UC Berkeley

The ability to transfer information from DNA to RNA to protein is a fundamental characteristic of life. The flow of information between biomolecules was first described using a model termed the “central dogma,” through which genetic information, stored in the four nucleotides of DNA, is transcribed to mRNA, and the resulting three base-pair codons are decoded during protein synthesis, or translation. Contemporarily, the organization of the central dogma is used as a scaffold for the diverse and varied mechanisms of cellular information flow across the tree of life. It follows that the genetic code is fundamental to the central dogma, as it defines the framework for how the codons of mRNA are interpreted and decoded into protein sequence. Given its universality and essentiality to protein synthesis, the genetic code was first proposed to be immutable, a “frozen accident in time” that was incapable of evolution. This is due in part to the diverse suite of translational machinery—ribosomes, release factors, tRNA’s, and their cognate tRNA synthetases—that is required to decode mRNA. Any change to the canonical encoding of twenty amino acids in sixty-one sense codons would necessarily require compensatory changes to this translational machinery, which was initially thought to be an insurmountable evolutionary challenge. Yet, the discovery of over thirty alternative genetic codes in the years since the genetic code was proposed 1961, demonstrates that it is subject to the same pressures that govern cellular evolution. An alternative genetic code encapsulates myriad types of codon reassignments, where either a sense or nonsense (i.e., stop) codon is captured and assigned to a meaning that deviates from the standard genetic code. These alternative genetic codes are found across all domains of the tree of life and can range in complexity. Most involve single codon changes, wherein the genetic code degeneracy allows for a sense codon to be reassigned to a pre-existing amino acid, as in the case of certain fungi, like Candida albicans and members of Saccharomycotina. In other, more extreme cases, like the yeast mitochondrial genome, six codons (five sense codons and one stop codon) have been reassigned to standard amino acids. The co-option of one of the three stop codons (TAG/UAG, TGA/UGA, and TAA/UAA) as a sense codon is a common form that alternative genetic codes take and is characteristic of several bacterial, archaeal, eukaryotic, and even phage genetic codes. Yet, in contrast to sense-to-sense codon reassignments, nonsense-to-sense genetic codes present a unique challenge in how cells adapt to complete or partial reassignment of one of its stop codons. In the former, complete stop re-assignment typically co-occurs with the gain of a codon-specific suppressor tRNA for the given codon, and, sometimes, loss of release factor activity. In theory, this buffers the cell from the challenge of balancing between truncated and elongated protein forms: a conundrum that arises when partial, or ambiguous, codon reassignment allows for a nonsense codon to be interpreted as both a stop and sense codon. Natural genetic code expansions represent a unique case of nonsense-to-sense codon reassignment, under which a stop codon is co-opted to either conditionally or stochastically encode an amino acid outside of the standard twenty. While there are only two known natural genetic code expansions to date—Selenocysteine (Sec) and Pyrrolysine (Pyl)—these two code expansions are emblematic of the molecular strategies that are used to navigate partial stop codon reassignment. In the former, charged Sec-tRNASec is formed through the post transcriptional modification of a charged Ser-tRNASec. Sec’s incorporation at UGA sites is contingent upon a proximal sequence motif, the Sec insertion sequence (SECIS) and is facilitated by Sec-tRNASec binding by a specialized elongation factor called SelB. Together, the coordination of these two processes allows for partial UGA reassignment to Sec by splitting UGA sites into two pools: 1) true stops, which lack the SECIS cue, and 2) true Sec sequences, which contain the proximal SECIS element. In contrast, Pyl is freely biosynthesized in the cell, and its incorporation at UAG sites is, to date, independent of either sequence motif or specialized elongation factor. Thus, how Pyl-encoding organisms navigate ambiguous interpretation of UAG sites, as either Pyl or stop, remains largely unknown. In this dissertation, I use Pyl as a system to address how ambiguity in information transfer (i.e., ambiguous UAG decoding) is productively maintained in the Pyl-encoding, methane-producing archaeon Methanosarcina acetivorans. Chapter 1 of this thesis discusses and reviews literature pertaining to the transcriptional regulation of methane metabolism in the domain Archaea, a class of organisms to which M. acetivorans belongs. This chapter begins by summarizing what is known of archaeal transcription and genomic organization. It proceeds to review known regulators, their targets, and their influence on methane metabolism in two model methane-producing archaea, (methanogens) with different energy conservation strategies. The first, Methanococcus maripaludis, relies on flavin-based electron transport to regenerate reducing equivalents in the absence of cytochromes and the quinone-like molecule, methanophenazine. We contrast regulatory responses during growth on reduced compounds like H2 and formate, highlighting outstanding questions related to signal detection and transduction. The second model organism addressed in this review, M. acetivorans, contains a complete electron transport chain. In this section we synthesize what is known of regulators, sensor kinases, and the interactions therein, which tune the expression of enzymes involved in methylotrophic and acetoclastic methanogenesis. Moreover, we discuss evidence for post-transcriptional regulation in M. acetivorans. We conclude by discussing how methodological advances can be applied to the questions raised in this chapter. Since transcriptional regulation represents one of many regulatory valves that tune the extent and degree of information transfer between DNA and mRNA, this chapter offers a framework for the regulatory trends discussed in subsequent chapters.Chapter 2 explores the world of Pyrrolysine and addresses how methanogenic archaea, like Methanosarcina acetivorans have adapted to an ambiguous genetic code. First, we examine the distribution and diversity of Pyl systems across the archaea (using a bioinformatic tool described in Appendix A). With data from this survey, we 1) establish the co-occurrence between pyl genes and associated metabolic machinery for methylamine methanogenesis, 2) evaluate the evolutionary history of Pyl biosynthetic and incorporation machinery, and 3) determine that while TAG usage decreases in Pyl users, patterns of stop codon usage are poor predictors for Pyl usage. We then test the functionality of diverse Pyl systems from uncultured archaea that break from Pyl’s co-occurrence trends and confirm that pyl genes in ANME-3 and Borrarchaeum weybense are likely vestiges of previously functional systems. Next, we sought to understand how genetic code ambiguity affects the physiology of Pyl-encoding methanogens like M. acetivorans. To this end, we provide conclusive evidence that M. acetivorans has an ambiguous genetic code where the amber codon (UAG) is decoded as both stop and Pyl. We use RNA-Seq to observe transcriptional changes that occur under pyl deletion and find that deletion mutants grown on methanol override substrate-specific transcriptional control and exhibit transcriptional profiles akin to wildtype cells grown on trimethylamine. Using our transcriptomic data, we present a model for how an ambiguous code, like that of M. acetivorans, might be tolerated. We posit that readthrough efficiency, or the suppression of UAG stop codons by Pyl, dictates the balance between the elongated and truncated forms of UAG-containing proteins. Under this model, we propose that Pyl demand outpaces Pyl supply during growth on trimethylamine, resulting in relatively low readthrough efficiency that drives UAG-containing proteins to their truncated forms. We conclude by proposing that readthrough efficiency is detected as a transcriptional cue via Pyl’s incorporation in regulatory proteins. Together, this chapter offers insights into how genetic code ambiguity can be productively integrated in an organism’s transcriptional regime. Chapter 3 tests the hypothesis that Pyl-containing sensor kinase, called RdmS, acts as a sensor module for readthrough efficiency, driving the transcriptional effects we observe in Chapter 2. If the failure to incorporate Pyl into RdmS is causal to transcriptional dysregulation observe in pyl deletion mutants, we expect a ∆rdmS strain to exhibit similar growth defects to ∆pyl strains on permissive substrates like methanol. Instead, ∆rdmS grows akin to wildtype cells on both methanol and trimethylamine. Simultaneously, we investigated whether truncations of rdmS to its sensory, ATPase, and first two PAS domains produced a detectable phenotype. While we observed alterations to growth in the ATPase and PAS2 domain complements, neither strain produced a significant defect relative to controls, likely due to the wide range of observed growth rates. To investigate whether RdmS’s role manifests at a transcriptional level, unobservable by growth, we performed RNA-Seq on a ∆pyl strain complemented with a constitutively long form of rdmS. Our results rule out RdmS as a candidate sensory module for readthrough efficiency. While it is unlikely that RdmS plays a role in Pyl homeostasis, our results offer additional insights into the in vivo physiology of RdmS and underscore its relationship to methylsulfide methyltransferases.Together, this dissertation dissects adaptations to code ambiguity across several steps in the central dogma. Using large-scale genomic surveys, we demonstrate that Pyl-based code expansion is widespread across the archaea, where it strongly co-occurs with a methylamine methanogenesis, a microbial metabolism of environmental importance. We establish that Pyl-encoding organism like Methanosarcina acetivorans jointly decode UAG residues as Pyl and stop, providing conclusive evidence that pools of UAG-encoding proteins are split into two forms: elongated and truncated. Insights derived from transcriptional profiling of a suite of Pyl mutants, complements, and wildtype strains enabled us to develop a model for how cells overcome the conundrum posed by ambiguous decoding of UAG as Pyl and stop by integrating this ambiguity in its transcriptional regime. Although the readthrough detection module remains elusive, we highlight candidates for future work and provide evidence ruling out RdmS’s involvement in this process. Through this, and future work, we will be able to draw a mechanistic throughline between genetic code ambiguity and its role in organismal physiology.