The role of transcript-specific translation in human neuronal differentiation

Stephen N. Floor ENCODE Users Meeting

8 June 2016

Control of protein translation achieves precise gene expression

Spatially or temporally localized function

Xue, S. et al. and Barna, M. (2015)

Control of protein translation achieves precise gene expression

Spatially or temporally localized function

Xue, S. et al. and Barna, M. (2015)

MAP2 mRNA

Batish, M. et al. and Tyagi, S. (2012)

Control of protein translation achieves precise gene expression

Spatially or temporally localized function

Xue, S. et al. and Barna, M. (2015)

Suppression of

proliferation

Wolfe A.L. and Singh, K. et al. and Wendel, H-G. (2014)

MAP2 mRNA

Batish, M. et al. and Tyagi, S. (2012)

Two models of translational control

I) Alleviate inhibition through the action of proteins (or RNAs)

II) Excise inhibitory regions by altered transcription or splicing

DEAD-box protein

AUGm7G

ORF

AUGm7G

ORF

Two models of translational control

I) Alleviate inhibition through the action of proteins (or RNAs)

II) Excise inhibitory regions by altered transcription or splicing

DEAD-box protein

AUGm7G

ORF

AUGm7G

ORF

0 10 20 30 40 50 60 70 8005

1000

2000

3000

4000

5000

Number of transcript isoforms per gene

Cou

nt

Ensembl Release 75 protein coding genes

GPR56 (82)

Median=5

Human genes have up to 80 annotated isoforms

0 10 20 30 40 50 60 70 8005

1000

2000

3000

4000

5000

Number of transcript isoforms per gene

Cou

nt

Ensembl Release 75 protein coding genes

GPR56 (82)

Median=5

Human genes have up to 80 annotated isoforms

Can we measure genome-wide isoform specific translation?

Transcript Isoforms in Polysomes sequencing (TrIP-seq)

HEK 293T cells

1 2 3 4 5678+# of ribosomes

Deep sequencing

Transcript Isoform Abundances

Cytoplasmic fraction

Cycloheximide treatment

Extract RNA

Ultracentrifugation

GradientFractionation

A 260

Small Large

Floor, S.N. and Doudna, J.A. (2016)

Diverse transcript-isoform specific polysome occupancy

Hierarchical clustering of Spearman distance; partitioning by dendrogram height 62,703 transcript isoforms represented

ntxs = 5563

ntxs = 5518

ntxs = 11705

ntxs = 9611

ntxs = 10128

Cluster 1

Cluster 2

Cluster 3

Cluster 4

Cluster 5

Cluster 6

Cluster 7

Cluster 8

ntxs = 3614

ntxs = 8861

ntxs = 7703

Cyto.80S 2 3 4 5 6 7 8+ Rel

ativ

e ex

pres

sion

Rel

ativ

e ex

pres

sion

Polysomes Cyto.

1.5

-1.5

Cyto.80S 2

2

-2

0

2

-2

0

2

-2

0

2

-2

0

2

-2

0

2

-2

0

2

-2

0

2

-2

0

3 4 5 6 7 8+# of ribosomes

# of ribosomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

0 5000 100000.0

0.2

0.4

0.6

0.8

1.0

CDS length (nt)

Cum

ulat

ive

frac

tion

Low polysomesHigh polysomes

p = 1.06e-54

Cum

ulat

ive

fract

ion

0

Coding Sequence length (nt)

0.20.40.60.81.0

0 5000 10000

-0.4

-0.2 0.0 0.2 0.4

Transcript halflifeMedian polysome abundance

Cytoplasmic abundanceTranscript exon countTranscript GC content

Transcript length

Effect Size (Cliff's d)

Increased in high polysomes

Increased in low polysomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

0 5000 100000.0

0.2

0.4

0.6

0.8

1.0

CDS length (nt)

Cum

ulat

ive

frac

tion

Low polysomesHigh polysomes

p = 1.06e-54

Cum

ulat

ive

fract

ion

0

Coding Sequence length (nt)

0.20.40.60.81.0

0 5000 10000

-0.4

-0.2 0.0 0.2 0.4

5' UTR structure (- G)Cap structure (- G)5' UTR GC content

5' UTR length

Transcript halflifeMedian polysome abundance

Cytoplasmic abundanceTranscript exon countTranscript GC content

Transcript length

Effect Size (Cliff's d)

Increased in high polysomes

Increased in low polysomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

0 5000 100000.0

0.2

0.4

0.6

0.8

1.0

CDS length (nt)

Cum

ulat

ive

frac

tion

Low polysomesHigh polysomes

p = 1.06e-54

Cum

ulat

ive

fract

ion

0

Coding Sequence length (nt)

0.20.40.60.81.0

0 5000 10000

-0.4

-0.2 0.0 0.2 0.4

Min. codon frequency windowAverage codon frequency

CDS structure (- G)CDS GC content

CDS length

5' UTR structure (- G)Cap structure (- G)5' UTR GC content

5' UTR length

Transcript halflifeMedian polysome abundance

Cytoplasmic abundanceTranscript exon countTranscript GC content

Transcript length

Effect Size (Cliff's d)

Increased in high polysomes

Increased in low polysomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

0 5000 100000.0

0.2

0.4

0.6

0.8

1.0

CDS length (nt)

Cum

ulat

ive

frac

tion

Low polysomesHigh polysomes

p = 1.06e-54

Cum

ulat

ive

fract

ion

0

Coding Sequence length (nt)

0.20.40.60.81.0

0 5000 10000

-0.4

-0.2

0.0

0.2

0.4

Min. cnsv. miRNA score in HEK

Min. conserved miRNA score

Conserved miRNA sites in HEK

Conserved miRNA sites

AU-element fraction

3' UTR structure (- G)

3' UTR GC content

3' UTR length

Min. codon frequency window

Average codon frequency

CDS structure (- G)

CDS GC content

CDS length

5' UTR structure (- G)

Cap structure (- G)

5' UTR GC content

5' UTR length

Transcript halflife

Median polysome abundance

Cytoplasmic abundance

Transcript exon count

Transcript GC content

Transcript length

Effect Size (Cliff's d)

Increased in high polysomes

Increased in low polysomes

CytoPolysomes

Dependence of transcript features on translatability

vs

ntxs = 19739 ntxs = 7703

CytoPolysomes

High polysomes

Low polysomes

0 5000 100000.0

0.2

0.4

0.6

0.8

1.0

CDS length (nt)

Cum

ulat

ive

frac

tion

Low polysomesHigh polysomes

p = 1.06e-54

Cum

ulat

ive

fract

ion

0

Coding Sequence length (nt)

0.20.40.60.81.0

0 5000 10000

Currently analyzing translation in differentiating neurons

Human ES cells Neural progenitor cells Neurons

Currently analyzing translation in differentiating neurons

Human ES cells Neural progenitor cells Neurons

Cytopla

sm

Nucleu

s

Monos

ome

Low po

lysom

es

High po

lysom

es0

5

10

15

20

Expr

essi

on (T

PM)

NANOG-001NANOG-002

hESC_nuc−1hESC_nuc−2hESC_cyto−1hESC_cyto−2NPC_nuc−1NPC_nuc−2Neuron_nuc−1Neuron_nuc−2NPC_cyto−2Neuron_cyto−1Neuron_cyto−2hESC_mono−1hESC_mono−2hESC_high−1hESC_high−2hESC_low−1hESC_low−2Neuron_high−1Neuron_high−2NPC_cyto−1NPC_high−1NPC_high−2NPC_mono−2Neuron_mono−1Neuron_mono−2Neuron_low−1Neuron_low−2NPC_mono−1NPC_low−1NPC_low−2

0 40000 80000Value

010

2030

Color Key

and Histogram

Count

Translated RNA clusters by cell type

hESC_nuc−1hESC_nuc−2hESC_cyto−1hESC_cyto−2NPC_nuc−1NPC_nuc−2Neuron_nuc−1Neuron_nuc−2NPC_cyto−2Neuron_cyto−1Neuron_cyto−2hESC_mono−1hESC_mono−2hESC_high−1hESC_high−2hESC_low−1hESC_low−2Neuron_high−1Neuron_high−2NPC_cyto−1NPC_high−1NPC_high−2NPC_mono−2Neuron_mono−1Neuron_mono−2Neuron_low−1Neuron_low−2NPC_mono−1NPC_low−1NPC_low−2

0 40000 80000Value

010

2030

Color Key

and Histogram

Count

Translated RNA clusters by cell type

Total RNA

hESC_nuc−1hESC_nuc−2hESC_cyto−1hESC_cyto−2NPC_nuc−1NPC_nuc−2Neuron_nuc−1Neuron_nuc−2NPC_cyto−2Neuron_cyto−1Neuron_cyto−2hESC_mono−1hESC_mono−2hESC_high−1hESC_high−2hESC_low−1hESC_low−2Neuron_high−1Neuron_high−2NPC_cyto−1NPC_high−1NPC_high−2NPC_mono−2Neuron_mono−1Neuron_mono−2Neuron_low−1Neuron_low−2NPC_mono−1NPC_low−1NPC_low−2

0 40000 80000Value

010

2030

Color Key

and Histogram

Count

Translated RNA clusters by cell type

Total RNA

hESC polysomal RNA

hESC_nuc−1hESC_nuc−2hESC_cyto−1hESC_cyto−2NPC_nuc−1NPC_nuc−2Neuron_nuc−1Neuron_nuc−2NPC_cyto−2Neuron_cyto−1Neuron_cyto−2hESC_mono−1hESC_mono−2hESC_high−1hESC_high−2hESC_low−1hESC_low−2Neuron_high−1Neuron_high−2NPC_cyto−1NPC_high−1NPC_high−2NPC_mono−2Neuron_mono−1Neuron_mono−2Neuron_low−1Neuron_low−2NPC_mono−1NPC_low−1NPC_low−2

0 40000 80000Value

010

2030

Color Key

and Histogram

Count

Translated RNA clusters by cell type

Total RNA

hESC polysomal RNA

Neuron/NPC polysomal RNA

3ʹ UTRs have cell type dependent effects on translation

hESCs NPCs Neurons

Hierarchical clustering of regularized log transformed data by Euclidean distance cluster partitioning by dendrogram height

47096 transcripts 49002 transcripts 47766 transcripts

3ʹ UTRs have cell type dependent effects on translation

hESCs NPCs Neurons

-0.2 0.0 0.2 0.4

Effect Size (Cliff's d)

Longer in high polysomes

Longer in low polysomes

3′ UTR

ORF

-0.2 0.0 0.2 0.4

Effect Size (Cliff's d)Cytopla

sm

Nucleu

s

Monos

ome

Low po

lysom

es

High po

lysom

es0

5

10

15

20

Expr

essi

on (T

PM)

NANOG-001NANOG-002

Figure 7: Preliminary TrIP-seq data from hPSCs, NPCs and Neurons.

(A) Transcript abundances from the NANOG gene in subcellular and

polysomal fractions. The well-translated 002 transcript is enriched

in polysomes while 001 is not. (B) Global analysis of the impact on

translation of open reading frame (ORF) or 3′ UTR length between transcript isoforms from the same gene. Effect size is measured by

Cliff’s delta, proportional to the Mann-Whitney U statistic. ORF length influences translation in all three cell types while 3′ UTRs inhibit translation selectively in neurons.

Well translated

ORF

length

hPSCs

NPCs

Neurons

3′ UTRlength

Poorly

translatedBA

Features contrasted between transcripts of the same gene

Hierarchical clustering of regularized log transformed data by Euclidean distance cluster partitioning by dendrogram height

47096 transcripts 49002 transcripts 47766 transcripts

Insights from ENCODE RNA binding protein data

-0.1 0.0 0.1

Effect Size (Cliff's d)

FMR1

ORF

5′ UTR

3′ UTR

More binding in high polysomes

More binding in low polysomes

Insights from ENCODE RNA binding protein data

-0.1 0.0 0.1

Effect Size (Cliff's d)

FMR1

ORF

5′ UTR

3′ UTR

More binding in high polysomes

More binding in low polysomes

-0.1 0.0 0.1

Effect Size (Cliff's d)

SF3B4

ORF

3′ UTR

More binding in high polysomes

More binding in low polysomes

Insights from ENCODE RNA binding protein data

-0.1 0.0 0.1

Effect Size (Cliff's d)

FMR1

ORF

5′ UTR

3′ UTR

More binding in high polysomes

More binding in low polysomes

-0.1 0.0 0.1

Effect Size (Cliff's d)

SF3B4

ORF

3′ UTR

More binding in high polysomes

More binding in low polysomes

eCLI

P D

ensi

ty

intronexon exon

Acknowledgements

Helen Hay Whitney

Foundation

Thanks to:

Kendall Condon

Yun Bai Akshay Tambe

Amy Lee

Doudna Lab

Computing:

Doudna lab Pachter lab

LBL UC Berkeley BRC

Collaborators:

DDX3: Yoon-Jae Cho

Howard Chang Kevan Shokat

Eckhard Jankowsky

hESC/NPC/Neurons:

Dirk Hockemeyer Helen Bateup