Summary No single gene exists in isolation, rather the genes in the genome make up an intricate network of interacting components that come together in different pathways and processes to generate a phenotype. Even small genetic changes can have far reaching consequences for an organism’s phenotype, let alone when large numbers of genes are different between strains, which is the case for those species with a pan-genome. For instance, a strain of the bacterial pathogen S. pneumoniae on average contains ~2100 genes, while the entire species harbors >4400 genes, which means that two random strains may differ by the presence and absence of hundreds of different genes. Within a biological network, many genes are dispensable, which within a pan- genome are mostly those genes that are variably present. In contrast, about 10-15% of genes in a bacterial genome are essential and keep an organism’s functionality intact under any circumstance. Due to their acute importance, essential genes are generally seen as rigid and largely immutable, consequently making them excellent targets for, for instance, antimicrobial therapies. However, by computationally interrogating thousands of S. pneumoniae strains, and 17 clinical strains experimentally, we have created a large dataset that shows that not all essential genes are ‘created equal’. Specifically, essential genes do not always seem to be present in all strains, and depending on a strain’s background, can sometimes be experimentally deleted. This raises the hypothesis that the essential gene concept is much more fluid than assumed and indicates that, under the right circumstances (i.e., genetic background), essential genes are evolvable and can switch to non-essential. In this proposal we aim to understand why some genes are essential, while others are not, we experimentally explore how essentiality can evolve, whether it is predictable, what the possible functional, phenotypic and/or evolutionary consequences are, and how we can take advantage of evolvable essentiality. Specifically, In Aim 1.1 several genomics tools are used to comprehensively map out evolvable essential genes in S. pneumoniae by sampling >85% of the genetic diversity in the pan-genome. In Aim 1.2 the evolvability of ~200 genes is explored with three validated strategies that reflect and uncover the ease in which an essential gene can become non-essential. And in Aim 1.3 we use machine learning to determine whether essential gene evolvability is predictable. In Aim 2.1 45 (evolvable) essential genes in cell wall synthesis and associated pathways are interrogated with CRISPRi-TnSeq to build a detailed interaction network. In Aim 2.2 we engineer paired-strains, where in one strain a gene is essential, and a near identical strain it is not. And in Aim 2.3 we use the paired- strains and employ different approaches to assign gene function and identify mechanistic consequences of evolvable essentials. In Aim 3.1 the paired-strains are used to dete...