The DEAD-Box Protein Dhh1p Couples Mrna Decay And

The DEAD-Box Protein Dhh1p Couples mRNA Decay and

Translation by Monitoring Codon Optimality

Aditya Radhakrishnan

A dissertation submitted to The Johns Hopkins University in conformity with the

requirements for the degree of Doctor of Philosophy.

Baltimore, Maryland

December, 2016

Recent experimental ﬁndings have substantially advanced the notion that the codon- dependent rate of translation elongation is a major determinant of mRNA stability.

While the role of the ribosome in identification and decay of aberrant mRNAs has been well established, how the process of translation elongation and mRNA decay communicate is less well understood. Here, we report that the yeast DEAD-box protein Dhh1, long implicated in regulation of translation and activation of mRNA decay, acts as a sensor of codon optimality that targets an mRNA for decay. First, we find that Dhh1 specifically associates with and degrades mRNAs of low codon optimality. We show that messages with greater numbers of slowly translating ribosomes are preferential targets for Dhh1 mediated decay. Moreover, we note that these mRNAs are degraded by Dhh1-specific mechanism separate from the standard cellular ribosome quality control apparatus. We find that overexpression of Dhh1 leads to accumulation of ribosomes specifically on mRNAs of poor optimality. We supplement this with high-throughput sequencing analysis to show that Dhh1 overexpression leads to ribosomal stalling on specific non-optimal codons. Taken together

ii Abstract iii with the ﬁnding that Dhh1 is found to associate with ribosomes in vivo, these data suggest that Dhh1 acts a sensor for translation elongation, eciently coupling codon optimality with mRNA decay.

Rachel Green, Ph. D. (Sponsor and Reader)

Professor

Department of Molecular Biology and Genetics

Johns Hopkins University School of Medicine

Je↵ry Corden, Ph. D. (Reader)

Professor

Department of Molecular Biology and Genetics

Johns Hopkins University School of Medicine Acknowledgments

To Rachel, my sincerest thanks. The opportunity to work in such an intellectually engaging environment has been nothing short of a blessing. To have a mentor who is both understanding and supportive of my idiosyncrasies, even more so. The training

I’ve received and experiences I’ve shared will forever stay with me. Particularly pertaining to presentation slides.

To the varied, wonderful people I’ve had the privilege of calling my friends through my time in graduate school. To my academic family in Biophysics, and my adoptive family in BCMB. To the Green lab, both new and old, my thanks for the best set of work colleagues a guy could ask for. To Kristin, my long su↵ering baymate, and

Anthony (V), my long su↵ering new baymate. To Chris, who ignited my passion for climbing stu↵, and to Boris who nurtured it. To Nick, who taught me how to make proﬁling samples, and to Karen, who taught me how to make proﬁling samples. To

Julie and Beth, who ensured the only thing keeping me from getting my work done was me. To Colin and Kazuki, who brightened my days with unexpected snark. To

Alan, with whom I could geek out about music theory. To Dan, who seemed to think

I actually know what I’m doing at a computer. To Fuad and Jamie and Laura and

iv Acknowledgments v

Karole, who all have helped me maintain a sense of fun and perspective through the latter stages of my PhD. And to Anthony, who has been through this journey at the same time as me, and who is someone I’m truly lucky to call my friend. To all of you, thank you. The meaningless chats, beer, and jaunts to the Daily Grind have contributed more to the success of the thesis than you realize.

To my family. To my new parents, and by new brother and sister-in-law, who endeavor to always make me loved. To all my extended family, ever ready to support and help out. To my parents, whose unhealthy obsession with my well-being and happiness has lead to me completing this PhD — as well as pretty much any other notable accomplishment or success I’ve encountered. And, ﬁnally, to my wife, as I’ve saved the best for last. I love you all. Contents

Abstract ii

Acknowledgments iv

List of Tables ix

List of Figures x

1 Introduction 1

1.1 Mechanisms of RNA Decay ...... 2

1.1.1 Deadenylation and exosome-mediated mRNA decay ...... 3

1.1.2 Decapping-mediated mRNA decay ...... 4

1.1.3 Decay of aberrant mRNAs ...... 5

1.2 General connections between translation and mRNA decay ...... 8

1.2.1 Codon selection informs mRNA stability ...... 9

1.3 Figures...... 11

2 Dhh1 represses translation by modulating ribosome occupancy 15

vi Contents vii

2.1 Dhh1 is at the nexus of mRNA decay and translation repression . . . 16

2.2 Tethering of me31b and orthologs eciently represses translation . . 18

2.3 Both RecA domains of Dhh1 are necessary for translation repression . 19

2.4 Dhh1-tethered mRNA predominantly sequesters ribosomes ...... 20

2.5 Dhh1-mediated ribosome occupancy is not a termination defect . . . 23

2.6 Materials and methods ...... 26

2.7 Figures...... 34

3 Dhh1 stimulates decay of mRNAs with low codon optimality 42

3.1 Codon optimality underlies ecient translation ...... 43

3.2 Metrics for codon optimality ...... 44

3.2.1 CAI and tAI: Supply and demand at the codon level . . . . . 46

3.3 Dhh1p stimulates the degradation of mRNAs with low codon optimality 49

3.4 Dhh1p binds preferentially to mRNAs of low codon optimality . . . . 52

3.5 Materials and methods ...... 54

3.6 Figures...... 58

4 Dhh1 monitors codon optimality through ribosome elongation 68

4.1 Decay is stimulated by increasing numbers of slow-moving ribosomes 69

4.2 Dhh1p physically binds to the eukaryotic ribosome ...... 71

4.3 Ribosome occupancy is enhanced upon Dhh1 binding ...... 72

4.4 Discussion ...... 73 Contents viii

4.5 Materials and methods ...... 77

4.6 Figures...... 81

Bibliography 90

Vita 109 List of Tables

2.1 Characterized functions of Dhh1 and orthologs ...... 17 2.2 Reporter ribosome proﬁling samples and GEO sample numbers . . . . 24

3.1 Additional ribosome proﬁling samples and GEO sample numbers . . . 43

4.1 Publically available yeast strains generated for this study ...... 80

ix List of Figures

1.1 There is heterogeneity in the stability of mRNA transcripts ...... 11 1.2 Canonical mRNA decay in S. cerevisiae ...... 12 1.3 Decay of aberrant mRNAs in S. cerevisiae ...... 13 1.4 Codon optimality vs. mRNA stability in mice and yeast ...... 14

2.1 Characteristic sequence motifs in Dhh1 ...... 34 2.2 The crystal structure of Dhh1 ...... 35 2.3 me31b and Dhh1 repress translation in D. melanogaster ...... 36 2.4 Dhh1 requires both RecA-like domains to repress translation . . . . . 37 2.5 Catalytic activity in Dhh1 is required to sediment reporter mRNAs with polyribosomes ...... 38 2.6 Tethering Dhh1 increases ribosome density on reporter mRNAs . . . 39 2.7 Nucleotide resolution into ribosome occupancy by ribosome proﬁling . 40 2.8 Dhh1 does not a↵ect translation termination ...... 41

3.1 Experimentally observed vs. computationally predicted half-lives. . . 58 3.2 Codon contributions to mRNA half-life ...... 59 3.3 Codon e↵ects on mRNA stability is a tunable phenomenon ...... 60 3.4 Codon composition of HIS3 reporter mRNAs ...... 61 3.5 Dhh1 selectively stimulates decay of mRNAs with low codon optimality 62 3.6 Dhh1-mediated decay of mRNAs depends on the level of codon optimality 63 3.7 Loss of Dhh1 stabilizes low optimality mRNAs genome wide . . . . . 64 3.8 Dhh1-mediated decay is not due to mRNA secondary structure . . . . 65 3.9 Dhh1 preferentially binds with low optimality mRNAs ...... 66 3.10 Dhh1 associates with low optimality mRNAs genome-wide ...... 67

4.1 Dhh1 senses polarity of non-optimal codons within mRNAs ...... 81 4.2 Dhh1-mediated degradation is dependent on inecient translation . . 82 4.3 Dhh1-mediated degradation is dependent on ribosome pausing upstream of non-optimal stretches ...... 83 4.4 Canonical RQC proteins do not sense polarity of non-optimal codons 84 4.5 Pull-down of Dhh1 suggests association with the ribosome ...... 85

x List of Figures xi

4.6 Catalytically active Dhh1 modulates ribosome occupancy on mRNAs with low codon optimality ...... 86 4.7 A-site occupancy by non-optimal codons is increased on Dhh1 overexpression ...... 87 4.8 Dhh1 preferentially sequesters ribosomes on messages with low codon optimality ...... 88 4.9 Dhh1 is a general sensor of ribosome speed during elongation . . . . . 89 Chapter 1

Introduction

Note: Parts of this chapter were published in:

Radhakrishnan, A. & Green, R. (2016). Connections Underlying Translation and

mRNA Stability. J. Mol. Biol. 428 (18), 3558-3564.

A coding mRNA lives its life in three distinct phases: birth by transcription,

production of protein through translation, and ﬁnally death through decay. As the

central nexus through which information ﬂows in gene expression, mRNAs play a key

role in regulation of gene expression.1 Given that protein synthesis depends on the

availability of mRNA,2–4 understanding how cells regulate the availability of mRNA is of paramount importance to understanding gene expression. Speciﬁcally, the steady- state level of coding mRNAs is governed by transcription (which, for the purposes of this discussion, we take to include all subsequent processing steps required for ecient export and proper translation) and decay.

Eukaryotic transcription is a highly complex event, requiring the concerted ef-

1 1.1. Mechanisms of RNA Decay 2

fort of numerous proteins, subject to multiple levels of spatial and temporal reg-

ulation.5 Further sequence-mediated processing events (e. g. 7-methylguanosine capping, intron removal through splicing, 3’ terminal cleavage and polyadenylation) act to regulate the number of mRNA transcripts that are exported from the nucleus to the cytoplasm. However, upon export of mRNA transcripts to the cytoplasm, the primary mechanism of mRNA regulation is decay. Indeed, recent genome-wide exper- iments6–10 have shown a clear role for decay in mRNA regulation, demonstrating that there is substantial heterogeneity in the stability of mRNA transcripts, both between organisms, as well as within the transcriptome of an organism (Figure 1.1).

However, the basis of this heterogeneity in mRNA stability, often spanning multiple orders of magnitude in a given organism, remains poorly understood. Despite substantial knowledge of the major pathways and enzymatic complexes responsible for mRNA degradation in both bacteria11, 12 and eukaryotes,13–17 we have only just begun to understand a subset of factors that dictate the heterogeneity in stability of mRNAs. This is not altogether surprising, as considerable diversity exists in the mRNA decay pathways; endonuclease-initiated mRNA decay serves as the committed step for decay in bacteria while exonucleolytic decay predominates in eukaryotes.

1.1 Mechanisms of RNA Decay

In eukaryotes, cytoplasmic mRNAs are protected from decay machinery through the terminal 7-methylguanosine (m7G) and polyadenosine, or poly(A), tail structures 1.1. Mechanisms of RNA Decay 3

added during processing and maturation in the nucleus. As the addition of these

features is coupled to transcription, any mRNA lacking these elements is committed

and condemned to decay. The m7G cap, a methylated guanine residue linked to the

ﬁrst nucleotide of the mRNA message through a 5’-5’ triphosphate moiety,18 protects

the mRNA from 5’ 3’ exonucleases which are incapable of hydrolyzing the unique ! triphosphate linkage.19 Likewise, removal of the poly(A) tail requires the action of speciﬁc deadenylase complexes, Ccr4-Not-Pop2 and Pan2-Pan3, to remove the long stretches of adenosine at the 3’ end of mRNAs (Figure 1.2).

1.1.1 Deadenylation and exosome-mediated mRNA decay

Truncation of the poly(A) tail is generally the ﬁrst step of mRNA decay and is often

rate-limiting with deadenylation rates matching those of decay rates in vivo.20 This truncation is mediated primarily by the deadenylation complex, CCR4-NOT, comprised of two exonucleases - Ccr4 and Pop2, as well as a family of sca↵olding proteins, whose functions range from sca↵olding (Not1) to roles in translational repression as well as proteosome assembly (Not4).21–23 While Ccr4 serves as the major catalytic subunit of the complex in S. cerevisiae,16 Pop2 (also known as Caf1) can substantially contribute to deadenylation in higher eukaryotes.24 Truncation of the poly(A) tail

is thought to be concomitant with the removal of poly(A) binding protein (PABP),

which otherwise coats the poly(A) tail, occluding any exonucleolytic degradation by

the exosome complex.25 1.1. Mechanisms of RNA Decay 4

Upon deadenylation, the mRNA can either undergo decapping-mediated mRNA decay (discussed below) or, less frequently, undergo decay through the exosome- mediated mRNA decay pathway (Figure 1.2). The exosome complex is a ring-like structure comprising nine structural subunits as well as a tenth, Rrp44, which serves as a 3’ 5’ exonuclease.26, 27 The exosome works in concert with a host of pro- ! teins, the Ski complex, that aid in exonucleolytic activity by unwinding mRNAs and

channeling mRNA substrates directly to the exosome.28, 29

1.1.2 Decapping-mediated mRNA decay

While a subset of deadenylated messages are targeted for decay by the exosome

complex, degradation by way of the decapping-dependent pathway is the far more

likely fate for the average mRNA.30 The ﬁrst step of this process is the removal of

the m7G cap structure from the 5’ end of the mRNA (Figure 1.2) by the Dcp1-Dcp2

enzyme complex. However, much as the process of deadenylation strips mRNAs of

PABP, the process of Dcp1-Dcp2 binding to the cap as well as subsequent decapping

ensures a host of proteins integral to translation initiation are no longer able to bind

to the cap structure - most notably, eIF4E. Given that eIF4E is required for ecient

translation initiation, the act of decapping to repress translation.

Dcp2 is the catalytic subunit of the enzyme complex,31 responsible for hydrolysis and removal of the cap structure, but this process is aided by a number of cofactors and activators. Some of these cofactors serve to directly activate decapping, such 1.1. Mechanisms of RNA Decay 5 as Edc1, Edc2, and Edc3. Notably, Edc3 demonstrates transcript speciﬁcity for decapping - a mechanism that remains unresolved.32, 33 Others, however, appear to enhance decapping by repressing translation or otherwise altering the composition of the mRNA-ribosome-protein (mRNP) complex. Most notable amongst this set of factors include Dhh1, Pat1, and the Lsm family of proteins.34–36 Upon removal of the cap structure, the mRNA rapidly undergoes exonucleolytic degradation by the 5’ 3’ ! exonuclease, Xrn1. Recent studies have shown that Xrn1 can and often does degrade messages co-translationally, often running up against slowly moving ribosomes.37, 38

1.1.3 Decay of aberrant mRNAs

Not all mRNAs are created equal. While normal mRNAs, namely, those that are eciently translated to create proteins, are degraded by the mechanisms outlined above, there exist aberrant mRNAs which are an immediate threat to the cell. These are mRNAs that, through genetic mutations, cleavage, or modiﬁcation can no longer can be eciently translated by the ribosome and thus lead to ribosomal stalling.

While many of these mRNAs could ostensibly be targeted for decay by the standard machinery, surveillance mechanisms have evolved to ensure ecient targeting of these mRNAs for degradation.

In all bacteria, ribosomes stalled on truncated mRNAs are “rescued” by tmRNA, a factor which couples ribosome release and recycling with decay of mRNA and the protein product.39, 40 In eukaryotes, more specialized mechanisms exist to deal with 1.1. Mechanisms of RNA Decay 6

aberrant translation events (Figure 1.3). Examples of these aberrant translation

events include premature stop codons in the middle of an open reading frame (ORF)

triggering nonsense-mediated decay (NMD), poly(A) tails at the end of mRNAs due

to the mRNA lacking a stop codon triggering non-stop decay (NSD), and dramatic

kinetic traps (e.g., hairpins or truncated mRNAs) that prevent further translation

triggering no-go decay (NGD).41, 42

In NMD, mRNAs appear to have exceptionally long 3’ untranslated regions (3’

UTRs) due to a mutation in the open reading frame to encode a stop codon.43 As ribosomes cannot proceed past this premature stop codon, full length proteins are not generated and potentially pathogenic peptide products can be formed.44 Under these conditions, the stalled ribosome is capable of recruiting accessory proteins, most notably the Upf proteins (Upf1, Upf2, and Upf3), which enable recruitment of decay factors.45–48 Interestingly, NMD targets appear, in S. cerevisiae, to be degraded by the general exonuclease, Xrn1, but NMD containing transcripts appear to be degraded multiple orders of magnitude faster.49 Further, the mechanism of how these ribosomes are able to discriminate premature stop codons from “normal” stop codons, and accordingly pause any further translation remains poorly understood.

NGD also involves ribosomes that are incapable of translating further along a message. In this case, ribosomes are prevented from further translation due to a myriad of reasons: the presence of large secondary structures,50 the absence of further

message due to cleavage of the mRNA,51 polybasic peptides being encoded for by the 1.1. Mechanisms of RNA Decay 7

mRNA,52 or even on stretches of a very poorly decoded codons (CGA).53 Here too,

sucient headway has been made on coupling translation to mRNA decay, speciﬁcally

in the role of rescue or stalled ribosomes by the factors Dom34 and Hbs1.54 Rather

surprisingly, mRNA decay in these contexts seems to employ an initial endonucleolytic

cleavage, rather than the exonucleolytic cleavage that is standard in most mRNA

decay contexts.

NSD is, in some respects, the opposite of NMD, where the lack of a stop codon

leads to the ribosome translating well past the end of the protein product into the

3’ UTR and the poly(A) tail (assuming that there are no in-frame stop codons in

the 3’ UTR). However, in most contexts, the ribosome does not continue translating

till the end of the message. Rather, ribosomes on poly(A) messages tend to engage

in frameshifting and sliding.55 Under these conditions, the exosome is recruited to

the ribosome and message by way of the Ski complex.56 Compared to standard

exosomal decay, NSD requires both the endonucleolytic and exonucleolytic functions

of Rrp44.57 Notably, in each case, the ribosome (and associated factors) is thought

to “sense” the defect in the mRNA and recruit additional machinery to implement

downstream events including mRNA decay, proteolysis of the nascent peptide, and

recycling of the stalled ribosome.58–61 Taken together, a clear picture of translation informing mRNA stability and decay emerges. 1.2. General connections between translation and mRNA decay 8

1.2 General connections between translation and mRNA decay

While mRNA quality control couples decay of a nonfunctional mRNA to an aberrant

translational event, there is a great deal of literature showing that decay is directly

coupled with translation in more general contexts on functional mRNAs. In bacteria,

decay is primarily governed by endonucleolytic activity, and a barren mRNA appears

to be an ideal substrate for the decay machinery. These ﬁndings come from studies

showing that when mRNAs are depleted of ribosomes, either by inhibiting ecient

translation initiation through perturbation of the Shine-Dalgarno sequence62, 63 or by using exogenous RNA polymerases to outrun the translational machinery,64 they are

more susceptible to decay. These ﬁndings in bacteria provide precedent for direct

coupling between the translational state of the mRNA and its decay.

In eukaryotes, there is also strong evidence for widespread coupling of decay with

translation. For example, while mRNA surveillance is primarily thought to act on

aberrant mRNA transcripts, there are numerous instances where mRNA surveillance

pathways are co-opted to link translation to decay on functional mRNA transcripts.

Speciﬁcally, genes with actively translated upstream ORFs in the transcript leader

have been shown to be subject to decay by NMD machinery.65–67 More generally, Hu

et al. showed that mRNAs that sediment deep in a polysome proﬁle with associated

ribosomes are already substantially decapped and partially degraded by the exonu-

clease Xrn1.37 These initial observations were extended by experiments aimed at 1.2. General connections between translation and mRNA decay 9

determining how widespread such co-translational decay might be across the genome.

Employing high-throughput 5’ P-sequencing (trapping the natural product of Xrn1-

mediated exonucleolytic decay), studies found that approximately a tenth of all cellu-

lar mRNAs in yeast were in the process of being degraded. Moreover, the position of

the 5’ ends of these decay intermediates exhibited 3-nucleotide periodicity, suggesting

that Xrn1 appears to be running into actively translating ribosomes.37, 38

Thus, general mRNA decay, much like decay of aberrant mRNAs, is impacted by the function of the ribosome. What these experiments do not establish with respect to general mRNA decay, however, is the ordering of these events. Does translation influence decay or does decay influence translation? More recent studies argue that rates of translation, dictated by the inclusion of specific codons, strongly influence the rates of decay.

1.2.1 Codon selection informs mRNA stability

Given the plethora of contexts discussed above, in which the translational state of the

ribosome is intimately linked with mRNA decay, recent ﬁndings by Presnyak et al.

indicating that codon bias broadly impacts mRNA stability in yeast are unsurprising.6

Based on both genome-wide mRNA stability measurements as well as reporter studies,

Presnyak et al. show that stable genes are generally enriched for a certain subset of the codon pool (optimal) while less stable genes are enriched for another subset of the codon pool (non-optimal). Moreover, consistent with these observations by Presnyak 1.2. General connections between translation and mRNA decay 10

et al., Xrn1 decay products analyzed with 5’ P-sequencing reveal a substantial increase

in ribosome occupancy on rare codons, suggesting that mRNA species targeted for

decay are those enriched in slowly translating ribosomes.38

These findings from yeast have been supported by recent studies in E. coli,8 where the stabilities of thousands of reporter mRNAs were found to strongly correlate with codon usage. Moreover, by analyzing existing data on mRNA stability in higher eukaryotes, we find that similar trends hold true (Figure 1.4). RNA stability measurements obtained in NIH3T3 mouse fibroblasts reveal a clear, albeit modest, correlation between codon optimality and RNA stability.7 These findings have also been

extended to other higher eukaryotes, where codon usage has been shown to be a key

determinant of maternal mRNA stability in zebraﬁsh embryos.9 Taken together, these data from yeast, bacteria, and metazoans suggest that mRNA stability is dictated by the eciency of translation throughout biology. 1.3. Figures 11

1.3 Figures

E. coli 3

Relative0 Density 10-1 100 101 102 103 104

S. cerevisiae 3

Relative0 Density 10-1 100 101 102 103 104

M. musculus 3

Relative0 Density 10-1 100 101 102 103 104 mRNA Half Life (minutes)

Figure 1.1: There is heterogeneity in the stability of mRNA transcripts

Genome-wide measurements of mRNA half-lives show that stability of di↵erent transcripts vary over orders of magnitude, both within an organism as well as between organisms.6–8 1.3. Figures 11

1.3 Figures

E. coli 3

Relative0 Density 10-1 100 101 102 103 104

S. cerevisiae 3

Relative0 Density 10-1 100 101 102 103 104

M. musculus 3

Relative0 Density 10-1 100 101 102 103 104 mRNA Half Life (minutes)

Figure 1.1: There is heterogeneity in the stability of mRNA transcripts

Genome-wide measurements of mRNA half-lives show that stability of di↵erent transcripts vary over orders of magnitude, both within an organism as well as between organisms.6–8 1.3. Figures 12

m7G Open Reading Frame PolyA

Ccr4/Pop2/Not(1-5) Pan2/Pan3

Dcp1/Dcp2 Exosome

Xrn1

Figure 1.2: Canonical mRNA decay in S. cerevisiae

In S. cerevisiae, decay occurs in three separate phases. First, the mRNA is deadenylated, followed by decapping. These two processes are closely linked and serve as the committing step for decay, as an mRNA lacking these features is immediately decayed by exonucleases. Finally, exonucleases degrade the mRNA. 1.3. Figures 13

Translationally arrested ribosomes

NMD NGD NSD

Ski7

Upf Dom34 1/2/3 Hbs1 Dom34 Hbs1 Decay of aberrant messenger RNA

Figure 1.3: Decay of aberrant mRNAs in S. cerevisiae

Decay of aberrant mRNAs occurs downstream of a non-productive translation event. This suggests a clear interplay between the translational state of a ribosome on an mRNA and the potential for that mRNA to be decayed. 1.3. Figures 14

S. cerevisiae 80 154 2076 1166 239 81 97 77

20 Half-Life (Minutes)

0 0.3 0.4 0.5 0.6 0.7 0.8 0.9

M. musculus 50 50 447 1239 1412 1157 308 15

20 Half-Life (Hours) 10

0 0.64 0.68 0.72 0.76 0.8 0.84 0.88 Codon Adaptation Index

Figure 1.4: Codon optimality vs. mRNA stability in mice and yeast

mRNA half-lives were obtained from previously published data for log phase yeast6 and cultured mouse ﬁbroblasts (NIH3T3)7 grown in SILAC medium, and codon adaptation index (CAI) values for genes were calculated per established methods.68 Colors represent the gradation from genes enriched in non-optimal codons (red) to optimal codons (green) showing increasing RNA stability with increasing codon optimality. Chapter 2

Dhh1 represses translation by modulating ribosome occupancy

Note: Parts of this chapter were published in:

Radhakrishnan, A., Chen, Y. H., Martin, S., Alhusaini, N., Green, R. & Coller, J.

(2016). The DEAD-Box Protein Dhh1p Couples mRNA Decay and Translation by

Monitoring Codon Optimality. Cell, 167 (1), 122132.

While the previous discussion of mRNA decay is important for contextualizing the

ﬁndings of this thesis, the motivating questions behind this work were far removed from mRNA decay. Rather, this work was motivated by the desire to reconcile two seemingly contradictory phenotypes engendered by the yeast protein, Dhh1 - translation repression concomitant with increased ribosome occupancy on mRNAs. Thus, what follows is a quick overview of the literature regarding Dhh1, followed by the initial experiments and results which guide the rest of this work.

15 2.1. Dhh1 is at the nexus of mRNA decay and translation repression 16

2.1 Dhh1 is at the nexus of mRNA decay and translation repression

Dhh1, as well as its orthologs in higher eukaryotes (Table 2.1), belong to a family of

proteins known as DEAD-box helicases. The DEAD-box protein family is large in

number and varied in function, though they are generally involved in some form of

remodeling of RNPs.69, 70 DEAD-box proteins are homologous to eIF4A, the found-

ing member of the family, a key factor in translation initiation. DEAD-box proteins

contain a set of highly conserved regions within a somewhat conserved structure, con-

sisting of two central RecA-like domains (Figure 2.1 and Figure 2.2). Interestingly,

despite being designated as helicases, no members of the family have been character-

ized as having substantial helicase activity.

First identiﬁed in D. melanogaster as me31b, Dhh1 is a yeast protein that has

been characterized as important in regulation of translation of mRNAs as well as

in mRNA decay. The ﬁrst established role of Dhh1 in yeast was that of activating

mRNA decapping, and thus, serving as a general promoter of mRNA decay.34, 71 In this role, Dhh1 has been shown to engage in interactions with general decay machinery including the previously described Ccr4-Not complex as well as through the enzyme responsible for decapping, Dcp1.72 Consistent with these ﬁndings, a deletion of Dhh1

leads to a general stabilization of cellular mRNAs. Interestingly, association of Dhh1

with the deadenylation as well as decapping complexes appears to occur independent

of an mRNA substrate.34 2.1. Dhh1 is at the nexus of mRNA decay and translation repression 17

Table 2.1: Characterized functions of Dhh1 and orthologs

DHH1 ME31B (DME1) XP54 RCK (P54) Organism S. cerevisiae D. melanogaster X. laevis H. sapiens Translational repression Yes35, 73, 74 Yes75 Yes76, 77 Yes78 Granule formation Yes73 Yes75, 79 Yes77, 80, 81 Yes78 mRNA decay activation Yes34, 73 Yes79 -- miRNA repression - Yes75 -Yes78, 82 Maternal-zygote transition - Yes75 Yes80, 81 -

Given that the initial ﬁndings demonstrated that Dhh1 was implicated in mRNA

decay, any changes in protein production were ascribed to decreases in cellular mRNA.

However, later studies study showed that Dhh1 was capable of repressing translation

independent of mRNA decay in vitro,35 afactfurtherdemonstratedusingavariety of reporter systems in vivo.73, 74 Consistent with these results, early ﬁndings further

showed that a mild overexpression of Dhh1 and Pat1 appeared to decrease polysomes

while drastically increasing the number of Dhh1 and mRNA molecules localized to

P-bodies, cytoplasmic foci thought to be where a fraction of mRNA decay occurs.

In light of these results, the ﬁnding by Sweet et al. that tethering Dhh1 to

an reporter mRNA lead to an accumulation of slowly translocating ribosomes on

the mRNA was surprising.74 That this phenotype was concomitant with translation

repression, was even more surprising. This ﬁnding suggested a role for Dhh1 down-

stream of translation initiation, associating with ribosomes and repressing translation

before further sequestration of the mRNA into P-bodies or decay. However, as there

was little insight into any mechanism or even what factors would lead to Dhh1 asso-

ciating with translating ribosomes, we attempted to study this phenomenon further. 2.2. Tethering of me31b and orthologs eciently represses translation 18

2.2 Tethering of me31b and orthologs eciently represses translation

We ﬁrst attempted to reproduce the phenotype of translation repression through

tethering of Dhh1. Upon successful reproduction of this ﬁnding, we intended to

investigate stalling and sequestration of ribosomes on reporter mRNAs upon tethering

of Dhh1. In order to do this, we employed a dual-luciferase protein expression reporter

system that had been previously used in our lab to characterize miRNA-mediated

translation repression in D. melanogaster.83 At the same time, we began to develop a similar reporter system for use in S. cerevisiae. We started work in Drosophila as the dual-luciferase system had been modiﬁed to allow tethering, through the BoxB-N tethering system.84–86

In our assay, we used Renilla luciferase as a reporter gene and Fireﬂy luciferase as a transfection control. To verify that the system was working as intend, we also tethered proteins that had been previously characterized to repress translation when tethered to the 3’ UTR (GW182 and dAgo1).83, 87 While the positive controls exhibited an

approximately 10-fold repression in luciferase activity, consistent with values previ-

ously published from our lab,83 the tethered me31b (Drosophila ortholog of Dhh1) managed to repress translation approximately 5-fold (Figure 2.3). We then asked if tethering the yeast ortholog, Dhh1, would repress translation.

We found an approximately 3-fold reduction in Renilla luciferase levels upon tethering Dhh1. This is not surprising in light of the 54% identity and 63% similarity 2.3. Both RecA domains of Dhh1 are necessary for translation repression 19 between the yeast and ﬂy orthologs. In fact, account for the high variability in the

N- and C-terminal regions of the various orthologs, if we consider only the 400 ⇠ amino acid core domain, these values increase to 71% and 85%, respectively. The high homology between the orthologs as well as similar levels of translation repression engendered upon tethering of Dhh1 and me31b suggest that they appear work in similar roles biologically. These ﬁndings are further supported by crystallographic studies that show nearly identical structures between the Dhh1 and the human ortholog, RCK (P54).88–90

2.3 Both RecA domains of Dhh1 are necessary for translation repression

Having cloned the dual luciferase assay system into yeast, we then switched to performing our experiments in yeast given the greater array of genetic tools available.

Using the same experimental set-up (Renilla luciferase as a reporter and Fireﬂy luciferase as a transformation control), we asked what e↵ect various truncations and mutations would have on the ability of Dhh1 to repress translation. First, much as in Drosophila, we ﬁnd that upon tethering full length Dhh1 (N-HA-Dhh1) to the reporter mRNA, there is a substantial decrease in protein output - approximately

7-fold (Figure 2.4). We also ran the previous experiment in reverse, asking how well me31b could repress translation in a yeast system. Consistent with the previous ﬁnd- ings, we see that the Drosophila ortholog (N-HA-me31b) also represses translation

(approximately 4-fold), albeit less well than the yeast protein. 2.4. Dhh1-tethered mRNA predominantly sequesters ribosomes 20

Given that Dhh1 is comprised of two RecA-like domains, with highly unstruc- tured N- and C-terminal tails, we asked which, if either, of the RecA-like domains was necessary for translation repression. Looking at only the N-terminal RecA-like domain (N-HA-Dhh1NT), we ﬁnd that this segment alone is not capable of repressing translation, in spite of the DEAD-box motif being found in this domain (Figure

2.4). Given this, it is unsurprising that the C-terminal RecA-like domain (N-HA-

Dhh1CT) is equally ine↵ective in repressing translation of the reporter luciferase.

However, a truncated construct of only the two RecA-like domains without the un- structured terminal regions is capable of repressing translation. Finally, mutating out the DEAD-box (mutating to DQAD) is sucient to abrogate the translation repression phenotype in either the full length (N-HA-Dhh1-DQAD) or truncated

(N-HA-tDhh1-DQAD) construct.

2.4 Dhh1-tethered mRNA predominantly sequesters ribosomes

To better understand if the tethering of Dhh1 to reporter mRNAs led to an increase in ribosome occupancy on mRNAs, we then turned our attention to looking at the reporter mRNA species. Unfortunately, while the luciferase reporter system proved to be robust for reporting on translation repression, technical issues relating to poor qPCR ampliﬁcation and smeared, poorly resolved bands in northern blots precluded use of Renilla luciferase samples for analysis of the tethered reporter mRNA. In light of these diculties, we switched to a tethering system with a reporter mCherry 2.4. Dhh1-tethered mRNA predominantly sequesters ribosomes 21

mRNA.

We started by attempting to reproduce an experiment in Sweet et al. where

the authors monitored polyribosomal association on a reporter mRNA when Dhh1

is tethered.74 However, while the authors of that paper employed using only the

N tethering element as a negative control, we decided to employ the Dhh1-DQAD mutant which we found to be incapable of repressing translation (Figure 2.4) as our negative control. Thus, we prepared extracts from yeast cells transformed with a reporter mCherry mRNA and N-Dhh1 with and without a catalytically active

DEAD-box. These extracts were layered on sucrose gradients and separated by ve- locity sedimentation. The gradients were then fractionated and mRNA was isolated from each sample allowing us to quantify the amount mCherry mRNA level in each fraction by way of Northern blotting. Through this experiment, we were able to reproduce the result observed in Sweet et al., that tethering functional Dhh1 leads to co-sedimentation of mRNA with heavy polyribosomes (Figure 2.5). We could further quantify this e↵ect by looking at the mRNA fold enrichment in each fraction. This was done by calculating the percentage of mRNA that was distributed in each fraction along a given gradient. We then calculated what the percent of mRNA was in a fraction with catalytically active Dhh1, relative to the percent mRNA in a fraction with catalytically inactive Dhh1. We then calculated this value across three biological replicates, observing the largest enrichment at the bottom of the gradient in fractions

13 through 16, concomitant with polysomes. 2.4. Dhh1-tethered mRNA predominantly sequesters ribosomes 22

However, that the reporter mRNA sediments in a gradient with heavy polyribosomes is hardly conclusive proof that tethering Dhh1 has has increased ribosome occupancy on the reporter mRNA. There are numerous other large particles that are capable of sediments deep within a gradient, namely P-bodies. While the original study by Sweet et al. had numerous controls to verify that these mRNAs were, in fact, loaded with ribosomes upon tethering of Dhh1, we felt the simplest and most elegant experiment would simply be to tether Dhh1 to an extremely small ORF where a message fully loaded with ribosomes would still only migrate midway in the sucrose gradient. Thus, if the tethered messages were to migrate to bottom of the gradient, the messages are are being shuttled to larger granular species.

We decided to use an endogenous gene as a reporter to minimize the likelihood of issues or challenges for the yeast translational machinery. Thus, we repeated the previously outlined experiment, using a small subunit of the oligosaccharyltransferase complex, Ost4. As Ost4 is only 36 amino acids long, we calculated that a fully loaded message was only capable of sequestering around four ribosomes. Reassuringly, the tethering assay shows a clear shift of mRNA into the middle of the gradient, suggesting that upon tethering of catalytically active Dhh1, the reporter mRNA is loaded with ribosomes, rather than shuttled to P-bodies or other granules in the cell (Figure 2.6).

Further, given that the message shifts to the trisomal and tetrasomal peak along the gradient, we suspected that these messages are fully loaded with ribosomes. This led us to believe that the basis of this phenotype may be a failure of termination, where 2.5. Dhh1-mediated ribosome occupancy is not a termination defect 23

Dhh1 inhibits translation termination, causing a pile-up of ribosomes upstream of the

stop codon.

2.5 Dhh1-mediated ribosome occupancy is not a termination defect

To test the exact nature of Dhh1-mediated ribosome occupancy changes, we required

a technique with greater resolution than polysome proﬁling. Thus, we decided to use

ribosome proﬁling to assess ribosome occupancy on our reporters. Ribosome proﬁling

is a high-throughput sequencing method that examines where ribosomes are bound

throughout across mRNAs by looking at mRNA regions that are “footprinted” by

ribosomes and thus, are protected from decay by RNases (Figure 2.7).91 Given that the number of footprints that map to a given region of mRNA is a function to the amount of time a ribosome spends at that position, the relative number of mapped reads directly correlates with ribosome occupancy along a message.92, 93

Thus, we chose to proﬁle yeast lysates transformed with the mCherry reporter construct with overexpression of catalytically active and catalytically inactive Dhh1.

However, as ribosome proﬁling only provides a the relative fraction of ribosomes on a given mRNA across a transcriptome, we further performed RNA-Seq. As we are concerned with the number of ribosomes per mRNA, we needed to ensure that increased ribosome occupancy was not simply due to increased transcription of the mRNA. The sequencing samples are summarized in Table 2.2.

From these samples, we mapped reads to the mCherry reporter in all four condi- 2.5. Dhh1-mediated ribosome occupancy is not a termination defect 24

Table 2.2: Reporter ribosome proﬁling samples and GEO sample numbers

Sample (GEO) Name GSM2148000 mCherry + NL-Dhh1-OE Proﬁling GSM2148001 mCherry + NL-Dhh1-OE mRNA-Seq GSM2148002 mCherry + NL-Dhh1-DQAD-OE Proﬁling 1 GSM2148003 mCherry + NL-Dhh1-DQAD-OE mRNA-Seq

tions allowing us to calculate the reads per million (RPM) for every nucleotide along

mCherry in the proﬁling samples.

Number of mapped reads (nt) RPM (nt) = 106 (2.1) Total number of mapped reads ⇥

As each sample tends to have tens of millions of reads, multiplying by 106 allows us to bring values into a range where they are more tractable. By comparison, RNA-Seq contains substantially less positional information, owing to random fragmentation of mRNAs during our sample preparation. As such, we talk about relative amounts of given mRNAs in the transcriptome by using reads per kilobase million (RPKM):

Total number of mapped reads (gene) RPKM (gene) = 109 Total number of mapped reads Gene length (in bases) ⇥ · (2.2)

Thus, we can calculate a normalized RPM along the length of our reporter genes in

each condition, where the RPM at every nucleotide is normalized by relative levels of

mRNA through dividing by gene-wide RPKM of the reporter mRNA. Plotting these

values along the length of the mCherry reporter, we ﬁnd that when catalytically active 2.5. Dhh1-mediated ribosome occupancy is not a termination defect 25

Dhh1 is tethered, there are an increased number of ribosomes per mRNA relative to the catalytically inactive sample (Figure 2.8). However, the distribution of ribosomes suggests that Dhh1 does not a↵ect translation termination. If eciency of termination was a↵ected, we would expect to see a distribution of ribosome density similar to the dotted line. Thus, we surmised that Dhh1 must be acting upstream of termination - suggesting elongation as a likely target. 2.6. Materials and methods 26

2.6 Materials and methods

Transfection, cell culture, and luciferase assays in Drosophila

S2 cell cultures were grown at room temperature using Express Five SFM medium

(Invitrogen) with the following supplements: 100 units/mL penicillin and strepto-

mycin (Cambrex Bioscience). The media was further supplemented with 10% head-

inactivated FBS (Invitrogen) as well as 8.25% v/v 200 mM L-glutamine (Invitrogen).

We used the E↵ectene transfection reagent kit (Qiagen) for the luciferase assays and

seeded cells in 24-well plates, transfecting in triplicate. Each well contained approx-

imately 10 ng Renilla luciferase reporter plasmid, 10 ng Fireﬂy luciferase control

plasmid, and 100 ng of e↵ector plasmid. Samples were processed three days post-

transfection with the Dual Luciferase Reporter Assay System (Promega).

Cloning of reporter constructs for yeast

The plasmid pRaugFFug (referred to as pRIF in the Green lab), which was used in

earlier studies from the Lorsch lab, was obtained.94, 95 The single plasmid expresses

both luciferase constructs under the control of an ADH1 promoter and HIS ter-

minator (Renilla luciferase) and a GPD promoter and a CYC1 terminator (Fireﬂy luciferase). The Renilla luciferase 3’ UTR was modiﬁed by insertion of 3 BoxB hairpin elements cloned out from the previously described drosophila Renilla luciferase, truncating the following region in Drosophila: 2.6. Materials and methods 27

GGGCCCTGAAGAAGGGCCCCTCGACTAGTCCAAATACAAACTGGGCCCTGAAGAAGGGCCCATATA

GGGCCCTGAAGAAGGGCCCTATCGAGGATATTATCTCGACTAAGTCCAACTACAAACTGGGCCCTG

AAGAAGGGCCCATATAGGGCCCTGAAGAAGGGCCC

Which contains 5 BoxB hairpin elements, highlighted in red, to the following sequence in yeast, which contains three:

AGGGCCCTGAAGAAGGGCCCCTCGACTAGTCCAAATACAAACTGGGCCCTGAAGAAGGGCCCATAT

AGGGCCCTGAAGAAGGGCCCA

Which was inserted exactly 100 nucleotides downstream of the stop codon, immediately preceding the HIS terminator.

Tethering constructs were cloned using the Gateway cloning system by generating the following general construct:

CACCATGGACGCTCAAACCAGAAGAAGAGAAAGAAGAGCTGAAAAGCAAGCTCAATGGAAGGCTGCT

AACTACCCATACGATGTTCCAGATTACGCTGGCGGCGGCGGATCCGCAGGGTCAGCAGGGTCAGCAG

GGTCACTCGAGTAA

Where the CACC overhang at the 5’ end of the message was to ensure correct incorporation of the construct into the pENTR/D-TOPO E. coli vector. The construct contains a N-terminal N peptide (red) required for tethering of the construct to the reporter mRNA, followed by an HA tag (blue). This is separated from the protein of interest by a glycine linker (green) and a spacer region (black) is ﬂanked by cloning sites for BamHI (purple) and XhoI (orange). Thus, using suitable handles and di- gestion, any protein of interest can be cloned into this system to tether to luciferase 2.6. Materials and methods 28

reporter mRNAs. Once this construct was cloned into pENTR, it was moved into a

yeast expression vector (pAG425) using the LR clonase reaction.

A similar strategy was employed for generation of the mCherry and OST4 reporter

mRNAs, with a similar base construct, lacking the N peptide, while maintaining a

ATG start codon downstream of the CACC overhang. Further, this construct was moved into a di↵erent yeast expression vector (pYES52) which contains a URA3 cassette for selection (as it is being used in a similar fashion as the pRIF reporter vector). By comparison, the pAG425 expression vector contains a LEU2 selectable marker.

Growth and preparation of yeast cells for luciferase assays

Cells used in luciferase assays were grown in dropout media (CSM -Leu - Ura) sup-

plemented with 2% galactose and ranose (w/v) at 30 C until mid-log phase (OD600

=0.45-0.5).Thedropoutmediawaschosentoselectforsuccessfultransforma-

tion of both pRIF (URA3 selectable marker) as well as pAG425 (LEU2 selectable

marker) into BY4741 yeast. To measure luciferase activity, 1µLofculturewasadded

to 50µLof1XPassiveLysisBu↵er (Promega) and then incubated for 60 minutes

at room temperature. Luciferase activity was then measured using a FLUOstar OP-

TIMA microplate reader using the luciferase head attachment. Technical samples

and biological samples were taken in triplicate. 2.6. Materials and methods 29

Growth and preparation for yeast for polysome analysis and northern blotting

Cells were grown in 500 mL dropout media (CSM -Leu -Ura) supplemented with 2%

galactose and ranose (w/v) at 30 Cuntilmid-logphase(OD600 +0.6).Cellswere

then collected by vacuum ﬁltration, ﬂash frozen as droplets in liquid nitrogen and

mixed with 1 mL of frozen beads of 1x lysis bu↵er (10 mM Tris pH 7.4, 100 mM

NaCl, 30 mM MgCl2, 0.5 mg/mL heparin, 1 mM DTT, 100 mg/mL cycloheximide,

1% Triton X-100). Cells and lysis bu↵er beads were ground and homogenized using

a Spex 6870 freezer mill. The resulting lysate was quantiﬁed using a NanoDrop

1000 spectrophotometer using the absorbance at 260nm. 20 absorbance units were

loaded onto a 10% to 50% (w/v) sucrose gradient, with varying sucrose concentration

amongst a background gradient bu↵er (20 mM Tris pH 8, 150 M KCl, 5 mM MgCl2,

0.5 mM DTT, 100 mg/ml CHX). The gradients were prepared by layering 6 mL of

10% w/v sucrose-gradient bu↵er mix on top of 6 mL of 50% w/v sucrose-gradient bu↵er mix and using a BioComp Gradient Master (Time: 1:48, Angle: 81.5, Speed:

17 rpm) to form the gradient.

Samples were spun in an SW-41 Ti rotor for 3 hours at 40,000 rpm at 4 C. Gra- dients were then fractionated using a Teledyne Isco Foxy R2 and RNA was extracted from individual fractions using two rounds of phenol:chloroform:isoamyl alcohol extraction (using 1x fraction volume). This was followed by isopropanol precipitation

(500 µL 300 mM NaOAC pH 5.2, 500 µL isopropanol). RNA pellets were isolated by centrifugation at 14,000 rpm for 30 minutes at 4 and subsequently washed with 70% 2.6. Materials and methods 30

ethanol. Following the ethanol wash, the pellets were isolated by centrifugation once

again. Pellets were then dried for 15 minutes and resuspended in 20 µL1xTEbu↵er

(10 mM Tris pH 8.0, 1 mM EDTA). 20 µLofloadingbu↵er (0.95 mL formamide, 10

µL EDTA, 40µLH2O, 0.025% w/v Bromophenol blue, 0.025% w/v Xylene cyanol)

was added to each sample. Samples were then incubated at 95 for 5 minutes before being placed on ice. 10 µL of sample from mCherry and OST4 gradients were loaded on 1% formaldehyde-agarose gels. The gels were then transferred onto Hybond-N+ nylon membranes using a Bio-Rad 785 vacuum blotter. The membrane was probed with 32P end-labeled oligonucleotides antisense to the 3’ UTR prior to the BoxB stem

loops (AR-N-16):

GGTGAAAGAGAAAAGAAAAAAATTGATCTATCGATTTCAATTCAATTCAATTTACTCGAG

Blots were imaged using the Typhoon FLA 9500 and quantiﬁed in ImageJ. All

experiments were performed in triplicate.

Growth and preparation of samples for ribosome proﬁling

Cells used in ribosome proﬁling and RNA-Seq were grown at 30 C in either YPD

or CSM Leu dropout media and with 2% glucose for those transformed with the

pAG425 plasmid and CSM Leu Ura and 2% galactose/ranose for those transformed

with both the pAG425 and pYES-DEST52 plasmids. Cells were harvested at early-

log phase (OD600 = 0.3-0.35). Footprints were prepared largely according to existing

methods.54 Upon vacuum ﬁltration, the cells were frozen in liquid nitrogen and ground 2.6. Materials and methods 31

in a Spex 6870 freezer mill along with frozen lysis bu↵er beads (20 mM Tris pH 8,

140 mM KCl, 1.5 mM MgCl2, 1% Triton X-100, 100 mg/mL CHX). For RNA-Seq

experiments, ribosomal RNA was subtracted using RiboZero Magnetic Gold (Yeast)

from Epicenter, and the remaining RNA was ligated to a universal adaptor. Upon

reverse transcription, circularization, and PCR ampliﬁcation, cDNA fragments were

sequenced on either an Illumina HiSeq2000 or HiSeq2500 machine at facilities at UC

Riverside or the Johns Hopkins Institute of Genetic Medicine.

For footprint libraries, the above steps were preceded by RNase treatment, 15

U of RNase I (Ambion) per OD260 unit of the lysate, and monosome species were

separated by a 10% to 50% (w/v) sucrose gradient, with varying sucrose concentration

amongst a background gradient bu↵er (20 mM Tris pH 8, 150 M KCl, 5 mM MgCl2,

0.5 mM DTT, 100 mg/ml CHX). The gradients were prepared by layering 6 mL of

10% w/v sucrose-gradient bu↵er mix on top of 6 mL of 50% w/v sucrose-gradient bu↵er mix and using a BioComp Gradient Master (Time: 1:48, Angle: 81.5, Speed:

17 rpm) to form the gradient.

(500 µL 300 mM NaOAC pH 5.2, 500 µL isopropanol). RNA pellets were isolated by centrifugation at 14,000 rpm for 30 minutes at 4 and subsequently washed with 2.6. Materials and methods 32

70% ethanol. Following the ethanol wash, the pellets were isolated by centrifugation once again. Pellets were then dried for 15 minutes and resuspended in 20 µL1xTE bu↵er (10 mM Tris pH 8.0, 1 mM EDTA). The extracted RNA was puriﬁed from a

15% denaturing PAGE gels between markers of the following sizes: 25-34 nt (Original

Samples), 15-35 nt (Replicates), with empty lanes left between all samples to prevent cross contamination.

Read preparation and sequence alignment

All gene boundaries and annotations used in analysis are from the R64-1-1 S288C reference genome assembly (sacCer3) from the Saccharomyces Genome Database

Project. A tab delimited ﬁle containing gene annotations was obtained from the

UCSC Table Browser (https://genome.ucsc.edu/cgi-bin/hgTables).

De-multiplexed sequences were ﬁrst processed to remove the Universal miRNA

Cloning Linker (NEB) adaptor sequences:

CTGTAGGCACCATCAATAGATCGGAA

using the CutAdapt tool (Martin, 2011). The remaining reads were then processed to remove low-quality reads (PHRED accuracy 97.5%). In addition, contaminating  non-coding and ribosomal RNAs were ﬁltered. This was accomplished by alignment to the ncRNA gene database FASTA ﬁle available at the Saccharomyces Genome

Database Project. This alignment was performed using Bowtie 1.1.2 (Langmead et al., 2009) using the following parameters: ‘-Sv 3 p 4 best’. 2.6. Materials and methods 33

The remaining reads were then aligned to the yeast genome using Bowtie 1.1.2 using he following parameters: ‘-Sm 1 p 4 best strata’. Reads were then mapped to the genome to nucleotide resolution using either 30-end mapping of all read lengths or 50-end mapping of 28-nt fragments. Once counts are tabulated at each nucleotide position, the values are normalized to reads per million (rpm) which involves dividing counts at each position by the total number of mapped reads. Python and R scripts used to generate the data and ﬁgures in this paper can be found on GitHub at https://github.com/greenlabjhmi/2016-Cell-Dhh1. 2.7. Figures 34

2.7 Figures

Walker A-2 Walker A-1 Walker B Motif III Walker A 96 153 124 230 171 197 89 147 169 120 194 225 162 138 241 108 145 181 167 117 214 191 221 85

Motif Q

Arg Finger Motif 281 381 403 352 372 306 310 331 285 336 258 263 387 357 313 288 339 266 389 362 292 318

Motif IV Motif V C-Term N-Term

Figure 2.1: Characteristic sequence motifs in Dhh1

DEAD-box helicases, like Dhh1, have 9 broadly conserved regions that help deﬁne their broad functional role (RNP remodeling). 2.7. Figures 35

Figure 2.2: The crystal structure of Dhh1

88 The crystal structure of Dhh1 rotated 180 about the vertical axis relative to one another (PDB: 1S2M rendered in PyMOL). The same color scheme used in Figure 2.1 is employed here. However, only the DEAD-box motif is shown with atomistic representation. The opposite face to the DEAD-box motif has been shown to interact with RNA.71 2.7. Figures 36

1.00

0.75

0.50

0.25 Relative Renilla Luciferase 0.00

N-FLAG

N-FLAG-Dhh1 N-FLAG-GW182 N-FLAG-dAgo1 N-FLAG-me31B

Figure 2.3: me31b and Dhh1 repress translation in D. melanogaster

Measuring Renilla luciferase activity normalized to Fireﬂy luciferase activity in Drosophila. Proteins of interest were targeted to the 3’ UTR of the reporter Re- nilla luciferase mRNA through a N peptide element cloned into the N-terminus of the protein and through 5 BoxB hairpin tethering elements placed in the 3’ UTR of the reporter mRNA. Luciferase activity was measured in triplicate. 2.7. Figures 37

1.00

0.75

0.50

0.25 Relative Renilla Luciferase 0.00

N-HA

N-HA-Dhh1 N-HA-me31bN-HA-tDhh1 N-HA-Dhh1NTN-HA-Dhh1CT N-HA-Dhh1-DQADN-HA-tDhh1-DQAD

Figure 2.4: Dhh1 requires both RecA-like domains to repress translation

Measuring Renilla luciferase activity normalized to ﬁreﬂy luciferase activity in yeast. Truncations of Dhh1p (as well as me31b) were targeted to the 3’ UTR of the reporter Renilla luciferase mRNA through a N peptide element cloned into the N-terminus of the protein and through 5 BoxB hairpin tethering elements placed in the 3’ UTR of the reporter mRNA. Luciferase activity was measured in triplicate. 2.7. Figures 38

80S 60S Polyribosomes 40S

HA-mCherry + λN-HA-Dhh1-DQAD

HA-mCherry + λN-HA-Dhh1

1 mRNA Fold Enrichment 0 0 5 10 15 Fraction

Figure 2.5: Catalytic activity in Dhh1 is required to sediment reporter mRNAs with polyribosomes

Catalytically active and inactive Dhh1 was tethered to a reporter mCherry message using the previously described N-BoxB tethering system. The positioning of the reporter mCherry along a sucrose gradient (representative sample seen at top, with the curve representing the intensity of absorbance at 254 nm) was assessed via northern blot. Samples with catalytically active Dhh1 tethered sediment to the bottom of the gradient, suggesting the mRNA is highly loaded with mRNAs. Quantifying the relative mRNA levels in each fraction across biological triplicates shows an enrichment in fractions 13 to 16. 2.7. Figures 39

80S Polyribosomes 60S

40S

HA-OST4 + λN-HA-Dhh1-DQAD

HA-OST4 + λN-HA-Dhh1

1 mRNA Fold Enrichment 0 0 5 10 15 Fraction

Figure 2.6: Tethering Dhh1 increases ribosome density on reporter mRNAs

Catalytically active and inactive Dhh1 was tethered to a reporter OST4 message using the previously described N-BoxB tethering system. The positioning of the reporter OST4 along a sucrose gradient (representative sample seen at top, with the curve representing the intensity of absorbance at 254 nm) was assessed via northern blot. Samples with catalytically active Dhh1 tethered sediment to the bottom of the gradient, suggesting the mRNA is highly loaded with mRNAs. Quantifying the relative mRNA levels in each fraction across biological triplicates shows an enrichment in fractions 8 to 11. 2.7. Figures 40

Figure 2.7: Nucleotide resolution into ribosome occupancy by ribosome proﬁling

Ribosome proﬁling uses high-throughput sequencing to selectively sequence mRNA translated by ribosomes. These mRNA fragments are selected for by degrading all non-protected regions of the transcriptome by RNase treatment. Following this, sequencing libraries are made and tens of millions of sequencing reads are generated. 2.7. Figures 41

Expected Ribosome Occupancy for Translation Termination Defect 200 RPM (Normalized) 7 mGpppN HA mCherry mRNA AAAAn

Dhh1p or Dhh1p DQAD

Figure 2.8: Dhh1 does not a↵ect translation termination

While tethering of catalytically active Dhh1 increases ribosome occupancy on the mCherry reporter mRNA, consistent with the previous polysome proﬁles and northern blots, this occupancy increase is general and across the length of the ORF. This suggests that our hypothesis of Dhh1 a↵ecting translation termination is incorrect as amessagewithinecient termination is expected to exhibit a proﬁle similar to the dashed line. Chapter 3

Dhh1 stimulates decay of mRNAs with low codon optimality

Note: Parts of this chapter were published in:

Radhakrishnan, A., Chen, Y. H., Martin, S., Alhusaini, N., Green, R. & Coller, J.

(2016). The DEAD-Box Protein Dhh1p Couples mRNA Decay and Translation by

Monitoring Codon Optimality. Cell, 167 (1), 122132.

As we had established that Dhh1-mediated ribosome occupancy was not due to adefectintranslationtermination,thenextplacewesoughttoinvestigatearole for Dhh1 was in translation elongation. To do this, we decided to perform ribosome proﬁling on a larger set of samples (Table 3.1) to analyze the e↵ects of Dhh1 overexpression as well as Dhh1 deletion on global ribosome dynamics. As we obtained this data, however, a paper by Presnyak et al. caught our attention that established a clear connection between codon choice in the open reading frame (ORF) of mRNAs and the half-life of the mRNA.6 While this was not a wholly original ﬁnding, with careful reporter analysis showing codon dependence of mRNA stability in the early

42 3.1. Codon optimality underlies ecient translation 43

Table 3.1: Additional ribosome proﬁling samples and GEO sample numbers

Sample (GEO) Name GSM2147982 Wild-Type Profiling GSM2147983 Wild-Type Profiling (Replicate) GSM2147984 Wild-Type mRNA-Seq GSM2147986 Wild-Type mRNA-Seq (Replicate) GSM2147987 Dhh1KO Profiling GSM2147988 Dhh1KO Profiling (Replicate) GSM2147990 Dhh1KO mRNA-Seq GSM2147991 Dhh1KO mRNA-Seq (Replicate) GSM2147993 Dhh1-OE Profiling GSM2147995 Dhh1-OE Profiling (Replicate) GSM2147996 Dhh1-OE mRNA-Seq GSM2147997 Dhh1-OE mRNA-Seq (Replicate) GSM2147998 Dhh1-DQAD-OE Profiling GSM2147999 Dhh1-DQAD-OE mRNA-Seq

1990s, this was the ﬁrst genome-wide analysis to focus on the role of codon opti-

mality.96 Thus, before proceeding further into the linkage between Dhh1 and codon optimality, we ﬁrst provide a quick overview of codon optimality.

3.1 Codon optimality underlies ecient translation

The idea that codon choice inﬂuences gene expression has long been understood.97, 98

The inherent degeneracy of the genetic code leads to the possibility that synony-

mous codons are recognized distinctly by the ribosome as a function of subtle di↵er-

ences in tRNA availability, demand, decoding ﬁdelity, and mRNA secondary structure

propensity. All of these factors can lead to variability in codon-speciﬁc rates of trans-

lation.99–102 Codon optimality is a term coined to discuss the non-uniform recognition

of each of the 61 codons by the ribosome based on supply and demand arguments.102 3.2. Metrics for codon optimality 44

Codon bias, which is the frequency at which distinct synonymous codons are

present within the genome is, in part, shaped by codon optimality.103 Codons that

are evolutionarily enriched in highly translated mRNA transcripts are often optimal

codons (i.e., triplets that are decoded by tRNAs of relatively higher abundance),

whereas codons that exhibit no such selective bias are typically non-optimal and

are decoded by tRNAs of relatively lower abundance. Since codon bias is distinct

for every genome and represents a balance between selection, mutation, and genetic

drift, codon optimality is often found to be distinct between species.104–106

In broad terms, it is generally accepted that the speed at which the ribosome decodes is a↵ected by the subtle distinctions in tRNA concentrations between synonymous sets of codons.107–110 Thus, tRNA abundance is a critical regulator of ribosome elongation rates and therefore can impact the eciency of protein folding, protein stability, protein activity, and coordinate expression of functionally related genes.102, 111–115

3.2 Metrics for codon optimality

In discussing codon optimality, it is important to employ a suitable metric to quantify

the “optimality” of a codon or an entire gene. The measure for optimality proposed

in the Presnyak et al. paper is to simply bin codons as being either optimal or non- optimal, where optimality is deﬁned when a codon exhibits a positive correlation coecient between fraction gene content and mRNA stability.6 Thus, if across the 3.2. Metrics for codon optimality 45

genome, genes with a greater fraction of a given codon are generally more stable

than genes with a lesser fraction, this gene is deﬁned to be optimal. From this, the

optimality of a gene is deﬁned as the percentage of codons that are “optimal,” with

this correlation coecient for each being called the Codon Stabilization Coecient

(CSC).

Given that the paper measures half-life data for thousands of genes, we decided to

reanalyze their data to calculate each codons actual contribution to mRNA stability.

We do this, by setting up the following system of equations:

⌧d,a fAAA,a fAAG,a fUUU,a ⌧d,AAA 2 3 2 ··· 3 2 3 6 ⌧d,b 7 6fAAA,b fAAG,b fUUU,b7 6 ⌧d,AAA 7 6 7 = 6 ··· 7 6 7 (3.1) 6 . 7 6 . . . 7 ⇥ 6 . 7 6 . 7 6 ...... 7 6 . 7 6 . 7 6 . . . 7 6 . 7 6 7 6 7 6 7 6 7 6 7 6 7 6 7 6 7 6 7 6 ⌧d,z 7 6fAAA,z fAAG,z fUUU,z7 6 ⌧d,UUU 7 6 7 6 ··· 7 6 7 4 5 4 5 4 5 Or simply stated, the half-life of some gene, ⌧d,a, is the sum of each codons contribution to stability, deﬁned as the fraction of the gene that is a given codon, fXXX,a, multiplied by the codon contribution to half-life, ⌧d,XXX . From this analysis, we are able to calculate the expected half-lives of the mRNA sample set in Presnyak et al. upon training with a subset of 1000 genes (Figure 3.1). That some mRNAs are found to have a negative half-life reveals the limitations of this approach. Nevertheless, a positive correlation is seen between the observed and predicted half-lives through this analysis with a correlation coecient value of 0.293. This suggests that codon usage 3.2. Metrics for codon optimality 46 certainly is a determinant for mRNA stability, however only one of a multitude of factors that are at play.

The more interesting finding from this analysis is the codon-specific contribution towards mRNA stability. Arranging from least to most stabilizing, we find that 23 codons make net negative contributions to the stability of the gene they are found in (Figure 3.2). Further, stabilizing codons tend to be more abundant across the genome, as expected. We can also calculate the half-life of an “average” mRNA

(assuming that it perfectly captures genomic frequencies), to ﬁnd a half-life of 12.7 minutes. However, much as with CSC, the empirical nature of this metric prevents it from being a useful metric when we discuss mechanistic basis for codon optimality related phenomenon.

3.2.1 CAI and tAI: Supply and demand at the codon level

The codon adaptation index (CAI) is a metric that assesses if a given gene’s codon usage is similar to that of highly expressed genes in the same organism.68 As such, the CAI ﬁrst requires a deﬁnition of a set of “optimal” genes, often comprising of ribosomal proteins and enzymes from key metabolic pathways that are thought to be well expressed. From this reference set, the comparative weight of each codon (wi) is calculated relative to its most abundantly used synonymous codon. Thus if two codons are synonymous for an amino acid, and one is employed 25% percent of the time, its weight would be: 3.2. Metrics for codon optimality 47

fi 0.25 wi = = =0.333 (3.2) max(fj) 0.75 ···

Conversely, the weight of the more abundant codon would be:

fi 0.75 wi = = =1.00 (3.3) max(fj) 0.75

Once these weights have been calculated given codon usage frequencies in a ref-

erence set of well expressed, and thus optimized genes, the CAI of any gene can be

calculated as:

1 L CAI = exp wi(l) (3.4) L ! Yl=1 In theory, the CAI of every gene describes how well the gene has adapted to

chemical environment of the cell. However, this metric also abstracts vital information

about what processes make certain codons “good” or “bad.” Rather, we are reliant

on a reference set.

The tRNA adaptation index (tAI) is superior in that it directly links mechanism

to codon optimality.116 Further, the tAI calculation is similar to that of CAI on the gene level – the di↵erence lies in the weighting factors. The tAI assumes that the weight of each codon is determined by two factors: 3.2. Metrics for codon optimality 48

ni w = (1 s )tGCN (3.5) i ij ij j=1 X

Where nj is the number of tRNA isoacceptors that recognize a given codon i and tGCNij is the gene copy number of the tRNA, j, recognizing the ith codon. Finally sij

is a parameter to capture the eciency of codon-anticodon decoding. Unfortunately,

it is this last parameter which fails the metric as the original publication sets these

values by performing a linear least-squares ﬁt between gene tAI values and published

protein abundance data. Thus, it is not usable for our purposes as we intend to

characterize codon e↵ects on gene expression and translation. Using these values

could lead to the potential pitfall of data overﬁtting.

However, an extension of this method, the stAI, avoids this pitfall by ﬁtting values

of sij such that correlation between stAI and another sequence dependent metric, the

directional codon bias (DCBS) is maximized.117, 118 Thus, this metric provides a way to quantify the “optimality” of a gene based on how well the codon sequence is capable of utilizing cellular supply of tRNAs (tGCNij)whileminimizing“dwell time” of the ribosome (1 - sij)usingnothingmorethatsequencestatisticsacrossall

genes in the cell. Thus, to discuss optimality amongst our reporter constructs, we will

predominantly use stAI though we will also refer to the CSC developed by Presnyak

et al. 3.3. Dhh1p stimulates the degradation of mRNAs with low codon optimality 49

3.3 Dhh1p stimulates the degradation of mRNAs with low codon opti-

mality

In light of the ﬁndings by Presnyak et al., we established a collaboration with the

Coller lab, who were responsible for the earlier Sweet et al. paper that initially

launched our investigation. Given the previously established roles for Dhh1 in mRNA

decapping and decay, and in light of our studies into its roles in repressing translation

as well as modulating ribosome occupancy on mRNAs, we decided to investigate if

Dhh1 played a role in the coupling between codon optimality and mRNA stability.

We began this investigation, by following up on the results of Presnyak et al. and

creating a set of 11 constructs that vary in codon optimality (calculated through CSC

and stAI) that nevertheless produce the same protein product (Figure 3.3). Design

of these reporters was performed by random substitutions to a fully optimal sequence

with non-optimal codons (Figure 3.4). This approach was used to minimize clustering

e↵ects in di↵erent regions of the mRNA.

We measured the mRNA decay rate using a temperature-sensitive allele of RNA

polymerase II (i.e., rpb1-1). Transcription was inhibited by quickly shifting cells from

apermissivetemperaturetoarestrictivetemperature(from24 to 37 ). Following this shift, time points were taken and mRNA levels were analyzed by Northern blot

(Figure 3.3B, left). To account for potential e↵ects of temperature shifts on protein synthesis rates, we repeated this experiment using the same set of reporter constructs, 3.3. Dhh1p stimulates the degradation of mRNAs with low codon optimality 50

under the control of an inducible GAL1 promoter. Cells were grown in galactose at

24 Ctomid-logphase.Transcriptionwastheninhibitedbyaddingglucosebutmain- taining the cells at 24 C. Following addition of glucose, time points were once again taken and mRNA levels were assessed using Northern blots (Figure 3.3B, right). Re- assuringly, these two methods provided nearly identical half-lives across the spectrum of constructs (Figure 3.3C).

We then asked if Dhh1 plays a role in coupling mRNA decay to codon optimality by seeing if Dhh1 a↵ects the decay of mRNA reporters of varying optimality. Once again, we employed reporter constructs that expressed the same polypeptide, but composed of entirely optimal (Opt, stAI = 0.539) or entirely non-optimal (Non-Opt, stAI = 0.167) codons, which were characterized previously.6 For this experiment,

we employed the GAL1 promoter so that we could measure half-lives upon shifting

carbon sources. As expected, in wild-type yeast, the Opt construct is far more stable

than the Non-Opt construct (Figure 3.5). Importantly, however, in the absence of

Dhh1, any di↵erence in stability of the Opt and Non-Opt constructs is negligible,

with dramatic stabilization of the Non-Opt construct. To check if this was a gen-

eral response of Dhh1-like factors, we checked if loss of Pat1, Ccr4, or Dcp2 would

engender similar results. While the absence of these factors all served to stabilize

the mRNA species (as would be expected when the general mRNA decay machinery

is compromised), they all maintained a di↵erential stability between the Opt and

Non-Opt constructs, suggesting Dhh1, and Dhh1 speciﬁcally, plays a role in coupling 3.3. Dhh1p stimulates the degradation of mRNAs with low codon optimality 51

codon optimality to mRNA decay. We then repeated our half-life measurements of

our panel of 11 reporters in the absence of Dhh1. As expected, loss of Dhh1 lead

to a preferential stabilization of the low optimality reporter constructs (Figure 3.6).

Comparatively, Dhh1 played little role in modulating the stability of highly optimal

constructs relative to wild-type.

To compliment these reporter studies, we analyzed our RNA-Seq data between our

wild-type and dhh1 conditions to see di↵erential stabilization of mRNA according to codon optimality could be observed genome-wide. This was achieved by binning transcripts by stAI and looking at steady state levels of mRNAs in dhh1 relative to wild-type. Here too, we ﬁnd that low stAI mRNAs are more abundant at steady state in absence of Dhh1 (Figure 3.7). Interestingly, mRNA levels when Dhh1 is overexpressed by a glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter do not display an inverse trend, with high optimal mRNAs comprising a greater fraction of steady state mRNAs. This is potentially due to limiting cellular levels of downstream decay factors. Indeed, endogenous Dhh1 concentrations within the cell are already largely in excess relative to other decapping factors.119

In order to account for concerns that stAI may simply be serving as a proxy measure for another feature of mRNAs, namely secondary structure or GC content, we performed the above analysis, except binning transcripts based on the GC content of the transcript. In performing this analysis, we see no signiﬁcant dependence on mRNA levels on GC content, suggesting that stAI appears to be reporting on codon 3.4. Dhh1p binds preferentially to mRNAs of low codon optimality 52 optimality (Figure 3.8). Finally, it is important to note that the genome-wide data represent a steady-state analysis of mRNA levels, which necessarily misses some of the texture of a kinetic analysis. Nevertheless, the data remain strikingly consistent with the kinetic observations made with reporter mRNAs.

3.4 Dhh1p binds preferentially to mRNAs of low codon optimality

Given we see that Dhh1 selectively degrades mRNAs with low stAIs, we predicted that Dhh1 must then preferentially associate with these messenger ribonucleoproteins

(mRNPs). To test this, we used the synthetic Opt and Non-Opt mRNA reporters, and performed an anity pull-down (Figure 3.9A). This was done by treating cells with low amounts of formaldehyde to ensure that RNA was crosslinked to associated proteins. From these samples, we prepared cell lysates and hybridized the mRNA samples to DNA oligonucleotides antisense to the common 3’ UTR of the Opt and

Non-Opt mRNA reporters. Given that this oligo was conjugated to a biotin, we were able to anity purify the ribonucleoprotein (RNP) complex using magnetic streptavidin beads. Upon repeated washes, the bound material was eluted using a low-salt bu↵er.

Through this approach, we ﬁnd that we are able to greatly enrich for our reporter mRNA relative to endogenous control mRNAs (Figure 3.9B). Following elution, we then probe for the presence of Dhh1 by western blotting, which reveals a 3-fold enrichment of Dhh1 on the Non-Opt mRNA relative to the Opt mRNA (Figure 3.9C). 3.4. Dhh1p binds preferentially to mRNAs of low codon optimality 53

To control for the fact that there might be greater amounts of Opt mRNA, and thus greater amounts of bound Dhh1, we then probe for poly(A)-binding protein (Pab1),

ﬁnding equivalent amounts on both mRNAs.

We then extended this reporter analysis to characterize association of Dhh1 with mRNA transcripts across the entire genome. We were able to perform this analysis on data generated through previous crosslinking and immunoprecipitation (CLIP) studies.120 These earlier studies tried to unsuccessfully deﬁne a Dhh1 binding motif, noting only that Dhh1 bound throughout the 5’ and 3’ UTRs as well as the ORF of most genes with little apparent enrichment in any region of any transcript. However, by reanalyzing this published Dhh1 CLIP data, we were able to ask if a transcript’s propensity to associate with Dhh1 was governed by its optimality. In fact, in both replicates of the CLIP experiment, we observe that Dhh1 is preferentially bound to genes with low stAI relative to high stAI genes (Figure 3.10). 3.5. Materials and methods 54

3.5 Materials and methods

Bioinformatics and statistical analysis

All analysis of mRNA levels and ribosome occupancy looks only at genes that have greater than 128 mapped counts on average across all 18 datasets generated. All genome analysis plots are made using the ggplot2 R package; however, binning by sTAI and %GC content was performed in Python. Only bins of greater than 100 counts were considered to ensure that distributions were well sampled and e↵ects of outliers are minimized. Python analysis scripts and R plotting scripts are also available through GitHub at https://github.com/greenlabjhmi/2016-Cell-Dhh1.

Statistical parameters are reported in the Figures and the Figure Legends. Half- lives of reporter mRNAs were obtained by quantifying the Northern blots signals from biological triplicates. Concerning all genomic-wide analysis, violin plots list the number of genes in each bin as to better represent the distribution of genes across the various metrics (i.e., sTAI, GC content, etc.). Further in these analysis, replicates

(where available) were pooled and change in the median of two distributions (e.g., low sTAI/high sTAI) was calculated using a two-tailed Mann-Whitney test, where statistical signiﬁcance is denoted by a p value less than 0.05. Northern blots taken across polysomes were performed in triplicate and fold enrichment of mRNA is shown with calculated standard error.

Proﬁling and RNA-seq libraries were generated as outlined previously. Deep 3.5. Materials and methods 55

sequencing data for CLIP of Dhh1120 was downloaded from the GEO (Series ID

GSE46142).

Transcriptional shuto↵s and steady state RNA northern blot analysis

For the rpb1-1 shuto↵s, cells were grown at 24 C in synthetic media containing 2%

glucose and lacking the appropriate amino acids. Once the cells reached mid-log

phase, they were shifted to 37 Ctoinhibittranscription,andcellaliquotswerehar-

vested at the time points indicated in (Figure 3.3). For the GAL1 promoter shuto↵s

and steady state analysis, cells expressing the appropriate plasmids were grown at 24

Cinsyntheticmediawith2%galactose/1%sucrosetoallowforexpressionofthe reporter mRNA. For the steady state analysis, cells were harvested at OD600 =0.4.

For the transcriptional shuto↵s, cells were shifted to synthetic media without sugar at an OD600 = 0.4, and then transcription was repressed by adding glucose to a ﬁnal

concentration of 4%. Cells were collected at the time points indicated in the ﬁgures.

Total RNA was extracted by phenol/chloroform and precipitated by 95% EtOH.

30-40 µg of RNA was separated on 1.4% agarose-formaldehyde gels, transferred to

nylon membranes and probed with 32P-labeled oligonucleotides antisense to HIS3

(oJC2564), poly (G) (oJC168), MS2 binding sites (oJC1006), PGK1 (oJC986) or

SCR1 (oJC168). Blots were exposed to PhosphorImager screens, scanned by Typhon

9400, and quantiﬁed with ImageQuant software. 3.5. Materials and methods 56

Protein isolation and western blot analysis

Cells were harvested at OD600 = 0.4 and protein was isolated by 5M Urea and so- lution A (125 mM Tris-HCl pH 6.8, 2% SDS). Equivalent OD280 unit of protein was

separated on 10% SDS polyacrylamide gels, transferred to PVDF membrane, blot-

ted with primary antibodies (anti-HA [BioLegend], anti-Pab1 [EnCor Biotechnology],

anti-GAPDH [Cell Biolabs] and anti-Rpl5) at 4 Covernightandincubatedwithsec- ondary antibodies (goat-anti-Mouse [Santa Cruz sc-2005] and goat-anti-Rabbit [Pierce

31460]) at room temperature for 1 hr. Signal was detected by chemiluminescence using Blue Ultra Autorad ﬁlm (GeneMate F-2029).

mRNA pull-down

Cells (200 ml) were harvested at OD600 = 0.4 after crosslinking with 0.25% formalde-

hyde for 5 min and quenching with 125 mM glycine for 10 min. Cell pellets were

lysed in 400 µl 1X polysome lysis bu↵er (10 mM Tris, pH 7.4, 100 mM NaCl, 30 mM

MgCl2, 1 mM DTT) by vortexing with glass beads. The hot needle puncture method followed by centrifugation at 2,000 rpm for 2 min at 4 Cwasusedtoremovecell

debris. Equal OD units (OD260)ofeachlysateweredilutedtoaﬁnalvolumeof5mL in the hybridization reaction bu↵er (ﬁnal concentrations 500 mM LiCl, 0.5% SDS, 50 mM EDTA, 10 mM Tris pH 7.5, 14% formamide and Fungal protease inhibitors).

125 µL of streptavidin Dynabeads (Invitrogen #65002) were washed three times with an equal volume of 1X B&W bu↵er (5 mM Tris-HCl pH 7.5, 0.5 mM EDTA 3.5. Materials and methods 57

and 1 M NaCl) and once with 0.1 M NaCl. The beads were then incubated with 4

nM of biotinylated oligos complementary to the tag sequence (1.67 nM of each oligo

oJC2071, oJC2072 and oJC2073) in 1X B&W bu↵er at room temperature for 15 min.

After immobilization of the biotinylated oligos, beads were washed twice with 1X

B&W bu↵er and incubated with the 5 mL cell lysate at room temperature overnight.

Beads were then washed twice with Wash bu↵er 1 (10 mM Tris-HCl pH 7.5, 1 mM

EDTA, 250 mM LiCl and 0.1% SDS) and three times with Wash bu↵er 2 (10 mM

Tris-HCl pH 7.5, 1 mM EDTA, 100 mM LiCl). RNA was eluted by adding 93.5 mL

DEPC water and heating at 70 C for 2 min. RNA and protein were precipitated

and analyzed by Northern and western blot, respectively. Speciﬁcally, RNA was

precipitated at 20 C overnight by 0.3 M NaOAc, 1 mL of glycogen (ThermoFisher

AM9515) and 95% EtOH, then resuspended with 500 mL LET/SDS (1% SDS in LET bu↵er). Crosslinking was reversed at 70 C for 15 min and RNA was extracted once with phenol/chloroform/LET followed by another round of heating at 70 C and RNA extraction.

RNA was precipitated in 0.3 M NaOAc, 1 ml of glycogen and 95% EtOH at 20 C overnight, then resuspended in 15 mL DEPC water. For protein precipitation, eluate was concentrated in SpeedVac on high heat (37 C) to 1/5 of the original volume and precipitated with 10% TCA at 20 Covernight.Proteinswerepelletedat14,000rpm for 10 min at 4 C followed by one wash with 80% Acetone. Pellets were air-dried

and resuspended in 1X SDS Sample bu↵er. 3.6. Figures 58

3.6 Figures

10 Predicted Half-Life (Minutes)

0 20 40 60 Observed Half-Life (Minutes)

Figure 3.1: Experimentally observed vs. computationally predicted half-lives. mRNA half-lives were predicted based on calculating contributions of individual codons as solved for by (Equation 3.1). Values were calculated over a randomly selected test set of 1000 genes. Shown here are the predicted values for the remaining 2,889. 3.6. Figures 59

-50

-100 Half-Life Contribution (Minutes) TAT ATA TAC CTA TTA CAT ATC ATT AAT CTC ACC ACT CCC TCC CCT CAC TCT CTT TTT TTC ATG GTA GAT ACA CCA TCA AAC CAA CCG AGC TGT AGT CTG ACG CAG TCG GTC GCT TGC TTG CGC GCC CGT GAC GTT AAA CGA AGA GCA AAG GAA AGG GCG GGC CGG GGT TGG GTG GAG GGA GGG

Genomic Frequency 1

0 TAT ATA TAC CTA TTA CAT ATC ATT AAT CTC ACC ACT CCC TCC CCT CAC TCT CTT TTT TTC GTA ATG GAT ACA CCA TCA AAC CAA CCG CGC AGC TGT AGT CTG ACG CAG TCG GTC GCT TGC TTG GCC CGT GAC GTT AAA CGA AGA GCA AAG GAA AGG GCG GGC CGG GGT TGG GTG GAG GGA GGG Codon

Figure 3.2: Codon contributions to mRNA half-life

Contributions of individual codons to mRNA stability as solved for by solving the system of equations (Equation 3.1) for test set of 1000 randomly selected genes. Below this is the relative genomic frequency of each codon in protein-coding regions of the genome. 3.6. Figures 60

Probe seq (24nt) A AUG UAG C 50 HIS3 mRNA 220 amino acids 0.12 40 0.08 rpb1-1 shut-off 0.04 GAL1 UAS shut-off 0.00 30 -0.04 stabilization

Average codon Average -0.08 coefficient (CSC) coefficient 20 -0.12 mRNA half-life (min.) mRNA 0.60 0.40 10 0.20

Average sTAI Average 0.00 0 10 20 30 40 50 60 70 80 90 100 0 0 10 20 30 40 50 60 70 80 90 100 Percent optimal codons Percent optimal codons

rpb1-1 shut-off GAL1 UAS shut-off B 0 3 6 9 12 15 20 30 40 50 60 min. 0 3 6 9 12 15 20 30 40 50 60 min.

0 0

10 10

20 20

30 30

40 40

50 50

60 60 Percent optimal codons Percent optimal codons 70 70

80 80

90 90

100 100

Figure 3.3: Codon e↵ects on mRNA stability is a tunable phenomenon

(A) Representation of the HIS3 mRNA reporter. Each reporter encodes the exact same polypeptide sequence but is composed of di↵erent codon composition of varying optimality. The average codon stabilization coecient (CSC) and species-speciﬁc tRNA adaptation index (stAI) for each construct are shown. (B) Northern blots of the HIS3 reporter series following transcriptional shuto↵ in a rpb1-1 strain (left). The right panel shows the same reporters re-cloned with the GAL1 inducible promoter. Shown are northern blots following transcriptional inhibition with glucose. (C) Graphs of the half-lives of the mRNA reporters in (B). 3.6. Figures 61

0% Optimality 10% Optimality 20% Optimality

0.6 0.6 0.6

0.3 0.3 0.3

0.0 0.0 0.0

CSC Value/Average sTAI CSC Value/Average −0.3 sTAI CSC Value/Average −0.3 sTAI CSC Value/Average −0.3 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 Position Along HIS3 Gene Position Along HIS3 Gene Position Along HIS3 Gene 30% Optimality 40% Optimality 50% Optimality

0.6 0.6 0.6

0.3 0.3 0.3

0.0 0.0 0.0 CSC Value/Average sTAI CSC Value/Average sTAI CSC Value/Average −0.3 sTAI CSC Value/Average −0.3 −0.3 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200 Position Along HIS3 Gene Position Along HIS3 Gene Position Along HIS3 Gene 60% Optimality 70% Optimality 80% Optimality

0.6 0.6 0.6

0.3 0.3 0.3

0.0 0.0 0.0

90% Optimality 100% Optimality 0.6 0.6

0.3 0.3

0.0 0.0

CSC Value/Average sTAI CSC Value/Average −0.3 sTAI CSC Value/Average −0.3 0 50 100 150 200 0 50 100 150 200 Position Along HIS3 Gene Position Along HIS3 Gene

Figure 3.4: Codon composition of HIS3 reporter mRNAs

Graphs for CSC values (black bars) and sTAI values (blue line) averaged across ﬁve codon-long windows within the ORF of the HIS3 reporters. The red line represents the average sTAI across the gene for the 0% optimal HIS3 reporter. The total percent optimality is shown above each graph. Note the 50 end of each reporter is tagged with FLAG of consistent codon composition. Moreover, an identical codon stretch is present in all 11 reporters that comprise the probe site used for Northern analysis. 3.6. Figures 62

A Synthetic ORF MPPKASPTGASSVLKAKAPSIPAKTVGKTLPKTVITKLSTVITLGAAGLIVPLSIGIGV*

7 mGpppN AAAAn opt or non-opt codons

B 60 OPT 50 NON-OPT * * 40 * * 30 * * 20 * * *

mRNA Half Life (Minutes) * 0 WT dhh1Δ pat1Δ ccr4Δ dcp2Δ

Figure 3.5: Dhh1 selectively stimulates decay of mRNAs with low codon optimality

(A) Representation of the synthetic mRNAs (SYN) and the encoded polypeptide sequence. Optimal (OPT) or non-optimal (NON-OPT) codons encoding the same peptide were used. The artiﬁcial peptide has no similarity to any known proteins. (B) The half-lives of SYN OPT and NON-OPT mRNAs in WT and di↵erent mutant strains were obtained from GAL1 shuto↵ experiments. Quantitations were normalized to the amount of SCR1 RNA. The asterisk denotes average of three experiments. 3.6. Figures 63

50 WT dhh1Δ 40

20 mRNA half-life (min.) mRNA

0 0 10 20 30 40 50 60 70 80 90 100 Percent optimal codons

6 /wt) Δ 5 dhh1

2 Fold stabilization ( 1

0 0 10 20 30 40 50 60 70 80 90 100 Percent optimal codons

Figure 3.6: Dhh1-mediated decay of mRNAs depends on the level of codon optimality

Half-lives of HIS3 reporters from Figure 3.4 (GAL1 UAS constructs) in WT or dhh1 cells. Right panel indicates fold stabilization in a dhh1 cells versus WT. 3.6. Figures 64

n 131 2311 1743 497 151 107 100 10.0 Rel. to WT) Δ

1.0

0.1 Steady State mRNA (dhh1 0.25 0.3 0.35 0.4 0.45 0.5 0.55 sTAI n 131 2311 1743 497 151 107 100 10.0

1.0

0.1

Steady State mRNA (Dhh1p OE Rel. to WT) 0.25 0.3 0.35 0.4 0.45 0.5 0.55 sTAI

Figure 3.7: Loss of Dhh1 stabilizes low optimality mRNAs genome wide

Steady-state levels of mRNAs transcripts were quantiﬁed by RNA sequencing (RNA-seq) in dhh1cells (reads per kb per million [RPKM]) relative to WT cells (RPKM). mRNA transcripts are binned by sTAI, a numerical proxy for overall optimality. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low-optimality mRNAs (sTAI = 0.25, median [Med.] = 1.52) are enriched relative to high optimality mRNAs (sTAI = 0.55, Med. = 0.72) upon Dhh1p 16 depletion (U = 1668, p 2.2 10 ).  ⇥ By comparison, over-expression of Dhh1 does not a↵ect steady-state levels of mRNA. A two-tailed Mann-Whitney test shows that low optimality mRNAs (sTAI = 0.25, Med. = 1.09) are not enriched relative to high optimality mRNAs (sTAI = 0.55, Med. = 1.07) upon Dhh1p overexpression, U = 5412, p = 0.4593. 3.6. Figures 65

A n n 168 2484 1914 404 61 8 168 2484 1914 404 61 8 10.0 10.0 Rel. to WT) Δ dhh1 1.0 1.0

0.1 0.1 Steady State mRNA ( 0.35 0.4 0.45 0.5 0.55 0.6 Steady State mRNA (Dhh1p OE Rel. to WT) 0.35 0.4 0.45 0.5 0.55 0.6 Fraction GC Content Fraction GC Content B 1.00 n = 5,040 r = 0.228 Counts 0.75 300

200 0.50 100

0.25 Fractoin GC Content of Gene

0.00 0.00 0.25 0.50 0.75 1.00 sTAI of Gene

Figure 3.8: Dhh1-mediated decay is not due to mRNA secondary structure

(A) Quantifying steady state levels of mRNAs by RNA-Seq in dhh1D cells (RPKM) relative to WT cells (RPKM) where transcripts are binned by fraction GC content. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low GC content mRNAs (GC Fraction = 0.3, Med. = 1.33) are not enriched relative to high GC content mRNAs (GC Fraction = 0.55, Med. = 1.36) upon Dhh1p depletion, U=5210,p=0.847.

Steady state levels of mRNA by RNA-Seq in WT cells where Dhh1p is constitutively overexpressed (OE) relative to WT cells where transcripts are binned by fraction GC content. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low GC content (GC Fraction = 0.3, Med. = 0.95) are not enriched relative to high GC content mRNAs (GC Fraction = 0.55, Med. = 1.06) upon Dhh1p overexpression, U = 4102, p = 0.2117.

(B) Species-speciﬁc tRNA adaptation index (sTAI) plotted against percent GC content for all protein encoding transcripts in yeast. 3.6. Figures 66

A Synthetic ORF MPPKASPTGASSVLKAKAPSIPAKTVGKTLPKTVITKLSTVITLGAAGLIVPLSIGIGV*

7 mGpppN AAAAn opt or non-opt codons Biotin

B Input Sup. Streptavidin 1/10 1/10 Elution Dynabeads oonn o n Synthetic ORF Reporter

Endogenous PGK1 mRNA C o n 300

250 α-Dhh1p 200

150 α-Pab1p 100

GAPDH 50 Percent Dhh1p Bound

Figure 3.9: Dhh1 preferentially binds with low optimality mRNAs

(A) Representation of the reporters and experimental design used for mRNA pull- down. A tag sequence was inserted in the 30 UTR of the SYN reporters for pull-down. (B) Northern blot for the SYN mRNAs pull-downs. PGK1 mRNA was probed as a control of speciﬁcity. o, optimal; n, non-optimal. (C) Western blot showing the amount of Dhh1, Pab1, and GAPDH pulled down by the SYN mRNAs. Quantitations of Dhh1 were normalized to mRNA levels from eluates in (B). 3.6. Figures 67

n 91 626 3022 1944 530 163 123 92 8

10.0

0.1 mRNA Enrichment in Dhh1p Clip-Seq

0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 sTAI

Figure 3.10: Dhh1 associates with low optimality mRNAs genome-wide

Reanalysis of previously performed CLIP sequencing (CLIP-seq) on Dhh1p calculating enrichment of transcripts bound to Dhh1p relative to WT conditions, where transcripts are binned by sTAI. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low-optimality mRNAs (sTAI = 0.25, Med. = 2.02) are preferentially bound to Dhh1p relative to high-optimality mRNAs (sTAI = 0.55, 9 Med. = 0.32) (U = 304, p = 7.1 10 ). ⇥ Chapter 4

Dhh1 monitors codon optimality through ribosome elongation

Note: Parts of this chapter were published in:

Radhakrishnan, A., Chen, Y. H., Martin, S., Alhusaini, N., Green, R. & Coller, J.

(2016). The DEAD-Box Protein Dhh1p Couples mRNA Decay and Translation by

Monitoring Codon Optimality. Cell, 167 (1), 122132.

Radhakrishnan, A. & Green, R. (2016). Connections Underlying Translation and mRNA Stability. J. Mol. Biol. 428 (18), 3558-3564.

Having established a role for Dhh1 in coupling codon optimality to mRNA decay, two questions remain. First, how does a 60 kDa protein discriminate between all transcripts in the cell based on codon usage? Second, how do we reconcile the roles of Dhh1 in modulating ribosome occupancy on mRNAs while repressing translation?

A parsimonious explanation for these observations is that the density of slow-moving ribosomes on an mRNA (dictated by codon optimality) is sensed by Dhh1p and

68 4.1. Decay is stimulated by increasing numbers of slow-moving ribosomes 69

communicated to the mRNA degradation machinery.

4.1 Decay is stimulated by increasing numbers of slow-moving ribosomes

We tested if an increase in number of slowly elongating ribosomes was linked to

mRNA decay by generating a series of reporters based on the highly optimal PGK1

mRNA, where into each derivative we placed an identical stretch of ten amino acids

of exceptionally low stAI (stAI = 0.101) at increasing distances from the initiating

AUG: 5%, 25%, 50%, 63%, and 77% down the length of the gene (Figure 4.1A).

Importantly, the non-optimal codon (NC) stretch is of suciently low stAI that it

is predicted to dramatically slow ribosomes at the site and in turn upstream; we see

that protein expression is strongly and equivalently reduced for all ﬁve constructs to

10% of that of the normal PGK1 mRNA (Figure 4.1B). Interestingly, this suggests that the majority of Dhh1-mediated translation repression occurs upstream of decay.

As in our previous experiments, we measured half-lives of these reporter constructs upon shifting from galactose to glucose. We observed a clear polarity in our measured half-lives, where placing the NC stretch further from the stop codon served to decrease the stability of the reporter (Figure 4.1C and D). Strikingly, this polarity in decay was abrogated on deletion of Dhh1. These results are consistent with an interpretation where poor optimality stretches signiﬁcantly downstream of the start codon are have agreatercapacitytosequesterstalledribosomesupstream.Then,asthenumberof ribosomes stalled on the message increases, the stability of the mRNA decreases. 4.1. Decay is stimulated by increasing numbers of slow-moving ribosomes 70

However, to verify this hypothesis, we ﬁrst needed to demonstrate that the observed polarity e↵ect was dependent on active translation of the mRNA by the ribosome. Thus, we inserted a stem-loop capable of inhibiting translation initiation in the 5’ UTR (Figure 4.2A). In this context, the di↵erent constructs had largely similar stabilities, suggesting that the di↵erential translation of the reporters led to the polarity of decay observed.

Further, we wanted to establish that ribosomes on the NC stretch are still actively engaging in translation. If ribosomes on the NC stretch were truly stalled, the reporter may be an NGD substrate, and the experiment might instead be looking at decay of aberrant mRNAs. To test this, we placed a premature termination codon

(PTC) immediately downstream of the NC stretch (Figure 4.3A). Thus, if ribosomes are actively translating through the low optimality region, they encounter a PTC and become a substrate for NMD. Upon measuring the half-lives of these reporter constructs, we observe an inverse polarity, consistent with the impact of NMD (where

PTC closer to the start of the gene are more eciently decayed than PTCs near the end of the gene). However, when the experiment is repeated in a upf1 background, the polarity of mRNA decay once again reverses back to its original state

(Figure 4.3B). Taken together, these results suggest that ribosomes on low optimality regions are translating, albeit slowly. Finally, mRNAs containing slowly translating ribosomes seemed to be e↵ectively targeted for decay primarily by Dhh1, as other canonical ribosome quality control (RQC) factors play no role in regulating decay of 4.2. Dhh1p physically binds to the eukaryotic ribosome 71 these mRNAs (Figure 4.4).

Collectively, these data indicate that the polarity of mRNA degradation is translation dependent and on ribosome-associated events localized between the AUG start site and the NC stretch. The simplest explanation for these observations is that the number of slow-moving ribosomes on an mRNA determines the level of mRNA degradation observed.

4.2 Dhh1p physically binds to the eukaryotic ribosome

While CLIP data suggest that Dhh1p may directly bind to mRNA, thus dictating downstream functional consequences, it seems possible that like other DEAD-box proteins,121, 122 Dhh1p could also interact directly with the ribosome to mediate function.

We tested this hypothesis by using a tandem-anity tag (TAP) to purify Dhh1p from yeast cells and identify associated complexes by mass spectrometry. Importantly, we observed eight prominent protein bands upon purification that we identified as ribosomal proteins (Figure 4.5). We next repeated our TAP purification and probed for specific RNA species by northern blot. We observe that both the 25S and 18S rRNAs co-purify with Dhh1p, while other transcripts such as the 7S RNA (SCR1) or tRNA do not. Together, these data indicate that Dhh1p physically interacts with the ribosome. 4.3. Ribosome occupancy is enhanced upon Dhh1 binding 72

4.3 Ribosome occupancy is enhanced upon Dhh1 binding

Given the connection that we have established between ribosome density and Dhh1p function in mRNA decay, we next asked whether on a global scale there is preferential e↵ect of Dhh1p on the ribosome occupancy on mRNAs of low codon optimality.

Ribosome proﬁling was performed in four S. cerevisiae strains: wild-type, dhh1, and constitutively over-expressed Dhh1 and Dhh1-DQAD. While an assessment of ribosome occupancy (the average number of ribosomes on a given transcript) between the four strains failed to reveal genes or ontological categories of interest, characterizing genes binned according to their overall optimality (sTAI) revealed interesting features. In the Dhh1 over-express strain, we see a clear pattern of increased ribosome occupancy on non-optimal genes (Figure 4.6). As a control, we performed a similar analysis, measuring ribosome occupancy changes in the Dhh1 overexpression strain relative to the catalytically inactive Dhh1-DQAD protein. Again, we observe enrichment of ribosomes on low-optimality mRNAs, suggesting that this di↵erential ribosome occupancy is dependent on the catalytic activity of Dhh1p.

We next took advantage of the nucleotide resolution of ribosome footprint profiling to see if increased occupancy on non-optimal genes could be resolved at the codon level. To perform this analysis, we looked at a subset of the reads from footprint profiling (28-nt fragments) in the mutant and wild-type strains to characterize A-site occupancy. We find that when Dhh1 is overexpressed, relative to wild-type, there is 4.4. Discussion 73

increased footprint density when non-optimal codons occupy the A site (Figure 4.7);

no trends based on codon optimality are seen in the dhh1 strain.

We next probed the connections among ribosome occupancy, Dhh1 function, and codon optimality. We employed a similar tethering experiment but using the short

ORF we used previously to verify that mRNA shifting deep into gradients was due to ribosome occupancy rather than incorporation into granules (Ost4). We made synonymous variants of the Ost4 ORF with either high optimality (sTAI = 0.454) or low optimality (sTAI = 0.203) and evaluated its association with polysomes. With this reﬁnement, we could see di↵erences in ribosome occupancy on ORFs as a function of codon optimality. Consistent with our model, we see a clear increase in ribosome occupancy on the HA-Ost4-Non-Opt mRNA relative to the HA-Ost4-Opt mRNA, dependent on the presence of functional Dhh1p (Figure 4.8).

4.4 Discussion

In this work, we attempted to gain mechanistic insight into the role of Dhh1 in simultaneously repressing translation while increasing ribosomal occupancy. We ultimately met with some degree of success, providing a mechanistic understanding of how the rates of translation are communicated to the mRNA degradation apparatus.

That Dhh1 is a sensor of ribosome speed across the transcriptome, coupling codon optimality to mRNA decay, helps explain these vastly di↵ering phenotypes. This result was ultimately achieved through a combination of carefully designed reporters 4.4. Discussion 74

in conjunction with volumes of data generated through high-throughput methods.

Through polysomal analysis, we were able to show that tethering of Dhh1 to

mRNAs speciﬁcally leads to ribosomal accumulation, dependent on the optimality

of the mRNA. Through luciferase assays and mutational analysis, we were able to

gain some functional insight into the portions of Dhh1 relevant to its functions in

translation repression. Through RNA-Seq experiments, we were able to characterize

steady state shifts in the transcriptome upon depletion of Dhh1. Finally, through

ribosome proﬁling, we were able to gain insight on the level of individual nucleotides

by demonstrating that Dhh1 overexpression leads to increased ribosome dwell times

on non-optimal codons. Taken together, our data shed insight on central themes

of self-regulation in the coordination of gene expression, a↵ecting not only protein

synthesis and mRNA decay.

This role for Dhh1p in regulating translation elongation is consistent with obser-

vations from other systems. For instance, in Drosophila, translationally repressed

oskar and nanos mRNAs are found on polyribosomes in a so-called masked state;

the Dhh1p-homolog Me31b is required for their masking.75, 123 Similarly, the fragile

X mental retardation protein (FRMP), a RNA binding protein known to associate

with polysomes was also found regulate translation through induced stalling of ribo-

somes.124 In light of high sequence conservation amongst orthologs, and the essential nature of Dhh1 in higher eukaryotes, it seems likely that such a critical role in modulating translational elongation is conserved throughout the eukaryotic lineage. 4.4. Discussion 75

However, in order to gain further context into the mechanism and role of Dhh1,

it becomes necessary to both broaden and narrow the scope of inquiry. The mu-

tational analysis presented here provides a starting point to analyze the e↵ects of

various portions of Dhh1 on mRNA stability, translational repression, and ribosome

stalling. Without careful biochemical analysis, any further insight into Dhh1’s func-

tion with respect to the ribosome will be limited. At the same time, new evidence

suggests that the diverse roles ascribed to Dhh1 may be further regulated on the

cellular level.125 Dynamic reorganization of entire swathes of mRNPs into cellular granules and liquid droplets may be a regular occurrence in the life of an mRNA.

Better understanding the role of Dhh1 in shuttling of mRNPs to cellular foci could be achieved through extending the tools developed herein for use with microscopy and cell biology techniques.

A ﬁnal point of consideration that may serve to recontextualize the ﬁndings of this work is the very concept of “optimality.” As presented in this work, optimality refers to the intrinsic propensity of certain codons to be more eciently translated.

However, cells exist in dynamic milieus and any concept of optimality must then necessarily be ﬂuid. A potentially telling line of inquiry, then, is the role of Dhh1 under cellular stress. Is coupling between optimality and decay preserved? What is the nature of this regulation on the molecular level? And on the cellular level?

It is clear, that in spite and because of the results presented here, that much yet to be established about the role of Dhh1. Nevertheless, in light of recent ﬁndings, we 4.4. Discussion 76 are ﬁnally able to meaningfully establish connections between ribosome function and normal mRNA decay. Given that the main function of an mRNA is the production of protein product through translation, such a central role for the ribosome in specifying its stability is reassuring. 4.5. Materials and methods 77

4.5 Materials and methods

Tandem anity puriﬁcation

Dhh1p-TAP and associated complexes were puriﬁed from yeast cells by TAP method

126 as described previously. Brieﬂy, cells (grown to OD600 =1.2-1.3)werepelletedfor5 min, then washed and resuspended in one volume of ice cold bu↵er (10 mM K-HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.5 mM DTT, protease inhibitors). The cells

were then passed 3 times through a French-press (1,000 to 1,200 psi) and the lysate

was centrifuged at 16,500 rpm for 20 min at 4 C. The supernatant was diluted in a

ﬁnal concentration of 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% NP-40 (IPPI50

Bu↵er), 50 mg/mL heparin, and incubated with Sepharose 6B beads for 30 min at 4

C then IgG Sepharose 6 Flow for 3 hr at 4 C.The suspension was passed through a Bio-Rad Poly-Prep chromatography column then the beads were washed with 30 mL of IPPI50 Bu↵er and 10 mL of TEV C Bu↵er (-EDTA) (10 mM Tris-HCl pH 8.0,

150 mM NaCl, 0.1% NP-40, 1 mM DTT). They were resuspended in 1 mL of TEV C

Bu↵er (-EDTA) and incubated with 120 units of TEV (tobacco etch virus protease) overnight at 4 C. The eluate was drained into a new column and washed with TEC

C-bu↵er. The TEV supernatant was incubated for 1 hr with calmodulin Sepharose at

4 C in IPPI50 CBB Bu↵er (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM Mg(Ac)2,

1mMimidazole,2mMCaCl2, 0.1% NP-40, 10 mM -mercaptoethanol). After 3 washes with IPPI50 CBB Bu↵er, the complexes were eluted from the beads using 0.5 4.5. Materials and methods 78

mL IPPI50 CEB Bu↵er (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM Mg(Ac)2,

1 mM imidazole, 20 mM EGTA, 0.1% NP-40, 10 mM -mercaptoethanol), 6 times.

Proteins were precipitated with 20% TCA on ice for 30 min, pelleted for 30 min at 4

C, washed with acetone-0.05 N HCl, then resuspended in SDS sample Bu↵er to run on SDS-PAGE. Gels were stained with Coomassie blue. The proteins were identiﬁed by mass spectrometry. For RNA analysis, eluates were precipitated in ethanol at

20 C overnight, then RNA was extracted with phenol/chloroform and analyzed by

Northern blot.

Data and software availability

The three scripts used in analysis and generation of ﬁgures for deep sequencing data

are found on https://github.com/greenlabjhmi/ 2016-Cell-Dhh1. “Pipeline.py” is

used to process raw FASTQ ﬁles and generate WIG ﬁles as outlined in the methods

previously. “DataGen.py” uses the previously generated WIG ﬁles to create the

processed data files required to generate the figures in the paper. These figures are

then plotted using the R command “Plot.R.”

The accession number for the raw data ﬁles as well as per nucleotide counts (WIG

ﬁles) for the ribosome proﬁling and RNA sequencing analyses reported in this pa-

per is NCBI Gene Expression Omnibus: GSE81269. The accession number for the

raw Dhh1p CLIP data reported in this paper is NCBI Gene Expression Omnibus:

GSE46142. 4.5. Materials and methods 79

Growth and preparation for yeast for polysome analysis and northern blotting

Samples were prepared and analyzed as previously described (Chapter 2).

Preparation of footprint and RNA-Seq libraries

Samples were prepared and analyzed as previously described (Chapter 2). 4.5. Materials and methods 80 Publically available yeast strains generated for this study Table 4.1: MATa, ura3, leu2, his3, met15,MATa, ura3, [Dhh1, leu2, LEU2] his3, met15,MATa, ura3, [Dhh1-DQAD, leu2, LEU2] his3, met15,MATa, ura3, [HA-mCherry-BoxB, leu2, URA3], his3, [LambdaN-HA-Dhh1, met15, LEU2] MATa, ura3, [HA-mCherry-BoxB, leu2, URA3], his3, [LambdaN-HA-Dhh1-DQAD, met15, LEU2] MATa, ura3, [HA-OST4-BoxB leu2, opt, his3, URA3], met15, [LambdaN-HA-Dhh1,MATa, ura3, [HA-OST4-BoxB LEU2] leu2, opt, his3, URA3], met15, [LambdaN-HA-Dhh1-DQAD,MATa, ura3, [HA-OST4-BoxB LEU2] leu2, non-opt, his3, URA3], met15, [LambdaN-HA-Dhh1, [HA-OST4-BoxB LEU2] non-opt, URA3], [LambdaN-HA-Dhh1-DQAD, LEU2] yAR01 yAR02 yAR03 yAR04 yAR05 yAR06 yAR07 yAR08 4.6. Figures 81

4.6 Figures

A B 100 non-optimal codons Leu Ile Ala Arg Arg Arg Arg Arg Ala Thr 80 TTA ATA GCG CGG CGG CGG CGG CGG GCG ACG

HA 60 NC5 7 mGpppN AAAA 40 NC25 7 mGpppN AAAA 20

NC50 7 mGpppN AAAA Relative protein levels 0

NC63 7 mGpppN AAAA α-HA

NC77 7 mGpppN AAAA α-Rpl5p PGK1 mRNA

NC5 NC25 NC50 NC63 NC77

UntransformedNC0 (1:10 Dil)

C minutes D mRNA Half-Life (min) 0 2 4 6 8 10 20 30 40 60 0 10 20 30 NC0

NC5

NC25

NC50

NC63

NC77 WT dhh1Δ

Figure 4.1: Dhh1 senses polarity of non-optimal codons within mRNAs

(A) Representation of PGK1 reporters with a stretch of ten non-optimal codons at increasing distances from the initiating AUG. NC, non-optimal codons; NC0, no stretch; NC5, 25, 50, 63, 77, non-optimal codon stretch 5%, 25%, 50%, 63%, and 77% away from the AUG. (B) Protein output of the di↵erent reporters was analyzed by western blot; relative levels are plotted above. Rpl5p was probed as a loading control. (C) Northern blots of the di↵erent PGK1 reporters after GAL-transcriptional shuto↵ showing the remaining mRNA at the indicated time-points after shuto↵. (D) Half- lives of the di↵erent PGK1 reporters calculated from the northern blots (quantitation was normalized to SCR1, and loading controls are not shown) in WT and dhh1 cells. 4.6. Figures 82

A 7 SL-NC0 mGpppN AAAA n

7 SL-NC5 mGpppN AAAA n

7 SL-NC77 mGpppN AAAA n PGK1 mRNA HA B NC5 NC77 NC0 SL- SL- NC0 NC5 NC77 SL-

Reporter

SCR1 RNA

120 100 80 60 40 20 Percent of NC0 0

Figure 4.2: Dhh1-mediated degradation is dependent on inecient translation

(A) A stem loop (SL) was inserted in the 50 UTR of the previously described PGK1 reporters containing non-optimal codons at variable positions to inhibit translation. (B) Northern blot for steady-state abundance of the reporters with and without SL, with relative levels shown below. SCR1 was probed as a loading control. 4.6. Figures 83

A non-optimal codons Leu Ile Ala Arg Arg Arg Arg Arg Ala Thr TTA ATA GCG CGG CGG CGG CGG CGG GCG ACG

7 NC5-PTC mGpppN AAAAn

7 NC25-PTC mGpppN AAAAn

7 NC50-PTC mGpppN AAAAn

7 NC63-PTC mGpppN AAAAn

7 NC77-PTC mGpppN AAAAn PGK1 mRNA UAA

B WT upf1Δ NC5-PTC NC25-PTC NC50-PTC NC63-PTC NC77-PTC NC5-PTC NC25-PTC NC50-PTC NC63-PTC NC77-PTC

Reporter

SCR1 RNA

100 100 Percent of NC5 80 80

60 60

40 40

20 20 Percent of NC77 0 0

Figure 4.3: Dhh1-mediated degradation is dependent on ribosome pausing upstream of non-optimal stretches

(A) A premature termination codon (PTC) was inserted immediately after the NC stretch of the reporters to prevent ribosome association downstream of the stretch. (B) Northern blot for steady-state abundance of the reporters with and without PTC, and relative levels are shown below. SCR1 was probed as a loading control. 4.6. Figures 84

120 WT dom34Δ 100 ltn1Δ 80 rqc1Δ hel2Δ 60

Percent of WT NC0 20

0 NC0 NC5 NC25 NC50 NC63 NC77 Figure 4.4: Canonical RQC proteins do not sense polarity of non-optimal codons

Relative levels of the PGK1 reporters in di↵erent strains deleted for essential factors involved in the ribosome quality control (RQC) pathways. 4.6. Figures 85 Input 1/20 Dhh1p-Tap Tap Dhh1p-Tap 25S rRNA

L3 L4 18S rRNA

L2 S5 L13 7S RNA L19 L24 S8 tRNA

Figure 4.5: Pull-down of Dhh1 suggests association with the ribosome

Dhh1-TAP puriﬁcation followed by mass spectrometry (left, Coomassie blue gel stain- ing) or northern blots and speciﬁc probing for di↵erent rRNAs or tRNA (right). 4.6. Figures 86

n 131 2311 1743 497 151 107 100 10.0

1.0

0.1

0.25 0.3 0.35 0.4 0.45 0.5 0.55

Dhh1p OE Ribosome Occupancy Relative to WT sTAI

n 131 2311 1743 497 151 107 100 10.0

1.0 Relative to Dhh1p DQAD OE Dhh1p OE Ribosome Occupancy 0.1 0.25 0.3 0.35 0.4 0.45 0.5 0.55 sTAI

Figure 4.6: Catalytically active Dhh1 modulates ribosome occupancy on mRNAs with low codon optimality

Plotting the ribosome occupancy (average number of ribosomes per mRNA) for transcripts under constitutive Dhh1p OE relative to WT conditions (top), binning transcripts by sTAI. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low-optimality mRNAs (sTAI = 0.25, Med. = 1.30) have increased ribosome occupancy relative to high-optimality mRNAs (sTAI = 0.55, Med. = 0.72) (U = 1364, p ¡ 2.2 1016)uponDhh1poverexpression. ⇥ Plotting the ribosome occupancy (average number of ribosomes per mRNA) for transcripts under constitutive Dhh1p OE relative to constitutive Dhh1p-DQAD OE (bottom), binning transcripts by sTAI. Shown are two biological replicates. A two-tailed Mann-Whitney test shows that low optimality mRNAs (sTAI = 0.25, Med. = 1.53) have increased ribosome occupancy relative to high optimality mRNAs (sTAI = 0.55, Med. = 0.71, U = 685, p 1016)uponcatalyticallyactiveDhh1poverexpression ⇥ relative to catalytically inactive Dhh1p overexpression. 4.6. Figures 87

15 nt 13 nt

2.0

CCG

CCC AGG ACG sTAI 1.5 1.00

0.75

1.0 0.50

0.25

0.5

0.0 Codon Footprint Density Relative to WT Rep 1 Rep 2 Δ Δ

dhh1 dhh1 Dhh1p OE RepDhh1p 1 OE Rep 2

Figure 4.7: A-site occupancy by non-optimal codons is increased on Dhh1 overexpression

Quantifying the ribosome footprint density in the A site under Dhh1 OE or dhh1 relative to WT. The identity of the codon in the A site was determined by using 28-nt fragments (top diagram) as outlined previously.91 4.6. Figures 88

A HA-OST4 mRNA MYPYDVPDYAISDEQLNSLAITFGIVMMTLIVIYHAVDSTMSPKN*

7 mGpppN AAAAn opt or non-opt codons Dhh1p or Dhh1p DQAD B

2x 3x 4x 60S 40S 80S non-opt opt

6 4 2 0 Enrichment Figure 4.8: Dhh1 preferentially sequesters ribosomes on messages with low codon optimality

(A) Schematic of the reporter used in polysome occupancy assays. (B) Northern blots were used to quantify the enrichment (relative fractional occupancy) of optimal and non-optimal HA-OST4 mRNA along a polysome gradient upon tethering catalytically active and inactive Dhh1p. Reported values are averaged across three samples and presented with SE. Shown are representative northern blots for the non-optimal and optimal mRNAs upon tethering of catalytically active and inactive Dhh1p. 4.6. Figures 89

Non-optimal codons Optimal codons Translational Repression Efficient Translation

Decay

Figure 4.9: Dhh1 is a general sensor of ribosome speed during elongation

In this model, codon optimality inﬂuences the transit speed of ribosomes which in turns a↵ects the association of the decay factor Dhh1. Ribosomes are slowed down on non-optimal stretches, recruiting Dhh1, which may further slow ribosome movement and leads to mRNA decapping and degradation. Bibliography

[1] Crick, F. (1970). Central dogma of molecular biology. Nature, 227 (5258),

561–563.

[2] Volkin, E. & Astrachan, L. (1956). Intracellular distribution of labeled ribonucleic

acid after phage infection of escherichia coli. Virology, 2 (4), 433–7.

[3] Gros, F., Hiatt, H., Gilbert, W., Kurland, C. G., Risebrough, R. W. & Watson,

J. D. (1961). Unstable ribonucleic acid revealed by pulse labelling of escherichia

coli. Nature, 190,581–5.

[4] Brenner, S., Jacob, F. & Meselson, M. (1961). An unstable intermediate carrying

information from genes to ribosomes for protein synthesis. Nature, 190,576–581.

[5] Shandilya, J. & Roberts, S. G. (2012). The transcription cycle in eukaryotes:

from productive initiation to RNA polymerase II recycling. Biochim. Biophys.

Acta, 1819 (5), 391–400.

[6] Presnyak, V., Alhusaini, N., Chen, Y. H., Martin, S., Morris, N., Kline, N.,

90 Bibliography 91

Olson, S., Weinberg, D., Baker, K. E., Graveley, B. R. & Coller, J. (2015). Codon

optimality is a major determinant of mrna stability. Cell, 160 (6), 1111–24.

[7] Schwanhausser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J., Wolf, J.,

Chen, W. & Selbach, M. (2011). Global quantiﬁcation of mammalian gene expres-

sion control. Nature, 473 (7347), 337–42.

[8] Boel, G., Letso, R., Neely, H., Price, W. N., Wong, K. H., Su, M., Lu↵, J. D.,

Valecha, M., Everett, J. K., Acton, T. B., Xiao, R., Montelione, G. T., Aalberts,

D. P. & Hunt, J. F. (2016). Codon inﬂuence on protein expression in e. coli

correlates with mrna levels. Nature, 529 (7586), 358–63.

[9] Mishima, Y. & Tomari, Y. (2016). Codon usage and 3’ utr length determine

maternal mrna stability in zebraﬁsh. Mol Cell, 61 (6), 874–85.

[10] Bazzini, A. A., Del Viso, F., Moreno-Mateos, M. A., Johnstone, T. G., Vejnar,

C. E., Qin, Y., Yao, J., Khokha, M. K. & Giraldez, A. J. (2016). Codon identity

regulates mRNA stability and translation eciency during the maternal-to-zygotic

transition. EMBO J. 35 (19), 2087–2103.

[11] Laalami, S., Zig, L. & Putzer, H. (2014). Initiation of mrna decay in bacteria.

Cell Mol Life Sci, 71 (10), 1799–828.

[12] Hui, M. P., Foley, P. L. & Belasco, J. G. (2014). Messenger rna degradation in

bacterial cells. Annu Rev Genet, 48,537–59. Bibliography 92

[13] Decker, C. J. & Parker, R. (1993). A turnover pathway for both stable and

unstable mRNAs in yeast: evidence for a requirement for deadenylation. Genes

Dev. 7 (8), 1632–1643.

[14] Hsu, C. L. & Stevens, A. (1993). Yeast cells lacking 5’–¿3’ exoribonuclease

1containmrnaspeciesthatarepoly(a)deﬁcientandpartiallylackthe5’cap

structure. Mol Cell Biol, 13 (8), 4826–35.

[15] Anderson, J. S. & Parker, R. P. (1998). The 3’ to 5’ degradation of yeast mrnas

is a general mechanism for mrna turnover that requires the ski2 devh box protein

and 3’ to 5’ exonucleases of the exosome complex. EMBO J, 17 (5), 1497–506.

[16] Tucker, M., Valencia-Sanchez, M. A., Staples, R. R., Chen, J., Denis, C. L. &

Parker, R. (2001). The transcription factor associated ccr4 and caf1 proteins are

components of the major cytoplasmic mrna deadenylase in saccharomyces cere-

visiae. Cell, 104 (3), 377–86.

[17] van Dijk, E., Cougot, N., Meyer, S., Babajko, S., Wahle, E. & Seraphin, B.

(2002). Human dcp2: a catalytically active mrna decapping enzyme located in

speciﬁc cytoplasmic structures. EMBO J, 21 (24), 6915–24.

[18] Shatkin, A. J. (1976). Capping of eucaryotic mRNAs. Cell, 9 (4 PT 2), 645–653.

[19] Furuichi, Y., LaFiandra, A. & Shatkin, A. J. (1977). 5’-Terminal structure and

mRNA stability. Nature, 266 (5599), 235–239. Bibliography 93

[20] Franks, T. M. & Lykke-Andersen, J. (2008). The control of mRNA decapping

and P-body formation. Mol. Cell, 32 (5), 605–615.

[21] Maillet, L., Tu, C., Hong, Y. K., Shuster, E. O. & Collart, M. A. (2000). The

essential function of Not1 lies within the Ccr4-Not complex. J. Mol. Biol. 303 (2),

131–143.

[22] Panasenko, O. O. & Collart, M. A. (2011). Not4 E3 ligase contributes to pro-

teasome assembly and functional integrity in part through Ecm29. Mol. Cell. Biol.

31 (8), 1610–1623.

[23] Preissler, S., Reuther, J., Koch, M., Scior, A., Bruderek, M., Frickey, T. &

Deuerling, E. (2015). Not4-dependent translational repression is important for

cellular protein homeostasis in yeast. EMBO J. 34 (14), 1905–1924.

[24] Viswanathan, P., Ohn, T., Chiang, Y. C., Chen, J. & Denis, C. L. (2004). Mouse

CAF1 can function as a processive deadenylase/3’-5’-exonuclease in vitro but in

yeast the deadenylase function of CAF1 is not required for mRNA poly(A) removal.

J. Biol. Chem. 279 (23), 23988–23995.

[25] Tucker, M., Valencia-Sanchez, M. A., Staples, R. R., Chen, J., Denis, C. L. &

Parker, R. (2001). The transcription factor associated ccr4 and caf1 proteins are

components of the major cytoplasmic mrna deadenylase in saccharomyces cere-

visiae. Cell, 104 (3), 377–86. Bibliography 94

[26] van Hoof, A., Frischmeyer, P. A., Dietz, H. C. & Parker, R. (2002). Exosome-

mediated recognition and degradation of mRNAs lacking a termination codon.

Science, 295 (5563), 2262–2264.

[27] Lebreton, A., Tomecki, R., Dziembowski, A. & Seraphin, B. (2008). Endonucle-

olytic RNA cleavage by a eukaryotic exosome. Nature, 456 (7224), 993–996.

[28] Chlebowski, A., Lubas, M., Jensen, T. H. & Dziembowski, A. (2013). RNA decay

machines: the exosome. Biochim. Biophys. Acta, 1829 (6-7), 552–560.

[29] Halbach, F., Reichelt, P., Rode, M. & Conti, E. (2013). The yeast ski complex:

crystal structure and RNA channeling to the exosome complex. Cell, 154 (4),

814–826.

[30] Balagopal, V. & Parker, R. (2009). Polysomes, P bodies and stress granules:

states and fates of eukaryotic mRNAs. Curr. Opin. Cell Biol. 21 (3), 403–408.

[31] Coller, J. & Parker, R. (2004). Eukaryotic mRNA decapping. Annu. Rev.

Biochem. 73,861–890.

[32] Schwartz, D., Decker, C. J. & Parker, R. (2003). The enhancer of decapping

proteins, Edc1p and Edc2p, bind RNA and stimulate the activity of the decapping

enzyme. RNA, 9 (2), 239–251.

[33] Kshirsagar, M. & Parker, R. (2004). Identiﬁcation of Edc3p as an enhancer of

mRNA decapping in Saccharomyces cerevisiae. Genetics, 166 (2), 729–739. Bibliography 95

[34] Coller, J. M., Tucker, M., Sheth, U., Valencia-Sanchez, M. A. & Parker, R.

(2001). The DEAD box helicase, Dhh1p, functions in mRNA decapping and inter-

acts with both the decapping and deadenylase complexes. RNA, 7 (12), 1717–1727.

[35] Coller, J. & Parker, R. (2005). General translational repression by activators of

mRNA decapping. Cell, 122 (6), 875–886.

[36] Pilkington, G. R. & Parker, R. (2008). Pat1 contains distinct functional domains

that promote P-body assembly and activation of decapping. Mol. Cell. Biol. 28

(4), 1298–1312.

[37] Hu, W., Sweet, T. J., Chamnongpol, S., Baker, K. E. & Coller, J. (2009). Co-

translational mrna decay in saccharomyces cerevisiae. Nature, 461 (7261), 225–9.

[38] Pelechano, V., Wei, W. & Steinmetz, L. M. (2015). Widespread co-translational

rna decay reveals ribosome dynamics. Cell, 161 (6), 1400–12.

[39] Keiler, K. C., Waller, P. R. & Sauer, R. T. (1996). Role of a peptide tagging sys-

tem in degradation of proteins synthesized from damaged messenger rna. Science,

271 (5251), 990–3.

[40] Moore, S. D. & Sauer, R. T. (2007). The tmrna system for translational surveil-

lance and ribosome rescue. Annu Rev Biochem, 76,101–24.

[41] Shoemaker, C. J. & Green, R. (2012). Translation drives mrna quality control.

Nat Struct Mol Biol, 19 (6), 594–601. Bibliography 96

[42] Roy, B. & Jacobson, A. (2013). The intimate relationships of mrna decay and

translation. Trends Genet, 29 (12), 691–9.

[43] Losson, R. & Lacroute, F. (1979). Interference of nonsense mutations with eu-

karyotic messenger RNA stability. Proc. Natl. Acad. Sci. U.S.A. 76 (10), 5134–

5137.

[44] Pulak, R. & Anderson, P. (1993). mRNA surveillance by the Caenorhabditis

elegans smg genes. Genes Dev. 7 (10), 1885–1897.

[45] Leeds, P., Peltz, S. W., Jacobson, A. & Culbertson, M. R. (1991). The product

of the yeast UPF1 gene is required for rapid turnover of mRNAs containing a

premature translational termination codon. Genes Dev. 5 (12A), 2303–2314.

[46] He, F. & Jacobson, A. (1995). Identiﬁcation of a novel component of the

nonsense-mediated mRNA decay pathway by use of an interacting protein screen.

Genes Dev. 9 (4), 437–454.

[47] Cui, Y., Hagan, K. W., Zhang, S. & Peltz, S. W. (1995). Identiﬁcation and

characterization of genes that are required for the accelerated degradation of mR-

NAs containing a premature translational termination codon. Genes Dev. 9 (4),

423–436.

[48] Chang, Y. F., Imam, J. S. & Wilkinson, M. F. (2007). The nonsense-mediated

decay RNA surveillance pathway. Annu. Rev. Biochem. 76,51–74. Bibliography 97

[49] Lejeune, F., Li, X. & Maquat, L. E. (2003). Nonsense-mediated mRNA decay in

mammalian cells involves decapping, deadenylating, and exonucleolytic activities.

Mol. Cell, 12 (3), 675–687.

[50] Doma, M. K. & Parker, R. (2006). Endonucleolytic cleavage of eukaryotic mR-

NAs with stalls in translation elongation. Nature, 440 (7083), 561–564.

[51] Gandhi, R., Manzoor, M. & Hudak, K. A. (2008). Depurination of Brome mosaic

virus RNA3 in vivo results in translation-dependent accelerated degradation of the

viral RNA. J. Biol. Chem. 283 (47), 32218–32228.

[52] Kuroha, K., Akamatsu, M., Dimitrova, L., Ito, T., Kato, Y., Shirahige, K. &

Inada, T. (2010). Receptor for activated C kinase 1 stimulates nascent polypeptide-

dependent translation arrest. EMBO Rep. 11 (12), 956–961.

[53] Chen, L., Muhlrad, D., Hauryliuk, V., Cheng, Z., Lim, M. K., Shyp, V., Parker,

R. & Song, H. (2010). Structure of the Dom34-Hbs1 complex and implications for

no-go decay. Nat. Struct. Mol. Biol. 17 (10), 1233–1240.

[54] Guydosh, N. R. & Green, R. (2014). Dom34 rescues ribosomes in 3’ untranslated

regions. Cell, 156 (5), 950–962.

[55] Koutmou, K. S., Schuller, A. P., Brunelle, J. L., Radhakrishnan, A., Djura-

novic, S. & Green, R. (2015). Ribosomes slide on lysine-encoding homopolymeric

A stretches. Elife, 4. Bibliography 98

[56] Frischmeyer, P. A., van Hoof, A., O’Donnell, K., Guerrerio, A. L., Parker, R. &

Dietz, H. C. (2002). An mRNA surveillance mechanism that eliminates transcripts

lacking termination codons. Science, 295 (5563), 2258–2261.

[57] Schae↵er, D. & van Hoof, A. (2011). Di↵erent nuclease requirements for exosome-

mediated degradation of normal and nonstop mRNAs. Proc. Natl. Acad. Sci.

U.S.A. 108 (6), 2366–2371.

[58] Becker, T., Armache, J. P., Jarasch, A., Anger, A. M., Villa, E., Sieber, H.,

Motaal, B. A., Mielke, T., Berninghausen, O. & Beckmann, R. (2011). Structure

of the no-go mrna decay complex dom34-hbs1 bound to a stalled 80s ribosome. Nat

Struct Mol Biol, 18 (6), 715–20.

[59] Shoemaker, C. J., Eyler, D. E. & Green, R. (2010). Dom34:hbs1 promotes

subunit dissociation and peptidyl-trna drop-o↵ to initiate no-go decay. Science,

330 (6002), 369–72.

[60] Brandman, O., Stewart-Ornstein, J., Wong, D., Larson, A., Williams, C. C., Li,

G. W., Zhou, S., King, D., Shen, P. S., Weibezahn, J., Dunn, J. G., Rouskin, S.,

Inada, T., Frost, A. & Weissman, J. S. (2012). A ribosome-bound quality control

complex triggers degradation of nascent peptides and signals translation stress.

Cell, 151 (5), 1042–54.

[61] Tsuboi, T., Kuroha, K., Kudo, K., Makino, S., Inoue, E., Kashima, I. & Inada, T. Bibliography 99

(2012). Dom34:hbs1 plays a general role in quality-control systems by dissociation

of a stalled ribosome at the 3’ end of aberrant mrna. Mol Cell, 46 (4), 518–29.

[62] Jain, C. & Kleckner, N. (1993). IS10 mRNA stability and steady state levels

in Escherichia coli: indirect e↵ects of translation and role of rne function. Mol.

Microbiol. 9 (2), 233–247.

[63] Wagner, L. A., Gesteland, R. F., Dayhu↵, T. J. & Weiss, R. B. (1994). An

ecient Shine-Dalgarno sequence but not translation is necessary for lacZ mRNA

stability in Escherichia coli. J. Bacteriol. 176 (6), 1683–1688.

[64] Iost, I. & Dreyfus, M. (1995). The stability of Escherichia coli lacZ mRNA

depends upon the simultaneity of its synthesis and translation. EMBO J. 14 (13),

3252–3261.

[65] Gaba, A., Jacobson, A. & Sachs, M. S. (2005). Ribosome occupancy of the

yeast CPA1 upstream open reading frame termination codon modulates nonsense-

mediated mRNA decay. Mol. Cell, 20 (3), 449–460.

[66] Ruiz-Echevarria, M. J. & Peltz, S. W. (2000). The RNA binding protein Pub1

modulates the stability of transcripts containing upstream open reading frames.

Cell, 101 (7), 741–751.

[67] Hurt, J. A., Robertson, A. D. & Burge, C. B. (2013). Global analyses of UPF1 Bibliography 100

binding and function reveal expanded scope of nonsense-mediated mRNA decay.

Genome Res. 23 (10), 1636–1650.

[68] Sharp, P. M. & Li, W. H. (1987). The codon adaptation index–a measure of

directional synonymous codon usage bias, and its potential applications. Nucleic

Acids Res, 15 (3), 1281–95.

[69] Cordin, O., Banroques, J., Tanner, N. K. & Linder, P. (2006). The DEAD-box

protein family of RNA helicases. Gene, 367,17–37.

[70] Linder, P. (2006). Dead-box proteins: a family a↵air–active and passive players

in RNP-remodeling. Nucleic Acids Res. 34 (15), 4168–4180.

[71] Fischer, N. & Weis, K. (2002). The dead box protein dhh1 stimulates the decap-

ping enzyme dcp1. EMBO J, 21 (11), 2788–97.

[72] Maillet, L. & Collart, M. A. (2002). Interaction between Not1p, a component of

the Ccr4-not complex, a global regulator of transcription, and Dhh1p, a putative

RNA helicase. J. Biol. Chem. 277 (4), 2835–2842.

[73] Carroll, J. S., Munchel, S. E. & Weis, K. (2011). The dexd/h box atpase dhh1

functions in translational repression, mrna decay, and processing body dynamics.

J Cell Biol, 194 (4), 527–37.

[74] Sweet, T., Kovalak, C. & Coller, J. (2012). The DEAD-box protein Dhh1 pro-

motes decapping by slowing ribosome movement. PLoS Biol. 10 (6), e1001342. Bibliography 101

[75] Nakamura, A., Amikura, R., Hanyu, K. & Kobayashi, S. (2001). Me31B silences

translation of oocyte-localizing RNAs through the formation of cytoplasmic RNP

complex during Drosophila oogenesis. Development, 128 (17), 3233–3242.

[76] Minshall, N. & Standart, N. (2004). The active form of Xp54 RNA helicase in

translational repression is an RNA-mediated oligomer. Nucleic Acids Res. 32 (4),

1325–1334.

[77] Minshall, N., Kress, M., Weil, D. & Standart, N. (2009). Role of p54 RNA

helicase activity and its C-terminal domain in translational repression, P-body

localization and assembly. Mol. Biol. Cell, 20 (9), 2464–2472.

[78] Naarmann, I. S., Harnisch, C., Muller-Newen, G., Urlaub, H., Ostareck-Lederer,

A. & Ostareck, D. H. (2010). DDX6 recruits translational silenced human reticu-

locyte 15-lipoxygenase mRNA to RNP granules. RNA, 16 (11), 2189–2204.

[79] Zabolotskaya, M. V., Grima, D. P., Lin, M. D., Chou, T. B. & Newbury, S. F.

(2008). The 5’-3’ exoribonuclease Pacman is required for normal male fertility and

is dynamically localized in cytoplasmic particles in Drosophila testis cells. Biochem.

J. 416 (3), 327–335.

[80] Minshall, N., Thom, G. & Standart, N. (2001). A conserved role of a DEAD box

helicase in mRNA masking. RNA, 7 (12), 1728–1742.

[81] Ladomery, M., Wade, E. & Sommerville, J. (1997). Xp54, the Xenopus ho- Bibliography 102

mologue of human RNA helicase p54, is an integral component of stored mRNP

particles in oocytes. Nucleic Acids Res. 25 (5), 965–973.

[82] Chu, C. Y. & Rana, T. M. (2006). Translation repression in human cells by

microRNA-induced gene silencing requires RCK/p54. PLoS Biol. 4 (7), e210.

[83] Djuranovic, S., Zinchenko, M. K., Hur, J. K., Nahvi, A., Brunelle, J. L., Rogers,

E. J. & Green, R. (2010). Allosteric regulation of Argonaute proteins by miRNAs.

Nat. Struct. Mol. Biol. 17 (2), 144–150.

[84] Franklin, N. C. (1985). ”n” transcription antitermination proteins of bacterio-

phages lambda, phi 21 and p22. J Mol Biol, 181 (1), 85–91.

[85] Franklin, N. C. (1985). Conservation of genome form but not sequence in the

transcription antitermination determinants of bacteriophages lambda, phi 21 and

p22. J Mol Biol, 181 (1), 75–84.

[86] Lazinski, D., Grzadzielska, E. & Das, A. (1989). Sequence-speciﬁc recognition

of rna hairpins by bacteriophage antiterminators requires a conserved arginine-rich

motif. Cell, 59 (1), 207–18.

[87] Hur, J. K., Zinchenko, M. K., Djuranovic, S. & Green, R. (2013). Regulation

of Argonaute slicer activity by guide RNA 3’ end interactions with the N-terminal

lobe. J. Biol. Chem. 288 (11), 7829–7840. Bibliography 103

[88] Cheng, Z., Coller, J., Parker, R. & Song, H. (2005). Crystal structure and

functional analysis of DEAD-box protein Dhh1p. RNA, 11 (8), 1258–1270.

[89] Matsui, T., Hogetsu, K., Akao, Y., Tanaka, M., Sato, T., Kumasaka, T. &

Tanaka, N. (2004). Crystallization and X-ray analysis of the N-terminal core do-

main of a tumour-associated human DEAD-box RNA helicase, rck/p54. Acta Crys-

tallogr. D Biol. Crystallogr. 60 (Pt 1), 156–159.

[90] Matsui, T., Hogetsu, K., Usukura, J., Sato, T., Kumasaka, T., Akao, Y. &

Tanaka, N. (2006). Structural insight of human DEAD-box protein rck/p54 into

its substrate recognition with conformational changes. Genes Cells, 11 (4), 439–

452.

[91] Ingolia, N. T., Ghaemmaghami, S., Newman, J. R. & Weissman, J. S. (2009).

Genome-wide analysis in vivo of translation with nucleotide resolution using ribo-

some proﬁling. Science, 324 (5924), 218–223.

[92] Ingolia, N. T., Lareau, L. F. & Weissman, J. S. (2011). Ribosome proﬁling of

mouse embryonic stem cells reveals the complexity and dynamics of mammalian

proteomes. Cell, 147 (4), 789–802.

[93] Ingolia, N. T., Brar, G. A., Rouskin, S., McGeachy, A. M. & Weissman, J. S.

(2012). The ribosome proﬁling strategy for monitoring translation in vivo by deep

sequencing of ribosome-protected mRNA fragments. Nat Protoc, 7 (8), 1534–1550. Bibliography 104

[94] Cheung, Y. N., Maag, D., Mitchell, S. F., Fekete, C. A., Algire, M. A., Takacs,

J. E., Shirokikh, N., Pestova, T., Lorsch, J. R. & Hinnebusch, A. G. (2007). Disso-

ciation of eIF1 from the 40S ribosomal subunit is a key step in start codon selection

in vivo. Genes Dev. 21 (10), 1217–1230.

[95] Kolitz, S. E., Takacs, J. E. & Lorsch, J. R. (2009). Kinetic and thermodynamic

analysis of the role of start codon/anticodon base pairing during eukaryotic trans-

lation initiation. RNA, 15 (1), 138–152.

[96] Parker, R., Herrick, D., Peltz, S. W. & Jacobson, A. (1991). Measurement of

mRNA decay rates in Saccharomyces cerevisiae. Meth. Enzymol. 194,415–423.

[97] Dix, D. B. & Thompson, R. C. (1989). Codon choice and gene expression:

synonymous codons di↵er in translational accuracy. Proc Natl Acad Sci U S A, 86

(18), 6888–92.

[98] Thomas, L. K., Dix, D. B. & Thompson, R. C. (1988). Codon choice and

gene expression: synonymous codons di↵er in their ability to direct aminoacylated-

transfer rna binding to ribosomes in vitro. Proc Natl Acad Sci U S A, 85 (12),

4242–6.

[99] Ikemura, T. & Ozeki, H. (1983). Codon usage and transfer rna contents:

organism-speciﬁc codon-choice patterns in reference to the isoacceptor contents.

Cold Spring Harb Symp Quant Biol, 47 Pt 2,1087–97. Bibliography 105

[100] Drummond, D. A. & Wilke, C. O. (2008). Mistranslation-induced protein mis-

folding as a dominant constraint on coding-sequence evolution. Cell, 134 (2),

341–52.

[101] Gingold, H. & Pilpel, Y. (2011). Determinants of translation eciency and

accuracy. Mol Syst Biol, 7,481.

[102] Pechmann, S. & Frydman, J. (2013). Evolutionary conservation of codon op-

timality reveals hidden signatures of cotranslational folding. Nat Struct Mol Biol,

20 (2), 237–43.

[103] Novoa, E. M. & Ribas de Pouplana, L. (2012). Speeding with control: codon

usage, tRNAs, and ribosomes. Trends Genet, 28 (11), 574–81.

[104] Bulmer, M., Wolfe, K. H. & Sharp, P. M. (1991). Synonymous nucleotide

substitution rates in mammalian genes: implications for the molecular clock and

the relationship of mammalian orders. Proc Natl Acad Sci U S A, 88 (14), 5974–8.

[105] Hershberg, R. & Petrov, D. A. (2008). Selection on codon bias. Annu Rev

Genet, 42,287–99.

[106] Man, O. & Pilpel, Y. (2007). Di↵erential translation eciency of orthologous

genes is involved in phenotypic divergence of yeast species. Nat Genet, 39 (3),

415–21.

[107] Dong, H., Nilsson, L. & Kurland, C. G. (1996). Co-variation of trna abundance Bibliography 106

and codon usage in escherichia coli at di↵erent growth rates. J Mol Biol, 260 (5),

649–63.

[108] Sorensen, M. A., Kurland, C. G. & Pedersen, S. (1989). Codon usage determines

translation rate in escherichia coli. J Mol Biol, 207 (2), 365–77.

[109] Tuller, T., Waldman, Y. Y., Kupiec, M. & Ruppin, E. (2010). Translation

eciency is determined by both codon bias and folding energy. Proc Natl Acad Sci

USA, 107 (8), 3645–50.

[110] Varenne, S., Buc, J., Lloubes, R. & Lazdunski, C. (1984). Translation is a

non-uniform process. e↵ect of trna availability on the rate of elongation of nascent

polypeptide chains. J Mol Biol, 180 (3), 549–76.

[111] Kim, S. J., Yoon, J. S., Shishido, H., Yang, Z., Rooney, L. A., Barral, J. M.

& Skach, W. R. (2015). Protein folding. translational tuning optimizes nascent

protein folding in cells. Science, 348 (6233), 444–8.

[112] Sander, I. M., Chaney, J. L. & Clark, P. L. (2014). Expanding anﬁnsen’s

principle: contributions of synonymous codon selection to rational protein design.

J Am Chem Soc, 136 (3), 858–61.

[113] Spencer, P. S., Siller, E., Anderson, J. F. & Barral, J. M. (2012). Silent substitu-

tions predictably alter translation elongation rates and protein folding eciencies.

J Mol Biol, 422 (3), 328–35. Bibliography 107

[114] Yu, C. H., Dang, Y., Zhou, Z., Wu, C., Zhao, F., Sachs, M. S. & Liu, Y.

(2015). Codon usage inﬂuences the local rate of translation elongation to regulate

co-translational protein folding. Mol Cell, 59 (5), 744–54.

[115] Zhang, G., Hubalewska, M. & Ignatova, Z. (2009). Transient ribosomal atten-

uation coordinates protein synthesis and co-translational folding. Nat Struct Mol

Biol, 16 (3), 274–80.

[116] dos Reis, M., Savva, R. & Wernisch, L. (2004). Solving the riddle of codon

usage preferences: a test for translational selection. Nucleic Acids Res, 32 (17),

5036–44.

[117] Roymondal, U., Das, S. & Sahoo, S. (2009). Predicting gene expression level

from relative codon usage bias: an application to Escherichia coli genome. DNA

Res. 16 (1), 13–30.

[118] Sabi, R. & Tuller, T. (2014). Modelling the eciency of codon-trna interactions

based on codon usage bias. DNA Res, 21 (5), 511–26.

[119] Ghaemmaghami, S., Huh, W. K., Bower, K., Howson, R. W., Belle, A., De-

phoure, N., O’Shea, E. K. & Weissman, J. S. (2003). Global analysis of protein

expression in yeast. Nature, 425 (6959), 737–41.

[120] Mitchell, S. F., Jain, S., She, M. & Parker, R. (2013). Global analysis of yeast

mrnps. Nat Struct Mol Biol, 20 (1), 127–33. Bibliography 108

[121] Geissler, R., Golbik, R. P. & Behrens, S. E. (2012). The dead-box helicase ddx3

supports the assembly of functional 80s ribosomes. Nucleic Acids Res, 40 (11),

4998–5011.

[122] Gingras, A. C., Raught, B. & Sonenberg, N. (1999). eif4 initiation factors:

e↵ectors of mrna recruitment to ribosomes and regulators of translation. Annu

Rev Biochem, 68,913–63.

[123] Braat, A. K., Yan, N., Arn, E., Harrison, D. & Macdonald, P. M. (2004).

Localization-dependent oskar protein accumulation; control after the initiation of

translation. Dev Cell, 7 (1), 125–31.

[124] Darnell, J. C., Van Driesche, S. J., Zhang, C., Hung, K. Y., Mele, A., Fraser,

C. E., Stone, E. F., Chen, C., Fak, J. J., Chi, S. W., Licatalosi, D. D., Richter,

J. D. & Darnell, R. B. (2011). Fmrp stalls ribosomal translocation on mrnas linked

to synaptic function and autism. Cell, 146 (2), 247–61.

[125] Mugler, C. F., Hondele, M., Heinrich, S., Sachdev, R., Vallotton, P., Koek,

A. Y., Chan, L. Y. & Weis, K. (2016). ATPase activity of the DEAD-box protein

Dhh1 controls processing body formation. Elife, 5.

[126] Rigaut, G., Shevchenko, A., Rutz, B., Wilm, M., Mann, M. & Seraphin, B.

(1999). A generic protein puriﬁcation method for protein complex characterization

and proteome exploration. Nat Biotechnol, 17 (10), 1030–2. ADITYA RADHAKRISHNAN

personal information address 1211 Light St Unit 115 Baltimore, MD 21230 email [email protected] phone (M) +1 (630) 991 3977 website http://radhakrishnan.me education

2011-2016 Johns Hopkins University PhD in School: Johns Hopkins Medical Institute Biophysics Thesis: Dhh1 regulates translation by monitoring codon optimality Committee Members: Rachel Green, James Berger, Christian Kaiser, Jeffry Corden, Sarah Woodson

2010-2011 Washington University in St. Louis Masters in GPA: 3.80/4.0 School: Engineering and Applied Science · Biomedical Thesis: Monte Carlo simulations of poly-L-proline via the ABSINTH framework Engineering Advisor: Rohit Pappu

2006-2010 Washington University in St. Louis Bachelors in GPA: 3.58/4.0 School: Engineering and Applied Science · Biomedical Engineering research experience 2011-2016 Graduate Student Under Prof. Rachel Green, Department of Molecular Biology and Genetics Johns Hopkins University School of Medicine, Baltimore, MD

Developed a system to study the effects of the DEAD-box protein Dhh1 in molecular biology and high throughput contexts, suggesting its role as a monitor of ribosomal elongation speed, thereby suggesting a molecular mechanism for the link between codon identity of mRNAs and their intrinsic decay rates. This was done through collaboration with the Coller Lab at Case Western Reserve University. In addition, I helped to develop tools to aid in the analysis of large bioinformatics datasets, for the high-throughput genomic studies, which are publicly available via GitHub.

2011-2016 Graduate Student Under Prof. Rohit Pappu, Department of Biomedical Engineering Washington University in St. Louis, St. Louis, MO

Implemented the framework required for accurate atomistic simulation of proline in the chemical simulation package ABSINTH. Worked with a team of researchers and an centralized, extant code base for this implementation. Wrote a scientific paper and presented data based on simulations performed on proline polymers post- implementation. Developed assets to facilitate simulation of proline in biologically important contexts (e.g. Huntingtin protein and the larger class of amyloid-fibril based neurodegenerative diseases). teaching experience 2012-2014 Teaching Assistant Johns Hopkins 250.689 - Physical Chemistry of Biological Macromolecules Worked with graduate · University students of high technical proficiency and developed an optional “scientific computa- tion boot camp” to instruct them in the use of Mathematica. Held office hours and assessed problem sets. Received highly positive reviews in student evaluations.

250.381 - Spectroscopy and Its Application in Biophysical Reactions Designed · and taught six lectures to undergraduates with limited prior technical knowledge: 3 on the mathematical foundations of quantum mechanics, 3 based in reading and analysis of current technical literature in biological spectroscopy. Held ofﬁce hours to directly address student issues. Received positive reviews in student evaluations.

publications

Radhakrishnan A, Chen Y, Martin S, Alhusaini N, Green R, Coller J. (2016) The DEAD-box protein Dhh1p couples mRNA decay and translation by monitoring codon optimality. Cell. 167: 122-32. DOI: 10.1016/j.cell.2016.08.053.

Radhakrishnan A, Green R. (2016) Connections underlying translation and mRNA stability. J. Mol. Biol. 428: 3558-64. DOI: 10.1016/j.jmb.2016.05.025.

Koutmou KS, Radhakrishnan A, Green R. (2015) Synthesis at the speed of codons. Trends Biochem. Sci. 40: 717-18. DOI: 10.1016/j.tibs.2015.10.005.

Koutmou KS, Schuller AP, Brunelle JL, Radhakrishnan A, Djuranovic S, Green R. (2015) Ribosomes slide on lysine-encoding homopolymeric A stretches. Elife. 19: 4. DOI: 10.7554/eLife.05534.

Radhakrishnan A, Vitalis A, Mao AH, Steffen AT, Pappu RV. (2012) Improved atomistic Monte Carlo simulations demonstrate that poly-L-proline adopts heterogeneous ensembles of conformations of semi-rigid segments interrupted by kinks. J. Phys. Chem. B. 116: 6862-71. DOI: 10.1021/jp212637r.

other information

Awards 2011-2013 Francis D. Carlson Fellowship · 2010 Alpha Eta Mu Beta (Biomedical Engineering Honorary) · 2008 WUSTL Summer Undergraduate Research Fellowship · Presentations & 2015 Dhh1 regulates translation by monitoring codon optimality. In: EMBO · Talks Conference Series: Protein Synthesis and Translational Control. Presentation. 2015 Exploring the mechanism of Dhh1-mediated general translation repression. · In: 59th Annual Meeting of the Biophysial Society. Poster abstract. 2013 Exploring the role of Dhh1-ribosome interactions in general translation · repression. In: EMBO Conference Series: Protein Synthesis and Translational Control. Poster abstract. 2011 Characterizing the Conformation Equilibria of Polyproline. In: 3rd Annual · Meeting of the Protein Folding Consortium. Poster abstract. 2011 Accurate atomistic modeling of conformational equilibria of proline-rich · sequences. In: 55th Annual Meeting of the Biophysial Society. Poster abstract.