Decoding enhancer grammar using regressive evolution

Abstract

The PAX6 gene encodes a transcription factor expressed in the developing eye, brain, spinal cord and pancreas. Deletions downstream of PAX6 disrupt its expression and cause abnormal eye development, indicating that non- coding sequences are essential for the complex expression pattern of PAX6. Multiple-species sequence alignments identify non-coding conserved elements of tens to thousands of base-pairs. Many such elements drive consistent, tissue-specific and PAX6-overlapping expression in transgenic reporter assays and are described as tissue-specific ‘enhancers’.

It is hypothesised that enhancers together drive the complex expression pattern of PAX6, implying that variants within conserved enhancers may alter PAX6 expression in specific tissues. However, only one pathogenic PAX6 enhancer variant with loss of tissue-specific expression has been reported to date. Identification of other putative deleterious variants above the background of neutral variants is hampered by a lack of understanding of the rules of enhancer sequence features, or ‘grammar’.

In this thesis I explore whether evolutionary analysis can identify sequences important for PAX6 enhancer grammar. It has been reported that moles have an accelerated rate of evolution, or loss of conservation, in their PAX6 eye enhancers, a feature that correlates with regressed eye development in these mammals. These mole-associated enhancer variants could uncover features in PAX6 enhancers that are important for activity.

I used reporter assays to investigate whether accelerated evolution in mole PAX6 enhancers alters their activity during eye development. First, I investigated differences using a luciferase reporter assay in eye-related cell- lines. The distal Cre21/HS3 PAX6 enhancer was active in the RPE1 cell line, but reporter expression was variable between mole and non-mole species alike. Other PAX6 eye enhancers were inactive across cell lines. I therefore utilised live imaging of enhancer-linked fluorescent reporters in zebrafish embryos to survey enhancer activity across a range of cell-types and developmental stages. I found that the PAX6 enhancer with the most accelerated evolution – the neuro-retinal enhancer (NRE) from naked mole- rat (Heterocephalus glaber) - remained active across zebrafish retinal development. I concluded that this enhancer retains tissue-specific activity in early retinal development despite its accelerated evolution. I then employed a dual-enhancer-reporter design to directly compare activity of the naked mole-rat and human NRE in single zebrafish embryos. A surprising observation of loss of enhancer function was found to result from genome mis-integration, as identified by long-read sequencing. Recommendations are made for future study of potential loss-of-function variants.

A second, distal PAX6 enhancer – SIMO – was investigated in the dual reporter assay, given that a pathogenic variant in this enhancer was previously reported to disrupt lens expression. The naked mole-rat SIMO sequence carries a variant adjacent to that mutation, but this did not disrupt lens expression in dual-reporter zebrafish lines. I further investigated this difference using an in vitro binding assay.

In a final exploratory chapter, I move from considering enhancers as individual conserved elements to studying their combined effect on gene expression. I utilised the deep conservation of PAX6 enhancers in the compacted genome of Fugu (Takifugu rubripes) and assayed the consistency of the expression pattern driven by the Fugu pax6b regulatory region across independent zebrafish lines. I demonstrate the potential for this assay to compare the effect of regulatory region rearrangements on the pattern of gene expression.