Yale University
EliScholar – A Digital Platform for Scholarly Publishing at Yale
Yale Medicine Thesis Digital Library
School of Medicine
January 2019
A Genomic Approach To Idiopathic Liver Disease
In Adults
Aaron Hakim
Follow this and additional works at: https://elischolar.library.yale.edu/ymtdl
This Open Access Thesis is brought to you for free and open access by the School of Medicine at EliScholar – A Digital Platform for Scholarly
Publishing at Yale. It has been accepted for inclusion in Yale Medicine Thesis Digital Library by an authorized administrator of EliScholar – A Digital
Platform for Scholarly Publishing at Yale. For more information, please contact elischolar@yale.edu.
Recommended Citation
Hakim, Aaron, “A Genomic Approach To Idiopathic Liver Disease In Adults” (2019). Yale Medicine Thesis Digital Library. 3501.
https://elischolar.library.yale.edu/ymtdl/3501
A GENOMIC APPROACH TO IDIOPATHIC LIVER DISEASE IN ADULTS
A Thesis Submitted to the
Yale University School of Medicine
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Medicine
by
Aaron Hakim
2019
ABSTRACT
Adult patients suffering from liver disease of unknown cause represent an
understudied and underserved population. Over the past 15 years, next-
generation sequencing technologies have matured into an inexpensive, effective,
and widely available set of tools to do genomic analysis. One of these
technologies, whole-exome sequencing (WES), allows for high throughput
sequencing of all of the genome’s protein coding regions (exons). In pediatric
cohorts, WES combined with deep clinical phenotyping has been shown to be an
effective and unbiased method of identifying rare protein-altering coding variants
in individual genes. WES has also contributed to the diagnosis and
individualization of medical care in oncologic patients. The use of WES for the
study of a broader spectrum of non-oncological diseases, among adults, remains
poorly understood. We assessed the utility of WES in the diagnosis and
management of adults with unexplained liver disease despite a comprehensive
conventional workup and with no history of alcohol overuse.
We performed exome sequencing and deep phenotyping in two independent
adult cohorts with unexplained liver disease. In the first cohort, we analyzed
nineteen unrelated adult patients with idiopathic liver disease recruited at Yale
New Haven Hospital. In a second cohort from Bridgeport Hospital, four unrelated
adult patients presenting with fatty liver disease, hypertriglyceridemia, insulin
resistance, and physical exam findings suggestive of lipodystrophy were
recruited for genomic analysis.
In cohort 1, analysis of the exome in nineteen cases identified four monogenic
disorders in five unrelated adults. Patient 1 suffered for 18 years from
devastating complications of undiagnosed Type 3 Familial Partial Lipodystrophy
due to a deleterious heterozygous variant in PPARG. Molecular diagnosis
enabled initiation of leptin replacement therapy with subsequent normalization of
liver transaminases, and amelioration of dyslipidemia. Patients 2 and 3 were
diagnosed with MDR3 deficiency (also known as PFIC3, progressive intrahepatic
familial cholestasis type 3) due to recessive mutations in ABCB4. Patient 4 with a
prior diagnosis of non-alcoholic steatohepatitis was found to harbor a
mitochondrial disorder due to a homozygous pathogenic variant in NDUFB3;
subsequent muscle biopsy revealed a deficiency of rotenone sensitive I+III
activity consistent with a mitochondrial disorder. This finding enabled initiation of
disease-preventative measures including supplementation with antioxidants.
Patient 5 is a lean patient with hepatic steatosis of unknown etiology who was
found to have a damaging heterozygous variant in APOB, consistent with familial
hypobetalipoproteinemia. In cohort 2, we identified a potential genetic diagnosis
in all four cases of suspected lipodystrophy, including a patient with an LMNA
mutation, a patient with two pathogenic heterozygous mutations in APOE, a
patient with a homozygous deleterious mutation in the leptin receptor (LEPR),
and a patient with a pathogenic heterozygous variant in PPARG.
In conclusion, WES provided a diagnosis with impact on clinical management in
a significant number of adults suffering from liver disease of unknown cause,
gaining insight into disease pathogenesis and identifying new therapeutic and
preventive medicine interventions. This study supports the use of WES in the
evaluation and management of adults with idiopathic liver disease in clinical
practice.
Published in part:
Hakim A, Zhang X, DeLisle A, Oral EA, Dykas D, Drzewiecki K, Assis DN,
Silveira M, Batisti J, Jain D, Bale A, Mistry PK, Vilarinho S. Clinical Utility of
Genomic Analysis in Adults with Idiopathic Liver Disease. Journal of Hepatology
2019 (in press, February 2019)
Presented in part:
Vilarinho S, Hakim A, Oral E, Zhang X, Mistry PK. A Genomic Approach to
Idiopathic Liver Disease in Adults: New Insights into Disease Pathogenesis and
New Interventions at Bedside. Oral Abstracts (Abstract 170). Hepatology
(Baltimore, Md) 2018;68:1-183.
ACKNOWLEDGEMENTS
The work presented in this thesis is a direct result of the incredible support and
phenomenal mentorship of my supervisor, Dr. Vilarinho. She has imparted a true
passion for bench to bedside translational research. I would also like to thank all
the patients and their families whose contribution to this study led to advancing
our understanding of liver disease, Dr. Michael Nathanson and Dr. Sachin K.
Majumdar for their efforts to refer patients to this study, and the staff of the Yale
Center for Genome Analysis. I am also indebted to my family for their unending
love and support.
Research reported in this publication was supported by the National Institute of
Diabetes and Digestive and Kidney Diseases of the National Institutes of Health
under Award Number K08DK113109. The content is solely the responsibility of
the authors and does not necessarily represent the official views of the National
Institutes of Health. This work was also supported in part by Yale Liver Center
P30DK034989, and AASLD Sheila Sherlock Clinical and Translational Research
Award in Liver Disease (to S.V.).
i
TABLE OF CONTENTS
TABLE OF CONTENTS………………………………….…………………………..i
LIST OF FIGURES…………………………………………………………………..1
LIST OF TABLES…………………………………………………………………….2
INTRODUCTION……………………………………………………………………..3
STATEMENT OF PURPOSE……………………………………………………….9
PATIENTS AND METHODS….……………………………………………………10
RESULTS……………………………………………………….…………….……..17
DISCUSSION…………………………………………………….………….………48
REFERENCES…………………………………………………….……….………..54
1
LIST OF FIGURES
Figure 1: The incidence of cryptogenic cirrhosis has been steadily declining…..4
Figure 2: Overview of whole-exome sequencing pipeline………………..………..7
Figure 3. Representative flowchart of genetic variant filtering strategy in this
study…………………………………………………………………………………….13
Figure 4: Example of principal component analysis to determine ethnicity
clustering…………………………………………………………………………..…..21
Figure 5: Liver histology findings in patient 1, cohort 1……………………………23
Figure 6: Representative plot read of disease-causing mutation identified in
Patient 1, cohort 1………………………………………………………………………………….25
Figure 7: Genetic findings in patient 1, cohort 1…………………………………..27
Figure 8: Illustrative representation of the role of PPARG, peroxisome proliferator-
activated receptor-gamma, in adipocyte differentiation and relationship to serum
leptin……………………………………………………………………………………28
Figure 9: Laboratory findings in patient 1, cohort 1…………………………….….29
Figure 10: Liver histology findings in patient 2, cohort 1………………………….31
Figure 11: Genetic findings in patient 2, cohort 1………………………………….32
Figure 12: Conservation findings in patient 2, cohort 1……………………………33
Figure 13: Liver histology findings in patient 3, cohort 1………………………….34
Figure 14. Liver biopsies of patient 4, cohort 1…………………………………….36
Figure 15. Liver biopsy of patient 5, cohort 1………………………………………38
Figure 16: Schematic representation of multidisciplinary Genome Rounds in
Adult Hepatology……………………………………………………………………………………52
2
LIST OF TABLES
Table 1. Gene name, accession number, and forward and reverse primer
sequences used for Sanger sequencing……………………………………………15
Table 2: Summary of study population characteristics and demographics in
cohort 1 (n = 19)……………………………………………………………………….18
Table 3: Sequencing coverage and quality metrics for patient cohort 1 (n =
19)………………………………………………………………………………….……19
Table 4: Demographics, clinical features, and genetic diagnosis identified in five
subjects in cohort 1, and its clinical implications…………………………………..20
Table 5: Diagnostic genetic variants identified in five subjects in cohort 1……..26
Table 6. Summary of outcomes for fourteen patients who remain unexplained
after whole exome sequencing analysis………………………………………..…..40
Table 7: Sequencing coverage and quality metrics for patient cohort 2 (n =
4)…………………………………………………………………………………………43
Table 8: Demographics, clinical features, and genetic diagnosis identified in four
subjects in cohort 2, and its clinical implications…………………………………..43
Table 9: Diagnostic genetic variants identified in cohort 2………………………..47
3
INTRODUCTION
Liver Disease of Unknown Etiology: A Historical Perspective
Liver disease is a major public health problem that affects approximately 30
million people and leads to over 40,000 deaths annually in the United States.1
Chronic liver disease (CLD) is often silent unless there is awareness of subtle
clinical signs, behavioral risk factors and/or investigation of abnormal liver
function tests. Untreated liver disease may progress to end-stage liver disease
(cirrhosis) and further decompensation with ascites, hepatic encephalopathy,
esophageal variceal hemorrhage, jaundice, and/or hepatocellular carcinoma,
leading to liver failure and death.2 Advances in our understanding of liver disease
have led to a marked decline in the attribution of CLD to “unknown etiology”
(Figure 1). Prior to 1965, cryptogenic cirrhosis, defined as cirrhosis of unknown
etiology after extensive clinical, laboratory, and histological analysis, accounted
for >50% of all cases of cirrhosis.3 The discovery of hepatitis B virus (1965)4,
hepatitis D virus (1977)5, and hepatitis C virus (1989)6 eventually led to the
recognition of their contributions to cirrhosis worldwide. The description of non-
alcoholic steatohepatitis as a clinical entity in 19807, and improved diagnostic
criteria for autoimmune hepatitis, first published in 19988, further reduced the
diagnosis of cryptogenic cirrhosis, as did improved diagnosis of iron overload
syndromes (i.e. HFE mutation)9, alpha-1-antitrypsin deficiency (A1ATD)10, and
Wilson’s disease11 (Figure 1). However, it is currently estimated that up to 30% of
cases of cirrhosis and up to 14% of adults awaiting liver transplantation suffer
from liver disease of unknown etiology.12,13 In tertiary medical centers, the
4
incidence of cirrhosis of unknown etiology has been estimated at 5-10%.14 These
patients often undergo a long and costly odyssey of diagnostic tests,
interventions and medical opinions, and represent an understudied and
underserved population. Understanding the etiology of CLD is essential to halt
the progression of liver dysfunction, as illustrated by the development of a
vaccine and anti-viral therapy for hepatitis B, and the highly effective, safe and
curative anti-viral therapies for hepatitis C.15
Figure 1: The incidence of cryptogenic cirrhosis has been steadily declining as new etiologies are
being recognized, as outlined above. A1ATD, alpha-1-antitrypsin deficiency; HBV, hepatitis B
virus; PFIC, progressive familial intrahepatic cholestasis (Byler disease); NASH, non-alcoholic
steatohepatitis; HCV, hepatitis C virus; HFE, human hemochromatosis protein; AIH, autoimmune
hepatitis; WES, whole-exome sequencing.
5
Current State of Genetic Analysis in Hepatology Clinical Practice
The taxonomy of CLD in clinical practice is based broadly on categories of
etiology such as exposure to toxins, viral infections, cholestatic, autoimmune,
metabolic and select genetic disorders. A significant limitation of this approach is
that it precludes consideration of a wider array of underlying genetic disorders
masquerading within these broad phenotypes. Indeed, the current state of
genetic analysis of adult liver disease in practice only involves the exclusion of a
limited number of inherited conditions through single gene tests, including
Wilson’s Disease (ATP7B), Hemochromatosis (HFE, HJV, HAMP, TFR2,
SCL40A1), and A1ATD (SERPINA1). In some circumstances, commercial gene
panel tests including the Jaundice Chip or EGL Cholestasis Panel are used,
which include up to 72 genes.16 However, these panel tests represent only a
small fraction of the ~20,000 protein coding genes of the human genome
reference sequence, completed in 2003.17
Whole-Exome Sequencing
Advances in human genetics and genomics have created an unprecedented
opportunity for gene discovery and diagnosis in the clinic. Over the past 15
years, next-generation sequencing technologies have matured into an
inexpensive, effective and widely available set of tools. One of these
technologies, whole-exome sequencing (WES), consists of targeted capture and
sequencing of the ~20,000 human protein-coding genes (Figure 2). Although
WES excludes the 99% of the genome that does not code for proteins, it is
estimated that approximately 85% of all mutations with large effects on disease-
6
related traits are located within exomes.18 Furthermore, WES has traditionally
been described as having excellent sensitivity and specificity (~98-99%),
especially when the mean per-base coverage is over 20x and with a minimal
local read depth of 13x.19 Importantly, various bioinformatics programs have
been optimized to translate raw WES data into manageable and intelligible
datasets: performing base calling (translating the raw signals from the
sequencers into A, C, T or G, with an accompanying quality score), alignment of
the reads to the human reference genome (searching for the best matching
segment), removal of PCR duplicates (generated during preparation of the
genomic library), variant calling (recording all positions that differ from the
reference genome), and variant annotation (using data from various public
databases such as ClinVar, Online Mendelian Inheritance in Man [OMIM], Exome
Aggregation Consortium database [ExAC], gnomAD, 1000 Genomes, National
Heart, Lung and Blood Institute’s [NHLBI] Exome Variant Server, HapMap, and
others to provide information about minor allele frequency, function of the gene,
degree of inter-species amino acid conservation, etc). Furthermore, there are
existing computational tools for in silico assessment of a given variant’s
pathogenic potential. As such, WES currently represents a remarkable balance
between cost (<$300 for a research exome without analysis, ~$3000 for a clinical
exome), time of analysis, and information collected, making it attractive and
suitable for clinical use and translational research studies. Nearly 3,000 genes
underlying over 4,000 Mendelian phenotypes have been discovered, and next-
7
generation sequencing approaches including WES account for more than three
times as many discoveries as conventional methods.20
Figure 2: Overview of whole-exome sequencing pipeline. SNV, single nucleotide variant; Indel,
insertion/deletion. Oligonucleotide probes designed to specifically hybridize to all exons in the
genome are in solution. The probes are linked to magnetic beads. Other exome capture systems
have probes attached to a microarray. Adapted from Gerald Goh and Murim Choi.21
Diagnostic Utility of Whole-Exome Sequencing
WES combined with deep clinical phenotyping has been increasingly applied as
a first-line diagnostic tool in clinical medicine, particularly for the diagnosis of
8
inborn metabolic and neurodevelopmental disorders, unexplained liver failure in
children22-26, as well as for the detection of causal mutations in cancer27,28. In
these contexts, exome sequencing can inform medical management, including
prognosis, choice of therapy, and accurate reproductive counselling. However, to
date, most studies that investigate the use of next generation sequencing
technologies in the diagnosis and individualization of medical care have been
performed in either pediatric or cancer patients. There is a paucity of information
on the clinical utility of these approaches for a broader spectrum of diseases
among adults. A number of small studies and one study in a large cohort support
the usefulness of exome sequencing for the diagnosis of early onset or familial
nephropathy29-31, sporadic chronic kidney disease32, and inherited cardiovascular
diseases33, however to date the utility of this approach in chronic liver disease
has not been elucidated. By using unbiased genomic analysis, we may begin to
understand parameters of adult clinical presentations that harbor an underlying
monogenic cause, and to develop a more comprehensive category of ‘genetic’
liver diseases in adults beyond the traditionally considered disorders such as
Wilson’s disease, A1ATD, or hemochromatosis. Here, we provide data to support
the utility of WES in the diagnosis and management of adults with liver disease of
unknown cause, with or without involvement of other diseases and/or unusual
clinical findings. We also extend our analysis to an independent cohort of
patients with fatty liver and physical exam findings suggestive of lipodystrophy, a
group of heterogeneous disorders characterized by the absence or reduction of
subcutaneous adipose tissue.
9
STATEMENT OF PURPOSE
To assess the utility of whole-exome sequencing in the diagnosis and
management of adults with unexplained liver disease.
MAIN OUTCOMES AND MEASURES
To obtain the diagnostic yield of WES and its direct impact in providing new
therapeutic options, targeted preventive medicine interventions, and adequate
family counselling.
10
PATIENTS AND METHODS
Human Subjects
Study protocol was approved by the Yale Human Investigational Committee, and
informed consent was obtained in accordance with institutional review board
standards. In cohort 1, nineteen adults with unexplained liver disease despite a
comprehensive evaluation at Yale New Haven Hospital (unrevealing hepatitis
viral serologies including negative HBsAg and anti-HBc, ferritin, iron studies,
ceruloplasmin, ANA, alpha-1-antitrypsin phenotype, abdominal imaging, liver
biopsy, etc) underwent further investigation using whole-exome sequencing.
Patients may have had other medical co-morbidities but did not have a history of
alcohol overuse. For some patients in the cohort, we questioned prior diagnosis
such as non-alcoholic fatty liver disease (NAFLD) in absence of typical metabolic
or body habitus features. Where possible, samples from available family
members were also obtained for segregation studies. In cohort 2, we recruited
four unrelated adult patients with fatty liver disease, insulin resistance/diabetes,
hypertriglyceridemia, and physical exam findings suggestive of lipodystrophy per
evaluation by an endocrinologist at Bridgeport Hospital.
DNA isolation, exome capture and sequencing
Genomic DNA was isolated from peripheral blood mononuclear cells or buccal
swabs using standard procedures. DNA fragments contained in exonic sequences
were captured and sequenced on the Illumina HiSeq platform.
11
Exome Sequencing Analysis
Exome sequencing data were mapped and aligned to the reference human
genome (reference sequence hg19) using BWA. Variants were called using
GATK34,35 and annotated using Annovar.36 Since the goal of this project was to
identify genetic causes for rare Mendelian conditions, we focused on variants
that are uncommon in the general population. In other words, the higher the
frequency of the variant, the lower the probability to be causal of a rare disease.
Variants were selected for minor allele frequency (MAF) <0.01 for homozygous
and compound heterozygous variants (recessive inheritance pattern) or <2x10-5
for heterozygous variants (dominant inheritance pattern). Variants with MAF >1%
are unlikely to cause recessive disorders with full penetrance (prevalence
1:10,000 or less) in the general population. If autosomal dominance is the
suspected pattern of inheritance, the favored MAF cutoff is more stringent
because a single allele is sufficient to cause disease. Allele frequencies were
determined using the genome aggregation database (gnomAD) databases,37
including the Exome Aggregation Consortium database (ExAC), 1000 Genomes,
and the National Heart, Lung and Blood Institute’s (NHLBI) Exome Variant
Server. After filtering out common variants, variants located in intronic and
intergenic segments of the genome were removed. Subsequently, protein-
altering variants were selected and prioritized based on their predicted
deleteriousness. Deleterious prediction methods might filter, for example, coding
variants that do not result in an amino acid change, substitutions that do not alter
the physicochemical properties of protein product despite the mutated amino
12
acid, or amino acid variants that are not well conserved across orthologues.
MetaSVM38 was used to infer the impact of missense mutations. Rare protein-
altering variants predicted to be deleterious were then selected if they occurred
as pathogenic variants described in NCBI Clin Var, and/or in genes previously
associated with liver-related diseases listed in the Online Mendelian Inheritance
in Man (OMIM) database. BLAT, a local alignment software embedded within the
UCSC Human Genome Browser39, was used to verify that a pathogenic variant
and its surrounding sequences mapped specifically to the target gene. Figure 3
outlines the genetic variant filtering strategy used in this study.
13
Figure 3. Representative flowchart of genetic variant filtering strategy in this study. Minor allele
frequencies were determined using the gnomAD database.
Adults with liver disease of
unknown etiology
Isolate individual genomic DNA
Exome Capture and Sequencing
Align WES data to human genome (hg 19) using BWA
Variants called using GATK and annotated with Annovar
Dominant Inheritance Pattern
(i.e. heterozygous variants)
Minor allele frequency <2x10-5 (gnomAD)
Recessive Inheritance Pattern
(i.e. homozygous or compound
heterozygous variants)
Minor allele frequency <0.01 (gnomAD)
Protein-altering variants:
- premature termination,
- frameshift,
- splice-site, or
- missense variants predicted to
be deleterious by MetaSVM
Pathogenic variants described in NCBI
Clin Var
and/or
Variants(s) in genes related to a liver-
related disease listed in Online
Mendelian Inheritance in Man (OMIM)
WES Data Analysis
Generation of
Annotated WES Data
14
Principal Component Analysis
Principal component analysis (PCA) was performed to determine the ancestry of
the patients in our cohort. All tag SNP genotypes (genotype of a subset of single
nucleotide polymorphisms within a linkage disequilibrium block) were obtained
from WES data and used as inputs, along with the same SNPs from subjects in
the HapMap project, to perform PCA with EIGENSTRAT software.40
Sanger Sequencing
Sanger sequencing of the identified PPARG variant (p.Gly161Val) in patient 1
was performed by PCR amplification of genomic DNA of the proband and her
parents. Sanger sequencing of the identified ABCB4 variants (p.Arg549Cys and
p.Ala934Thr) in patient 2 was performed by PCR amplification of genomic DNA
of the proband, her mother and her son. Sanger sequencing of the identified
ABCB4 variant (p.Ter1280Arg) in patient 3 was performed by PCR amplification
of genomic DNA of the proband. Sanger sequencing of the identified NDFUB3
variant (p.Trp22Arg) in patient 4 was performed by PCR amplification of genomic
DNA of the proband and her parents. Sanger sequencing of the heterozygous
splice-site variant (c.2067+1G>A) in APOB in patient 5 was confirmed by PCR
amplification of genomic DNA of the proband. Forward and reverse primers for
each variant are described in Table 1.
15
Table 1. Gene name, accession number, and forward and reverse primer sequences used for
Sanger sequencing.
Gene
Symbol
NCBI
reference
sequence
Amino acid
change
Forward Primer
Reverse Primer
PPARG
NM_015869
p.Gly161Val
5’-
CAGGCCAGTATACC
TTTCGC-3’
5’-
GGATCCGACAGTT
AAGATCACA-3’
ABCB4
NM_000443
p.Gly161Val
5’-
ATGTGGTGGTCCTT
CAGCTT-3’
5’-
CTTCAAGAGCTGAT
CCATGTTTTCT-3’
ABCB4
NM_000443
p.Ala934Thr
5’-
ACCAAATCGAAAAC
AACCGGCA-3’
5’-
AGGAGGCTGAAGA
GATGGTTACA-3’
ABCB4
NM_000443
p.Ter1280Arg
5’-
ATCAAGACAGGTGT
CACTTCTAACT -3’
5’-
GAATGGGAGAGTC
AAGGAGCAT -3’
NDFUB3
NM_002491
p.Trp22Arg
5’-
GTGTTAATCTTTTCC
TTACAGACATGG-3’
5’-
CATTGAAAAGCAAC
ATAGACACTTG-3’
APOB
NM_000384
c.2067+1G>A
5’-
GGAAGTGCCTGGTG
GTTCTT-3’
5’-
TTCCATCACTTGAC
CCAGCC-3’
16
Orthologues
Full-length orthologous protein sequences from both vertebrate and invertebrates
were obtained from GenBank. Protein sequences were aligned using the
ClustaW or Clustal Omega algorithm.
Author Contributions
A.D., E.O., D.A., M.S., J.B., D.J., P.K.M., and S.V. participated in patient
recruitment and/or patient’s ascertainment and management; A.H. performed the
exome sequencing analysis for all patients in the study; D.D., K.D., and A.B.
assisted with DNA extraction, next-generation and Sanger sequencing analysis;
X.Z., and D.J. analyzed pathologic specimens. A.H. wrote the thesis.
17
RESULTS
Study population characteristics and whole-exome sequencing in cohort 1
Nineteen adults with unexplained liver disease and no history of alcohol overuse
were recruited from Yale New Haven Health after an unrevealing conventional
work-up performed by a hepatologist. These individuals presented between the
ages of 22 and 73 years-old with a variety of liver disorders (Table 2) with or
without other co-morbidities. We performed individual whole-exome sequencing
of germ line DNA isolated from each patient. Targeted bases were sequenced by
a mean of 90 reads, with 94% of targeted bases having more than eight
independent reads, and 92% having more than fifteen independent reads,
conferring high confidence calling of homozygous and heterozygous variants
across the exome (Table 3). Genomic analysis identified a monogenic disorder in
five patients of this adult population cohort (~25%), gaining insight into liver
disease pathogenesis and with direct impact on clinical management (Table 4).
Ethnicity was determined using Principal Component Analysis (Figure 4).
