UC Berkeley

Abstract

Patterns of computed conserved noncoding sequence loss

following the paleopolyploidies in the maize and Brassica lineages

and their functional consequences

by

Sabarinath Subramaniam

Doctor of Philosophy in Plant Biology

University of California, Berkeley

Professor Michael Freeling, Chair

Following polyploidy, duplicate genes are often deleted, and if

they are not, then duplicate regulatory regions are sometimes lost.

By what mechanism is this loss, and what is the chance that such a

loss removes function? To explore these questions, we followed

individual Arabidopsis thaliana-Arabidopsis thaliana conserved

noncoding sequences (CNSs) into the Brassica ancestor, through a

paleohexaploidy and into Brassica rapa. Thus, a single

Brassicaceae CNS has six potential orthologous positions in

Brassica rapa; a single arabidopsis CNS has three potential

homeologous positions. We reasoned that a CNS, if present on a

singlet Brassica gene, would be unlikely to lose function compared

to a more redundant CNS, and this is the case. Redundant CNSs

go nondetectable often. Using this logic, each mechanism of CNS

loss was assigned a metric of functionality. By definition, proved

deletions do not function as sequence. Our results indicated that

CNSs that go nondetectable by base substitution or large insertion

are almost certainly still functional (redundancy does not matter

much to their detectability frequency), while, those lost by inferred

deletion or indels are about 75% likely to be nonfunctional.

Overall, an average nondetectable, once-redundant CNS > 30 bps

in length has a 72% chance of being nonfunctional, and that makes

sense because 97% of them sort to a molecular mechanism with

"deletion" in its description, but base substitutions do cause loss.

Similarly, proved-functional G-boxes go undetectable by deletion

82% of the time. Fractionation mutagenesis is a procedure that

uses polyploidy as a mutagenic agent to genetically alter RNA

expression profiles, and then to construct testable hypotheses as to

the function of the lost regulatory site. We show fractionation

mutagenesis to be a "deletion machine" in the Brassica lineage.