the human biofluid rna atlas

EVA HULSTAERT, MDAACR – Advances in Liquid Biopsies

January 13-16, 2020Miami, Florida

The Human Biofluid RNA Atlas

no conflicts of interest to declare

expanding the liquid biopsy field beyond the blood stream

plasmaserum

sweat

gastric fluidbile

peritoneal fluid

colostrummature breast milk

amniotic fluid

synovial fluid

pancreatic cyst fluid

cerebrospinal fluid

tearsaqueous humor

saliva

plasmaserum

bronchial lavagesputum

urinestoolseminal plasma

expanding the liquid biopsy field beyond the blood stream

RNA sequencing allows detailed analysis of the transcriptome

mRNAs

circRNAs

AAAA

mRNA capture seq

miRNAs

tRNAs

piRNAs

small RNA seq

200 µL biofluid RNA eluate

12 RC

76 sequin

9 LP

92 ERCC

spike-in RNAs as processing controls and normalization tool

.

highly variable RNA concentrations amongst different biofluids

small RNA

mRNA

Hulstaert et al, in reviewPreprint on BioRxiv

CHARTING EXTRACELLULAR TRANSCRIPTOMES IN THE HUMAN BIOFLUID RNA ATLAS

23496B23496B

Hulstaert E1-3, Morlion A1,2, Avila Cobos F1,2, Verniers K1,2, Nuytens J1,2, Vanden Eynde E1,2, Yigit N1,2, Anckaert J1,2, Geerts A4, Hindryckx P4, Jacques P5,6, Brusselle G7, Bracke KR7, Maes T7, Malfait T7, Derveaux T8, Ninclaus V8, Van Cauwenbergh C8, Roelens K9, Roets E9, Hemelsoet D10, Tilleman K11, Brochez L2,3, Kuersten S12, Simon L13, Karg S14, Kautzky-Willers A15, Leutner M15, Nöhammer C16, Slaby O17-19, Schroth GP12, Vandesompele J1,2*, Mestdagh P1,2*

23496B

1Center for Medical Genetics, Department of Biomolecular Medicine, Ghent University, Belgium, 2Cancer Research Institute Ghent (CRIG), Ghent University, Belgium, 3Department of Dermatology, Ghent University Hospital, Belgium, 4Department of Gastroenterology, Ghent University Hospital, Belgium, 5Department of Rheumatology, Ghent University Hospital, Belgium, 6VIB Inflammation Research Center, Ghent University, Belgium, 7Department of Respiratory Medicine, Ghent University Hospital, Belgium, 8Department of Ophthalmology, Ghent University Hospital, Belgium, 9Department of Obstetrics, Women’s Clinic, Ghent University Hospital, Belgium, 10Department of Neurology, Ghent University Hospital, Belgium, 11Department of Reproductive Medicine, Ghent University Hospital, Belgium, 12Illumina, San Diego, CA, USA, 13University of Texas Health Science Center, School of Biomedical Informatics, Center for Precision Health, Houston, TX, USA, 14Helmholtz Zentrum München, German Research Center for Environmental Health, Institute of Computational Biology, Germany, 15Department of Internal Medicine III, Clinical Division of Endocrinology and Metabolism, Unit of Gender Medicine, Medical University of Vienna, Austria, 16Austrian Institute of Technology, Center for Health and Bioresources, Molecular Diagnostics, Vienna, Austria, 17Masaryk Memorial Cancer Institute, Department of Comprehensive Cancer Care, Brno, Czech Republic, 18Department of Pathology, University Hospital Brno, Czech Republic, 19Central European Institute of Technology, Masaryk University, Brno, Czech Republic *equally contributing authors

introduction

23496B23496B

Extracellular RNAs present in biofluids have emerged as potential biomarkers for disease. Where most studies focus on plasma or serum, other biofluids may contain more informative RNA molecules, depending on the type of disease. Here, we present an unprecedented atlas of messenger, circular and small RNA transcriptomes of a comprehensive collection of 20 human biofluids. By means of synthetic spike-in controls, we compared RNA content across biofluids, revealing a more than 10 000-fold difference in RNA concentration. The circular RNA fraction is increased in nearly all biofluids compared to tissues. Each biofluid transcriptome is enriched for RNA molecules derived from specific tissues and cell types. In addition, a subset of biofluids, including stool, sweat, saliva and sputum, contains high levels of bacterial RNAs. Our atlas enables a more informed selection of the most relevant biofluid to monitor particular diseases. To verify the biomarker potential in these biofluids, four validation cohorts representing a broad spectrum of diseases were profiled, revealing numerous differential RNAs between case and control subjects. Taken together, our results reveal novel insights in the RNA content of human biofluids and may serve as a valuable resource for future biomarker studies.

2 µL Sequin spikes2 µL RC spikes

200 µL input volume*

2 µL ERCC spikes2 µL LP spikes

12 µL eluate

SAMPLE COLLECTION

amniotic fluid

aqueous humor

bile

BAL

breast milk

CSF

colostrum

gastric fluid


ascites

blood plasma

saliva

serum

sputum

synovial fluid

sweat

tears

urine

seminal fluid

stool

discovery cohort

case/control cohort

RNA ISOLATION gDNA REMOVAL

PRPPPPPFP

sterile recipient

CALEX Cap Device

LIBRARY PREPARATIONRNA SEQUENCING

AAA

TruSeq RNA Exome

TruSeq Small RNA

DATA NORMALIZATIONDATA ANALYSIS

sputumCOPD

control

CSFglioblastoma

hydrocephalus

urinebladder cancer

control

salivadiabetes

control

* ** only in case/control cohort

pool of 4-5 donors

endogenous RNA

Sequin spikes

small RNA biotypes

exogenous RNA(Genboree)

endogenous RNA

RC spikes

circRNA assessment

exogenous RNA(Metamap)

tissues of originassessment

differential expressionGSEA**

2 µL Sequin spikes2 µL RC spikes

200 µL input volume*

2 µL ERCC spikes2 µL LP spikes

12 µL eluate

study flowchart

1 donor

except for tears (Schirmer strip method)

results

0

1

2

3

4

5

tear

sse

min

al p

lasm

abi

leco

lost

rum

syno

vial f

luid

brea

stm

ilkpa

ncre

atic

cyst

flui

dPR

Psp

utum stoo

lam

niot

ic flu

idsa

liva

PPP

gast

ric fl

uid

seru

msw

eat

BAL

stoo

l Cal

exas

cites

aque

ous

hum

orPF

Pur

ine

CSF

rela

tive

RNA

cont

ent (

log1

0)

assessment of the tissues of origin in the different biofluids


pro

state

eso

pha

gus

pa

ncre

as

Tissue or cell type contribution per biofluid

Granulocytes

Monocytes

B−cells

T−cells

NK cells

Adipose

Breast

Colon

Esophagus

Lung

Lymph node

Pancreas

Prostate

Small intestine

Spleen

Stomach

Testicle

Thyroid

Trachea

Amniotic fluid

Aqueous humor

Ascites

BAL

Bile

Breastmilk

CSF

Colostrum

Gastric fluid

PFP

PPP

PRP

Pancreatic cyst fluid

Saliva

Seminal plasma

Serum

Sputum

Stool

Stool Calex

Sweat

Synovial fluid

Tears

Urine

−4

−2

0

2

4

Scaled log2 fold change

seminal plasma

urine

salivapancreatic cyst fluid

amniotic fluidaqueous humorascitesBALbilebreast milkCSFcolostrumgastric fluidPFPPPPPRP

serumsputumstoolstool Calexsweatsynovial fluidtears

gra

nuloc

ytes

mo

noc

ytes

B-ce

lls

T-ce

lls

NK c

ells

ad

ipo

se

bre

ast

co

lon

lung

lymp

h nod

e

sma

ll intestine

sple

en

stom

ac

h

testic

le

thyroid

trac

hea

biofluid

tissue



pro

state

eso

pha

gus

pa

ncre

as


Granulocytes

Monocytes

B−cells

T−cells

NK cells

Adipose

Breast

Colon

Esophagus

Lung

Lymph node

Pancreas

Prostate

Small intestine

Spleen

Stomach

Testicle

Thyroid

Trachea

Amniotic fluid

Aqueous humor

Ascites

BAL

Bile

Breastmilk

CSF

Colostrum

Gastric fluid

PFP

PPP

PRP


Saliva

Seminal plasma

Serum

Sputum

Stool

Stool Calex

Sweat

Synovial fluid

Tears

Urine

−4

−2

0

2

4


seminal plasma

urine

salivapancreatic cyst fluid



gra

nuloc

ytes

mo

noc

ytes

B-ce

lls

T-ce

lls

NK c

ells

ad

ipo

se

bre

ast

co

lon

lung

lymp

h nod

e

sma

ll intestine

sple

en

stom

ac

h

testic

le

thyroid

trac

hea




Granulocytes

Monocytes

B−cells

T−cells

NK cells

Adipose

Breast

Colon

Esophagus

Lung

Lymph node

Pancreas

Prostate

Small intestine

Spleen

Stomach

Testicle

Thyroid

Trachea

Amniotic fluid

Aqueous humor

Ascites

BAL

Bile

Breastmilk

CSF

Colostrum

Gastric fluid

PFP

PPP

PRP


Saliva

Seminal plasma

Serum

Sputum

Stool

Stool Calex

Sweat

Synovial fluid

Tears

Urine

−4

−2

0

2

4


urine

saliva



seminal plasma

pro

state

eso

pha

gus

pa

ncre

as

gra

nuloc

ytes

mo

noc

ytes

B-ce

lls

T-ce

lls

NK c

ells

ad

ipo

se

bre

ast

co

lon

lung

lymp

h nod

e

sma

ll intestine

sple

en

stom

ac

h

testic

le

thyroid

trac

hea


different cell types contribute to the RNA content of pancreatic cyst fluid

RNA seq ofpancreatic cyst fluid

scRNA seq of10 pancreatic cell types


Baron et al, Cell Syst, 2016

Baron et al, Cell Syst, 2016

different cell types contribute to the RNA content of pancreatic cyst fluid

RNA seq ofpancreatic cyst fluid

scRNA seq of10 pancreatic cell types


0

25

50

75

100

pancreatic cyst fluid 1(necrotizing pancreatitis)

pancreatic cyst fluid 2(side-branch intra papillary mucinous neoplasia)Sample

Perc

enta

ge o

f cel

l typ

e pr

esen

t in

the

sam

ple

cell typeacinar

activated stellate

alpha

beta

delta

ductal

endothelial

gamma

macrophage

quiescent stellate

nnls with marker selection, TMM scalingComposition of pancreatic cyst fluid samples based on deconvolution with single cell seq data of pancreas

pe

rce

nta

ge

0

25

50

75

100

sample 1 sample 2

cell type

acinaractivated stellatealpha

beta

deltaductalendothelialgammamacrophagequiescent stellate

differentially expressed mRNAs in urine of bladder cancer patients vs healthy controls

867 upregulated mRNAs

47 downregulated mRNAs

0

25

50

75

100

COPD no COPDSputum type

Sequ

in sp

ike n

orm

alize

d re

ads

Relative RNA content in sputumA

0

25

50

75

100

125

Bladder cancer ControlUrine type

Sequ

in sp

ike n

orm

alize

d re

ads

Relative RNA content in urineC

2.5

5.0

7.5

10.0

Glioblastoma HydrocephalusCSF type

Sequ

in sp

ike n

orm

alize

d re

ads

Relative RNA content in CSFE

CCL20

ADA

MMP1

HS3ST2CCR7

SPRR2EAL031985.2

FLGRPSAP29

LINC02280

0.0

2.5

5.0

7.5

−4 −2 −1 0 1 2 4 6 8log2 fold change

−log

10 Q

−valu

e

Volcano plot (COPD vs no COPD)B

MNDA

MMP1

PLAC4RGS18

KRT5

MDH1B

ATP5MDP1

PLBD1−AS1NEFHP1

TGM4

0

2

4

6

−6 −4 −2 −1 0 1 2 4 6log2 fold change

−log

10 Q

−valu

e

Volcano plot (bladder cancer vs control)D

CD163

FKBP5

BCAS1

MBP

KIF1A

MOBP

PLP1

0

1

2

3

4

5

−6 −4 −2 −1 0 1 2 4log2 fold change

−log

10 Q

−valu

e

Volcano plot (glioblastoma vs hydrocephalus)F

Wilcoxon, p = 0.007 Wilcoxon, p = 0.068 Wilcoxon, p = 0.71


volcanoplot (bladder cancer versus control)

circRNAs are enriched in biofluids compared to tissues

mRNA circRNA

circRNAs are enriched in biofluids compared to tissues

0

25

50

75

100

biof

luid

tissu

e

Circ

RN

A fra

ctio

n (1

00*c

irc/(c

irc+l

in))

mRNA circRNA

me

dia

n c

ircRN

Afra

ctio

n

100

75

50

25

0

biofluidsn = 42

tissuesn = 35

84%

18%

Take-home messages and future perspectives

• Dare to collect and investigate “alternative” fluids

• RNA is cool!

• Our findings enable a more informed selection of the most relevant biofluid to monitor individual cancer types

• Broader sample collections in different cancer types are ongoing

• We are open to new collaborations!

[email protected]

@ehulstaert

Ghent University Hospital, BelgiumPieter MestdaghJo VandesompeleFrancisco Avila CobosAnnelien MorlionJasper AnckaertKimberly VerniersNurten YigitEvelyn Vanden EyndeJustine NuytensAnja GeertsPeggy JacquesGuy BrusselleThierry DerveauxCaroline Van CauwenbergheThomas MalfaitDimitri HemelsoetPieter HindryckxKristien RoelensEllen RoetsKen Bracke

University of Texas Health Science Center, USALukas Simon

Helmholtz Zentrum Mu ̈nchen, GermanySebastian Karg

Austrian Institute of Technology, AustriaChrista NohammerAlexandra Kautzky-WillersMichael Leutner

Masaryk Memorial Cancer Institute, Czech RepublicOndrej Slaby

Illumina BiogazelleScott Kuersten Nele NijsGary Schroth Carolina Fierro

[email protected]

the human biofluid rna atlas

Documents