a decrease somewhere between $-$0.49 and $-$0.42 events each year. +Here, the opposite effect was seen, with recombination rate found to decrease with maternal age, with a larger effect size. +The crossover count was estimated to decrease by $-$0.49 to $-$0.42 events each year. Differences in the direction of the effect here may be due to real differences between populations, when considering the French-Canadian as a genetic isolate, or simply due due to a lack of power with only 195 meioses. %%% Another study reported a slight decrease in crossover count with maternal age in 338 meioses, finding a decrease of $-$0.29 events per year\cite{Bleazard2013}. In a direct test of the production line hypothesis, \citet{Rowsey2014} examine more than 8,000 oocytes, finding no significant change in crosover count among oocytes. -Thus, it appears that the number of crossovers in a given oocyte is not governed as a function of order of meiotic entry, and there is a lack of evidence for any effects of the production line hypothesis on crossover count. +Thus, it appears that the number of crossovers in a given oocyte is not controlled as a function of order of meiotic entry, and there is a lack of evidence for any effects of the production line hypothesis on crossover count. While evidence for an age effect on crossover count in females is conflicting, these studies all agree that there is no age effect present in males. -A recent meta-analysis by \citet{Martin2015} provided much need insight into the age effect issue. +A recent multi-cohort analysis by \citet{Martin2015} provided much need insight into the age effect issue. Using a combination of nine cohorts comprising $>$6,000 meioses, the authors report a modest but significant increase in the crossover count with age. The authors used a comprehensive and systematic approach that avoids the methodological differences between previous studies. %%% -An interesting suggestion from this study, was that of possible confounding factors upon the maternal age effect. +An interesting suggestion from this study was that of possible confounding factors upon the maternal age effect. First, that assisted reproductive technologies, including IVF, may provide an artificial selection for oocytes with a greater number of crossovers. Second was the possibility that oral contraception, which suppresses ovulation, could somehow alter the age-count association, especially if the production-line hypothesis holds true for humans. However, neither of these possibilities could be controlled for with the power of this study and remain undetermined. +Several other studies approach this question from another direction. Evidence in mice point to sex differences in cell cycle checkpoints that control for, and terminate cells with excessive DNA damage, or without chiasmata\cite{Cohen2006}. The first of at least two checkpoints in prophase I is triggered by lack of synapsis, or breaks in the DNA that have been missed by repair machinery. A second checkpoint, in late pachytene, senses chromsomes without chiasma, and female mice appear to frequently bypass this checkpoint, while male mice do not. Thus is appears that female mice may have a less stringent set of checkpoints, and that this may account for the increased incidence of meiotic abnormalities\cite{Cohen2006}. -Additional evidence from an analysis of single oocytes provides evidence that maternal recombination rate is highly variable within a single individual, with 41.6 $\pm$ 11.3 crossover per oocyte\cite{Ottolini2015}. -This study revealed a selection against transmission of non-recombinant oocytes at meiosis II, which were more likely to be found in the second polar body instead of the transmitted oocyte. +An analysis of single oocytes provides evidence that maternal recombination rate is highly variable within a single individual, with 41.6 $\pm$ 11.3 crossover per oocyte\cite{Ottolini2015}. +This study revealed a selection against transmission of non-recombinant oocytes at meiosis II. +These achiasmate products were more likely to be found in the second polar body instead of the transmitted oocyte. This evidence outlines a potential mechanism by which non-recombinant, potentially aneuploid oocytes could be eliminated from the germ line. Furthermore, recombination, thought to be limited to prophase I, was shown to influence events in meiosis II, much later than previously thought. @@ -661,30 +654,6 @@ \subsection{Maternal age effect on recombination rate.} Aging effects on the recombination process point to a carefully regulated process that has potential to degrade with time. Though much work has been done, both theoretical and experimental, no clear explanations have emerged, and the question of the cause of heterochiasmy remains unanswered. -%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -\section{Recombination in non humans} -%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% - - -Recombination in humans is of great interest to us as a species, and much effort has been focused here. -However it is of great interest to learn about recombination in all species to put human recombination in an evolutionary context. -Studies of recombination have been done in a wide variety of organisms to date, a partial list of which is summarized in Table \ref{tab:introHeterochiasmy}. - -Chimpanzees, the most recent common ancestor to humans, have a LD-based recombination map\cite{Auton2012a}, but no pedigree maps are yet available, leaving open questions regarding sex differences. - -Pedigree maps have been generated in mice\cite{Broman2002}, the most recent of which was generated by \citet{Cox2009}, using 3,546 meioses, but a low density of markers. -More recently, data from the Collaborative Cross\cite{CollaborativeCrossConsortium2012}, an inbred population generated from eight founder strains, has been used to generate sex-specific maps within the mouse genome\cite{Liu2014}. -The researchers here leveraged the breeding funnel approach from the Collaborative Cross, gathering genotype data from sibling pairs, and using computational techniques to infer recombination events. - - -Yeast has been the subject of a number of studies, due to their ease of use as a model organism and much of what we know today comes from yeast. - -Recombination maps are available in a number of other species, but most are limited to LD studies, due to the steep resource requirements of pedigree studies. -Most recently, study of recombination was released in yeast\cite{Lam2015} and birds\cite{Singhal2015}. -These studies provided an evolutionary perspective on recombination initiation and hotspot evolution. - %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% @@ -696,13 +665,12 @@ \section{Hotspots} \subsection{Initial discovery of hotspots} Sperm typing produced the first set of well-characterized hotspots in humans, initially focusing within the MHC\cite{Jeffreys2000,Jeffreys2001}. -The correlation of hotspot locations identified by sperm typing in this regions with breakdown of LD measurements provided support for the use of LD methods to find hotspots of recombination genome-wide, without the expense and limitations of single-cell assays\cite{Jeffreys2001}. -In 2005, \citet{Myers2005} produced a fine-scale recombination map in the human genome using LD methods. +The correlation of hotspot locations identified by sperm typing in this regions with the breakdown of LD measurements provided support for the use of LD methods to find hotspots of recombination genome-wide, without the expense and limitations of single-cell assays\cite{Jeffreys2001}. +Hotspots were found genome wide using LD methods\cite{Mcvean2004}, and it was in 2005 that \citet{Myers2005} produced a fine-scale recombination map in the human genome using LD methods. Along with this map, hotspots were found to be a ubiquitous feature of the human genome, with a set of $\sim$25,000 found, occurring roughly every 50 kb. -The Hapmap LD study expanded the characterized set of hotspots to more than 30,000 throughout the human genome\cite{hapmap2007}. -These LD studies of recombination also estimated the proportion of recombination occurring within various fractions of the total genome sequence, finding that recombination was intensely concentrated, with 80\% of all recombination occupying less than 20\% of sequence. - -% Extensive analysis by sperm typing has indicated that hotspots are generally 1-2 kb in width\cite{Jeffreys2004a,Arnheim2003}. +The Hapmap LD map expanded the characterized set of hotspots to more than 30,000 throughout the human genome\cite{hapmap2007}. +These LD maps of recombination also estimated the proportion of recombination occurring within various fractions of the total genome sequence, finding that recombination was intensely concentrated, with 80\% of all recombination occupying less than 20\% of sequence. +Extensive analysis by sperm typing has indicated that hotspots are generally 1-2 kb in width\cite{Jeffreys2004a,Arnheim2003}. \subsection{Discovery of PRDM9} @@ -710,8 +678,7 @@ \subsection{Discovery of PRDM9} Since hotspots were first looked at in detail, and the discovery that they were common and spread across the entire genome, questions regarding a possibly regulatory mechanism for hotspots have persisted. \citet{Myers2005} found that hotspots shared many common features including repeat elements THE1A/B, and a common sequence motif (CCTCCCT) located at the centers of many hotspots. A further famly of motifs were identified via the use of the Hapmap phase 2 LD map\cite{hapmap2007}, encompassed by the degenerate 13 bp motif (CCNCCNTNNCCNC)\cite{Myers2008}, estimated to be involved in up to 40\% of all hotspots. - -Spencer2006 : hotspots associated with increased GC content, and GC increasing mutations, a likely result of the action of biased gene conversion. +Hotspots were found to be associated with increased GC content, GC increasing mutations, a likely result of the action of biased gene conversion\cite{Spencer2006}. In 2010, a series of papers by three separate groups converged on the identification of PRDM9 from different approaches. @@ -728,7 +695,7 @@ \subsection{Discovery of PRDM9} \subsection{PRDM9 alleles} The PRDM9 zinc finger array has tremendous variability and there are a number of alleles present. -The major alleles A and B, are both present in a high percentage of European individuals, and differ by only a single amino acid change\cite{Baudat2010}. +The major alleles A and B, are both present in a high percentage of European individuals, and differ by only a single amino acid change, while the I allele encodes a longer zinc finger array\cite{Baudat2010}. The A and B alleles are predicted to recognize the degenerate 13-mer motif, but the I allele is not\cite{Baudat2010}. This effect is seen at the DSB-level as well, with PRDM9 A and B variants contributing to generate similar DSB hotspots\cite{Pratto2014}. @@ -745,71 +712,6 @@ \subsection{PRDM9 alleles} PRDM9 heterozygosity was also found to influence hotspot strength. -%%%%%%%%%%%%%%%%%%%% -\subsection{Hotspots in other species} -%%%%%%%%%%%%%%%%%%%% -%%%%%%%%%%%%%%%%%%%% -% \subsection{Conservation between species} combine -%%%%%%%%%%%%%%%%%%%% - -Hotspots have been discovered in a number of other species. -Recombination in chimpanzees is strongly influenced by hotspots, although there is a notable absence of a strong DNA motif for PRDM9 binding, in contrast to humans\cite{Auton2012a}. -One possibility is that chimpanzee PRDM9 has less specific binding, and targets a much wider variety of target sequences than humans. - -Mice contain approximately 15,000-20,000 hotspots, also under the regulation of PRDM9\cite{Brick2012,Smagulova2011} -%Mice: Brick2012 (Hotspots via ChIP); Smagulova2011. 15-20,000 total - - -% hotspot stability: -In addition, PRDM9 has been subject to rapid evolution across a wide variety of species and taxa\cite{Oliver2009,Ponting2011}. -Humans and chimpanzees have a complete absence of hotspot sharing, despite a high degree of overall DNA sequence identity\cite{Ptak2005,Winckler2005,Auton2012a}. -Evidence points specifically to the rapidly evolving zinc finger DNA binding array as an explanation for the lack of shared hotspots between humans and chimpanzees\cite{Myers2010}, and a wide variety of mammals\cite{Oliver2009,Ponting2011,Thomas2009}. -Even within different human populations, there are substantial differences in hotspot specification and usage, driven primarily by differences in PRDM9 alleles (discussed above). - -Evidence suggests that, in the absence of PRDM9, hotspots continue to persist within the genome. -% discuss species w/o prdm9... -Furthermore, without the rapid changes in hotspot specification driven by PRDM9 evolution, these hotspots tend to be stable in evolutionary time. -A recent study compared hotspot locations of four species of yeast, and found that hotspots were frequently shared, providing evidence for hotspots sharing that spans approximately 15 million years of evolutionary divergence\cite{Lam2015}. -In two species of birds, the zebra finch and the long-tailed finch, hotspots were again found to be shared, despite several million years of divergence\cite{Singhal2015}. - -Evidence for hotspots has also been found in dogs, which lack PRDM9, and these hotspots are characteristically different from those found in humans. -Dog hotspots appear to have a lowered intensity and occupy a wider range (4-18 kb)\cite{Axelsson2012,Auton2013} when compared to humans, although care must be taken in the interpretation of these data. -Additionally, dog hotspot appear to be localized near gene promoter regions\cite{Auton2013}, a seemingly common feature of PRDM9-absent species. - -%%%%%%%%%%%%%%% -% discussion? : -% Furthermore, without PRDM9, hotspots appear to be localized towards gene promoter regions, which tend to coincide with regions of open chromatin. -% Discuss hotspot paradox vs stable hotspots theory (latter saying hotspots are confined to specific chromosome features (promoters/GC content), as in yeast and dogs) -%%%%%%%%%%%%%%% - - -% PRDM9 absent / present - - -%%%%%%%%%%%%%%%%%%%% -\subsubsection{Species lacking PRDM9} -%%%%%%%%%%%%%%%%%%%% - -PRDM9 appears to be a essential component of recombination in a large number of species, however it is not a ubiquitous feature. -PRDM9 is absent in a number of species, including birds, lizards, amphibians, dogs, and fruit flies\cite{Ponting2011,Oliver2009}. - -The canid family provides an interesting look at recombination, since evidence exists that multiple truncating mutations occurred in canine PRDM9, rendering the gene inactive. -The first sign that PRDM9 might be missing in dogs came with the publication of the first draft sequence of the domestic dog genome, in a boxer, in 2005\cite{Lindblad-Toh2005}. -Since then a number of studies have looked at dogs and their close relative within the family Canidae to determine when and how PRDM9 became inactivated. -PRDM9 was found to be disrupted in the closest relative of dogs, wolves, as well as coyotes\cite{Munoz-Fuentes2011}, revealing that inactivation was not a result of domestication, or a limited event. -Additional studies found multiple PRDM9 mutations in both the Island Fox and Andean Fox\cite{Auton2013}, but not the cat and panda\cite{Axelsson2012}. -This indicates that the mutations must have happened at some point after the divergence of canids from panda, which occurred approximately 49 Mya\cite{Oliver2009,Axelsson2012} -Despite the loss of this gene, canids are able to successfully complete meiosis and recombination and produce fertile offspring, raising questions as to the requirement of PRDM9 in meiosis. - - -Intriguingly, a recent study in humans identified a healthy mother carrying a homozygous knockout of PRDM9 predicted to render the protein inactive\cite{Narasimhan2016}. -In PRDM9 knockout mice, meiosis is not able to complete properly\cite{Brick2012}. -However, this mother had three healthy children, one of which carried the mutation. -In this transmission, crossovers at PRDM9 binding locations reduced in number. -This observation by \citet{Narasimhan2016} raises the possibility of a backup mechanism for the completion of meiosis in the absence of PRDM9 in the human genome. - - - %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% @@ -818,28 +720,25 @@ \section{Recombination and disease} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -The most direct association of recombination with disease is the association with aneuploidy due to meiosis I errors\cite{Hassold2001,Hassold2007}. -As recombination has such a high impact on the structure of the genome, it also has the possibility to cause genomic instability if errors occur during the break and repair of DNA. +The most direct association of recombination with disease is with aneuploidy due to meiosis I errors\cite{Hassold2001,Hassold2007}. +Otherwise, since recombination has such a high impact on the structure of the genome, it also has the possibility to cause genomic instability if errors occur during the break and repair of DNA. Defects in recombination have been associated with a number of disorders caused by genomic instability. Nonallelic homologous recombination (NAHR), also known as ectopic exchange results in structural genomic rearrangements and is a major cause of recombination-associated copy number alteration. For example, 22q11.2 deletion syndrome is believed to be caused by improper pairing and crossing over between low copy repeats (LCRs) that flank the region, causing it to be deleted in the recombinant chromosome\cite{Emanuel2008}. Ectopic exchange contributes to a number of other diseases associated with genomic rearrangements\cite{Stankiewicz2002,Liu2012}. \citet{Berg2010} reported that variation found at the PRDM9 locus contributed to genome instability. -\citet{Pentao1992} found that rearrangements between repeat regions associate with the the deleteion of a 1.5 Mb region, associated with Charcot-Marie-Tooth disease type 1A (CMT1A). +\citet{Pentao1992} found that rearrangements between repeat regions associate with the deletion of a 1.5 Mb region, associated with Charcot-Marie-Tooth disease type 1A (CMT1A). Deletion of the 7q11 region in sperm causes Williams-Beuren syndrome\cite{Turner2008}. PRDM9 has also been implicated in genomic instability associated disease. PRDM9 hotspot motifs have been found in regions associated with disorders of genomic instability\cite{Myers2008}. In a sperm-typing study, non-A alleles of PRDM9 were found to be protective against the risk of rearrangements leading to CMT1A and hereditary neuropathy with liability to pressure palsies (HNPP)\cite{Berg2010}. Here, a protective effect against genome rearrangement was seen in men homozygous for the N/N allele, with a lesser effect in heterozygous A/N individuals. - In addition, PRDM9 has been associated with children with B-cell precursor acute lymphoblastic leukemia (B-ALL)\cite{Hussin2013}. Here, a rare PRDM9 allele (C) was found in a number of families, and was tied to the abnormal X of crossover events, and associated with B-ALL. - -%Berg2011: - Coupled with the finding of a healthy PRDM9 knockout in humans\cite{Narasimhan2016}, these disease association raise questions as to the requirement and effects of PRDM9 in meiosis. +%Berg2011: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Interference} @@ -873,24 +772,22 @@ \subsubsection{Crossover interference models} Several biological explanations have been proposed to account for the action of crossover interference in the genome, and are reviewed here. An early suggestion was that steric interference was responsible for interference. -In this case, a cluster of proteins that are necessary to facilitate resolution of the DSB would bind, the prevent further attachment at nearby sites. +In this case, the cluster of proteins that are necessary to facilitate resolution of the DSB would bind to the DNA, preventing further attachment at nearby sites. This idea was mostly discounted after cytological studies failed to observe complexes of a sufficient size to enable interference over long enough distances. -One aspect that models seek to address is that of crossover homeostasis, and crossover assurance. -Crossover assurance captures the idea that as least one crossover is required per chromosome, which is thought to prevent non-disjunction. -In a typical meiotic progression, multiple DSBs are created in the DNA and only a small fraction of themresolve to crossovers, with the remainder repaired as non-crossover gene conversions. +Models seek to account for two key aspects governing crossover placement, crossover homeostasis, and crossover assurance. +Crossover assurance captures the idea that as least one crossover is required per chromosome, which is thought to be required to prevent non-disjunction. +Crossover homeostasis is somewhat related, but deals with the ratio of crossovers to non-crossovers. +In a typical meiotic progression, multiple DSBs are created in the DNA and only a small fraction of them resolve to crossovers, with the remainder repaired as non-crossover gene conversions. Studies in yeast have shown that the final number of crossovers remains at the same level even when reducing the number of precursor DSBs\cite{Martini2006}. Thus, it appears that there is some regulatory mechanism in place to ensure at least one crossover per chromosome. Another implication of crossover homeostasis is that it has the potential to change the ratio of crossovers to gene conversion events. Since crossovers tend to be maintained at a constant level, the number of gene conversions decreases. -%%% crossover homeostatis -%%% crossover assurance - \paragraph{The mechanical stress model.} \citet{Kleckner2004} proposed a model in which interference is governed by mechanical forces. -This model starts from the basis that chromatin configuration changes over the course of meiosis, undergoing expansion and contraction cycles. +This model starts from the basis that chromatin configuration changes over the course of meiosis, undergoing cycles of expansion and contraction. These expansions and contractions create waves of periodic increased or decreased stress that propagate along the chromosome. DSBs create ``cracks'' in the DNA, reducing the amount of stress directly near the crack. The amount of relief slowly decreases with greater distance away from the break until the stress returns to normal levels. @@ -899,32 +796,30 @@ \subsubsection{Crossover interference models} First, it ensures that at least one crossover will occur on each chromosome, as long as enough stress is generated. Second, it allows for specification of the location of stress and therefore DSBs and crossover, based on, for example, chromatin configuration. Finally, this model allows for interference to act at multiple time points in meiosis using the same mechanism. -However, the mechanical stress model assumes that all crossovers are interfering, and does not explain the observation that some organism appear to have two pathways of interference. +However, the mechanical stress model assumes that all crossovers are interfering, and does not explain the observation that some organisms appear to have two pathways of interference. \paragraph{The polymerization model.} -The polymerization model describes the maturation of DSBs to crossovers in terms of a time-dependent polymerization event spreading along the synaptonemal complex\cite{King1990}. -The progression of DSBs to chiasmata to mature crossovers is described in terms of a precursor structure binding, triggering this maturation. +The polymerization model describes the maturation of DSBs to cross-overs in terms of a time-dependent polymerization event spreading along the synaptonemal complex\cite{King1990}. +The progression of DSBs to chiasmata to mature crossovers is described in terms of a precursor structure binding, which triggers this maturation. In the model, early recombination precursors form randomly along the chromosome while it is tied into the synaptonemal complex. Some of these early events mature into chiasmata and become crossovers. As this maturation occurs, a polymerization event is triggered that moves in both directions away from the crossover. -The expanding polymer blocks the binding, or forces detachment, of further precursor structures that would lead to crossover, thus creating an interference effect. -The ejected precursors could then re-bind at another location on the chromosome. +The expanding polymer blocks the binding (or forces detachment) of further precursor structures that would lead to crossover, thus creating an interference effect. +The ejected precursors can then re-bind at another location on the chromosome. This model makes a distinction between early and late crossover events, guarantees at least one crossover occurs, and captures the distance-dependency of the interference effect. However, no evidence of a polymer structure has been observed in cytological studies. -Furthermore, studies in mice have shown that the synaptonemal complex is not required for interference to act\cite{DeBoer2007}. - -%\subsubsection{Mathematical models of crossover interference} +Furthermore, studies in mice have shown that the formation of the synaptonemal complex is not required for interference to act\cite{DeBoer2007}. \paragraph{The counting model.} The counting model does not depend on SC length, but instead proposes that crossovers must be separated by a discrete number of non-crossover gene conversion in between\cite{Foss1993,Foss1995}. In this model, a fixed number of events (presumed to correspond to DSBs) are placed randomly upon a chromosome. Then, the events are classified in one of two ways, either crossovers or gene conversions. -Crossovers form the minority of the mature events with a specific number of gene conversion in between. +Crossovers form the minority of the mature events with a specific number of gene conversions in between. Therefore, no two crossovers can occur directly beside one another, and some degree of distance-dependent spacing is maintained. -Data in Drosophia were found to fit this model well, it was found that it poorly predicted interference in budding yeast and humans\cite{Foss1995}. -Therefore, extensions to this model were made to allow a second class of non-interfering crossovers to exist along-side those that are interfereing (see ``The simple gamma model'' on page \pageref{cointTPM}). +Data in Drosophia were found to fit this model well, but it was found that it poorly predicted interference in budding yeast and humans\cite{Foss1995}. +Therefore, extensions to this model were made to allow a second class of non-interfering crossovers to exist along-side those that are interfering\cite{Housworth2003} (see ``The simple gamma model'' and ``The two pathway model'' below). % on page \pageref{cointTPM}). A key element of the counting model is that interference here depends on genetic distance rather than physical, or SC distance. This allows interference to be estimated in different species with differing genomic properties, such as chromosome size. @@ -940,9 +835,11 @@ \subsubsection{Crossover interference models} \paragraph{The simple gamma model.} \label{cointTPM} +The gamma model is an extension of the counting model, in which inter-crossover distances are moeled instead of counts, which are not observed in inferential studies of interference. The gamma model starts with the assumption that all crossovers are capable of interfering with each other, and chromatid interference is neutral. The location of chiasmata on a tetrad bundle are determined by a stationary renewal process(*). -The inter-chiasmata distances are are modeled to follow a gamma distribution with a single parameter, $\nu$, representing the strength of interference. +Each chiasmata is considered an ``arrival'' that resets the probability controlling the distance from the previous arrival. +The inter-chiasmata distances are are represented by a gamma distribution with a single parameter, $\nu$, representing the strength of interference. In this model, $\nu <$ 0 corresponds to negative interference, $\nu=$ 0 no interference, and $\nu>$0 positive interference. The shape and rate parameters of the gamma distribution are not independent, but instead represented with shape of $\nu$ and rate 2$\nu$. Therefore, the gamma distribution has a mean of 0.5 Morgans, and a standard deviation of $1 / ( 2 \sqrt{\nu} )$. @@ -951,7 +848,7 @@ \subsubsection{Crossover interference models} In typical recombination studies, the locations of the chiasmata are unknown and only the crossover locations are observed. From a four strand tetrad, crossing over occurs twice as often as what is observed on a single product, so chiasmata are thinned to become crossovers. Since chromatid interference is neutral, each chiasma has a 0.5 probability of becoming a crossover. -The mean inter-chiasma distance of 0.5 therfore translates into an average distance of 1 Morgan between crossovers. +The mean inter-chiasma distance of 0.5 therefore translates into an average distance of 1 Morgan between crossovers. This model allows crossover locations from transmitted genotype data to be used to estimate the parameters of a gamma distribution that represent the strength of crossover interference. @@ -973,24 +870,30 @@ \subsubsection{Crossover interference models} \subsubsection{Data on crossover interference} \paragraph{Cytological measurement of interference} -A cytological study in mice measured distances between foci marking crossover events in terms of relative distance along the SC, and models these distances using a gamma distribution\cite{DeBoer2006a}. +Cytological methods have been used to study interference, relying on the identification of markers for DSBs, crossovers, and the SC during meiosis. +A cytological study in mice measured distances between foci marking crossover events in terms of relative distance along the SC. +These distances were modeled using a gamma distribution\cite{DeBoer2006a}. The researchers looked at MLH1 foci, which mark crossovers at the pachytene stage, as well as MSH4 loci, which occur just prior, in the zygotene stage. -Positive interference was observed in both foci at both stages, however it was much stronger in the later staage, marked by MLH1 foci at pacytene. +Positive interference was observed in both foci at both stages, however it was much stronger in the later stage, marked by MLH1 foci at pacytene. A follow-up to this study found that interference was not affected by the lack of an intact SC, and that interference remained present even without complete synapsis\cite{DeBoer2007} -Furthermore, interference between x foci indicate that interference can act between DSBs and is not limited to the subset that resolve as crossovers\cite{Baudat2007}. +Furthermore, interference can act between DSBs and is not limited to the subset that resolve as crossovers\cite{Baudat2007}. These studies suggest that, at least in mice, interference acts across at least two stages of meiotic prophase, and is not dependent on the full assembly of the synaptonemal complex. Furthermore, these observations appear to support the mechanical stress model of interference. - -Since this study, this method has been used in number of other organisms\cite{Barchi2008,Basheva2008}. -% \cite{Barchi2008,Basheva2008,Vozdova2013,Borodin2008,Borodin2008b,Borodin2009,Mary2014} +Since this study, this method has been used in number of other organisms %\cite{Barchi2008,Basheva2008}. +including mice\cite{Barchi2008}, dogs\cite{Basheva2008},pigs\cite{Mary2014}, +cattle\cite{Vozdova2013},wildebeest\cite{Vozdova2013},cats\cite{Borodin2008a},shrews\cite{Borodin2008},and mink\cite{Borodin2009}. % Barchi2008: mice. cytological interference is reduced in ATM KO mice. % Basheva2008: dogs. positive COint, nu=6.5 %%% Table on cytological interference? % Many of these found similar levels of crossover interference, suggesting that interference is a conserved feature. +In addition these studies provide evidence that the centromere is not a barrier to interference in any of the studied species and acts over the entire chromosome, a finding reinforced through inferential studies from human pedigrees\cite{Broman2000,Fledel-Alon2009}. \paragraph{Strength across the genome.} +Conflicting results have been reported on how the strength of interference varies across the genome. +%, and among chromosomes of differing sizes. +That is, does the chromosome length govern the strength of interference? A cytological study of crossover interference in human males\cite{Lian2008} found that interference strength was high in smaller chromosome, and decreased with larger chromosomes. However, this data was re-analyzed by \citet{Housworth2009}, this time using the two-pathway model, who found that interference strength was constant across all chromosomes. @@ -1000,15 +903,13 @@ \subsubsection{Data on crossover interference} % Getz2008: support two pathway model of interference., and contradit Baudat2007, suggesting that interference acts in two pathways, and that either can result in a CO or NCO (in yeast). % The finding that interference can act between DSBs in mice\cite{Baudat2007} suggests that interference may affect the placement of gene conversion events as well. -Youds2011: review on co/nco choice +% Youds2011: review on co/nco choice %%%%%%%%%% \paragraph{Interference inferred from pedigree studies.} -The centromere is not a barrier to interference\cite{Broman2000,Fledel-Alon2009}, and acts over the entire chromosome. - The recombination initiation maps generated by \citet{Pratto2014} provided important data regarding the initiation of DSBs that will lead to crossover events. -One suggestion from this study is that interference could act between nearby hotspots, inhibiting the formation of a second DSB nearby, supporting the idea DSB-DSB interference found in mice. +One suggestion from this study is that interference could act between nearby hotspots, inhibiting the formation of a second DSB nearby, supporting the idea of DSB-DSB interference found in mice\cite{Baudat2007}. The crossovers identified from single oocytes by \citet{Hou2013} allows valuable data to be inferred regarding both crossover interference, affecting the spacing between pairs of crossovers. @@ -1018,9 +919,9 @@ \subsubsection{Data on crossover interference} A reanalysis of this data using genetic distance is presented in this thesis in Chapter \ref{ch:cointExtras}. -\subsubsection{Other methods of measuring coint} -Coefficient of coincidence. -% The ratio of probability +% \subsubsection{Other methods of measuring coint} +% Coefficient of coincidence. +% % The ratio of probability %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsection{Chromatid interference} @@ -1043,6 +944,93 @@ \subsection{Chromatid interference} % Petkov2007 + + +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% +\section{Recombination in non humans} +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% + + +Recombination in humans is of great interest to us as a species, and much effort has been focused here. +However it is valuable to learn about recombination in other species to put human recombination in an evolutionary context. +Studies of recombination have been done in a wide variety of organisms to date, a partial list of which is summarized in Table \ref{tab:introHeterochiasmy}. +Recombination maps are available in a number of non-human species, but many are limited to LD studies, or low-resolution linkage analysis, due to the high resource requirements of pedigree studies, and the lack of availability of high quality genome builds or assay methods. + +Chimpanzees, the most recent common ancestor to humans, have a LD-based recombination map\cite{Auton2012a}, but no pedigree maps are yet available, leaving open questions regarding sex differences. +This sex averaged map shows that chimpanzee recombination is broadly similar to that of humans, with increased rates near the telomeres. +Recombination in chimpanzees is strongly influenced by hotspots, although there is a notable absence of a strong DNA motif for PRDM9 binding, in contrast to humans\cite{Auton2012a}. +One possibility is that chimpanzee PRDM9 has less specific binding, and targets a much wider variety of target sequences than humans. + +Pedigree maps have been generated in mice\cite{Broman2002}, the most recent of which uses 3,546 meioses, but a low density of markers\cite{Cox2009}. +Mice contain approximately 15,000-20,000 hotspots, also under the regulation of PRDM9\cite{Brick2012,Smagulova2011} +More recently, data from the Collaborative Cross\cite{CollaborativeCrossConsortium2012}, an inbred population generated from eight founder strains, has been used to generate sex-specific maps within the mouse genome\cite{Liu2014}. +The researchers here leveraged the breeding funnel approach from the Collaborative Cross, gathering genotype data from sibling pairs, and using computational techniques to infer recombination events. + +Hotspots have been discovered in a number of other species, both with and without PRDM9. +% hotspot stability: +PRDM9 has been subject to rapid evolution across a wide variety of species and taxa, contributing to rapidly diverging hotspot locations between species\cite{Oliver2009,Ponting2011}. +Humans and chimpanzees have a complete absence of hotspot sharing, despite a high degree of overall DNA sequence identity\cite{Ptak2005,Winckler2005,Auton2012a}. +Evidence points specifically to the rapidly evolving zinc finger DNA binding array to explain the lack of shared hotspots between humans and chimpanzees\cite{Myers2010}, and between a wide variety of other mammals\cite{Oliver2009,Ponting2011,Thomas2009}. +Even within different human populations, there are substantial differences in hotspot specification and usage, driven primarily by differences in PRDM9 alleles (discussed above). + +PRDM9 appears to be a essential component of recombination in a large number of species, however it is not a ubiquitous feature for meiosis. +PRDM9 is absent in a number of species, including birds, lizards, amphibians, dogs, and fruit flies\cite{Ponting2011,Oliver2009}. +Intriguingly, a recent study in humans identified a healthy mother carrying a homozygous knockout of PRDM9 predicted to render the protein inactive\cite{Narasimhan2016}. +In PRDM9 knockout mice, meiosis is not able to complete properly\cite{Brick2012}. +However, this mother had three healthy children, one of which carried the mutation. +In this transmission, crossovers at PRDM9 binding locations were reduced in number but recombination seemed otherwise normal. +This observation by \citet{Narasimhan2016} raises the possibility of a backup mechanism for the completion of meiosis in the absence of PRDM9 in the human genome. + + +Perhaps most interesting and relevant to this thesis is recombination in dogs. % , which are interesting because of their loss of PRDM9. +The canid family provides an interesting subject for recombination studies, since evidence exists that multiple truncating mutations occurred in canine PRDM9, rendering the gene inactive. +Linkage maps in dogs have been available for a number of decades\cite{Mellersh1997,Neff1999}, but +the first sign that dog PRDM9 might be missing came with the publication of the first draft sequence of the domestic dog genome, in a boxer, in 2005\cite{Lindblad-Toh2005}. +Since then a number of studies have looked at dogs and their close relative within the family Canidae to determine when and how PRDM9 became inactivated. +PRDM9 was found to be disrupted in the closest relative of dogs, wolves, as well as coyotes\cite{Munoz-Fuentes2011}, revealing that inactivation was not a result of domestication, or a limited event. +Additional studies found multiple PRDM9 mutations in both the Island Fox and Andean Fox\cite{Auton2013}, but not the cat and panda\cite{Axelsson2012}. +This indicates that the mutations must have happened at some point after the divergence of canids from the panda, which occurred approximately 49 Mya\cite{Oliver2009,Axelsson2012} +Despite the loss of this gene, canids are able to successfully complete meiosis and recombination and produce fertile offspring, raising questions as to the requirement of PRDM9 in meiosis. +Evidence for hotspots has been found in dogs and these hotspots are characteristically different from those found in humans. +Dog hotspots appear to have a lowered intensity and occupy a wider range (4-18 kb)\cite{Axelsson2012,Auton2013} when compared to humans, although care must be taken in the interpretation of these data. +Additionally, dog hotspot appear to be localized near gene promoter regions\cite{Auton2013}, a seemingly common feature of PRDM9-absent species. + +% A pedigree study in dogs \cite{Wong2010}... +%Yeast has been the subject of a number of studies, due to their ease of use as a model organism and much of what we know today comes from yeast. + +Evidence suggests that, in the absence of PRDM9, hotspots continue to persist within the genome. +Most recently, two studies of recombination were released in yeast\cite{Lam2015} and birds\cite{Singhal2015}, two species lacking PRDM9. +These studies provided an evolutionary perspective on recombination initiation and hotspot evolution. +%, showing a conservation in hotspot usage over millions of years of evolutionary divergence. +% discuss species w/o prdm9... +Hotspot locations of four species of yeast were compared, and it was found that hotspots were frequently shared, with a high overlap, providing evidence for hotspot sharing that spans millions of years of evolutionary divergence\cite{Lam2015}. +Without the rapid changes in hotspot specification driven by PRDM9 evolution, these hotspots tend to be stable in evolutionary time. +In addition, in two species of birds, the zebra finch and the long-tailed finch, hotspots were again found to be shared, despite several million years of divergence\cite{Singhal2015}. + + +%%%%%%%%%%%%%%%%%%%% +% \subsection{Hotspots in other species} +%%%%%%%%%%%%%%%%%%%% + + +%%%%%%%%%%%%%%% +% discussion? : +% Furthermore, without PRDM9, hotspots appear to be localized towards gene promoter regions, which tend to coincide with regions of open chromatin. +% Discuss hotspot paradox vs stable hotspots theory (latter saying hotspots are confined to specific chromosome features (promoters/GC content), as in yeast and dogs) +%%%%%%%%%%%%%%% + +% PRDM9 absent / present + +%%%%%%%%%%%%%%%%%%%% +% \subsubsection{Species lacking PRDM9} +%%%%%%%%%%%%%%%%%%%% + + + + + %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Gene conversion} @@ -1072,40 +1060,42 @@ \section{Gene conversion} % Admixture Due to the small size and nature of gene conversion events, their detection within population genetic data has proven to be difficult. -Given the small size of gene conversion events of 50-1000 bp, it is unlikely that a given event would ovelap a typed SNP. +Given the small size of gene conversion events of 50-1000 bp, it is unlikely that a given event would overlap a typed SNP. Instead the event would occur in a region in which the donor and recipient homologues have the same sequence. The resulting conversion will result in no change to the genome, and would not be detectable outside of molecular methods to observe the event while it occurs. Furthermore, even if a gene conversion overlaps a SNP, it would only be observed if the SNP is heterozygous. Within the human genome, even the highest density SNP arrays, consisting of $\sim$2 million markers, will cover sites spaced on average 1,500 bp apart. Not all of these positions are heterozygous, and therefore is a further limitation. -Most gene conversions therefore cannot be detected with this method. +Most gene conversions therefore will be missed with this method. In addition, genotyping error is a major factor that can complicate the detection of these events. Genotyping error rates for SNP arrays are typically under 0.05\%\cite{Imai2010}, however at this rate, with an array of 2 million SNPs, 1,000 of these will be improperly typed. +%%% Ptak2004 Despite the limitations, it is possibly to use genome wide pedigree data to detect gene conversion, as demonstrated in a recent study. -\citet{Williams2015} use 34 three generation pedigrees to study and detect gene converion within 98 meioses using first SNP arrays, with follow up validation via sequence data. +\citet{Williams2015} use 34 three generation pedigrees to study and detect gene conversion within 98 meioses using first SNP arrays, with follow up validation via sequencing. The use of three generation pedigrees provides a way to overcome the risk of false positive calls arising from genotyping error. -First, in order to confidently call an event, the conversion must be detected in the first generation, and again in the grandchildren. +First, in order to confidently call an event, the conversion must be detected in the first generation, and be transmitted through to the grandchildren. In addition, the parent must transfer the alternate allele from a putative gene conversion site to one of the other children. This latter requirement ensures that both of the parent's alleles are correctly typed. Gene conversion tract lengths were estimated at 100-1000 bp, but these estimates may be biased towards longer lengths due to the low SNP array density. -In addition, gene conversion was found to cluster within 20-30 kb invervals in several cases, a feature not previously seen. +In addition, gene conversion was found to cluster within 20-30 kb intervals in several cases, a feature not previously seen. \citet{Williams2015} demonstrated the success of a pedigree approach to gene conversion detection, providing valuable information on transmission across single generations. However, the sensitivity of this approach is limited. From a set of 98 meioses, with approximately 20-60 crossovers expected in each, one would expect to find $\sim$3,500 crossovers. -Given that gene conversion occurs 10 times more frequently that crossover, this results in $\sim$35,000 gene conversion events in the entire dataset. +Given that gene conversion occurs 10 times more frequently that crossover\cite{Jeffreys2004,Baudat2007,Cole2012}, this results in $\sim$35,000 gene conversion events in the entire dataset. Even accounting for the proportion of invisible events, only $\sim$100 gene conversions were detected. The analysis of gene conversion in pedigree data provides valuable data, however it is incomplete and should be combined with results from other approaches. Sperm typing has proven to be a powerful technique with which to study gene conversion events. \citet{Jeffreys2004} described three human hotspots in detail. -% 4-15 x CO, 55-290 bp tract length +This study found that gene conversion occurs 4-15 time as frequently as crossover with the genome. +In addition gene conversion events occupied a tract that was estimated to range from 55 to 290 bp, highlighting the variability in length. +The gene conversion tract length was further estimated to be around 500 bp\cite{Cole2012} in follow up studies. -% \cite{Cole2014} length diff --git a/latexmkrc b/latexmkrc index bfb8f72..0223e16 100644 --- a/latexmkrc +++ b/latexmkrc @@ -12,4 +12,5 @@ sub run_makeglossaries { push @generated_exts, 'glo', 'gls', 'glg'; push @generated_exts, 'acn', 'acr', 'alg'; -$clean_ext .= ' %R.ist %R.xdy'; \ No newline at end of file +push @generated_exts, 'bbl'; +$clean_ext .= ' %R.ist %R.xdy'; diff --git a/main.tex b/main.tex index 333493b..a03b715 100644 --- a/main.tex +++ b/main.tex @@ -1,6 +1,6 @@ \documentclass[11pt,twoside,openright,letterpaper]{memoir} \setsecnumdepth{subsection} % include numbering on subsections -\setcounter{tocdepth}{4} +\setcounter{tocdepth}{3} % Use 1/2-inch margins. % \usepackage[margin=1in,lmargin=1.5in,showframe]{geometry} % ,showframe % for margin boundary box