Klinik Für Urologie Und Urologische Onkologie

Medizinische Hochschule Hannover

Klinik für Urologie und Urologische Onkologie

Identifizierung epigenetischer Alterationen als Biomarker für Genese, Progression und Therapieansprechen des Nierenzellkarzinoms.

INAUGURALDISSERTATION zur Erlangung des Grades einer Doktorin der Naturwissenschaften - Doctor rerum naturalium - (Dr. rer. nat.)

Kumulative Dissertation, bestehend aus 6 Beiträgen vorgelegt von

Natalia Dubrowinskaja

aus Atschinsk Russland

Hannover 2019

Angenommen durch den Senat: 01.10.2019

Präsident: Prof. Dr. med. Michael P. Manns

Wissenschaftliche Betreuung: Prof. Dr. med. Markus Antonius Kuczyk

Wissenschaftliche Zweitbetreuung: Prof. Dr. rer. nat. Jürgen Alves

1. Referent/in: Prof. Dr. med. Markus Antonius Kuczyk

2. Referent/in: Prof. Dr. rer. nat. Jürgen Alves

3. Referent/in: Prof.‘in Dr. med. Brigitte Schlegelberger

Tag der mündlichen Prüfung: 01.10.2019

Prüfungsausschuss

Vorsitz: Prof. Dr. rer. nat. Jürgen Alves

1. Prüfer/in: Prof. Dr. med. Markus Antonius Kuczyk

2. Prüfer/in: Prof. Dr. rer. nat. Jürgen Alves

3. Prüfer/in: Prof.‘in Dr. med. Brigitte Schlegelberger

Verzeichnisse

I Inhaltsverzeichnis 1 Einleitung ...... 5 1.1 Epidemiologie des Nierenkarzinoms ...... 5 1.2 Aktueller Stand der Therapie...... 5 1.2.1 Therapieansätze...... 6 1.2.2 Therapieaussichten...... 7 1.2.3 Klinische Therapieentscheidungen – Möglichkeiten und Limitationen...... 7 1.3 Molekulare Pathogenese des Nierenzellkarzinoms...... 8 1.4 Molekulare Biomarker für das Nierenzellkarzinom – aktueller Stand...... 12 1.5 DNA-Methylierung als klinischer Biomarker...... 13 1.6 Zielsetzung der Arbeit...... 15 2 Ergebnisse...... 16 2.1 Übersicht der zugrundeliegenden Publikationen...... 16 2.1.1 Beigefügte Publikationen...... 18 2.2 Zusammenfassung der Ergebnisse...... 73 2.2.1 Allgemeine Kriterien für die durchgeführten Untersuchungen...... 73 2.2.2 Ergebnisse funktioneller Analysen...... 76 2.2.3 Deskriptive Ergebnisse...... 76 2.2.4 Translationale Ergebnisse...... 79 3 Diskussion ...... 83 3.1 MIR124-3...... 84 3.2 MIR9-1...... 86 3.3 GATA3...... 87 3.4 GATA5...... 88 3.5 CST6...... 90 3.6 LAD1...... 91 3.7 NEFH...... 93 3.8 CRHBP...... 94 3.9 Zusammenfassende Diskussion...... 96

3.9.1 Neue Erkenntnisse...... 96 3.9.2 Aktuelle klinische Anforderungen bei RCCs...... 97 3.9.3 Aspekte für zukünftige Untersuchungen...... 98 3.9.4 Perspektiven für die klinikorientierte RCC-Forschung...... 101 4 Zusammenfassung...... 102 5 Summary...... 104 6 Literaturverzeichnis...... 105 Danksagung...... 133 Eidesstattliche Erklärung...... 134 Lebenslauf...... 135

II Tabellenverzeichnis

Tabelle 1. Eigenanteilerklärung…………………………………………………………...….18 Tabelle 2. Übersicht über das experimentelle Design……………………….………...... …..75 Tabelle 3. Übersicht über die wichtigsten deskriptiven Ergebnisse...………………...……...78 Tabelle 4. Übersicht über die wichtigsten Ergebnisse zur Prognose………..……………...... 80 Tabelle 5. Übersicht über die wichtigsten Ergebnisse zum Therapieansprechen……...…...... 82 Tabelle 6. Übersicht über die Publikationen, Gene und Zitierungen……………………...….83

III Abkürzungsverzeichnis

AKT Proteinkinase B BAP1 BRCA1 Associated Protein-1 CGIs CpG islands (en.); CpG-Inseln (dt.) ccRCC Clear Cell Renal Cell Carcinoma (en.); klarzelliges Nierenzellkarzinom (dt.) CRH Corticotrophin-Releasing Hormone CRHBP Human Corticotrophin Releasing Factor Binding Protein CST6 Cystatin E/M ctDNA Circulating Tumor DNA (en.); zirkulierende Tumor-DNA (dt.) dt. deutsch en. englisch EMT Epithelial-Mesenchymale Transition GATA3 GATA Binding Protein 3 GATA5 GATA Binding Protein 5 IMDC International Metastatic Renal Cell Carcinoma Database Consortium Model (en.) ITH Intratumorale Heterogenität KPP Klinisch-Pathologische Parameter LAD1 Ladinin 1 LOH Loss Of Heterozygosity MET MET Proto-Oncogene miRNA microRNA mRCC metastasiertes Nierenzellkarzinom mRNA messenger RNA MSKCC Memorial-Sloan-Kettering-Cancer-Center Model mTOR mechanistic Target of Rapamycin (en.); Ziel des Rapamycins im Säugetier (dt.) NF-H Neurofilament Heavy Chain OS Overall Survival (en.); Gesamtüberleben (dt.) PCR Polymerase Chain Reaction (en.); Polymerase-Kettenreaktion (dt.)

PFS Progression Free Survival (en.); progressionsfreies Überleben (dt.) QMSP Quantitative Methylspezifische PCR RT-PCR Real-Time quantitative PCR (en.); Echtzeit-PCR siRNA small interfering RNA SNP Single Nucleotide Polimorphism (en.); Einzelnukleotid-Polymorphism (dt.) TCGA The Cancer Genome Atlas VEGF Vascular Endothelial Growth Factor VHL Von Hippel-Lindau-Syndrom

1 Einleitung

1.1 Epidemiologie des Nierenkarzinoms

Mit über 400 000 Fällen weltweit, welche ca. 2,2 % der Gesamtheit aller onkologischen Neuerkrankungen entsprechen, lag Nierenkrebs (renal cell carcinoma (RCC) - en.) - im Jahr 2018 auf Platz 14 der Inzidenzfälle1. Die Mortalität betrug dabei über 175 000 Fälle und lag dementsprechend auf dem 16 Platz1,2. In Deutschland wurden im Jahr 2014 offiziell 14 960 Neuerkrankungen und 5 278 Todesfälle registriert2. Die Inzidenz bei Männern lag etwa zwei Mal höher als bei Frauen1-3. Mit einem Anteil von 70 % ist das klarzellige Nierenzellkarzinom (Clear cell renal cell carcinoma (ccRCC) - en.) mit Abstand der häufigste Typ, 10-15 % sind papilläre Karzinome, ca. 5% - chromophobe und weniger als 1 % sind den Karzinomen der Sammeltubuli zuzuordnen4. Das durchschnittliche Erkrankungsalter von 68 Jahren bei Männern und 72 Jahren bei Frauen weist daraufhin, dass das Alter zu den wichtigsten Risikofaktoren bei der Ausbildung der Nierenkarzinome RCC gezählt werden kann1-4. Als gesicherte Risikofaktoren gelten ferner einige modifizierbare Lebensstilfaktoren wie Tabakkonsum, Adipositas, Bluthochdruck und Diabetes. Zu den protektiven Lebensstilfaktoren zählen der Alkoholkonsum und körperliche Aktivität5. Etliche chemische Agenzien wie Arsen, Asbest, Benzinbestandteile, Blei, Halogenkohlenwasserstoffe, Kadmium und Trichloroethylen gelten als krankheitsassoziiert und somit als zusätzliche Risikofaktoren3,4.

1.2 Aktueller Stand der Therapie.

In über 30 % der Fälle liegt bereits bei der Erstdiagnose eine Metastasierung der Tumore vor6. Derartige Befunde bedeuten für den Patienten eine sehr ungünstige Prognose mit einer 5-jährigen Überlebensrate von ca. 12 %7. Während die Patienten mit Tumoren im frühen Stadium nach der Resektion gute Genesungsaussichten haben, weist die Gruppe mit metastasierten Tumoren (mRCC), die zudem oft inoperabel sind, eine sehr hohe Mortalitätsrate auf. Nach der Primärtumorentfernung setzt die Metastasierung in 20-30 % der Fälle wieder ein, wobei zugleich sich oft eine Resistenz gegenüber „klassischen“ Chemo- und 5

Strahlentherapieansätzen entwickelt8,9. Die medikamentöse Systemtherapie spielt bei der Behandlung der mRCC-Patienten, vornehmlich vor einem palliativen Hintergrund, eine wichtige Rolle. Mit Hilfe der zur Verfügung stehenden Therapeutika wird angestrebt, vor allem das Gesamtüberleben (overall survival (OS) - en.) und progressionsfreie Überleben (progression free survival (PFS) - en.) metastasierender Patienten zu verlängern und deren Lebensqualität zu verbessern10.

1.2.1 Therapieansätze.

Die gezielten Krebstherapeutika. Mehrere Vertreter dieser vergleichsweise neuen Gruppe sind aktuell die wirksamsten Medikamente bei der Behandlung von inoperablen oder metastasierenden Nierenzellkarzinomen und befinden sich im klinischen Einsatz bzw. in den fortgeschrittenen Testphasen. Drei unterschiedliche Wirkungsansätze haben sich derzeit herauskristallisiert. Zum einem wird, angesichts der Tatsache, dass circa 60 % der Karzinome hohe Konzentration an VEGFs (Vascular Endothelial Growth Factors - en.) aufweisen, versucht an mehreren Stellen die Prozesse der Tumorvaskularisation gezielt zu beeinträchtigen. Konkret wird die Produktion der proangiogenetischen Faktoren durch die Tyrosinkinaseinhibitoren (TKI) gemindert, unterdrückt oder deren Bindung an die involvierten Rezeptoren verhindert bzw. deren Aktivität blockiert11,12. Zum anderem wird, mittels sogenannter mTOR-Inhibitoren (mechanistic Target of Rapamycin - en.), die allgemeine Funktionalität der Tumorzellen beeinflusst. Die Inaktivierung des mTORs, durch therapeutische Kleinmolekülinhibitoren der Klasse Makrolide, hemmt allgemein die stromabwärts liegenden Signalwege, was die Proliferationsstörung der Tumorzellen, die Stimulierung der Autophagie oder Apoptose und Beeinträchtigung der Tumorangiogenese bewirkt13-15. Des Weiterem sind diverse Substanzen, die als Inhibitoren verschiedener Checkpoints des Immunsystems (Immun-Checkpiont-Hemmer (ICH)) agieren und der immunevasiven Eigenschaft der Tumore entgegenwirken, ein Gegenstand pharmazeutischer Forschung16,17. Die ICHs weisen teilweise eine so derart hohe Toxizität auf, dass viele der klinischen Studien

(beispielsweise Teststudien der CTLA4-Inhibitoren (cytotoxic T-lymphocyte-associated protein 4)) in der Phase II abgebrochen werden9.

1.2.2 Therapieaussichten.

In den Jahren der Anwendung von gezielten Krebstherapeutika bei Nierenzellkarzinomen ist die 5-Jahres-Überlebensrate von 20 % auf 60 % gestiegen. Die 5-Jahres-Überlebensrate von 10 % und das mediane Gesamtüberleben von 6-12 Monaten bleiben bei primär metastasierten Tumoren weiterhin sehr niedrig18. Trotz der vergleichsweise höheren Rate der Therapieansprecher (eine Vergleichsstudie mit IL-2 (Interleukin-2)), ist das komplette Ansprechen dennoch sehr selten19. Auch wird lediglich eine signifikante Verlängerung der Gesamtüberlebenszeit beobachtet, jedoch nicht des progressionsfreien Überlebens19,20. Das Resistenzproblem der Tumore, welches mutmaßlich aus deren hoher epi- und genetischer Heterogenität resultiert, bleibt auch gegenüber den gezielten Krebstherapeutika bestehen – oft bereits in dem ersten Jahr der Therapie21,22. Überdies weisen alle Medikamente eine sehr hohe Toxizität auf, welche für die Patienten erhebliche Nebenwirkungen zufolge hat. Die klinischen Ärzte stehen somit vor der Aufgabe, eine möglichst individuelle und effiziente Therapie durchzuführen.

1.2.3 Klinische Therapieentscheidungen – Möglichkeiten und Limitationen.

An dieser Stelle ist kurz der Unterschied zwischen den Begriffen „Prognose“ und „Prädiktion“ beziehungsweise „prognostische“ und „prädiktive“ Faktoren zu verdeutlichen. Die Prognose dient der Einschätzung des Krankheitsausgangs ohne die Berücksichtigung der eventuellen zukünftigen adjuvanten Therapie. Die Prädiktion wird als die Prognose des Krankheitsverlaufes unter einer bestimmten Therapie und unter Einbezug der individuellen Risikofaktoren des Patienten definiert. Prädiktive Faktoren helfen bei der Einschätzung der Therapieeffektivität, was angesichts der Tatsache, dass speziell die ccRCC-Patienten bei der gezielten Therapie unterschiedliche Überlebenszeiten aufweisen, von großem klinischem Interesse ist23.

Als Grundlage bei der Behandlungswahl von Nierenkarzinompatienten dient in Deutschland die „S3-Leitlinie Diagnostik, Therapie und Nachsorge des Nierenzellkarzinom.“24. Den Klinikern stehen anatomische25, klinische26,27 und histologische Faktoren28,29 zur Verfügung um eine prognostische oder prädiktive Aussage treffen zu können. Da die Aussagekraft der Einzelfaktoren in der Praxis nur begrenzt für eine individuelle Prognose und Therapiewahl ausreicht, werden von Medizinern seit über 16 Jahren prognostische Modelle als Entscheidungshilfen, zwecks Risikostratifizierung, benutzt: das MSKCC (Memorial-Sloan-Kettering-Cancer-Center Model - en.) mit fünf unabhängigen klinischen Faktoren30 und das IMDC (International Metastatic Renal Cell Carcinoma Database Consortium Model - en.) mit sechs ebenso klinischen Kriterien30. Sie sind offiziell in S3-Leitlinien zur Bestimmung der Risikogruppe verankert. Verglichen mit vier anderen Prognose-Modellen (allerdings noch aus der Zytokinära der RCC-Therapie - darunter auch MSKCC), weist das IMDC-Modell die beste Aussagekraft mit Hinsicht auf 2-Jahres- Prognose krankheitsbedingter Todesfälle vor30. Somit ist das IMDC-Modell zurzeit (zusammen mit den histologischen Daten) auch für die Prädiktion der gezielten Krebstherapie das geeignetste Modell. Dennoch sind auch die Entwickler des IMDC-Modells der Meinung, dass rein klinische prognostische Faktoren bei Therapieprädiktion für metastatische Tumoren ihre Grenzen erreicht haben. Es erfordert neue Biomarker, die bei Individualisierung der Therapie, vor allem bei einer Voraussage von Therapieversagen bzw. im Falle einer erforderlichen „Switch- Therapie“ (Sequenztherapie um die Therapieresistenz zu umgehen) zu der Erhöhung der Erfolgsquote klinischer Behandlungen beitragen könnten31-33.

1.3 Molekulare Pathogenese des Nierenzellkarzinoms.

Das für die meisten Nierentumore als typisch geltende Bild des gestörten Glukosemetabolismus und Fettablagerungen weist auf die Beteiligung der Gene hin, die für die Stoffwechselvorgänge relevant sind34. Das lässt einerseits den Nierenkrebs als eine metabolische Krankheit verstehen34, andererseits sind zusätzliche Mechanismen notwendig, um eine tatsächliche Tumorbildung mit für ccRCC spezifischen metabolischen Phänotypen auszulösen35.

Ebenfalls wurde seit langem die Beteiligung des Immunsystems an den Prozessen der Tumorgenese der Niere untersucht. In der Tumormikroumgebung (tumor microenvironment (TME) - en.) wurde speziell bei den Nierentumoren eine sehr hohe Ansammlung und Aktivität diverser Abwehrzellen beobachtet36,37. Jedoch findet gleichzeitig eine Maskierung der Tumorzellen gegenüber dem Immunsystem, sowie eine verstärkte Apoptose der T-Zellen und daraus folgende Anergie statt37,38. Des Weiteren können die veränderten Differenzierungen dendritischer Zellen, die Induktion der Anergie assoziierten Gene in den T-Zellen39 und tumorassoziierte Makrophagen (TAMs) zur Insuffizienz des Immunsystems und der Tumorangiogenese und -infiltration beitragen40,41. Es wurde eine prognoserelevante tumorsubtypspezifische Zusammensetzung der TAM-Phänotypen in RCCs gezeigt42. Die molekularen Ursachen hinter den Nierenzellkarzinomen, vor allem im Kontext der genetischen Komponenten, sind Gegenstand zahlreicher wissenschaftlicher Untersuchungen. Aktuell wird davon ausgegangen, dass ca. 4 % der RCC-Fälle auf eine genetische Prädisposition zurückzuführen sind43. Ein Teil der erblichen Nierentumore gehört zu gut beschriebenen vererbbaren Syndromen, die Mehrzahl wurde allerdings noch nicht vollständig und eindeutig genetisch charakterisiert43. Zu den gut beschriebenen RCC-Syndromen mit assoziierten Mutationen zählen von Hippel-Lindau-Syndrom (VHL-Mutationen), erbliches papilläres RCC (MET-Mutationen), Birt-Hogg-Dubé- (BHD) -Syndrom (FLCN-Mutationen (Folliculin)) und hereditäre Leiomyomatose mit RCC (HLRCC) (FH-Mutationen (Fumarathydratase))43. Das am häufigsten betroffene Gen bei den ccRCCs ist VHL. Ca. 40-50 % der Träger verschiedener VHL-Alterationen erkranken an Nierenkrebs. Insgesamt ist das Gen bei ca. 60-80 % aller erblichen sowie sporadischen ccRCC-Fällen epigenetisch oder genetisch stillgelegt4,43,44. Dazu zählen unter anderem Allel- und Bialleldeaktivierungen, Allel- und Bialleldeletionen, DNA-Mutationen und DNA-Methylierungsalterationen4,43,44. Als ein Adaptorprotein tritt VHL-Protein (pVHL) zwar als Interaktionspartner für über 30 verschiedene in die Tumorgenese involvierten Proteine auf, eine Assoziation zwischen einem Funktionsverlust des pVHL und progressionsfreiem Überleben bzw. Gesamtüberleben wurde bei den metastasierten Patienten jedoch nicht festgestellt4,43,44. Dieser Umstand wurde dahingehend interpretiert, dass die Tumorentstehung und -progression nicht einzig als Folge des Funktionsverlusts des pVHL angesehen werden können45, sondern es vielmehr von mehreren genetischen und epigenetischen Ereignissen in Kombination verursacht wird46.

Weitere, bei sporadischen ccRCCs am häufigsten mutierte Gene sind PBRM1 (Polybromo 1), SETD2 (SET domain containing 2) und BAP1 (BRCA1 associated protein-1), wobei in der Summe bei 10-50 % der Patienten somatische Mutationen beobachtet wurden47,48. Diese Gene kodieren für histon- und chromatinregulierende Proteine und sind auch wie das VHL-Gen auf dem Chromosom 3 lokalisiert, welches in mehr als 90 % der ccRCCs den Verlust des kurzen Arms aufweist48. Die Mutationen in jedem dieser Gene sind signifikant mit dessen Expressionsverlust assoziiert. Diese Mutationen korrelieren untereinander und sind mit vorgeschrittenem Tumorstadium, hohem Tumorgrad, schlechter Tumordifferentiation und Nekrose signifikant assoziiert48,49. In früheren Arbeiten wurden die sich gegenseitig ausschließenden Mutationen von BAP1 und PBRM1 und deren unabhängige Aussagekraft bezüglich schlechter Prognose diskutiert50. Auch bei der Klassifizierung der ccRCCs in sieben klonale evolutionäre Subtypen spielen diese drei Gene bei fünf der Subtypen eine sehr wichtige Rolle48,49. In ca. 20 % der ccRCCs wurden Mutationen, die den Signaltransduktionsweg des mTOR- Komplexes 1 deregulieren, in den Genen MTOR, TSC1 (tuberous sclerosis proteins 1), PIK3CA (Phosphatidylinositol-4,5-Bisphosphate 3-Kinase Catalytic Subunit Alpha) und PTEN (phosphatase and Tensin homolog) identifiziert50. Verschiedene weitere spontane somatische Mutationen, darunter auch die des bereits erwähnten MET-Gens, konnten mit sporadischen Fällen des Nierenkarzinoms in Verbindung gebracht werden43,51. Als tumorrelevante identifizierte Gene sind unter anderem KDM5C (lysine-specific demethylase 5C), UTX (ubiquitously transcribed tetratricopeptide repeat, X chromosome), MITF (Melanocyte Inducing Transcription Factor), SDHD (succinate dehydrogenase complex subunit D), TFE3 (transcription factor E3), TFEB (transcription factor EB) und TP53 (tumor protein p53) beschrieben worden47,51,52. Hinsichtlich der Unterteilung der tumorrelevanten Gene in Kategorien nach deren Funktion im Organismus wären die am häufigsten betroffenen Gene (u.a. VHL, PBRM1, SETD2, BAP1, MTOR) den sogenannten Caretaker-Genen (Hausmeister-Gene), die für die Genomintegrität verantwortlich sind, zuzuordnen53. Zugleich wurden die Deletionen, Mutationen usw. in diesen Genen als Treiber-Aberrationen bei frühen Ereignissen der klonalen Tumorevolution in den ccRCCs identifiziert49,54,55. Diese Erkenntnisse würden im Einklang mit der Theorie über die Kolinearität zwischen der Entstehungszeit der tumorrelevanten Gene während der Evolution und deren Rolle bei der Tumorprogression stehen. Diese beinhaltet dass die

10 tumorassoziierten Mutationen der zwei Gengruppen – Caretakers und Gatekeepers (Pförtner- Gene) in zwei zeitlich nach einander folgenden Peaks stattfinden und die phylogenetische Abfolge der Genentstehung (Caretakers bei der Einzeller-Entstehung und Gatekeepers bei der Metazoa-Entstehung) widerspiegeln53. Eine ausgeprägte intratumorale Heterogenität (ITH) der klarzelligen Nierentumore wird als einer der entscheidendsten Aspekte für die häufige Metastasierung und Therapieresistenz diskutiert47,52,56,57. Aus den in silico-Analysen der in der TCGA-Datenbank zu Verfügung stehender Multi-omic Daten der Einzelbiopsien geht hervor, dass die Patienten mit ähnlichen klinischen Befunden auf der molekularen Ebene individuelle heterogene Profile aufweisen und die ITH ein häufiges Ereignis ist54,58. Das heterogene individuelle Mutationsspektrum erschwert die Suche nach statistischer Assoziation mit den klinischen Verläufen. Die tiefergreifenden Untersuchungen von Multiregiongewebsbiopsien mit Hilfe von ultratiefen Sequenzierungen des Genoms oder Exoms zeigten, dass die Diversität und Komplexität der ITH mit der Anzahl der analysierten Biopsien steigt49,54. Aktuell wird die subklonale Evolution der Tumorzellen der Niere intensiv erforscht, um die klonalen Architekturmuster mit klinischen prognoserelevanten Subtypen in Verbindung zu bringen55. Des Weiteren werden inzwischen auch vor dem populations-genetischen Hintergrund die Signatur-Analysen des ccRCCs durchgeführt55. Interessanterweise sind die genetischen Alterationen absolut gesehen vergleichsweise weniger häufige Ereignisse59. Zugleich kodieren einige der mit am häufigsten mutierten Gene wie z.B. PBRM1, SETD2 und BAP1 in die epigenetischen Prozesse involvierte chromatinmodulierende Enzyme. Überdies sind diese Gene selbst häufig epigenetisch alteriert59. Tatsächlich sind die epigenetischen Ereignisse bei den Nierentumoren häufiger zu finden als die somatischen Mutationen. Dazu gehören unter anderem die Veränderungen der Histonmodifikationen, der DNA-Methylierung und veränderte miRNA-Levels60. Mittlerweile wurden von mehreren Gruppen Hypermethylierungen diverser Loci in dem Nierenkarzinom gezeigt und einige davon teilweise unabhängig voneinander bestätigt61. Daraus folgend lässt sich auf die Relevanz die epigenetischen Alterationen bei der Entstehung und Progression von klarzelligen Nierentumoren schließen35. Dementsprechend wäre es denkbar, dass ferner Assoziationen solcher Alterationen mit Prognose und Prädiktion der Nierenzelltumoren existieren.

1.4 Molekulare Biomarker für das Nierenzellkarzinom – aktueller Stand.

Die limitierte Aussagekraft klinischer Modelle bezüglich einer Stratifikation der ccRCC- Patienten und deren Therapieansprechen ist vor allem auf hohe intratumorale genetische und epigenetische Heterogenität zurück zu führen56,62-64. Vor diesem Hintergrund könnte eine feinere Auflösung individueller molekularer Profile der Patienten mit ähnlichen klinischen und histologischen Phänotypen fehlende prognostische bzw. prädiktive Informationen liefern. Daher ist es weit akzeptiert dass neue biomolekulare Marker benötigt werden65. Dieser neue experimentelle Ansatz wird aktuell intensiv erforscht und ist Gegenstand zahlreicher Untersuchungen und Studien. Die Umsetzung in die Praxis steht jedoch noch aus. Da mittlerweile eine große Anzahl an vorgeschlagenen biomolekularen Kandidatenmarker existiert, werden hier nur beispielhaft einige Ergebnisse aus der Literatur aufgeführt. Die Kandidatenmarker lassen sich nach Art des zu analysierendes Materials und nach der Art der Analyseverfahren und des Weiteren nach deren diagnostischen, prognostischen und prädiktiven Nutzen unterscheiden. Die Anwendung von immunhistochemischen Protein- und zytogenetischen Analyseverfahren an Geweben66-70 und nicht invasiven Proteintests mit Hilfe bildgebender Verfahren71-73 könnten bei der Prognose eine Rolle spielen. Es gibt experimentelle Ansätze, aus dem nicht invasiv bzw. mit geringen Komplikations- häufigkeit zu beschaffenden Material, wie beispielsweise Urin oder Blut, die Informationen prognostischer bzw. prädiktiver Natur zu gewinnen74-76, welche aber vorerst widersprüchliche Resultate lieferten77-82. Diverse Kandidatenmarker für die Prognose wurden bei der Analyse an den Tumorgeweben im Bereich der DNA-Mutationen52,83, der mRNA84-87 und DNA-Methylierungen88,89, zum Teil als Einzelfaktoren oder auch in Kombination, identifiziert58,90-92. Auf genetischer Ebene konnten zum Beispiel die Assoziationen zwischen einigen SNP- Varianten (single nucleotide poyimorphism - en.; Einzelnukleotid-Polymorphismen) und dem Therapieausgang gezeigt werden93-98. Für die Veränderungen in der mRNA-Expression wurden in ersten kleineren Studien Korrelationen mit dem Therapieansprechen, also ein prädiktives Potenzial, nachgewiesen99,100.

Auf der epigenetischen Ebene wurden beispielsweise für Gene wie BNC1 (Basonuclin 1), CDO1 (Cysteine Dioxygenase Type 1), COL14A1 (Collagen Type XIV Alpha 1 Chain)101,102, GREM1 (Gremlin 1)103, GATA5 (GATA Binding Protein 5)104, RASSF1A (Ras Association Domain Family Member 1)105-107, SFRP1 (Secreted Frizzled Related Protein 1)108 signifikante Korrelationen der DNA-Methylierung mit klinisch-pathologischen Parametern (KPP; clinicopathological parameters (CPPs) - en.) - und einer ungünstigen Prognose bei RCC beobachtet. Ebenso wurde vor kurzem ein promoter methylation classifier ((PMC) panel (Promoter-Methylierungsklassifizierer-Panel (dt.)) aus 172 CpG-Stellen für die Prognose der nicht-metastasierten ccRCC-Patienten vorgeschlagen109. Die Liste der Loci, die tumorspezifische DNA-Methylierungsänderungen aufweisen bzw. starke Assoziationen mit den klinikrelevanten Fragestellungen zeigen, wird ständig mit neuen potenziellen Kandidaten erweitert. Als Schlussfolgerung aus den neuen Erkenntnissen letzter Jahre, die auf potenzielle klinische Relevanz epigenetischer Alterationen hindeuten, wurde in unserer Arbeitsgruppe beschlossen sich verstärkt mit dem Thema der DNA-Methylierungen in den urogenitalen Tumoren zu beschäftigen.

1.5 DNA-Methylierung als klinischer Biomarker.

DNA-Methylierung allgemein. Zwei Hauptfunktionen methylierter Cytosine, die aus der Hauptfurche des DNA-Doppelstrangs herausragen, werden postuliert - das Fernhalten der Transkriptionsfaktoren und das Anlocken von methylbindenden Proteinen, die mit dem Genschweigen und mit Histonmodifikationen assoziiert sind. Es wurde eine breite Spannweite an den DNA-Methylierungsvarianten unter den verschiedenen Spezies gefunden. Im Gegensatz zu den Wirbellosen wurde eine Sonderstellung der Wirbeltiere mit für sie typischen Genmethylierungen beobachtet. Bei den Letzteren wird ein Teil der Gene in Korrelation mit der Präsenz von CpG-Methylierungen transkribiert (und evtl. durch die auch kontrolliert), und ein anderer Teil ist davon unabhängig. Dies lässt die Methylierung als einen weiteren, zusätzlich hinzugekommenen, Mechanismus moderater Natur, der toleriert und / oder bei der Kontrolle der Genexpression benutzt wird, beschreiben110,111.

DNA-Methylierung in Tumoren. Ähnlich der Genmutationen können auftretende tumorspezifische Methylierungen zu der Aggregation der tumorbegünstigenden Signalwege führen, woraus sich auch der Tumorphänotyp ableiten lässt112. Trotz einer 5 - 10 -fachen Steigerung der Mutationsrate (im Detail überwiegen Rearrangements die Punktmutationen113) liegt die Frequenz der Methylierungsänderungen in Tumoren weit über die Frequenz der Genmutationen114-116. Dabei entstehen tumorspezifische Methylierungsmuster113,117-120, und der Verlust der DNA-Methylierung trägt mutmaßlich zu der chromosomalen Instabilität bei121. Diese Beobachtungen stehen im Einklang mit dem Krebsentstehungsmodell der epigenetischen Krebsvorläuferzelle110,121-124. Auch der globale Methylierungsverlust kann mit dem Stadium der Progression und Metastasierung korrelieren125-128. Früher im Text exemplarisch aufgeführte Beispiele der Korrelation der Methylierungsalterationen mit wichtigen tumorrelevanten Charakteristika veranschaulichen das Potenzial der DNA-Methylierung als vielversprechende translationale Kandidatenbiomarker. Obwohl die Anwesenheit von chromosomalen Aberrationen und Mutationen in RCCs als häufiges Ereignis gut beschrieben ist, konnte bis heute keiner der untersuchten genetischen Kandidaten eine ausreichende prognostische bzw. prädiktive Aussagekraft vorweisen. Dabei ist es zu beachten, dass die molekularen Schritte der Tumorentstehung aus Keimzellen nicht zwingend gleiche Mutationen wie die nicht-familiären Tumore aufweisen müssen. Anderseits schließt das Fehlen klinikrelevanter Aussagen anhand der genetischen Alterationen das Finden relevanter Informationen bei den gleichen Genen auf epigenetischer Ebene nicht aus. Somit können auch aus dem Bereich der Genetik abgeleitete Daten eine gute Ausgangsplattform sein, um die epigenetische Seite zu erforschen113. Praktibilität der DNA-Methylierungsanalysen. Zu erwähnen wären die unkomplizierten Abläufe und die Möglichkeit der Verfahrensstandardisierung für die klinische Routine bereits vorhandener Methoden wie beispielsweise QMSP (quantitative methylspezifische PCR) oder Pyrosequenzierung. Dafür benötigtes Material kann ohne Beeinträchtigung der Routinediagnostik und ohne zusätzliche Belastung des Patienten untersucht werden. Die Materialkonservierung kann flexibel gehandhabt werden - sowohl aus frisch eingefrorenem Gewebe, als auch aus FFPE-Gewebe (formalinfixiertes paraffineingebettetes Gewebe) und - Gewebsschnitten ist die Gewinnung hochqualitativer DNA in standardisierten Verfahren möglich.

1.6 Zielsetzung der Arbeit.

Gemeinsame Zielsetzung der vorliegenden Arbeit und der in diesem Rahmen publizierten Studien war die Identifizierung neuer molekularer Marker mit einer potenziellen Relevanz für die Prognose bzw. Prädiktion der Nierenzellkarzinome. Die Relevanz entsprechender biometrisch identifizierten Biomarker-Kandidaten sollte in retrospektiven Querschnittsanalysen unter Verwendung aktueller Laborverfahren, wie der Quantifizierung von mRNA-Expression sowie den quantitativen Nachweis von DNA- Methylierungsänderungen experimentell überprüft werden. Die quantitativen Daten sollten die Grundlage für eine objektive statistische Evaluation schaffen, um so Hinweise auf die Relevanz der untersuchten Alterationen zu geben.

2 Ergebnisse.

2.1 Übersicht der zugrundeliegenden Publikationen.

1) Gebauer K, Peters I, Dubrowinskaja N, Hennenlotter J, Abbas M, Scherer R, et al. Hsa- mir-124-3 CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell carcinoma. Br J Cancer. 2013 Jan 15;108(1):131-8.129

Mittels QMSP-Messungen wurde eine Erhöhung der relativen DNA-Methylierung in einem Locus innerhalb einer CpG-Insel des MIR124-3-Gens in der Kohorte der ccRCC- Tumorgewebsproben im Vergleich mit der Kohorte korrespondierender peritumoraler Geweben festgestellt. Die Subgruppe der fortgeschrittenen Tumore zeigte eine nochmals signifikant höhere relative Methylierungswerte. Die statistische Assoziation höherer relativer Methylierung mit einem verkürzten rückfallfreien Überleben identifizierte die Methylierung als signifikanten Modellparameter.

2) Peters I, Dubrowinskaja N, Kogosov M, Abbas M, Hennenlotter J, von Klot C, et al. Decreased GATA5 mRNA expression associates with CpG island methylation and shortened recurrence-free survival in clear cell renal cell carcinoma. BMC Cancer. 2014 Feb 17;14:101130.

Der mittels RT-PCR gemessener GATA5-mRNA-Expressionslevel wurde als statistisch signifikant reduziert in der Kohorte der ccRCC-Tumorgewebsproben im Vergleich mit der Kohorte korrespondierender peritumoraler Geweben gezeigt. Außerdem zeigte die Kohorte der ccRCC-Tumorgewebsproben, verglichen mit der Kohorte korrespondierender peritumoraler Gewebe, statistisch eine erhöhte relative Methylierung innerhalb eines Locus einer CpG-Insel des GATA5-Gens. Die statistische Untersuchung zu einer möglichen Assoziation des reduzierten mRNA-Expressionslevels mit schlechtem klinischem Ergebnis identifizierte mRNA-Level als signifikanten Modellparameter.

3) Peters I, Gebauer K, Dubrowinskaja N, Atschekzei F, Kramer MW, Hennenlotter J, et al. GATA5 CpG island hypermethylation is an independent predictor for poor clinical outcome in renal cell carcinoma. Oncol Rep. 2014 Apr;31(4):1523-30131.

In der Kohorte der RCC-Tumorgewebsproben zeigten mittels QMSP gemessene relative DNA-Methylierungen für ausgewählte Loci innerhalb der CpG-Inseln der GATA3- und GATA5-Gene eine statistische Assoziation mit dem Metastasierungsstatus. Die statistische Untersuchung zu einer möglichen Assoziation höherer relativer Methylierung mit einem verkürzten progressionsfreien Überleben identifizierte GATA5-Methylierung als signifikanten Modellparameter.

4) Peters I, Dubrowinskaja N, Abbas M, Seidel C, Kogosov M, Scherer R, et al. DNA Methylation Biomarkers Predict Progression-Free and Overall Survival of Metastatic Renal Cell Cancer (mRCC) Treated with Antiangiogenic Therapies. PLoS One. 2014 Mar 14;9(3):e91440132.

Die Prädiktion des Therapieverlaufs bei metastasierten RCC-Patienten war Gegenstand der obigen Arbeit. Die mittels QMSP gemessene relative DNA-Methylierung zeigte einen statistisch signifikanten Zusammenhang mit dem Therapieansprechen/-erfolg für die CST6- und LAD1-Genloci.

5) Dubrowinskaja N, Gebauer K, Peters I, Hennenlotter J, Abbas M, Scherer R, et al. Neurofilament Heavy polypeptide CpG island methylation associates with prognosis of renal cell carcinoma and prediction of antivascular endothelial growth factor therapy response. Cancer Med. 2014 Apr;3(2):300-9133.

Mittels QMSP-Messungen wurde eine durchschnittliche Erhöhung der relativen DNA- Methylierung in einem Locus innerhalb einer CpG-Insel des NEFH-Gens in der Kohorte der RCC-Tumorgewebsproben im Vergleich mit der Kohorte korrespondierender peritumoraler Geweben festgestellt. Die statistische Untersuchung zu einer möglichen Assoziation höherer relativer Methylierung mit progressionsfreiem Überleben und Gesamtüberleben identifizierte die Methylierung als signifikanten Modellparameter. Zusätzlich wurde in statistischer Untersuchung ein signifikanter Zusammenhang der Methylierung mit dem Therapieansprechen/-erfolg für den NEFH-Locus gezeigt.

6) Tezval H, Dubrowinskaja N, Peters I, Reese C, Serth K, Atschekzei F, et al. Tumor Specific Epigenetic Silencing of Corticotropin Releasing Hormone - Binding Protein in Renal Cell Carcinoma: Association of Hypermethylation and Metastasis. PLoS One. 2016 Oct 3;11(10):e0163873134.

Mittels QMSP wurde eine durchschnittliche Erhöhung der relativen DNA-Methylierung in einem Locus des CRHBP-Gens in der Kohorte der ccRCC-Tumorgewebsproben, im Vergleich zu der Kohorte korrespondierender peritumoraler Gewebe festgestellt. Gleichzeitig korrelierte die erhöhte relative Methylierung invers in ccRRC-Tumorgewebsproben mit einem reduzierten mRNA-Expressionslevel. Die erhöhte relative Methylierung zeigte statistische Assoziation mit dem Metastasierungsstatus und fortgeschrittener Krankheit. Eine Validierung der obengenannten Ergebnisse wurde mit Hilfe einer statistischen in silico-Evaluation unter Einbezug der TCGA-Datenbank (The Cancer Genome Atlas- en.) durchgeführt.

Tabelle 1. Eigenanteilerklärung.

1. 2. 3. 4. 5. 6. Publikations-Nr. Konzeption des 5 10 10 10 50 10 Forschungsansatzes (%) Planung der 20 40 50 50 70 40 Untersuchungen (%) 30 60 60 70 90 40 Datenerhebung (%) Datenanalyse und 20 40 25 20 50 30 -interpretation (%) Schreiben des Manuskripts 10 20 20 20 50 40 (%)

2.1.1 Beigefügte Publikationen.

Hier folgen die Originalpublikationen.

FULL PAPER

British Journal of Cancer (2013) 108, 131–138 | doi: 10.1038/bjc.2012.537

Keywords: renal cell carcinoma; microRNA; mir-124-3; DNA methylation; prognosis

Hsa-mir-124-3 CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell carcinoma

K Gebauer1, I Peters1, N Dubrowinskaja1, J Hennenlotter2, M Abbas3, R Scherer4, H Tezval1, A S Merseburger1, A Stenzl2, M A Kuczyk1 and J Serth*,1 1Department of Urology, Hannover Medical School, OE6247, Carl-Neuberg-Str.1, 30625 Hannover, Germany; 2Department of Urology, Eberhard-Karls University, 72074 Tu¨ bingen, Germany; 3Department of Pathology, Hannover Medical School, 30625 Hannover, Germany and 4Department of Biometry, Hannover Medical School, 30625 Hannover, Germany

Background: Whether methylation of the microRNA (mir)-124-3 CpG island is of relevance for the clinical course of a solid cancer and whether it shows association with clinicopathology or survival of patients with renal cell cancer (RCC) is not known as yet.

Methods: In a cross-sectional study, relative methylation of mir-124-3 was measured in 111 RCC samples and 77 paired normal appearing tissues using quantitative methyl-specific PCR. Results were statistically compared with tumour histology, clinicopathological parameters and disease recurrence.

Results: We found tumour-specific hypermethylation of mir-124-3 in samples of RCCs with clear cell histology (ccRCC) compared with paired normal appearing tissues (Po0.0001). Methylation was significantly increased in tumours with state of advanced disease (Po0.0001). Higher relative methylation was associated with worse recurrence-free survival in both univariate (hazard ratio ¼ 9.37; P ¼ 0.0005) as well as bivariate Cox regression analyses considering age, sex, diameter of tumours and state of advanced disease, metastasis and lymph node metastases as covariates (hazard ratios ¼ 5.9–18.2; P-values of 0.0003–0.008).

Conclusion: We identified mir-124-3 CpG islands (CGI) methylation as a relevant epigenetic mark for ccRCC thus underlining the need for functional studies of potentially affected signalling pathways in kidney tumour models. Methylation of mir-124-3 is suggested as an independent prognosticator for ccRCC.

With over 111 000 new cases and 43 000 deaths reported in 2008, stratification of targeted therapy schemes and the identification of kidney cancer is among the top 10 causes of cancer death of men in relevant signalling pathways for RCC development. The most the developed countries (Jemal et al, 2011). Nephron-sparing frequent histology of RCC is clear cell RCC (ccRCC) exhibiting surgery and nephrectomy are still the gold standard for therapy of mutations in the von Hippel-Lindau (VHL) and polybromo 1 localised stages of renal cell carcinoma (RCC), but despite recent (PBRM1) (Varela et al, 2011) genes in at least 30% of tumours advances in targeted therapy, the prognosis for patients with analysed. von Hippel-Lindau is involved in various cellular progressive disease is still poor. Moreover, few diagnostic methods mechanisms such as tumour suppression, response to hypoxia, are available to predict progression of the disease. The identifica- protein degradation and interaction with intra- as well tion of biomarkers for RCC could provide a basis for both future as extracellular matrix proteins (Nordstrom-O’Brien et al, 2010)

*Correspondence: Dr J Serth; E-mail: [email protected] Received 16 July 2012; revised 6 November 2012; accepted 8 November 2012 & 2013 Cancer Research UK. All rights reserved 0007 – 0920/13 www.bjcancer.com | DOI:10.1038/bjc.2012.537 131 BRITISH JOURNAL OF CANCER Mir-124-3 methylation and RCC prognosis and PBRM1 has been identified as part of chromatin structure Study design and patients. Cross-sectional analyses were con- modifying protein complexes (Reisman et al, 2009; Thompson, ducted on 111 RCC samples and 77 samples from paired 2009). histologically normal appearing renal tissues. Sample collection However, from other studies it has become apparent, that was approved by the local ethics committee and informed consent numerous epigenetic alterations are also involved in renal was obtained from each patient. TNM classification of all tissues carcinogenesis (Baldewijns et al, 2008; Morris et al, 2008, 2010, was evaluated according to the Union for International Cancer 2011). Most of these alterations refer to changed DNA methylation Control 2002 classification (Sobin and Compton, 2010). Two as observed in CGI of various genes and some of them have been pathologists assessed all specimens considering tumour classifica- suggested in functional analyses as candidate tumour suppressor tion, grade and histological subtype. Localised RCC was defined as genes (TSG) function (Dreijerink et al, 2001; Morris et al, 2010, pTp2, lymph node (N) and metastasis (M) negative (N0, M0) and 2011). Changes in clinical outcome of RCC patients have been a grading (G) of 1 and 1–2. Advanced tumours were classified as attributed so far either to multimarker methylation panels pTX3 and/or lymph node positive (N þ ), positive for distant (Urakami et al, 2006) or to gene-specific CGI methylation of metastasis (M þ ) or G2–3 and G3. The time to disease progression genes such as basonuclin 1 (BNC1), collagen type XIV a 1 corresponds to the point in time patients demonstrating either a (COL14A1), DAL-1/4.1B, GATA-binding protein 5 (GATA5), local recurrence or a synchronous/metachronous metastasis as secreted frizzled-related protein 1 (SFRP1) and signal peptide detected by CT scan. Clinical and histopathological characteristics CUB domain EGF-like 3 (SCUBE3) (Yamada et al, 2006; Morris of tumour specimens are summarised in Table 1. et al, 2010; Peters et al, 2012). Interestingly, a subset of these candidate prognosticators were independent from other clinico- DNA isolation, bisulphite conversion of DNA and quality pathological parameters including DAL-1/4.1B (independent from measurements. DNA isolation from frozen sections and stage and grade of tumours) COL14A1 (size, grade and stage), histopathological examination of control sections were carried gremlin 1 (GREM CGI region 1, age and sex) and SFRP1 (size, out as described previously (Peters et al, 2012). Tumour tissue grade, stage, age and gender) (Yamada et al, 2006; Morris et al, sections represented 100% tumour tissue and included only minor 2010; van Vlodrop et al, 2010; Atschekzei et al, 2012). fractions of normal cells. Bisulphite conversion of DNA was Alteration of microRNA (miR) expression as a result of DNA performed as described before (Peters et al, 2007). To control methylation has recently also been described as a common bisulphite conversion efficiency, all samples were subjected to mechanism in human tumourgenesis (Esquela-Kerscher and quantitative methylation-specific PCRs (qMSP) using the concept Slack, 2006). Considering that miR’s may interact with a large reported by Campan et al (2009) and van der Horst et al (2004). number of target mRNAs leading to degradation and loss or Briefly, detection of the methylation-independent repetitive reduction of corresponding gene expression epialterations of miR, sequence ALU-C4 (QC1) and a single copy b-Actin (ACTB) CGIs may be of great potential relevance for the prediction of (QC2) sequence served to measure the yield of conversion and to cancer progression (Han et al, 2007; Selbach et al, 2008). In line provide for controlled input of bisulphite converted template DNA with this hypothesis, miR-9 methylation has already been identified into qMSP analyses, while the repetitive sequence ALU-YB8 (QC3) as an epigenetic mark of potential prognostic relevance for RCC. allows to quantify unconverted DNA. Fully methylated bisulphite Apart from that only miR-34a and miR-34b/c CGIs were found to converted control DNA (M) and unmethylated bisulphite be methylated in RCC as yet (Hildebrandt et al, 2010; Vogt et al, converted control DNA (U) were generated as described recently 2011). (Weisenberger et al, 2008). Here we asked, whether methylation of a CGI annotated to Quantitative methylation-specific PCR (qMSP) analysis. Methy- mir-124-3 and found to be relevant in many tumours (Kent et al, lation analyses of bisulphite-treated genomic DNA was performed 2002) can be also detected in kidney tumours and is of relevance by a quantitative real-time fluorimetric 50 exonuclease PCR assay. for the clinical outcome of patients with RCC. MiR-124 matures Our analyses determined a subregion within the CpG island (CGI) from three precursor variants located on chromosomes 8p23.1 annotated to the hsa-mir-124-3 locus on chromosome 20 as (mir-124-1), 8q12.3 (mir-124-2) and 20q13.33 (mir-124-3). Hyper- indicated in Figure 1A. The qMSP-specific primers 50-GGTCGGG methylation of miR-124 has been initially demonstrated to occur in TCGGGTTAGTAGG-30 (forward) and 50-CGCAAACCGACTAC colon, breast and lung cancers as well as leukaemia and lymphoma GAACCG-30 (reverse) as well as the Taqman probe 50-FAM-CCA (Lujambio et al, 2007) and seems to play an important role in CGAAATCCACGCTACAAACGCCCA-BHQ-3 were designed tumour development. However, a possible clinical relevance of according to Weisenberger et al (2005) and using the Beacon mir-124-3 methylation as independent prognosticator has been Designer software (PREMIER Biosoft, Palo Alto, CA, USA). reported so far only for haematological malignancies (Agirre et al, Real-time PCRs were carried out in duplicate on an ABI 7900HT 2009; Roman-Gomez et al, 2009), while corresponding information (Life Technologies, Foster City, CA, USA) in 384-well plates. for solid tumours in general as well as for RCC in particular are not Amplification included initial incubation for 10 min at 95 1C available. Here, we analysed whether methylation of the mir-124-3 followed by 45 cycles of denaturation for 15 s at 95 1C, 1 min locus is relevant for RCC and, moreover, whether changes are annealing and elongation at 60 1C in a reaction volume of 9 ml, associated with the clinicopathology and disease progression of consisting of 1.2 mM of each primer, 400 nM probe and 1 Â RCC patients. Taqman Universal Master Mix II no UNG. For PCR setup, the FasTrans automatic Liquid Handling System (AnalyticJena, Jena, Germany) was used. Measurements were carried out blinded MATERIALS AND METHODS against type, order and clinicopathological status of samples by the experimenter. Cell lines. Human kidney tumour cell lines and primary renal Calculation of relative methylation levels. The relative degree of proximal tubular epithelial cells (RPTEC) were cultured not longer methylation was determined using the method of Weisenberger as 2 months following purchase and identity control by the et al (2005). Sample-specific total input of converted DNA in each manufacturer as described previously (Peters et al, 2012). Human sample was measured by detection by means of QC1 while breast cancer and cervical cancer cell line DNAs were a kind gift by calibrator samples M and U served as positive and negative Thilo Do¨rk (Department of Gynaecology & Obstetrics, Hannover controls, in each run. Relative methylation values were calculated Medical School, Hannover, Germany). according to the 2 À DDCT method (Livak and Schmittgen, 2001).

132 www.bjcancer.com | DOI:10.1038/bjc.2012.537 Mir-124-3 methylation and RCC prognosis BRITISH JOURNAL OF CANCER

methylated diploid genome. Hence, both potential cancer-related Table 1. Tumour patients characteristics alterations such as regional amplification of DNA as well as slight All ccRCC surv. variation in measured Ct values may lead to the measurement of % ccRCC % % 4 RCC group methylation values of 100%. Therefore, the methylation values were designated as ‘relative methylation level’. Samples exhibiting Total cases 111 80 37 lower template DNA input than a QC1 Ct value of 21.1 were Histology excluded. The analytical sensitivity of the mir-124-3 assay was estimated calculating the limit of quantitative detection (LQ) from ccRCC 80 72.1 80 100 37 100 the variance measurements of blank values as suggested by Currie papRCC 23 20.7 0 0 0 0 (1968). First the non-methylated control U was measured 12 times Chrom. 43.60 0 0 0 and all runs were found to be undetermined. To consider the RCC variance of ‘blank values’ present within sample group, L was Not class. 4 3.6 0 0 0 0 Q then calculated by addition of the mean of background methylation Sex levels as determined from 153 undetermined samples and the corresponding 10-fold standard deviation. Female 38 34.2 30 37.5 15 40.5 Male 73 65.8 50 62.5 22 59.5 Statistical analyses. Explorative statistical data analyses were conducted by the use of the statistical software R 2.12 Age (R Development Core Team, 2011). Po0.05 were considered to Median 65 64 65 indicate statistical significance. Relative methylation levels are (years) converted to the natural logarithmic scale before conducting further statistical calculations. Linear regression analysis was Distant metastasis performed for linearity and PCR efficiency measurements of the M0 86 77.5 60 75 28 75.7 qMSP assay. Statistical comparison of the mean relative methyla- M þ 25 22.5 20 25 9 24.3 tion levels observed for paired tumour and adjacent normal appearing tissue were carried out using the paired t-test. Lymph node metastasis For group comparisons of independent tissue samples representing N0 98 88.3 73 91.3 35 94.6 different clinicopathological classifications, univariate logistic N þ 13 11.7 7 8.8 2 5.4 regression analysis was conducted, providing both statistical significance and odds ratio (OR) serving as a measure of the T-classification observed effect size. Analysis of recurrence-free survival was achieved using Cox’s proportional-hazards regression model. pT1 11 9.9 8 10 1 2.7 pT1a 32 28.8 21 26.3 12 32.4 The grade and state of lymph node metastasis were not considered pT1b 19 17.1 14 17.5 7 18.9 in bivariate analyses due to low number of cases in subgroups pT2 7 6.3 5 6.3 2 5.4 following dichotomization. Note that G and N were part of the pT3 5 4.5 2 2.5 1 2.7 localised and advanced disease classifications (see above). pT3a 9 8.1 7 8.8 2 5.4 Optimum threshold calculations for dichotomization of methyla- pT3b/c 23 20.7 21 26.3 11 29.7 tion levels were performed using the R package ‘maxstat’. pT4 1 0.9 0 0 0 0 NA 4 3.6 2 2.5 1 2.7

Differentiation RESULTS

G1 22 19.8 19 23.8 5 13.5 Measurement of technical controls. We analysed the specificity G1–2 14 12.6 9 11.3 4 10.8 of the mir-124-3 qMSP analysis by duplicate measurements of G2 56 50.5 38 47.5 21 56.8 converted methylated (M), converted non-methylated (U) and G2–3 8 7.2 4 5 2 5.4 non-converted DNA control samples. We exclusively observed Ct G3 11 9.9 10 12.5 5 13.5 values of 45 (undetermined) for the U and non-converted DNA State of disease samples, while the M sample demonstrated Ct values of about 29 (Figure 1B). Non-converted DNA was detected neither in QC1 Loc. 59 53.2 39 48.8 17 45.9 a control PCR nor in mir-124-3-specific qMSP, thus demonstrating disease that only methylated and converted DNA has been measured by Adv. 51 45.9 41 51.3 20 54.1 b the mir-124-3 methylation assay. To determine PCR efficiency and disease linearity of the methylation detection assay, we analysed a two-fold NA 1 0.9 0 0 0 0 dilution series of the M control within the U control DNA Paired samples adjusting for constant total converted DNA input. Linear regression analysis revealed a slope of DCT ¼À3.3 per 10-fold All RCC 77 dilution and a coefficient of correlation of r ¼ 0.97 (P ¼ 0.001), ccRCC 58 indicating a high efficiency and linearity of the assay (Figure 1C). Abbreviation: ccRCC ¼ clear cell renal cell carcinoma. Estimation of the analytical sensitivity LQ of the mir-124-3 assay a pTp2, N0, M0 and G1 þ G1–2. according revealed a relative methylation level of 5.7E À 4 (mean of b X þ þ pT 3 and/or N ,M or G2–3 G3. blank values ¼ 3.0E À 5, s.d. ¼ 5.3E À 5). Briefly, Ct values of samples were normalised to QC1 and mir-124-3 methylation in tumour cell lines and normal primary referenced to the positive control M, allowing calculation of DDCt cells. We first evaluated whether the mir-124-3 qMSP assay is and the relative methylation. Notably, the resulting methylation capable of detecting methylation in cell lines used as surrogates for values are related to the copy number of methylated sequences tumours with known methylation (breast cancer) as well as for detected in the calibrator sample corresponding to a fully normal tissues and tissues of both localised and metastatic human www.bjcancer.com | DOI:10.1038/bjc.2012.537 133 BRITISH JOURNAL OF CANCER Mir-124-3 methylation and RCC prognosis

Chromosome 20p13.33

61 807 000 61 808 000 61 809 000 61 810 000

CpG island

qMSP mir-124-3 TSS

10 QC1: internal 20 R = −0.97 8 Control 1,2 19 Slope = –3.3 6 3 18 Delta Rn 4 4,5 17 10 mir-124-3 16

8 Delta Ctvalue 1,2 15 6

Delta Rn 3,4,5 14 4

10 20 30 40 0.03 0.1 0.3 1 Cycle number Relative methylation

Figure 1. (A) Structure of the mir-124-3 CGI and location of the qMSP assay relative to the mir-124-3 transcription start site (TSS). Vertical lines represent CpG sites within the CGI. Chromosomal positions refer to the GRCh37/hg19 annotation in the UCSC genome browser (Lander et al, 2001; Kent et al, 2002). (B) Exemplary primary data of quantitative methylation-specific and control PCR measurements in duplicate for A498 (1); methylated control DNA (2); unmethylated control DNA (3); unconverted DNA (4) and a blank control (5). (C) Normalised mir-124-3 assay threshold values (Ct) for a two-fold dilution series of the methylated control in non-methylated control DNA for determination of assay linearity and efficiency. cancers of other origin (kidney, prostate and bladder cancer; Prec Figure 2). As expected, methylation could be detected in five out of Rptec Primary cells eight (63%) breast cancer cell lines. Interestingly, five of six (83%) of the renal cancer cell lines showed high relative methylation RCC−MF levels ranging from 0.006 to 1.14, while normal primary cells from RCC−HS RCC−GS kidney (RPTEC) and prostate (PreC) exhibited undetectable or low RCC methylation, respectively. Notably, three of four (75%) bladder 786 A498 cancer cell lines were identified to show high relative methylation ACHN levels (Figure 2). PC3 mir-124-3 is hypermethylated in RCC. Analysis of relative DU145 ProstateCa methylation levels in paired tumour and adjacent normal tissues LN−cap regardless of histology revealed higher values for the tumour tissues (mean ¼ 6.5E À 4) compared with normal tissue samples EJ28 (mean ¼ 2.5E À 5; Po0.0001; paired t-test), demonstrating relative HB−CLS2 BladderCa methylation levels in the range of the blank values. The CLS439 corresponding analysis of the clear cell histology subset of tissues RT112 disclosed also higher relative methylation for neoplastic tissues MDA−MB231 ¼ À À (mean 7.1E 4 compared with paired normal tissues (2.5E 5; MCF−10a Po0.0001, paired t-test; Figures 3A and B). In view of previous MCF−7 results obtained by others and describing that methylation found in HCC−1937 tumour tissues may be related with their normal counterparts and HCC−1806 therefore supposed to be indicative of an epigenetic risk for tumour HCC−1599 BreastCa development, we additionally carried out a correlation analysis for HCC−1395 the paired tissue samples. As a result, we found a weak but HCC−38 significant statistical association (r ¼ 0.29, P ¼ 0.026, Pearson’s correlation analysis). 0.0 0.5 1.0 1.5 Relative methylation level mir-124-3 methylation and statistical association with clinicopathologic parameters. Analysis of all RCC samples revealed a Figure 2. Relative methylation levels measured for cancer cell lines and significant higher relative methylation levels for tumours with clear normal primary cells. cell histology (mean ¼ 7.1E À 4) compared with papillary tumours (mean ¼ 6.5E À 5, P ¼ 0.011; OR ¼ 0.843; 95% CI ¼ 0.74–0.97). Therefore, further statistical analyses were exclusively carried out metastasis (Po0.0001), state of localised vs advanced disease for the ccRCC tissue subset. As a result, we identified significant (Po0.0001) and high- (G2–3, G3) or low-grade (G1, G1–2, G2) differences in methylation levels for the parameters state of distant tumour (P ¼ 0.0063; Figure 4A; Table 2). Methylation of ccRCCs

134 www.bjcancer.com | DOI:10.1038/bjc.2012.537 Mir-124-3 methylation and RCC prognosis BRITISH JOURNAL OF CANCER

0 10 −2

5 −6 lnRML −10

Diff. T - pNT (lnRML) 0

−14

pNT T

Figure 3. Analysis of paired normal appearing and tumour tissues for the ccRCC subgroup. (A) Assorted pairwise differences for natural logarithms of relative methylation levels (lnRML) observed in tumour (T) and paired normal appearing tissues (pNT). (B) Direct comparison of natural logarithms of relative methylation levels for paired normal appearing (pNT) and tumour (T) samples.

−5

−10 Relative methylation level (ln) −15 Kernel distribution M0 M+ N0 N+ G low G high Loc. Adv.

1.0 1.0 0 Cutoff 1 Cutoff 2 Cutoff 1 0.8 0.8

−5 0.6 0.6 Cutoff 2 0.4 0.4 −10 0.2 0.2 Fraction recurrence-free surv. Relative methylation level (ln) −15 0.0 0.0 Kernel distribution 010 30 50 70 010305070 Recurrence-free survival (months)

Figure 4. (A) Distribution of methylation values detected in ccRCC and box plot analysis for subset-specific relative methylation levels of clinicopathological parameters; negative (M0) or positive (M þ ) for distant metastasis, negative (N0) or positive (N þ ) for lymph node metastasis, low (Go ¼ 2) or high (G42) grade tumours and localised (pTX2, N0, M0 and G1 þ G1–2) or advanced (pTX3 and/or N þ ,Mþ or G2–3 G3) disease. Bold horizontal lines show group medians; boxes and whiskers show the 25 and 75% and 10 and 90% quartiles, respectively. (B) Distribution of methylation values detected in the ccRCC survival subgroup and location of the statistical optimum (cutoff 1) or the limit of quantitative detection LQ (cutoff 2) cutoff values. (C) Kaplan–Meier plots showing relative survival of patients dichotomised by use of cutoff 1 or cutoff 2 values as indicated in (B). Note that some of the symbols displaying censored cases coincide due to similar follow-up periods of patients.

was not statistically associated with sex, age and lymph node identified the methylation status to be associated with increased metastasis status but showed a trend for the tumour diameter risk for disease recurrence (P ¼ 0.0005; hazard ratio (HR) ¼ 9.37; (P ¼ 0.062). CI ¼ 2.68–32.8; Table 3). The corresponding Kaplan–Meier analysis showed that in this case four out of four patients (100%) mir-124-3 methylation is associated with worse recurrence-free were identified with disease recurrence within o30 months and survival of patients. To assess whether mir-124-3 methylation is methylation above the cut point (Figure 4B,C). In contrast, 7 of 33 associated with the clinical course of patients with ccRCC, we first (21%) were detected with low methylation and disease recurrence performed a Cox regression analysis using a statistically calculated within the same period. Comparison of HRs from univariate optimum cutoff value corresponding to a relative methylation level Cox regression analysis exhibited the values 9.37, 4.86 and 4.28 of 0.0015 for dichotomization of patients (Figure 4B). We for the parameters methylation, metastatic state or advanced www.bjcancer.com | DOI:10.1038/bjc.2012.537 135 BRITISH JOURNAL OF CANCER Mir-124-3 methylation and RCC prognosis

Table 2. Statistical association of mir-124-3 CGI methylation with clinicopathological parameters of ccRCC patients

RMLa

Median Median P-valueb OR 95% CI

ccRCC

Dist. metastasis (M0/M þ ) 5.5E–5 0.03 0.0010 1.30 1.11–1.51 Lymph node met. (N0/N þ ) 1.0E–3 0.01 0.2253 1.13 0.93–1.39 Grade (low/high) 6.5E–5 0.03 0.0063 1.28 1.07–1.52 Diameterc 3.4E–5 3.8E–3 0.0620 1.14 0.99–1.30 Localised/advanced dis. 1.7E–5 0.01 o0.0001 1.36 1.19–1.57

Abbreviations: OR ¼ odds ratio; CI ¼ confidence interval; ccRCC ¼ clear cell renal cell cancer; CGI ¼ CpG island. aRelative methylation level. bUnivariate logistic regression. cDichotomised using median as cut point. Bold numbers indicate statistical significance (Po0.05).

Table 3. Univariate statistical association of mir-124-3 CGI methylation Table 4. Association of mir-124-3 CGI methylation and and clinicopathological parameters with recurrence-free survival of clinicopathological parameters with recurrence-free survival in bivariate patients survival analysis

P-valuea HR 95% CI P-valuea HR 95% CI mir-124-3 methylation 0.0005 9.37 2.68–32.8 mir-124-3 methylation 0.0003 13 3.3–51.7 0.0017 Metastasis 0.0032 4.86 1.7–14 Metastasis 6.05 1.9–18.6 mir-124-3 0.0081 Lymph node status 0.8751 1.18 0.15–9.02 methylation 5.87 1.6–21.8 Localised/advanced 0.1055 3.06 0.8–11.8 Localised/advanced disease 0.0260 4.28 1.19–15.4 mir-124-3 methylation 0.0004 10.90 2.9–40.3 Gender 0.2080 2.11 0.66–32.8 Gender 0.1410 2.45 0.74–8.07 Ageb 0.7058 0.82 0.28–2.35 mir-124-3 methylation 0.0005 18.20 3.6–92.4 Diameterb 0.4278 1.67 0.47–5.92 Ageb 0.1429 0.35 0.9–1.14

Abbreviations: CI ¼ confidence interval; HR ¼ hazard ratio; CGI ¼ CpG island. mir-124-3 methylation 0.0006 18.20 3.45–95.5 aUnivariate Cox regression analysis. Diameterb 0.1986 2.46 0.6–9.7 bDichotomisation by the median of parameter. Abbreviations: CI ¼ confidence interval; HR ¼ hazard ratio; CGI ¼ CpG island. Bold numbers indicate statistical significance (Po0.05). aBivariate Cox regression analyses. bDichotomised by the use of the median of parameter values. Bold numbers indicate statistical significance (Po0.05). disease (Table 3). To investigate whether the association of methylation status and recurrence-free survival could be a statistically independent parameter, we performed pairwise bivariate Cox proportional-hazards regression considering the DISCUSSION clinicopathological parameters distant metastasis, state of localised or advanced disease, gender, age and tumour diameter as Our analyses demonstrated that increased methylation levels of a covariates. Our analysis revealed that methylation dichotomised subregion in the mir-124-3 CGI are associated with adverse with the optimised cutoff value remained as a significant parameter clinicopathological parameters of ccRCC patients including in bivariate Cox regression models exhibiting HR values of metastasis, grade, state of advanced disease and greater tumour 5.87–18.2 and P-values between 0.0003 and 0.008 (Table 4). From diameter. Our data, therefore suggest that this epialteration could the clinicopathological parameters, only distant metastasis remained be involved in cancer progression in patients with ccRCC. This as a significant factor in the bivariate Cox regression models. view is supported by our findings that methylation is significantly Considering that statistical optimisation revealed a small subset associated with recurrence-free survival in univariate and in of patients at higher risk for disease recurrence, we asked whether bivariate analyses as well. Association of mir-124-3 CGI methyla- lowering the cutoff value thus increasing the number of positive tion and recurrence-free survival in solid tumours, to our tumours, could give also prognostic information for a greater knowledge, has not been described as yet. Previous studies number of patients. We chose LQ as the lowest estimated reported a shortened disease-free survival for patients with acute methylation level that can be reliably quantitatively measured for lymphoblastic leukaemia (ALL) using a multi-miR loci approach dichotomization, hence including the maximum number of including also mir-124-3 (Roman-Gomez et al, 2009). Methylation patients into the risk group (Figure 4B and C). As a result, we of the mir-124-3 locus was then identified as an independent found that these patients also exhibited a worse recurrence-free prognosticator for both disease-free and overall survival in ALL survival in univariate (P ¼ 0.025; HR ¼ 4.32; 95% CI ¼ 1.2–15.5) (Agirre et al, 2009). Contribution of mir-124-3 epigenetics to but not in the pairwise bivariate Cox regression analyses. The tumour progression is also indicated by gradually increasing corresponding Kaplan–Meier plot showed that within a follow-up methylation levels found in a series of premalignant and malignant period of 70 months only 4 (24%) of 17 patients from the low cervical tissues of differing biological aggressiveness (Wilting et al, methylation group, but 11 (55%) of 20 patients with higher 2010). Although DNA hypermethylation is a frequently observed methylation demonstrated disease recurrence (Figure 4C, cutoff 2). characteristic in RCC (Baldewijns et al, 2008) only few loci

136 www.bjcancer.com | DOI:10.1038/bjc.2012.537 Mir-124-3 methylation and RCC prognosis BRITISH JOURNAL OF CANCER currently including COL14A1 (Morris et al, 2010), GREM1 (van in line with results of recent analyses of haematological Vlodrop et al, 2010), SFRP1 (Atschekzei et al, 2012) and DAL-1/ malignancies. Considering that first functional analyses in 4.1B (Yamada et al, 2006), were associated with RCC prognosis, Lujambio et al (2007) described mir-124-3 methylation to effect provided clinicopathological parameters have been considered as cellular proliferation and chromosome stability via cyclin D kinase covariates in multi- or bivariate analyses. In view of these results, it 6(CDK6), our results may point to the existence of comparable seems remarkable that mir-124-3 methylation contributed high and mechanisms also taking place in RCC development. Some evidence fairly independent hazards for disease recurrence in our bivariate for the relevance of CDK6 signalling in RCC development is given analyses accounting for the states of distant metastasis or localised/ by public mRNA expression data showing a significant increase of advanced disease. So, methylation-related risk values exceed those CDK6 mRNA levels in ccRCC tissues when compared with normal obtained for metastasis, thus suggesting mir-124-3 methylation as a paired samples (P ¼ 0.001; paired t-test; data from Lenburg et al, promising candidate for an independent prognosticator. However, 2003). This would be consistent with a hypothesised silencing of although a high sensitivity of 100% for detection of disease mir-124-3 expression due to CGI methylation. Moreover, from a recurrence has been observed, only a fraction of patients kinase silencing screen in RCC cell lines deficient for VHL, CDK6 subsequently demonstrating disease recurrence would have been has been identified as one of the targets to be functionally identified at the time of surgery. Interestingly, making use of the associated with the viability of kidney cancer cells and, in addition, second cutoff value, a larger group of patients of B38% might be can be affected by a small molecule inhibitor (Bommi-Reddy et al, identified with a low risk of disease recurrence. We therefore 2008). Hence, our clinical analysis emphasises the need for further conclude that mir-124-3 methylation should be further evaluated in functional analysis of the CDK6 signalling in RCC. In case of that prospective studies using independent and enlarged patients methylation-dependent silencing of the mir-124-3 expression can groups. Moreover, it would be of interest whether this potential be shown to be associated with concurrent activation of CDK6 biomarker gives prognostic information that is independent from signalling, the measurement of mir-124-3 methylation would other DNA methylation-based potential biomarkers or can be even provide a rationale for treatment of such RCCs with so far improved by generation of combined methylation markers. experimental small molecule CDK6 inhibitors. In conclusion, Our methylation analyses using cell lines of different tumour mir-124-3 CGI is hypermethylated in ccRCC and associated with entities showed that mir-124-3 methylation likely also occurs in poorer clinical outcome. Our results emphasise the need for further bladder cancers, a tumour that has not been reported as yet to functional characterisation of mir-124-3 alterations in clear cell show this epigenetic mark. In contrast, none of the cell lines RCC such as CDK6 signalling. Methylation of the mir-124-3 locus deriving from prostatic metastases demonstrated a higher degree of represents a candidate biomarker for further prospective molecular methylation. stratification of patients possibly for targeted therapy using small The methylation analysis of the mir-124-3 locus overall revealed molecule effectors. apparently relative low methylation values in tumours. Considering that histopathological assessment of control sections confirmed that tissues submitted to DNA extraction demonstrated a tumour REFERENCES cell content of at least 50%, both the absolute levels of mean methylation values and the measured increases of relative Agirre X, Vilas-Zornoza A, Jimenez-Velasco A, Martin-Subero JI, Cordeu L, methylation levels in metastasised or advanced tumours cannot Garate L, San Jose-Eneriz E, Abizanda G, Rodriguez-Otero P, Fortes P, be attributed to sample-specific variation of tumour cell propor- Rifon J, Bandres E, Calasanz MJ, Martin V, Heiniger A, Torres A, tions. Comparable low mean methylation values, on the other Siebert R, Roman-Gomez J, Prosper F (2009) Epigenetic silencing of the hand, may be a consequence of the qMSP method used for tumor suppressor microRNA Hsa-miR-124a regulates CDK6 expression methylation analysis when taking into account that only sequences and confers a poor prognosis in acute lymphoblastic leukemia. Cancer Res exhibiting a uniformly dense methylation are detected and intra- 69(10): 4443–4453. CGI heterogeneity of methylation may occur in principle as has Ando T, Yoshida T, Enomoto S, Asada K, Tatematsu M, Ichinose M, already been reported for another gene (van Vlodrop et al, 2010). Sugiyama T, Ushijima T (2009) DNA methylation of microRNA genes in gastric mucosae of gastric cancer patients: its possible involvement On the other hand, our methylation values may also reflect that in the formation of epigenetic field defect. Int J Cancer 124(10): only a subpopulation of tumour cells carries densely methylated 2367–2374. miR-124-3 loci. A limitation of our study is therefore, that neither Atschekzei F, Hennenlotter J, Janisch S, Grosshennig A, Trankenschuh W, explanation can be supported as yet. While our statistical findings Waalkes S, Peters I, Dork T, Merseburger AS, Stenzl A, Kuczyk MA, clearly suggest miR-124-3 methylation as a marker for biologically Serth J (2012) SFRP1 CpG island methylation locus is associated with aggressive tumours, it remains to be clarified whether it represents renal cell cancer susceptibility and disease recurrence. Epigenetics 7(5): an early or late event in the carcinogenesis of RCC. 447–457. A relative increase in methylation levels of the mir-124-3 CGI Baldewijns MM, van Vlodrop IJ, Schouten LJ, Soetekouw PM, de Bruine AP, has also been reported for non-cancerous gastric tissues of van Engeland M (2008) Genetics and epigenetics of renal cell cancer. Helicobacter pylori-positive individuals when compared with Biochim Biophys Acta 1785(2): 133–155. corresponding normal tissue samples of H. pylori-negative patients Bommi-Reddy A, Almeciga I, Sawyer J, Geisen C, Li W, Harlow E, Kaelin Jr (Ando et al, 2009). This correlation with a known risk factor for WG, Grueneberg DA (2008) Kinase requirements in human cells: III. gastric cancer led to the hypothesis that mir-124-3 methylation also Altered kinase requirements in VHL-/- cancer cells detected in a pilot could contribute for cancer risk. Both, analyses of methylation in synthetic lethal screen. Proc Natl Acad Sci USA 105(43): 16484–16489. paired precancerous and cancer tissues indicating epigenetic Campan M, Weisenberger DJ, Trinh B, Laird PW (2009) MethyLight. inheritance of the methylation status into tumour tissue (Shen Methods Mol Biol 507: 325–337. Currie LA (1968) Limits for qualitative detection and quantitative et al, 2005) as well as a direct risk estimation by a case–control determination. Application to radiochemistry. Anal Chem 40(3): 586–593. study (Ando et al, 2009; Atschekzei et al, 2012) have been applied Dreijerink K, Braga E, Kuzmin I, Geil L, Duh FM, Angeloni D, Zbar B, to assess epigenetic cancer risk contribution for other loci. Thus, Lerman MI, Stanbridge EJ, Minna JD, Protopopov A, Li J, Kashuba V, our observation of a weak statistical correlation of mir-124-3 Klein G, Zabarovsky ER (2001) The candidate tumor suppressor gene, methylation in normal appearing and paired tumour tissues RASSF1A, from human chromosome 3p21.3 is involved in kidney suggest further appropriate risk studies. tumorigenesis. Proc Natl Acad Sci USA 98(13): 7504–7509. Our analyses suggest mir-124-3 methylation as a biomarker for Esquela-Kerscher A, Slack FJ (2006) Oncomirs - microRNAs with a role in independent identification of a subset of high-risk patients and are cancer. Nat Rev Cancer 6(4): 259–269. www.bjcancer.com | DOI:10.1038/bjc.2012.537 137 BRITISH JOURNAL OF CANCER Mir-124-3 methylation and RCC prognosis

Han L, Witmer PD, Casey E, Valle D, Sukumar S (2007) DNA methylation Shen L, Kondo Y, Rosner GL, Xiao L, Hernandez NS, Vilaythong J, regulates MicroRNA expression. Cancer Biol Ther 6(8): 1284–1288. Houlihan PS, Krouse RS, Prasad AR, Einspahr JG, Buckmeier J, Hildebrandt MA, Gu J, Lin J, Ye Y, Tan W, Tamboli P, Wood CG, Wu X Alberts DS, Hamilton SR, Issa JP (2005) MGMT promoter methylation (2010) Hsa-miR-9 methylation status is associated with cancer and field defect in sporadic colorectal cancer. J Natl Cancer Inst 97(18): development and metastatic recurrence in patients with clear cell renal cell 1330–1338. carcinoma. Oncogene 29(42): 5724–5728. Sobin LH, Compton CC (2010) TNM seventh editionwhat’s new, what’s Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D (2011) Global changed: communication from the International Union Against Cancer cancer statistics. CA Cancer J Clin 61(2): 69–90. and the American Joint Committee on Cancer. Cancer 116(22): Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, 5336–5339. Haussler D (2002) The human genome browser at UCSC. Genome Res Thompson M (2009) Polybromo-1: the chromatin targeting subunit of the 12(6): 996–1006. PBAF complex. Biochimie 91(3): 309–319. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, Devon K, Urakami S, Shiina H, Enokida H, Hirata H, Kawamoto K, Kawakami T, Dewar K, Doyle M, FitzHugh W et al. (2001) Initial sequencing and Kikuno N, Tanaka Y, Majid S, Nakagawa M, Igawa M, Dahiya R (2006) analysis of the human genome. Nature 409(6822): 860–921. Wnt antagonist family genes as biomarkers for diagnosis, staging, and Lenburg ME, Liou LS, Gerry NP, Frampton GM, Cohen HT, Christman MF prognosis of renal cell carcinoma using tumor and serum DNA. Clin (2003) Previously unidentified changes in renal cell carcinoma gene Cancer Res 12(23): 6989–6997. expression identified by parametric analysis of microarray data. BMC van der Horst EH, Leupold JH, Schubbert R, Ullrich A, Allgayer H (2004) Cancer 3: 31. TaqMan-based quantification of invasive cells in the chick embryo Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data metastasis assay. Biotechniques 37(6): 940–942, 944, 946. using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. van Vlodrop IJ, Baldewijns MM, Smits KM, Schouten LJ, van Neste L, van Methods 25(4): 402–408. Criekinge W, van Poppel H, Lerut E, Schuebel KE, Ahuja N, Herman JG, Lujambio A, Ropero S, Ballestar E, Fraga MF, Cerrato C, Setien F, Casado S, de Bruine AP, van Engeland M (2010) Prognostic significance of Gremlin1 Suarez-Gauthier A, Sanchez-Cespedes M, Git A, Spiteri I, Das PP, (GREM1) promoter CpG island hypermethylation in clear cell renal cell Caldas C, Miska E, Esteller M (2007) Genetic unmasking of an carcinoma. Am J Pathol 176(2): 575–584. epigenetically silenced microRNA in human cancer cells. Cancer Res Varela I, Tarpey P, Raine K, Huang D, Ong CK, Stephens P, Davies H, 67(4): 1424–1429. Jones D, Lin ML, Teague J, Bignell G, Butler A, Cho J, Dalgliesh GL, Morris MR, Gentle D, Abdulrahman M, Clarke N, Brown M, Kishida T, Yao Galappaththige D, Greenman C, Hardy C, Jia M, Latimer C, Lau KW, M, Teh BT, Latif F, Maher ER (2008) Functional epigenomics approach to Marshall J, McLaren S, Menzies A, Mudie L, Stebbings L, Largaespada DA, identify methylated candidate tumour suppressor genes in renal cell Wessels LF, Richard S, Kahnoski RJ, Anema J, Tuveson DA, Perez- carcinoma. Br J Cancer 98(2): 496–501. Mancera PA, Mustonen V, Fischer A, Adams DJ, Rust A, Chan-on W, Morris MR, Ricketts C, Gentle D, Abdulrahman M, Clarke N, Brown M, Subimerb C, Dykema K, Furge K, Campbell PJ, Teh BT, Stratton MR, Kishida T, Yao M, Latif F, Maher ER (2010) Identification of candidate Futreal PA (2011) Exome sequencing identifies frequent mutation of the tumour suppressor genes frequently methylated in renal cell carcinoma. SWI/SNF complex gene PBRM1 in renal carcinoma. Nature 469(7331): Oncogene 29(14): 2104–2117. 539–542. MorrisMR,RickettsCJ,GentleD,McRonaldF,CarliN,KhaliliH,BrownM, Vogt M, Munding J, Gruner M, Liffers ST, Verdoodt B, Hauk J, Steinstraesser Kishida T, Yao M, Banks RE, Clarke N, Latif F, Maher ER (2011) Genome- L, Tannapfel A, Hermeking H (2011) Frequent concomitant inactivation wide methylation analysis identifies epigenetically inactivated candidate of miR-34a and miR-34b/c by CpG methylation in colorectal, pancreatic, tumour suppressor genes in renal cell carcinoma. Oncogene 30(12): 1390–1401. mammary, ovarian, urothelial, and renal cell carcinomas and soft tissue Nordstrom-O’Brien M, van der Luijt RB, van Rooijen E, van den Ouweland sarcomas. Virchows Arch 458(3): 313–322. AM, Majoor-Krakauer DF, Lolkema MP, van Brussel A, Voest EE, Giles Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, Fiala E, Ehrlich M, RH (2010) Genetic analysis of von Hippel-Lindau disease. Hum Mutat Laird PW (2005) Analysis of repetitive element DNA methylation by 31(5): 521–537. MethyLight. Nucleic Acids Res 33(21): 6823–6836. Peters I, Eggers H, Atschekzei F, Hennenlotter J, Waalkes S, Trankenschuh W, Weisenberger DJ, Trinh BN, Campan M, Sharma S, Long TI, Grosshennig A, Merseburger AS, Stenzl A, Kuczyk MA, Serth J (2012) Ananthnarayan S, Liang G, Esteva FJ, Hortobagyi GN, McCormick F, GATA5 CpG island methylation in renal cell cancer: a potential biomarker Jones PA, Laird PW (2008) DNA methylation analysis by digital bisulfite for metastasis and disease progression. BJU Int 110(Pt 2): E144–E152. genomic sequencing and digital MethyLight. Nucleic Acids Res 36(14): Peters I, Rehmet K, Wilke N, Kuczyk MA, Hennenlotter J, Eilers T, Machtens 4689–4698. S, Jonas U, Serth J (2007) RASSF1A promoter methylation and expression Wilting SM, van Boerdonk RA, Henken FE, Meijer CJ, Diosdado B, Meijer analysis in normal and neoplastic kidney indicates a role in early GA, le Sage C, Agami R, Snijders PJ, Steenbergen RD (2010) Methylation- tumorigenesis. Mol Cancer 6: 49. mediated silencing and tumour suppressive function of hsa-miR-124 in R Development Core Team (2011) R: A language and environment for cervical cancer. Mol Cancer 9: 167. statistical computingR Foundation for Statistical Computing: Vienna, Yamada D, Kikuchi S, Williams YN, Sakurai-Yageta M, Masuda M, Austria, ISBN 3-900051-07-0, URL http://www.R-project.org/. Maruyama T, Tomita K, Gutmann DH, Kakizoe T, Kitamura T, Kanai Y, Reisman D, Glaros S, Thompson EA (2009) The SWI/SNF complex and Murakami Y (2006) Promoter hypermethylation of the potential tumour cancer. Oncogene 28(14): 1653–1668. suppressor DAL-1/4.1B gene in renal clear cell carcinoma. Int J Cancer Roman-Gomez J, Agirre X, Jimenez-Velasco A, Arqueros V, Vilas-Zornoza A, 118(4): 916–923. Rodriguez-Otero P, Martin-Subero I, Garate L, Cordeu L, San Jose-Eneriz E, Martin V, Castillejo JA, Bandres E, Calasanz MJ, Siebert R, Heiniger A, Torres A, Prosper F (2009) Epigenetic regulation of microRNAs in acute This work is published under the standard license to publish agree- lymphoblastic leukemia. J Clin Oncol 27(8): 1316–1322. ment. After 12 months the work will become freely available and Selbach M, Schwanhausser B, Thierfelder N, Fang Z, Khanin R, Rajewsky N (2008) Widespread changes in protein synthesis induced by microRNAs. the license terms will switch to a Creative Commons Attribution- Nature 455(7209): 58–63. NonCommercial-Share Alike 3.0 Unported License.

138 www.bjcancer.com | DOI:10.1038/bjc.2012.537 Peters et al. BMC Cancer 2014, 14:101 http://www.biomedcentral.com/1471-2407/14/101

RESEARCHARTICLE Open Access Decreased GATA5 mRNA expression associates with CpG island methylation and shortened recurrence-free survival in clear cell renal cell carcinoma Inga Peters1, Natalia Dubrowinskaja1, Michael Kogosov1, Mahmoud Abbas2, Jörg Hennenlotter3, Christoph von Klot1, Axel S Merseburger1,ArnulfStenzl3, Ralph Scherer4, Markus A Kuczyk1 and Jürgen Serth1*

Abstract Background: GATA-5, a zinc-finger transcription factor and member of the GATA family proteins 1–6, is known to be involved in cellular differentiation. We recently found that tumor-specific hypermethylation of the GATA5 CpG island (CGI) occurs in renal cell carcinoma (RCC) and is associated with an adverse clinical outcome. In this study, we investigated whether epigenetic GATA5 alterations may result in changes in GATA5 mRNA expression levels and correlate with the observed prognostic impact of epigenetic changes in GATA5 in RCC. Methods: Quantitative real-time reverse-transcribed polymerase chain reaction was applied to measure relative GATA5 mRNA expression levels in 135 kidney tissue samples, including 77 clear cell RCC (ccRCC) tissues and 58 paired adjacent normal renal tissue samples. Relative GATA5 expression levels were determined using the ΔΔCt method and detection of three endogenous control genes then compared to previously measured values of relative methylation. Results: The mean relative GATA5 mRNA expression level exhibited an approximately 31-fold reduction in tumor specimens compared with corresponding normal tissues (p < 0.001, paired t-test). Decreased GATA5 mRNA expression was inversely correlated with increased GATA5 CGI methylation (p < 0.001) and was associated with shortened recurrence-free survival in ccRCC patients (p = 0.023, hazard ratio = 0.25). Conclusion: GATA5 mRNA expression is decreased in ccRCC, likely due to gene silencing by methylation of the GATA5 CGI. Moreover, reduced GATA5 mRNA levels were associated with a poor clinical outcome, indicating a possible role of GATA5 for the development of aggressive ccRCC phenotypes. Keywords: GATA5, Renal cell carcinoma, mRNA, Prognosis, DNA methylation

Background As a member of the GATA family of transcription fac- Renal cell carcinoma (RCC) is one of the top ten causes tors, GATA5 is known to be functionally involved in cel- of cancer deaths in industrialized countries, and its inci- lular lineage and cell differentiation during embryonic dence has consistently increased during the past decades development of the heart, lung, urogenital tract, and gut [1]. Clear cell renal cell carcinomas (ccRCCs) are the epithelium [2]. Altered expression of GATA5 has been most frequently occurring histological entity, comprising associated with intestinal epithelial cell differentiation approximately 75% of all RCC. [3]. GATA5 is assumed to be a selective transcriptional regulator of mucin genes in gastrointestinal tissues [4], and regulates the promoter of the sodium-hydrogen ex- * Correspondence: [email protected] changer isoform 3 that is expressed in intestinal and 1Department of Urology and Urologic Oncology, Hannover Medical School, renal epithelium via Sp-family transcription factors [5]. Carl-Neuberg-Str.1, Hannover 30625, Germany Full list of author information is available at the end of the article

© 2014 Peters et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. Peters et al. BMC Cancer 2014, 14:101 Page 2 of 6 http://www.biomedcentral.com/1471-2407/14/101

Of note, a previous study found GATA5 hypermethyla- Table 1 Clinicopathological parameters of ccRCC patients tion and associated epigenetic silencing to be involved in Clinicopathological n% carcinogenesis of gastric and colorectal cancers [6]. Epi- parameters genetic alterations of GATA5 were also described in Cases in total 77 100 other tumor tissues and were linked to the development Sex Female 29 38 of ovarian, lung, pancreatic, and esophageal cancer Male 48 62 [7-10]. In a recent study aimed at identifying new DNA Median age, years 64 methylation targets in ccRCC, we detected for the first time a tumor-specific hypermethylation of the GATA5 Median tumor size, cm 5.5 CpG island (CGI) in RCC [11]. Hypermethylation was Primary tumor classification pT1 6 8 also associated with advanced disease and shortened pT1a 19 25 recurrence-free survival (RFS) of patients. pT1b 14 18 In this study, we asked whether mRNA expression pT2 3 4 levels of GATA5 are reduced as suggested by our previ- pT3 2 3 ous DNA methylation analysis, and if the mRNA levels are associated with adverse clinical parameters, further pT3a 9 12 underlining the relevance of epigenetic GATA5 alter- pT3b/c 22 29 ations in ccRCC carcinogenesis. pT4 0 0 not known 2 3 Material and methods Lymph node status N0 68 88 Tissue specimens N1 9 12 One hundred and thirty-five kidney tissues, including 77 ccRCCs and 58 paired adjacent normal renal tissue sam- N2 0 0 ples, were included in this study. RCC tissues were ob- Status of metastasis M0 56 73 tained from open or laparoscopic nephrectomies and M1 21 27 partial resection. Paired tissue samples with adjacent Grade normal tissue (adN) were obtained from a subgroup of - Low risk G1 14 18 our 77 ccRCCs cohort. Adjacent normal tissues, i.e. G1-2 8 10 morphologically normal kidney were isolated with minimum of 0.5 cm to 2 cm distant from the primary tumor - Intermediate risk G2 40 52 lesion. Samples were snap-frozen in liquid nitrogen im- - High risk G2-3 5 6 mediately following surgery and stored at −80°C. Ethical G3 10 13 approval of the university ethical committee (Prof. H. D. Localized disease pT ≤ 2, N0, M0 and G1; G1-2 34 44 Tröger, Hannover Medical School, Carl-Neuberg-Str. 1, Advanced disease pT ≥ 3 and/or N1, M1 or G2-3; G3 43 56 Hannover, Germany) and informed consent from all pa- Paired samples 58 75 tients were obtained. Localized disease was defined as Abbreviations: ccRCC clear cell renal cell carcinoma, n number. pT ≤ 2, lymph node involvement and metastasis negative (N0/M0), and grading (G) G1 and 1–2, whereas advanced RCC was defined as pT ≥ 3, N1 and/or M1, and previously [12]. Duplicate measurements for qRT-PCR G2-3 and G3. Patients with G2 were assigned to the analysis were performed using 384 sample plates, an au- intermediate risk group and were not considered as a tomated liquid handling system (FasTrans, AnalyticJena, parameter for low vs. high grade group comparisons. Jena, Germany), and the ABI 7900 Fast Sequence Detec- Follow up data were available for 35 patients, and RFS tion System as described previously [12]. Experimenters was defined as the interval up to the time that disease were blinded to any patient clinicopathological or sur- progression could be detected by computer tomography vival information. The TaqMan expression assays used scans. Clinical and histopathological parameters are were GATA-5 (Hs00388359_m1), HPRT1 (Hs999999 summarized in Table 1. 09_m1), GUSB (Hs00939627_m1), and RPL13A (Hs0304 3885_g1) (all assays were from Life Technologies, Foster RNA isolation, cDNA synthesis, and quantitative real-time City, CA, USA). HPRT1, GUSB, and RPL13A were in- PCR analysis cluded as endogenous references. The cDNA obtained Isolation of total RNA from tissue specimens and from from RPTEC primary cell transcripts served as biological renal proximal tubular epithelial cells (RPTEC) as controls. For each qRT-PCR run, blank and no-template controls, cDNA synthesis, and quantitative real-time controls were included. Relative expression levels were PCR analysis (qRT-PCR) were carried out as described calculated using the delta-deltaCT (ΔΔCt) method Peters et al. BMC Cancer 2014, 14:101 Page 3 of 6 http://www.biomedcentral.com/1471-2407/14/101

[13,14], and the SDS 2.3 Manager and dataAssist V2.0 Loss of GATA5 mRNA expression correlates with CGI software (Life technologies) as described previously [12]. hypermethylation The endogenous controls, HPRT1, RPL13A and GUSB, We compared lnRQ expression values and natural loga- were combined by dataAssist V2.0 software and “arith- rithms relative methylation (lnRML) values of GATA5 in metic mean” was used as a method of normalization. all samples (Figure 2). Regression analysis revealed an inverse relationship between relative expression levels and methylation of GATA5 (coefficient of regression = −0.41, Statistics and survival analysis p < 0.001). The comparison of paired tissues, indicated by Natural logarithms of relative expression (lnRQ) values solid lines, shows that high methylation values with con- were used for statistical calculations. All statistics were current low expression is frequently observed in tumor done using the statistical software, R 2.15.2 [15]. The tissues, whereas corresponding adN samples largely had t-test paired was used for statistical analyses of expres- high expression levels and low methylation. sion differences in paired tumor and adjacent normal tissues samples, whereas univariate Cox regression GATA5 mRNA expression is associated with tumor diameter models were used for statistical analysis of RFS. The Logistic regression analysis for comparison of tumor threshold for dichotomization of expression values was subgroups detected a significant difference in mRNA ex- calculated using the selected rank statistics of R package, pression levels only for the tumor diameter (p = 0.02, which provides the minimum p-value for log rank statis- odds ratio = 0.63; 95% CI: 0.42–0.93) whereas other clin- tics [15]. P-values < 0.05 were considered to be statisti- icopathological parameters like sex, gender, distant cally significant. The Kaplan-Meier method was used for metastasis, lymph node metastasis and tumor grade ex- survival analyses. hibited no statistically significant association.

Results Loss of GATA5 mRNA expression is associated with GATA5 mRNA expression is decreased in ccRCC decreased recurrence-free survival The analyses of relative GATA5 mRNA expression levels Cox regression survival analyses using a statistically cal- revealed significantly decreased expression in tumor speci- culated optimum cut off value for relative GATA5 mens (TU; mean lnRQ = −1.7; ±SD = 1.63) compared with mRNA expression (lnRQ = −3.52) showed that a lower the corresponding adN (mean lnRQ = 1.73; ±SD = 1.32; expression status was associated with increased risk for p < 0.001; paired t-test). Figure 1A illustrates the differ- shorter time to disease recurrence (p = 0.023, hazard ences in expression values observed for paired tumor and ratio (HR) = 0.25, 95% CI: 0.07–0.82; Table 2). Within adN, indicating a strong reduction of up to 31-fold in the 30 months, four out of five patients (80%) whose tumor expression levels, largely in tumor tissues. The compari- specimens demonstrated expression values below the son of the distribution of relative expression values be- cut off value were identified with disease recurrence tween both tissue groups showed only a small overlap (Figure 3). The status of localized and advanced disease (Figure 1B). (p = 0.03, HR = 4.18; 95% CI: 1.15–15.2), status of

Figure 1 GATA5 mRNA expression in paired clear cell renal cell carcinoma and adjacent normal tissues. A) Comparison of the relative GATA5 expression (RQ) values in adjacent normal (adN) and tumor (TU) tissues from ccRCC patients (p < 0.001). B) Scatterplot analysis illustrating the distribution of relative expression values (RQ) observed for TU and adN in ccRCC specimens. Bold lines indicate the median of relative expression values. Peters et al. BMC Cancer 2014, 14:101 Page 4 of 6 http://www.biomedcentral.com/1471-2407/14/101

Figure 3 Kaplan-Meier plot for illustrating recurrence-free survival. The solid line shows the Kaplan-Meier curve for 5 patients (pts.) with GATA5 mRNA expression lower than or equal to the cut off of −3.52 (natural logarithm), indicating patients with a shortened Figure 2 Association of GATA5 CGI methylation and relative recurrence-free survival. The dashed line illustrates the Kaplan-Meier mRNA expression in tumor and adjacent normal tissues. Solid curve for patients with mRNA expression levels above the cut off lines connect the subgroup of paired tissues (tumor tissue = solid including 30 pts. Disease progression of ccRCC within a period of triangle; adjacent normal = solid squares). The regression line approximately two years was observed in seven cases with a low (dashed line) and 95% CI (grey shaded) are presented. Note that relative mRNA expression level, whereas high GATA5 mRNA expression tissues exhibiting concurrent high methylation and low mRNA phenotypes showed only four progression events within that time interval. expression can be found in the upper left corner, whereas the occurrence of low methylation and higher relative expression in tissues is displayed in the lower right corner. expression was not associated with RFS while mRNA levels were detected as a significant parameter in bivariate metastasis (p = 0.009, HR = 4.27; 95% CI: 1.43–12.8), and Cox regression models including the statuses of lymph tumor grade (p < 0.001, HR = 9.48; 95% CI: 2.92–30.8) node metastasis, age and gender (Table 3). Moreover, were also shown to be associated with RFS (Table 2). we carried out a multivariate analysis demonstrating Pairwise bivariate Cox regression analyses were first carried out to investigate whether an association between Table 3 Bivariate statistical association of GATA5 mRNA expression status and clinicopathological parameters and expression and clinicopathology with recurrence-free RFS exists. In bivariate statistical models, considering survival the statuses of advanced disease, metastasis, and tumor p-value° HR 95% CI grade as covariates, we found in each case that GATA5 GATA5 mRNA expression 0.137 0.390 0.11-1.35 Localized vs Advanced 0.080 3.340 0.87-12.9 Table 2 Univariate statistical association of GATA5 mRNA expression and clinicopathological parameters with GATA5 mRNA expression 0.113 0.370 0.11-1.27 recurrence-free survival Status of metastasis 0.036 3.410 1.08-10.8 p-value° HR 95% CI GATA5 mRNA expression 0.511 0.640 0.17-2.41 GATA5 mRNA expression 0.023 0.25 0.07-0.82 Tumor grade 0.001 8.320 2.34-29.6 Localized vs. Advanced 0.030 4.18 1.15-15.2 GATA5 mRNA expression 0.032 0.260 0.08-0.90 Status of metastasis 0.009 4.27 1.43-12.8 Lymph node metastasis 0.657 1.420 0.29-6.77 Tumor grade <0.001 9.48 2.92-30.8 GATA5 mRNA expression 0.035 0.270 0.08-0.91 Lymph node status 0.398 1.920 0.42-8.72 Age* 0.214 0.470 0.14-1.55 Age* 0.155 0.420 0.13-1.38 GATA5 mRNA expression 0.023 0.250 0.08-0.83 Gender 0.489 1.510 0.47-0.82 Gender 0.512 1.480 0.46-4.84 °Univariate Cox regression analysis. °Bivariate Cox regression analysis. *Dichotomization by the median of parameter. *Dichotomization by the median of parameter. bold numbers: p-value < 0.05 (statistically significant). bold numbers: p-value < 0.05 (statistically significant). hazard ratio (HR). hazard ratio (HR). confidence interval (CI). confidence interval (CI). Peters et al. BMC Cancer 2014, 14:101 Page 5 of 6 http://www.biomedcentral.com/1471-2407/14/101

that mRNA expression of GATA5 (p = 0.12, HR = 0.32; of tumors and RFS in univariate Cox regression analysis, 95% CI: 0.07–1.34) is not significantly associated with showing a hazard ratio of 0.25, which resembles the recip- RFS including the covariates status of advance disease rocal hazard ratio observed for the corresponding methy- (p = 0.29, HR = 0.18; 95% CI: 0.01-4.41), status of metasta- lation analysis. sis (p = 0.14, HR = 6.60; 95% CI: 0.55–78.68), tumor grade Interestingly, a subset of patients with very low mRNA p = 0.002, HR = 0.09; 95% CI: 2.46–56.14, age (p = 0.36, expression levels also demonstrated a shortened HR = 2.54; 95% CI: 0.05–2.85) and gender (p = 0.39, HR = recurrence-free survival in univariate Cox regression 2.03; 95% CI: 0.39–10.38). analysis. However, taking into account that only a small number of tumors have been identified, these results re- Discussion quire future extended evaluation studies including multi- Members of the GATA1-6 transcription factor family variate analyses. contribute to stem cell differentiation in embryonic tissue, and GATA5 is involved in intestinal epithelial cell Conclusion differentiation in adults [3]. Moreover, previous analyses Decreased GATA5 mRNA expression in ccRCC may be GATA5 found hypermethylation in human malignancies caused by epigenetic silencing, and is likely associated such as gastric and colorectal cancers and demonstrated with a poor clinical outcome. Our results underline the that epigenetic silencing of the gene occurred in various need for further functional studies to characterize the GATA5 human cancer cell lines, providing evidence that interaction of GATA5 and cellular signaling in ccRCC alterations may represent epigenetic alterations of wider with respect to the observed changes in expression and relevance for carcinogenesis [4,6]. methylation levels, and its association with tumor In a recent study aimed at identifying new DNA methy- progression. lation targets in ccRCC, we detected tumor-specific hypermethylation of the GATA5 CGI in RCC [11]. Competing interest Hypermethylation was also associated with advanced The authors’ declare that they have no competing interest. disease and shortened RFS of patients, which had not ’ been previously reported for any other human cancer. Authors contributions IP wrote the manuscript, prepared the figures and participated in the study Hypermethylation of GATA5 in RCC indicated that design. ND and MK carried out the methylation analyses and participated in the expression of GATA5 might be epigenetically si- the sequence analyses. JH and MK assembled histopathological, lenced in tumor cells, leading to a biologically more ag- clinicopathological and survival data. JH performed isolation and characterization of tissue samples and assembly of patients. MA evaluated gressive tumor phenotype. In the current study, we the histopathologies of given tissues. MK, CvK, AM and AS assisted with demonstrate that GATA5 mRNA expression is strongly general scientific discussion. JS identified the candidate promoter, conceived reduced in ccRCC and, moreover, that a subgroup of tis- of the study, developed the study design and analytical assays, constructed and ran the clinical database, performed statistical analyses together with RS sues shows a clear relationship between methylation of and participated in manuscript preparation. All authors read and approved the GATA5 CGI and reduced mRNA expression, indicat- the final manuscript. ing that epigenetic silencing of GATA5 occurs in a substantial fraction of ccRCC. A significant relationship Acknowledgements We thank Christel Reese and Magrit Hepke for technical assistance. between GATA5 hypermethylation and reduced GATA5 mRNA expression within a human tissue, to the best of Author details 1 our knowledge, has not been previously demonstrated. Department of Urology and Urologic Oncology, Hannover Medical School, Carl-Neuberg-Str.1, Hannover 30625, Germany. 2Department of Pathology, Thus, our results support the notion that epigenetic Hannover Medical School, Hannover, Germany. 3Department of Urology, silencing due to DNA methylation is a relevant process Eberhard Karls University of Tuebingen, Tuebingen, Germany. 4Department of in RCC. Biometry, Hannover Medical School, Hannover, Germany. A subgroup of tissues showed only moderate reduc- Received: 18 August 2013 Accepted: 12 February 2014 tion of expression, although GATA5 methylation was de- Published: 17 February 2014 tectable, indicating that other biological mechanisms, e.g. histone alterations, play a role in tumor development References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D: Global cancer in ccRCC. Additional functional investigations are re- statistics. CA Cancer J Clin 2011, 61(2):69–90. quired to clarify these aspects. 2. Morrisey EE, Ip HS, Tang Z, Lu MM, Parmacek MS: GATA-5: a transcriptional GATA5 methylation is associated with various clinico- activator expressed in a novel temporally and spatially-restricted pattern during embryonic development. Dev Biol 1997, 183(1):21–36. pathological parameters as well as RFS. Hence, we hy- 3. Gao X, Sedgwick T, Shi YB, Evans T: Distinct functions are implicated for pothesized that reduced mRNA expression levels in the GATA-4, -5, and −6 transcription factors in the regulation of intestine ccRCC would also show an association with unfavorable epithelial cell differentiation. Mol Cell Biol 1998, 18(5):2901–2911. 4. Ren CY, Akiyama Y, Miyake S, Yuasa Y: Transcription factor GATA-5 clinical parameters. Indeed, we found that decreased selectively up-regulates mucin gene expression. J Cancer Res Clin Oncol GATA5 mRNA expression is associated with the diameter 2004, 130(5):245–252. Peters et al. BMC Cancer 2014, 14:101 Page 6 of 6 http://www.biomedcentral.com/1471-2407/14/101

5. Kiela PR, LeSueur J, Collins JF, Ghishan FK: Transcriptional regulation of the rat NHE3 gene. Functional interactions between GATA-5 and Sp family transcription factors. J Biol Chem 2003, 278(8):5659–5668. 6. Akiyama Y, Watkins N, Suzuki H, Jair K-W, van Engeland M, Esteller M, Sakai H, Ren C-Y, Yuasa Y, Herman JG, et al: GATA-4 and GATA-5 transcription factor genes and potential downstream antitumor target genes are epigenetically silenced in colorectal and gastric cancer. Mol Cell Biol 2003, 23(23):8429–8439. 7. Wakana K, Akiyama Y, Aso T, Yuasa Y: Involvement of GATA-4/-5 transcription factors in ovarian carcinogenesis. Cancer Lett 2006, 241(2):281–288. 8. Guo M, Akiyama Y, House MG, Hooker CM, Heath E, Gabrielson E, Yang SC, Han Y, Baylin SB, Herman JG, et al: Hypermethylation of the GATA genes in lung cancer. Clin Cancer Res 2004, 10(23):7917–7924. 9. Guo M, House MG, Akiyama Y, Qi Y, Capagna D, Harmon J, Baylin SB, Brock MV, Herman JG: Hypermethylation of the GATA gene family in esophageal cancer. Int J Cancer 2006, 119(9):2078–2083. 10. FuB,GuoM,WangS,CampagnaD,LuoM,HermanJG,Iacobuzio-DonahueCA: Evaluation of GATA-4 and GATA-5 methylation profiles in human pancreatic cancers indicate promoter methylation patterns distinct from other human tumor types. Cancer Biol Ther 2007, 6(10):1546–1552. 11. Peters I, Eggers H, Atschekzei F, Hennenlotter J, Waalkes S, Trankenschuh W, Grosshennig A, Merseburger AS, Stenzl A, Kuczyk MA, et al: GATA5 CpG island methylation in renal cell cancer: a potential biomarker for metastasis and disease progression. BJU Int 2012, 110(2 Pt 2):E144–E152. 12. Waalkes S, Atschekzei F, Kramer MW, Hennenlotter J, Vetter G, Becker JU, Stenzl A, Merseburger AS, Schrader AJ, Kuczyk MA, et al: Fibronectin 1 mRNA expression correlates with advanced disease in renal cancer. BMC Cancer 2010, 10:503. 13. Livak KJ, Schmittgen TD: Analysis of relative gene expression data using real-time quantitative PCR and the 2(−Delta Delta C(T)) Method. Methods 2001, 25(4):402–408. 14. Schmittgen TD, Livak KJ: Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc 2008, 3(6):1101–1108. 15. Team RDC: R: A Language and Environment for Statistical Computing. Vienna Austria; 2011. http://www.r-project.org/.

doi:10.1186/1471-2407-14-101 Cite this article as: Peters et al.: Decreased GATA5 mRNA expression associates with CpG island methylation and shortened recurrence-free survival in clear cell renal cell carcinoma. BMC Cancer 2014 14:101.

Submit your next manuscript to BioMed Central and take full advantage of:

• Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit ONCOLOGY REPORTS 31: 1523-1530, 2014

GATA5 CpG island hypermethylation is an independent predictor for poor clinical outcome in renal cell carcinoma

INGA PETERS1, KAI GEBAUER1, NATALIA DUBROWINSKAJA1, FARANAZ ATSCHEKZEI1, MARIO W. KRAMER1, JOERG HENNENLOTTER4, HOSSEIN TEZVAL1, MAHMOUD ABBAS2, RALPH SCHERER3, AXEL S. MERSEBURGER1, ARNULF STENZL4, MARKUS A. KUCZYK1 and JUERGEN SERTH1

Departments of 1Urology and Urologic Oncology, 2Pathology and 3Biometry, Hannover Medical School, D-30625 Hannover; 4Department of Urology, Eberhard-Karls University, D-72076 Tuebingen, Germany

Received November 3, 2013; Accepted December 2, 2013

DOI: 10.3892/or.2014.3030

Abstract. Transcriptional inactivation and CpG island (CGI) tumor. The use of targeted therapies has improved treatment of methylation of GATA transcription factor family members metastatic RCC, but survival remains significantly decreas ed GATA3 and GATA5 have been reported for a few types of in late-stage or metastatic RCC patients (2). human cancer. Whether high-density CGI methylation of The molecular carcinogenesis of clear cell renal cell GATA3 or GATA5 is associated with the clinical course of carcinoma (ccRCC) includes von Hippel-Lindau gene altera- patients with renal cell cancer (RCC) has not been clarified. tions as gatekeeper mutations that are followed by additional Quantitative methylation-specific PCR assays were carried genetic changes for full development of the cancer (3). In view out to analyze 25 tumor cell lines including 6 RCC lines and of the epigenetic progenitor cancer model, such mutations 119 RCC and 87 adjacent normal tissues for the presence of may be substituted by epigenetic alterations that cause gene densely methylated sequences. Methylation values were statis- silencing and thus contribute to the accumulation of epigenetic tically compared with clinicopathological and recurrence-free and genetic alterations, as has been found for several human survival (RFS) data for patients. Comparison of GATA3 and malignancies (4). Indeed, a considerable number of loci GATA5 methylation in different tumor cell lines revealed undergoing DNA methylation have been identified in ccRCC a marker-specific methylation characteristic with high and at a high frequency. For example, the secreted frizzled-related frequent signals for both methylation marks in RCC lines. protein (SFRP1) and RAS-associated domain family 1 CpG GATA3 and GATA5 CGI relative methylation levels were found island (CGI) hypermethylation have been found in 34-68% to be strongly associated with the state of metastasis (P=0.003 and 28-76% of RCCs, respectively (5-7). Hypermethylation of and P<0.001, respectively) and advanced disease (P=0.024 and the SCUBE3 gene is associated with clinicopathological para- P<0.001, respectively). Moreover, an independent decrease meters and poorer survival (8). A genome-wide CGI methylation in RFS in Cox proportional hazard analysis was found for analysis by Ricketts et al (9) showed that CGI hypermethyl- tumors exhibiting high GATA5 methylation (P<0.001, hazard ation of several genes (including SLC34A2 in 63%, OVOL1 in ratio, 19.3; 95% confidence interval, 4.58-81.6). Epigenetic 40%, DLEC in 20%, TMPRSS2 in 26%, SST in 31% and BMP4 alterations in GATA family members may be associated in 35% of RCC) is associated with transcriptional silencing, with aggressive tumor phenotypes in RCC, and in the case of reactivation after demethylation in RCC cell lines and down- GATA5, may serve as a new independent molecular marker for regulation of expression in RCC. aggressiveness and disease progression. Recently, we identified GATA5, a member of the GATA transcription factor family (GATA1 to GATA6), as a new Introduction target for CGI hypermethylation in RCC, also demonstrating a statistical association with disease progression Renal cell carcinoma (RCC) is the tenth most common cancer and decreased survival. However, since combined bisulfite in men worldwide (1) and the third most common genitourinary restriction analysis detection was applied for methylation detection, only site-specific average methylation could be assessed (10). Heterogeneous methylation as determined in the CGI of GREM1 in RCC (11) may lead to varying statis- Correspondence to: Dr Juergen Serth, Department of Urology tical associations with clinicopathological parameters; thus, and Urologic Oncology, Hannover Medical School, Carl-Neuberg- our previous findings ofGATA5 CGI methylation as a poten- Strasse 1, D-30625 Hannover, Germany tial prognosticator for RCC would be strengthened if another E-mail: [email protected] GATA5 methylation locus could be identified to demonstrate association with an unfavorable prognosis. Detecting highly Key words: GATA3, GATA5, renal cell cancer, DNA hypermethylated sequences located in a different subregion of the methylation, survival, prognosis GATA5 CGI would provide further evidence for a crucial role of GATA5 in RCC progression. 1524 PETERS et al: GATA3 AND GATA5 CGI METHYLATION IS ASSOCIATED WITH SURVIVAL IN RCC

In addition, comparing expression and methylation data Table I. Clinicopathological data of patients. from public databases (12), we noted that GATA3, as a member of the GATA transcription factor family, might also represent Clinicopathological parameters GATA3 (%) GATA5 (%) a potential target for CGI hypermethylation. The GATA1, GATA2 and GATA3 members of the GATA transcription Cases in total (all RCC) 119 (100) 109 (100) factor family are functionally involved in cellular lineage Histology determination (13) while the GATA4, GATA5 and GATA6 are ccRCC 86 (72) 78 (72) mainly involved in epithelial differentiation and are suggested papRCC 24 (20) 22 (20) to play a critical role in tumorigenesis of cancer with endo- Chromophobe/mixed RCC 5 (4) 5 (5) or mesodermal origins (13). Furthermore, both mechanisms exhibit extensive changes in neoplastic development in different Not classified 4 (3) 4 (4) cancer types (14) and loss of GATA3 expression in breast Gender cancer patients has been significantly associated with poor Female 42 (35) 37 (34) clinical outcome and advanced tumor disease (15). Comparing Male 77 (65) 72 (66) normal and tumor renal tissues, decreased GATA3 protein and mRNA expression levels have already been observed, Age (years) supporting the hypothesis that GATA3 may be epigenetically Median 65 (55) 65 (60) silenced in RCC (16). Tumor size To clarify the relevance of GATA3 and GATA5 methyla- In diameter (cm) 4.6 4.5 tion in RCC, we measured CGI methylation of both genes in normal human primary tubule epithelial cells and in renal Primary tumor classification tumor cell lines, as well as in renal cancer tissues and a subset pT1 11 (9) 11 (10) of paired adjacent normal tissues, using quantitative methyla- pT1a 35 (29) 32 (29) tion-specific PCR (qMSP). We found that higher methylation pT1b 19 (16) 19 (17) is more likely to be found in tumors of patients with advanced pT2 8 (7) 6 (6) and metastatic disease and in case of GATA5 is also associated pT3 5 (4) 4 (4) with poorer survival of RCC patients. pT3a 11 (9) 8 (7) Materials and methods pT3b/c 25 (21) 24 (22) pT4 1 (1) 1 (1) Tissue specimens. Cross-sectional analyses were conducted on Not known 4 (3) 4 (4) 119 RCC samples and 87 samples from paired histologically Lymph node status normal-appearing tissues, i.e., adjacent normal renal tissue. Tissue samples were collected from patients who had under- N0 104 (87) 96 (88) gone radical or nephron-sparing nephrectomy and stored as N+ 15 (13) 13 (12) previously described (17). TNM classification of all tissues Metastasis classification was evaluated according to the Union for International Cancer M0 92 (77) 85 (78) Control 2010 classification, and grading was assessed as prev i- M+ 27 (23) 24 (22) ously described (18,19). Localized RCC was defined as pT ≤2, lymph node (N) and metastasis (M) negative (N0 and M0), and Grade a grading (G) of 1 and 1-2. Advanced tumors were classified Low risk group as p ≥T3 and/or lymph node positive (N+), positive for distan t G1 23 (19) 22 (20) metastasis (M+) or G2-3 and G3. Time to disease recurrence G1-2 16 (13) 14 (13) was designated as the point at which patients had either a G2 60 (50) 57 (52) local recurrence or a synchronous/metachronous metastasis as High risk group detected by computerized tomography scan. The local ethics G2-3 9 (8) 7 (6) committee approved sample collection, and informed consent G3 11 (9) 9 (8) was obtained from each patient. Clinical and histopathological Localized disease parameters of tissues are summarized in Table I. Purchase, culturing, storage and identity control of cell lines and primary pT ≤2, N0, M0 and G1; G1-2 63 (53) 58 (53) cells were carried out as previously described (17). Advanced disease pT≥3 and/or N+, M+ or G2-3;G3 55 (46) 50 (46) Isolation of DNA and bisulfite conversion.DNA was extracted Not known 1 (1) 1 (1) from frozen tissue sections using a standard phenol/chloroform Paired samples extraction method. Bisulfite conversions and histopatholog ical examination of control sections were conducted as previously All RCC 87 (73) 77 (71) reported (20). ccRCC 66 (55) 57 (52)

Quantitative methylation-specific real-time PCR analysis fo ccRCC, clear cell renal cell carcinoma; papRCC, papillary renal cell GATA3 and GATA5 CGI methylation. Methylation analyses carcinoma. ONCOLOGY REPORTS 31: 1523-1530, 2014 1525

Figure 1. Description of investigated CpG islands of GATA3 and GATA5 and assay controls. (A-a) Structure of the GATA3 CGI locus and location of the qMSP assay relative to the transcription start site. GATA3 is located on chromosome 10p14. CpG sites are illustrated with vertical lines within the CpG island. Information refers to UCSC Genome Browser database and GenBank (12,22). (A-b) GATA5 structure and location of the qMSP assay relative to the transcription start site. Vertical lines represent CpG sites within the CpG island. Base positions refer to the GRCh37/hg19 annotation in the UCSC Genome Browser and GenBank (12,22). (B) Primary data of quantitative methylat ion-specific and control PCR measurements in methylated con trol DNA (1); unmethylated control DNA (2); unconverted DNA (3); and a blank control (4) for GATA3 (a) and GATA5 (b) analysis. (C) Normalized GATA3 (a) and GATA5 (b) assay threshold values (Ct) for a 2-fold dilution series of the methylated control in non-methylated control DNA for determination of assay linearity and efficiency. of bisulfite-treated genomic DNA for CGI methylation of AACCGCGACTCCTACCAATTCATTCG-BHQ-3' were GATA3 and GATA5 was performed by quantitative real-time designed using Beacon Designer™ software (Premier fluorimetric 5' exonuclease methylation-specific PCR assays. Biosoft, Palo Alto CA, USA). Intra-CGI location of both Methylation analysis was carried out as described elsewhere qMSP assays, designed within an area of high GC percentage, (21). The qMSP-specific primers 5'-TGTATCGGGACGGA is shown in Fig. 1A-a (GATA3) and in Fig. 1A-b (GATA5). ATCGTT-3' (forward) and 5'-ACGCGCGCTCTAACCCTT-3' Table II shows the base positions of investigated CpG sites in (reverse) as well as the Taqman® probe 5'-FAM-AAATAT the corresponding CGI referenced in the USCS Genome 1526 PETERS et al: GATA3 AND GATA5 CGI METHYLATION IS ASSOCIATED WITH SURVIVAL IN RCC

Table II. Detailed chromosomal information of GATA3 and GATA5.

GATA3 GATA5

Chromosome 10p14 20q13.33 GeneID 2625 140628 CpG Island No. of CpG sites 509 247 Base position (bp) 8091375-8098329 61049362-61051897 bp of CpG sites investigated by qMSP 8097735, ~744, ~750, ~796, 61051188, ~210, ~223, ~232, ~801, ~811, ~831, ~849 ~236, ~241, ~253, ~255, ~262

Chromosomal information and base position (bp) location of GATA3 and GATA5 qMSP relevant CpG sites. Information refers to the UCSC Genome Browser annotation GRCh37/hg 19.

Figure 2. Measurement of relative methylation in different cancer cell lines. Levels of relative methylation values in cancer cell lines and normal primary cells for (A) GATA3 and (B) GATA5.

Browser (12,22). Real-time PCR was carried out in duplicate based on the method of Weisenberger et al, recently described using a FasTrans automatic Liquid Handling System (Analytik in detail (21,23). Statistical analyses and definition of the Jena, Jena, Germany) and an ABI 7900HT (Life Technologies, cut-off value for dichotomization used in survival analysis Foster City, CA, USA) in 384-well plates as previously were also carried out as previously described (17). reported (17). An experimenter who was blinded to type, For univariate statistical analyses, all groups were dichoto- order and clinicopathological status of samples carried out mized according to their clinicopathological parameters, i.e., measurements. localized (Loc.) vs. advanced (Adv.) disease, metastasis negative (M0) vs. positive (M+), lymph node metastasis-negative Calculation of relative methylation indices and statistical vs. lymph node metastasis-positive (N0/N+), and low-grade analysis. Calculation of the relative degree of methylation was (G1, G1-2) vs. high-grade (G2-3, G3) tumors. ONCOLOGY REPORTS 31: 1523-1530, 2014 1527

Table III. Statistical analyses of GATA3 and GATA5 CGI methylation and correlation with clinicopathological parameters in paired t-test and univariate logistic regression analysis.

GATA3 GATA5 Paired t-test P-value P-value adN/TU all RCC 0.006 <0.001 ccRCC 0.001 <0.001

Univariate logistic regression analysis OR (95% CI) OR (95% CI) ccRCC/papRCC 0.006 0.77 (0.63-0.94) 0.015 0.80 (0.67-0.96) Localized/advanceda all RCC 0.024 1.32 (1.04-1.68) <0.001 1.55 (1.29-1.88) ccRCC 0.277 1.16 (0.89-1.50) <0.001 1.46 (1.19-1.80) Metastasis: M0/M+ all RCC 0.003 1.59 (1.05-2.43) <0.001 1.65 (1.29-2.11) ccRCC 0.179 1.38 (0.86-2.20) <0.001 1.64 (1.23-2.17) Grade: low/highb all RCC 0.658 1.06 (0.82-1.37) 0.003 1.47 (1.14-1.88) ccRCC 0.542 0.92 (0.68-1.21) 0.009 1.54 (1.11-2.14) Lymph node metastasis: N0/N+ all RCC 0.187 1.36 (0.86-2.14) 0.03 1.32 (1.03-1.68) ccRCC 0.572 1.17 (0.68-2.01) 0.35 1.15 (0.85-1.56) adN, adjacent normal tissue; TU, tumor tissue; ccRCC, clear cell renal cell carcinoma; papRCC, papillary renal cell carcinoma; OR, odds ratio; CI 95%, confidence interval. aLocalized tumor is pT ≤2, lymph node (N) and metastasis (M) negative (N0/M0) and grading (G) G1 and G1-2. Advanced tumor is pT ≥3 and/or N+, M+ or G2-3 and G3. bLow grade tumor (G1 and G1-2). High grade tumor (G2-3 and G3).

Results and cervical cancer cell lines), which in part have already been reported to demonstrate tumor specific hypermethyl- Measurement of technical controls and analysis of GATA3 ation. Methylation for GATA3 was found in 5/8 (63%) breast and GATA5 CGI methylation in human normal cells and cancer cell lines, as expected from previous reports describing cancer cell lines. The specificity of theGATA3 and GATA5 GATA3 methylation in breast cancer tissue samples. Notably, qMSP analyses was evaluated by duplicate measurements of all 6 renal cancer cell lines showed high relative methylation converted methylated (M), converted non-methylated (U) and indices while normal primary cells from kidney (RPTEC), non-converted DNA control samples. For U and non-converted prostate cancer, and mammary tissues demonstrated low or DNA samples, we exclusively measured Ct values of 45 (unde- undetectable methylation (Fig. 2A). Similarly, GATA5 CGI termined) whereas the M sample demonstrated Ct values of methylation was not detectable or was low in normal primary ~32 for GATA3 (Fig. 1B-a) and Ct values of ~29 for GATA5 cells but demonstrated higher relative methylation indices only (Fig. 1B-b). None of the control or CGI-specific qMSP assays for 4/6 renal cancer cell lines (Fig. 2B). gave signals for non-converted DNA, thus demonstrating that only methylated and converted DNA was detected. PCR effi- GATA3 and GATA5 CGI is hypermethylated in RCC. ciency and linearity of the methylation detection assays were Comparison of GATA3 and GATA5 methylation in matched assessed using a log dilution series of the M control within tumor (TU) and adjacent normal (adN) tissues demonstrated the U control DNA and adjusting for constant total converted tumor-specific hypermethylation (Fig. 3A and B). Using DNA amount in PCR reactions. Linear regression analyses the paired t-test for statistical analysis (Table III), we found demonstrated a slope of ∆Ct = -3.3 per 10-fold dilution and a significant differences forGATA3 methylation in the RCC coefficient of correlation of r=-0.99 for both genes (P=0.001), tissue groups (P=0.006) as well as in the ccRCC subset indicating linearity of the assays (Fig. 1C-a and 1C-b). (P=0.001). The corresponding analysis of GATA5 methylation We assessed whether the GATA3 and GATA5 qMSP assays also demonstrated higher methylation in tumor tissues both for are capable of methylation detection in normal human primary the RCC group (P<0.001) and the ccRCC subset (P<0.001). tubule epithelial cells and in cancer cell lines, each respectively used as a proxy for normal tissues and localized and metastatic Analysis of GATA3 and GATA5 CGI methylation and asso- human cancers of other origin (kidney, prostate, bladder, breast ciation with clinicopathological parameters. Univariate 1528 PETERS et al: GATA3 AND GATA5 CGI METHYLATION IS ASSOCIATED WITH SURVIVAL IN RCC

Figure 3. Associations of DNA methylation of GATA3 and GATA5 with clinicopathology and recurrence-free survival. Comparison of the natural logarithms of relative methylation values for (A) GATA3 and (B) GATA5 in adjacent normal (adN) and tumor (TU) tissues of a ccRCC patient cohort (P=0.001, P<0.001). (C) Box plot illustration of GATA3 CGI methylation. GATA3 methylation was significantly increased in advanced (Adv.) and metastasized (M+) renal cell cancer patients compared to localized (Loc.) and non-metastasized (M0) disease (P=0.024 and P=0.003, respectively). Distribution of relative methylation values is illustrated in the Kernel distribution graph. (D) Distribution of GATA5 methylation values in all RCCs in association with clinicopathological parameters: Loc. and Adv. (P<0.001), M1 and M0 (P<0.001), lymph node status (N0/N+; P=0.03), and high-risk (G >2) and low-risk (G ≤ 2) grade (P=0.003). Distribution of relative methylation values is illustrated in the Kernel distribution graph. (E) Kaplan-Meier plot showing the relative recurrence-free survival of clear cell renal cancer patients with GATA5 hypermethylation. Results were dichotomized by a cut-off of -2.447. The dashed line indicates the patients with relative methylation value higher than the cut-off of -2.447, demonstra ting a significantly decreased recurrence-free survival.

logistic regression analysis (Table III) of dichotomized (GATA3 and GATA5) was significantly higher in advanced groups revealed a statistically significant association weenbet vs. localized (P=0.024 and P<0.001, respectively) and in methylation of GATA3 and GATA5 CGI with advanced and metastasis-negative (M0) vs. metastasis-positive (M+) tumors metastasized RCC disease. Mean methylation for both CGIs (P=0.003 and P<0.001, respectively; Fig. 3C and D) of the ONCOLOGY REPORTS 31: 1523-1530, 2014 1529

Table IV. Uni- and bivariate Cox regression model analysis of 95% (CI) confidence interval, 4.89-65.1] and ccRCC (P<0.001; GATA5 CGI methylation. HR = 13; 95% CI, 3.57-47.4; Table IVA) tissue groups. The Kaplan-Meier analysis with a calculated optimum cut-off of A, Univariate Cox regression analysis of GATA5 CGI methyl- -2.447 for dichotomization showed that higher CGI methyla- ation and association with recurrence-free survival in patients tion of GATA5 is associated with a decreased RFS in patients with clear cell renal cell carcinoma with ccRCC (Fig. 3E). A pairwise bivariate Cox regression model demonstrated that the GATA5 CGI methylation status P-value HR 95% CI remained a significant and strong parameter in the bivariate models when the status of metastasis, advanced tumor disease, Methylation <0.001 13.0 3.57-47.4 grade, and age were considered as co-variables (Table IVB). Status of metastasis (M0/M+)0.012 4.07 1.36-12.2 Localized vs. advanceda 0.061 3.44 0.94-12.5 Discussion Grade (low/high)b <0.001 8.46 2.49-28.7 Age-Medianc 0.362 0.59 0.19-1.82 GATA1, GATA2 and GATA3 from the GATA transcription factor family are involved in cellular lineage and hematopoietic development while GATA4, GATA5 and GATA6 are B, GATA5 CGI methylation analysis in a bivariate Cox regres- involved in epithelial and endodermal differentiations (13,24). sion model and its association with recurrence-free survival GATA proteins have been suggested to play a crucial role in keeping cells in the undifferentiated state (13). Moreover, P-value HR 95% CI previous experiments (10) as well as in silico analyses detecting reduced GATA3 and GATA5 mRNA expression Methylation <0.001 19.3 4.58-81.6 levels suggested that GATA3 and GATA5 are potential targets Status of metastasis (M0/M+)0.004 5.8 1.73-19.4 of epigenetic alteration in RCC. The present study has taken a Methylation 0.002 9.55 2.36-38.7 translational approach to investigate the presence and clinical Localized vs. advanced 0.355 1.96 0.47-8.23 relevance of CpG island methylation of both genes for RCC. Tumor cell lines (renal, bladder, prostate and breast cancer) Methylation 0.04 5.35 1.1-26.1 revealed distinct CGI methylation patterns for GATA3 and Grade (low/high) 0.09 3.80 0.80-18.1 GATA5 methylation but showed no obvious overall correla- Methylation <0.001 29.7 5.72-154 tion between the epi-alterations. Notably, both methylation Age-Medianc 0.043 0.23 0.05-0.96 markers were frequently observed in kidney-derived tumor cell lines and also demonstrated tumor-specific hypermeth- aLocalized tumor is pT ≤2, lymph node (N) and- metastasisylation in RCC(M) innega concordance with results for our paired tive. (N0/M0) and grading (G) 1 or 1-2. Advancedgroup. tumor The ispresent pT study≥3 identified both genes as candidates and/or N+, M+ or G2-3bLow and grade G3. tumor (G1 and G1-2). with a possible relevance for RCC development. Therefore, our c High grade tumor (G2-3 and G3). Values dichotomized by the data are in line with a recent functional study demonstrating median of parameter. HR, hazard ratio; CI, 95%, confidence interval.that methylation-dependent silencing of GATA3 expression Clinicopathological factors were dichotomized in both regression models. is correlated with the loss of transforming growth factor-β receptor III and tumorigenesis in ccRCC tissues and cell lines, although its role in disease progression and patient survival remained to be elucidated (25). Our study revealed that both GATA3 and GATA5 showed RCC tissue group. In addition, GATA5 showed a significantly a highly significant association between CGI methylation and higher CGI methylation status in the high-grade tumor and advanced as well as metastatic RCC. Furthermore, GATA5 positive lymph node metastasis (N+) groups compared to low- CGI methylation exclusively demonstrated a statistical asso- grade tumor tissues (P=0.003) or negative lymph node status ciation with grade and lymph node status of the primary (P=0.03; Fig. 3D). Comparison of CGI methylation of GATA3 tumor. In addition, bivariate Cox regression analysis adjusted and GATA5 in ccRCC and papillary renal cell carcinoma for advanced disease, metastatic status, and grade revealed showed significant statistical differences for the meanGATA3 a high and fairly stable HR for GATA5 methylation in the (P=0.006) and GATA5 (P=0.015) relative methylation indices bivariate statistical survival models overall, identifying this observed in both histological entities (Table III). epigenetic mark as a new candidate for independent prognosis of decreased RFS. GATA5 CGI methylation is independently associated with Although a great number of hypermethylated loci have decreased recurrence-free survival. Univariate Kaplan- been identified in RCC (9), to date, only a subset of CGIs has Meier and bivariate Cox proportional hazard analysis were been functionally or clinically characterized. A recent study conducted to elucidate a possible relationship between GATA3 found that a large portion of clinically relevant epigenetic and GATA5 CGI methylation and recurrence-free survival alterations identified in RCC also exhibit functional changes (RFS) of RCC patients. GATA3 analysis showed no statis- in kidney cancer (8). Hence, pre-selection of CGIs based on tical relationship with survival. In contrast, univariate Cox their statistical association with clinical factors could repre- regression analysis revealed GATA5 methylation as a strong sent an efficient means of narrowing the pool of candidate parameter in the RCC [P<0.001; hazard ratio (HR) = 17.8; epi-alterations affecting the onset or course of RCC. Only a 1530 PETERS et al: GATA3 AND GATA5 CGI METHYLATION IS ASSOCIATED WITH SURVIVAL IN RCC limited number of methylation-based independent candidate 7. Peters I, Rehmet K, Wilke N, et al: RASSF1A promoter methylation and expression analysis in normal and neoplastic kidney prognosticators including BNC1, COL14A1, SFRP1, SCUBE3, indicates a role in early tumorigenesis. Mol Cancer 6: 49, 2007. GREM1 and DAL-1/4.1b (6,8,10,11,26) have thus far been 8. Morris MR, Ricketts CJ, Gentle D, et al: Genome-wide meth- reported. Therefore, our results identify GATA5 as a new ylation analysis identifies epigenetically inactivated andidatec tumour suppressor genes in renal cell carcinoma. Oncogene 30: candidate prognosticator gene and suggest its functional 1390-1401, 2011. relevance in the progression of RCC. 9. Ricketts CJ, Morris MR, Gentle D, et al: Genome-wide CpG We observed a noticeable difference of approximately two island methylation analysis implicates novel genes in the patho- genesis of renal cell carcinoma. Epigenetics 7: 278-290, 2012. orders of magnitude in the median relative methylation values 10. Atschekzei F, Hennenlotter J, Janisch S, et al: SFRP1 CpG island detected for GATA3 and GATA5 CGIs in tumor compared to methylation locus is associated with renal cell cancer suscepti- adjacent normal renal tissues. Considering that histological bility and disease recurrence. Epigenetics 7: 447-457, 2012. assessment of control sections ensured a minimum tumor 11. van Vlodrop IJ, Baldewijns MM, Smits KM, et al: Prognostic significance ofGremlin1 (GREM1) promoter CpG island hyper- cell content of at least 50% and that identical samples have methylation in clear cell renal cell carcinoma. Am J Pathol 176: been measured, a variation in tumor cell content as a possible 575-584, 2010. 12. Kent WJ, Sugnet CW, Furey TS, et al: The human genome explanation can be ruled out. Instead, we infer that a different browser at UCSC. Genome Res 12: 996-1006, 2002. methylation characteristic is present in both CGIs, as detected 13. Patient RK and McGhee JD: The GATA family (vertebrates and by qMSP specifically measuring completely methylated invertebrates). Curr Opin Genet Dev 12: 416-422, 2002. 14. Chou J, Provot S and Werb Z: GATA3 in development and cancer sequences. Moreover, as the present study only considered differentiation: cells GATA have it! J Cell Physiol 222: 42-49, single regions within the analyzed CGIs, we cannot rule out 2010. that other methylation marks may exist that exhibit significant 15. Abba MC, Nunez MI, Colussi AG, Croce MV, Segal-Eiras A and Aldaz CM: GATA3 protein as a MUC1 transcriptional regulator associations with clinicopathological parameters, bearing in in breast cancer cells. Breast Cancer Res 8: R64, 2006. mind that a recent report has shown such intra-CGI varia- 16. Tavares TS, Nanus D, Yang XJ and Gudas LJ: Gene microarray tions (11). analysis of human renal cell carcinoma: the effects of HDAC inhibition and retinoid treatment. Cancer Biol Ther 7: 1607-1618, The present study identified GATA3 and GATA5 meth- 2008. ylation as a common and cancer-specific event in RCC. The 17. Gebauer K, Peters I, Dubrowinskaja N, et al: Hsa-mir-124-3 association with late-stage disease and for GATA5 with short- CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell ened RFS suggests these targets as biomarkers for biological carcinoma. Br J Cancer 108: 131-138, 2013. aggressiveness of RCC and, in case of GATA5, as a candidate 18. Sobin LH and Compton CC: TNM seventh edition: what's new, prognosticator. what's changed: communication from the International Union Against Cancer and the American Joint Committee on Cancer. Cancer 116: 5336-5339, 2010. Acknowledgements 19. Thoenes W and Storkel S: Pathology of benign and malignant renal cell tumors. Urologe A 30: W41-W50, 1991 (In German). 20. Peters I, Eggers H, Atschekzei F, et al: GATA5 CpG island meth- We thank Margrit Hepke and Christel Reese for the technical ylation in renal cell cancer: a potential biomarker for metastasis assistance. and disease progression. BJU Int 110: E144-E152, 2012. 21. Weisenberger DJ, Siegmund KD, Campan M, et al: CpG island methylator phenotype underlies sporadic microsatellite insta- References bility and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 38: 787-793, 2006. 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E and Forman D: 22. Lander ES, Linton LM, Birren B, et al: Initial sequencing and Global cancer statistics. CA Cancer J Clin 61: 69-90, 2011. analysis of the human genome. Nature 409: 860-921, 2001. 2. Ljungberg B: Prognostic factors in renal cell carcinoma. Urologe 23. Weisenberger DJ, Campan M, Long TI, et al: Analysis of repeti- A 43 (Suppl 3): S119-S120, 2004. tive element DNA methylation by MethyLight. Nucleic Acids 3. Kaelin WG Jr: The von Hippel-Lindau tumor suppressor gene Res 33: 6823-6836, 2005. and kidney cancer. Clin Cancer Res 10: 6290S-6295S, 2004. 24. Graf T and Enver T: Forcing cells to change lineages. Nature 4. Feinberg AP, Ohlsson R and Henikoff S: The epigenetic progen- 462: 587-594, 2009. itor origin of human cancer. Nat Rev Genet 7: 21-33, 2006. 25. Cooper SJ, Zou H, Legrand SN, et al: Loss of type III 5. Dahl E, Wiesmann F, Woenckhaus M, et al: Frequent loss of transforming growth factor-β receptor expression is due to meth- SFRP1 expression in multiple human solid tumours: association ylation silencing of the transcription factor GATA3 in renal cell with aberrant promoter methylation in renal cell carcinoma. carcinoma. Oncogene 29: 2905-2915, 2010. Oncogene 26: 5680-5691, 2007. 26. Yamada D, Kikuchi S, Williams YN, et al: Promoter hypermeth- 6. Morris MR, Ricketts C, Gentle D, et al: Identification of ylation of the potential tumor suppressor DAL-1/4.1B gene in candidate tumour suppressor genes frequently methylated in renal clear cell carcinoma. Int J Cancer 118: 916-923, 2006. renal cell carcinoma. Oncogene 29: 2104-2117, 2010. DNA Methylation Biomarkers Predict Progression-Free and Overall Survival of Metastatic Renal Cell Cancer (mRCC) Treated with Antiangiogenic Therapies

Inga Peters1, Natalia Dubrowinskaja1, Mahmoud Abbas2, Christoph Seidel3, Michael Kogosov1, Ralph Scherer4, Kai Gebauer1, Axel S. Merseburger1, Markus A. Kuczyk1, Viktor Gru¨ nwald5, Ju¨ rgen Serth1* 1 Department of Urology and Urologic Oncology, Hannover Medical School, Hannover, Germany, 2 Department of Pathology, Hannover Medical School, Hannover, Germany, 3 Department of Oncology/Hematology/Bone MarrowTransplantation/Pneumology, University Medical Center Eppendorf, Hamburg, Germany, 4 Department of Biometry, Hannover Medical School, Hannover, Germany, 5 Clinic for Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, Germany

Abstract VEGF-targeted therapy increases both the progression-free (PFS) and overall survival (OS) of patients with metastasized renal cell cancer (mRCC). Identification of molecular phenotypes of RCC could improve risk-stratification and the prediction of the clinical disease course. We investigated whether gene-specific DNA hypermethylation can predict PFS and OS among patients undergoing anti-VEGF-based therapy. Primary tumor tissues from 18 patients receiving targeted therapy were examined retrospectively using quantitative methylation-specific PCR analysis of CST6, LAD1, hsa-miR-124-3, and hsa-miR-9- 1 CpG islands. PFS and OS were analyzed for first-line and sequential antiangiogenic therapies using the log rank statistics. Sensitivity and specificity were determined for predicting first-line therapy failure. Hypermethylation of CST6 and LAD1 was associated with both a shortened PFS (log rank p = 0.009 and p = 0.004) and OS (p = 0.011 and p = 0.043). The median PFS observed for the high and low methylation groups of CST6 and LAD1 was 2.0 vs.11.4 months. LAD1 methylation had a specificity of 1.0 (95% CI 0.65–1.0) and a sensitivity of 0.73 (95% CI 0.43–0.90) for the prediction of first-line therapy. CST6 and LAD1 methylation are candidate epigenetic biomarkers showing unprecedented association with PFS and OS as well as specificity for the prediction of the response to therapy. DNA methylation markers should be considered for the prospective evaluation of larger patient cohorts in future studies.

Citation: Peters I, Dubrowinskaja N, Abbas M, Seidel C, Kogosov M, et al. (2014) DNA Methylation Biomarkers Predict Progression-Free and Overall Survival of Metastatic Renal Cell Cancer (mRCC) Treated with Antiangiogenic Therapies. PLoS ONE 9(3): e91440. doi:10.1371/journal.pone.0091440 Editor: Zhengdong Zhang, Nanjing Medical University, China Received October 7, 2013; Accepted February 11, 2014; Published March 14, 2014 Copyright: ß 2014 Peters et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The authors have no support or funding to report. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction tumors [8,9]. However, the progression-free (PFS) and overall survival (OS) of patients with mRCC are independent of the loss of Renal cell cancer (RCC) is one of the top ten causes of cancer VHL function [10]. In contrast, blood-based analysis of single deaths in industrial countries [1]. Though recent improvements in nucleotide polymorphisms (SNPs) potentially affecting sunitinib targeted therapy have resulted in prolonged survival of patients target genes and ligands identified two polymorphisms in VEGFR3 with metastatic RCC (mRCC), the overall outcome is still poor that are associated with the PFS, but not OS, of patients [2,3]. undergoing targeted therapy [11]. Measurement of serum Due to different available compounds and a growing number of carbonic anhydrase IX (CA9) protein levels in metastatic ccRCC new agents affecting molecularly targeted structures, such as patients revealed significantly decreased OS among patients with vascular endothelial growth factor (VEGF) and mammalian target higher CA9 serum concentrations [12]. An individual advantage of rapamycin (mTOR) signaling [4] an optimal sequence of of tissue- or blood-based measurements for patients undergoing targeted therapies might exist for patients, potentially increasing therapy is perceptible, but the accuracy, sensitivity, and specificity survival with mRCC treatment. Although prognostic models, such of these methods have not yet been reported or validated [13]. as the MSKCC (Memorial Sloan Kettering Cancer Center) and Although large patient cohorts have been subjected to exom- Heng scoring systems, have been reported to be independent wide mutational analyses, only a limited number of genes other predictors of clinical outcome, [5,6] discrimination between than VHL and polybromo 1 (PBRM1) have been identified to have outcomes is still limited. Tumor-specific biologically based mutations in RCC, and most with low frequency [14]. Therefore, parameters have been suggested to improve these issues [7]. the limited number of frequently mutated genes reduces the Most RCCs have clear cell (ccRCC) histology and exhibit probability of identifying mutation-based predictors with appro- functional inactivation of the von Hippel-Lindau (VHL) gene due priate sensitivity and specificity. DNA methylation of CpG islands to mutations or epigenetic silencing in approximately 80% of (CGIs) in a substantial number of regulatory or tumor suppressor

Patient No. Age (years) Sex RCC type iTNM status First-line treatment PFS (months) OS (months) *Response

1 68 F clear cell T1bNxMx Sunitinib 1.34 2.47 NE 2 66 M clear cell T4N1Mx Sunitinib 1.77 1.77 PD 3 57 M clear cell T3bNxMx Sunitinib 2.76 3.62 PD 4 59 M chromophobe T4N2Mx Sunitinib 1.70 9.76 PD 5 72 F clear cell T3bN0Mx Sorafenib 5.52 23.64 SD 6 48 F clear cell T3NxMx Sunitinib 2.26 2.99 PD 7 62 M clear cell T2aNxMx Sunitinib 2.63 3.25 PD 8 80 F clear cell n.a. Sunitinib 0.88 26.14 SD 9 50 M clear cell T2NxMx Sorafenib 11.86 13.68 SD 10 71 F clear cell T1bN1M1 Axitinib 13.70 19.04 SD 11 69 F clear cell T1bNxMx Bevacizumab 6.21 11.07 SD 12 57 M clear cell T3N0Mx Sunitinib 11.50 29.77 SD 13 49 F clear cell T3aNxM1 Sunitinib 11.27 26.86 PR 14 60 M clear cell T3bNxMx Sunitinib 0.42 0.76 NE 15 54 M clear cell T2aNxMx Sorafenib 3.03 13.05 PD

ac 04|Vlm su e91440 | 3 Issue | 9 Volume | 2014 March 2 16 53 M clear cell T1NxMx Sorafenib 43.66 59.28 SD 17 64 M clear cell n.a. Sunitinib 30.31 59.24 CR 18 51 M papillary T1aNxMx Sunitinib 1.08 3.42 PD

Note: *Response: according to RECIST 1.1 criteria. Sex: Male, Female. NE: not evaluable du e to RECIST 1.1 criteria. CR: complete response.

PR: partial response. mRCC in Failure Therapy of Prediction Epigenetic SD: stable disease. PD: progressive disease. Age: At the beginning of first-line therapy. n.a: not available. iTNM: initial TNM status of primary RCC. doi:10.1371/journal.pone.0091440.t001 Epigenetic Prediction of Therapy Failure in mRCC genes has been identified as a functional surrogate of mutations for basic research. A written statement of our ethic committee for and has been reported to be a frequent event in RCC [15]. this study was obtained. Moreover, functional loss of VHL has been found to associate with broadened appearance of epigenetic alterations [16], and all of the Patient Characteristics and Treatment Regimens second-frequent mutations are related to altered chromatin/ Clinicopathological data, corresponding tissues, and follow-up histone stabilization or modification, mechanisms linked to CGI data including the PFS and OS of patients with mRCC who were methylation and gene expression [17]. Therefore, the frequent treated with first-line VEGF-targeted therapy were collected detection of epigenetic alterations in RCC differentiates RCC between November 2005 and October 2011 in the Clinic of tumor biology and provides candidates for novel diagnostic, Hematology and the Department of Urology and Urologic prognostic, or predictive markers [18]. CGI methylation in several Oncology at Hannover Medical School (Table 1). The MSKCC genes has already been identified as candidate prognosticators score or ECOG performance status were not available. Patients independent from clinicopathological parameters [19–21]. How- received the following treatment regimens in the first-line setting: ever, epigenetic biomarkers predicting the clinical course of sunitinib (n = 12, 67%), sorafenib (n = 4, 22%), axitinib (n = 1, mRCC patients subjected to targeted therapy, have, to the best of 5.5%), and bevacizumab (n = 1, 5.5%). our knowledge, not been reported. PFS was defined as the time from the beginning of the first day We hypothesized that CGI methylation is related to the of systemic therapy to the detection of a progressive event response to therapy, as well as the survival of mRCC patients according to RECIST 1.1 criteria on a computer tomography undergoing antiangiogenic therapy. We investigated four candi- (CT) scan [24]. OS was defined as the period from the first day of date genes, three of which, cystatin E/M (CST6) and the micro systemic therapy until the patient’s death or censored at the last RNA genes miR-9-1 and miR-124-3, were identified recently with follow-up. The terminus ‘‘not evaluable’’ in Table 1 describes tumor-specific CGI hypermethylation and a possible association patients with a PFS ,2 months due to an early cessation of with the prognosis of RCC patients [19,21,22]. The Ladinin 1 therapy caused by toxicity or early death before the first (LAD1) gene was identified recently by our group as a new recommended CT scan after therapy was initiated. The initial candidate methylation marker in RCC showing univariate TNM classification of primary tumors was evaluated according to association with adverse clinicopathological parameters such as the Union for International Cancer Control 2002 classification tumor grade, lymph node metastasis, status of distinct metastasis [25]. Patient follow-up included up to three sequence changes in and advanced disease (unpublished data). the therapy regimen. LAD1 encodes an anchoring filament protein, a component of The terms prognostic` and predictive` were used according to the the basement membrane that likely contributes to the stability of definition by the National Cancer Institute [26]. the epithelial-mesenchymal interaction [23]. The present study investigated whether a DNA methylation Tissue Specimens, Isolation, and Bisulfite Conversion of mark can predict the response of targeted antiangiogenic therapy Tumor DNA of mRCC patients and describes the identification of two DNA Independent control of histopathology, tumor cell content of methylation markers in the CST6 and LAD1 CGIs as candidate routine pathological specimens, and the selection of tissue areas for epigenetic predictors of the PFS and OS of mRCC patients tissue extraction were determined by the pathologist. Subsequent- undergoing targeted therapy. ly, cylinders 1.5 mm in length and 2 mm in height were stamped out from the formalin-fixed and paraffin-embedded (FFPE) tissue Materials and Methods blocks using an 18 Charrie`re core stamp and subjected to DNA Ethics Statement isolation. Genomic DNA was extracted using the automated MagNA Pure LC 2.0 system and MagNA Pure LC DNA isolation Informed consent was obtained from each patient, and the local kit II - tissue (Roche Diagnostics Deutschland, Roche Applied ethics committee (Ethic Committee; Prof. H. D. Tro¨ger, Science, Mannheim, Germany). The quality of extracted DNA Hannover Medical School, Carl-Neuberg-Str. 1, Hannover, was assessed using spectrophotometry, gel electrophoresis, and Germany; Study_No: 1213–2011) specifically approved this study. quantitative PCR, which characterized the yield, purity, and grade Patients agreed in a written form for utilization of tissue specimen of degradation of isolated DNA. Bisulfite conversion of DNA was

Table 2. Gene informations.

CST6 LAD1 miR-9-1 miR-124-3

Chromosome 11q13 1q32.1 1q22 20q13.33 Name Cystatin E/M Ladinin 1 micro RNA9-1 micro RNA124-3 GeneID 1474 3898 407046 406909 CpG Island # number of CpG sites 59 54 99 424 # base position (bp) 65779312–65777967 201368561–201369032 156390404–156391581 61806255–61810867 bp of CpG sites 65779535, ,541, ,600, ,604, 201368651, ,669, ,672, ,689, 156390684, ,701, ,745, 61809002, ,007, ,026, investigated by ,612, ,620, ,630, ,640, ,693, ,696, ,700, ,704, ,713, ,747, ,753, ,758, ,035, ,044, ,059, qMSP ,644, ,647 ,725, ,733 ,764, ,783 ,065, ,072

Note: Gene informations according to the USCS Genome Browser [27]. doi:10.1371/journal.pone.0091440.t002

PLOS ONE | www.plosone.org 3 March 2014 | Volume 9 | Issue 3 | e91440 Epigenetic Prediction of Therapy Failure in mRCC

Figure 1. Survival analyses. A and D. istribution of the relative methylation values of CST6 (A) and LAD1 (D) in mRCC patients. A cutoff value is presented for dichotomization. B and E. Kaplan-Meier plots of the progression-free survival of mRCC patients dichotomized by high and low methylation of CST6 (B) and LAD1 (E). C and F Kaplan-Meier plots of the overall survival of mRCC patients dichotomized by high and low methylation of CST6 (C) and LAD1 (F). doi:10.1371/journal.pone.0091440.g001 carried out using the EZ DNA Methylation-Gold Kit (Zymo in CGI methylation determined by real-time detection and Research Corporation, Irvine CA, USA) and 1 mg of isolated independent internal control measurements to the corresponding DNA. Fully methylated and converted DNA, as well as difference in the fully methylated DNA control samples as unmethylated bisulfite-converted DNA controls, were used as described previously [21,28]. reported previously [21]. Statistical Analysis of Survival Quantitative Methylation-specific Real-time PCR Analysis Kaplan-Meier plots were used to present relative survival in the Quantitative real-time fluorimetric 59 exonuclease PCR (qMSP) PFS and OS analyses following dichotomization of tumors into assays were performed to quantify the CGI methylation levels of high and low methylation phenotypes. The median survival and CST6, LAD1, hsa-miR-124-3, and hsa-miR-9-1. The methylation corresponding 95% confidence intervals (CIs) were reported. analysis of hsa-miR-124-3 was carried out as described previously Differences in PFS and OS were tested using log-rank statistics and [21]. qMSP systems were established for CST6, LAD1, and hsa- median survival ratios calculated. P-values ,0.05 were considered miR-9-1 using Beacon Designer software (PREMIER Biosoft, Palo significant. To allow a comparison with the literature, univariate Alto CA, USA). The base positions of investigated CGI sites for Cox regression analyses were performed to estimate hazard ratios CST6, LAD1, hsa-miR-124-3, and hsa-miR-9-1 are presented in (HRs). To calculate sensitivity and specificity for therapy failure, a Table 2. The base positions refer to the USCS Genome Browser PFS cutoff value of 6 months was used for dichotomization [29] [27]. The qMSP systems were characterized as described for the into therapy responders and non-responders. hsa-miR-124-3 methylation measurements [21]. Duplicate real- The heat map and receiver operating characteristic curves were time PCRs were performed on an ABI 7900HT (Life technologies, constructed using the heatmap2 and ROCR function in the R Foster City, USA) in 384-well plates as described previously [21]. package (version 2.11.0.1) with a default clustering algorithm and Experimenters were blinded to the histopathological and clinical gplot package [30]. status of the samples. Relative methylation levels were calculated as an analogue of the DDCt method by normalizing the difference

PLOS ONE | www.plosone.org 4 March 2014 | Volume 9 | Issue 3 | e91440 Epigenetic Prediction of Therapy Failure in mRCC

Table 3. Survival analyses.

A) PFS Median survival (months, 95% CI) Median survival ratio (high/low) p-value* low methylation high methylation HR (95% CI)**

CST6 0.009 11.4 (6.2–NE) 2.0 (1.3–NE) 0.175 4.1 (1.3–12.6) LAD1 0.004 11.4 (3.0–NE) 2.0 (1.7–NE) 0.175 6.4 (1.6–26.0) miR-124-3 0.339 11.9 (6.2–NE) 2.6 (1.7–11.5) 0.218 1.8 (0.5–6.6) miR-9-1 0.319 4.6 (1.3–NE) 2.7 (1.8–NE) 0.587 1.7 (0.6–4.7)

B) OS Median survival (months, 95% CI) Median survival ratio (high/low)

p-value* low methylation high methylation HR (95% CI)**

CST6 0.011 22.9 (13.1–NE) 3.4 (2.5–NE) 0.148 4.1 (13.0–13.4) LAD1 0.043 16.4 (11.7–NE) 3.4 (3.0–NE) 0.207 2.9 (1.0–8.6) miR-124-3 0.786 13.7 (11.7–NE) 9.8 (3.2–29.8) 0.715 0.8 (0.2–3.1) miR-9-1 0.624 12.4 (3.4–NE) 14.4 (3.2–NE) 1.161 1.3 (0.5–3.6)

Abbreviations: PFS: Progression-free survival. OS: Overall survival. NE: not estimable. HR: Hazard ratio. CI: Confidence interval. *: log-rank statistical analysis. **: Univariate Cox regression for purpose of comparision. low methylation cutoff ,8.75. high methylation cutoff $8.75. doi:10.1371/journal.pone.0091440.t003

Results Analysis of Sensitivity and Specificity To determine the sensitivity and specificity of CST6 and LAD1 Bimodal Distribution of Relative Methylation Levels in the methylation analyses for predicting first-line therapy failure, CGIs of Candidate Genes methylation values were dichotomized into low and high Quantitative methylation analyses of the CGIs of CST6, LAD1, methylation phenotypes using the same cutoff value of 0.02% as hsa-miR-124-3, and hsa-miR-9-1 revealed the presence of a specified above. PFS values were dichotomized using a cutoff of 6 bimodal distribution of relative methylation values (Figure 1A months, a parameter that was previously suggested to better and D; data not shown for hsa-miR-124-3 and hsa-miR-9-1). distinguish between therapy responders and non-responders [29]. Applying a single cutoff value of 0.02% (corresponding to 28.75 High methylation of LAD1 and CST6 was a characteristic of failed in the ln-scale used for Kernel density plots in Figure 1A and D) therapy (Figure 2A). In the case of LAD1, all eight patients with for relative methylation, high and low methylated epigenotypes high methylation were non-responders. The specificity was 1.0 were uniformly distinguished for all of the analyzed genes and used (95% CI 0.65–1.0) and sensitivity 0.73 (95% CI 0.43–0.90) for the for consistent dichotomization in survival analyses. detection of therapy failure using LAD1 methylation (Table 4), whereas the specificity was 0.86 (95% CI 0.49–0.97) and sensitivity Analysis of PFS 0.82 (95% CI 0.52–0.95) for CST6 methylation (Table 4). Kaplan-Meier and log rank analysis of PFS in high and low methylated tumors demonstrated a significant difference for both Discussion CST6 and LAD1. High methylation was associated with a median The clinical outcomes of patients with mRCC have improved survival of 2.0 months, compared to 11.4 months among patients since VEGF-targeted therapies and mTOR inhibitors were made with low methylation (p = 0.009 and p = 0.004, Table 3A). In available [2,3]. However, the stratification of patients using miR-124-3 miR-9-1 contrast, neither nor demonstrated a statistical biomarkers could allow a better understanding of drug resistance relationship with PFS (p = 0.339 and p = 0.319). and identify an optimized patient-specific sequence of antiangiogenic therapies, improving individual survival [7]. Moreover, the Analysis of OS side effects of anti-VEGF-based regimens, such as diarrhea, rash, Kaplan-Meier analysis and log rank statistics revealed that high hand-foot syndrome, hypertension, and asthenia, which often methylation of CST6 and LAD1 was associated with impaired OS. severely impair quality of life during treatment can be minimized. A median OS of 22.9 and 3.4 months (p = 0.011, Table 3B) was We found that DNA hypermethylation of CST6 and LAD1 in obtained for low and high CST6 methylation, respectively. A primary RCC tumor tissue is significantly associated with the PFS median OS of 16.4 and 3.4 months (p = 0.043, Table 3B) was of patients receiving anti-VEGF-based medication as a first-line obtained for low and high LAD1 methylation. therapy and also the OS of patients sequentially treated with anti- VEGF targeted drugs and mTOR inhibitors in second- and third- line therapy. Our methylation markers predicted therapy failure with high specificity and good sensitivity.

PLOS ONE | www.plosone.org 5 March 2014 | Volume 9 | Issue 3 | e91440 Epigenetic Prediction of Therapy Failure in mRCC

Figure 2. Heat map illustration of therapy response and Receiver Operating Characteristic Curves for CST6 and LAD1. A. Heat map of normalized relative methylation values (natural logarithm) detected in LAD1, CST6, miR-9-1, and miR-124-3 CGIs for each patient. Low to high methylation values are encoded as violet (low) to red (high) hues. The dashed and solid lines describe the median and individual methylation values, respectively. Patient numbers given on the left correspond to the numbering presented in Table 1. Therapy response (0) and therapy failure (1) are indicated for each patient on the right. Notably, all of the patients (no. 1–8) exhibiting high methylation of LAD1 and 9 of 10 patients (no. 1–9, 14) exhibiting high methylation (red colored) of CST6 were part of the non-responder (1) group. B. The receiver operating characteristics (ROC) curves illustrated the discrimination of methylation measurements and the area under the curve (AUC) shows that even with our small patient cohort, a robust result for the accuracy of both methylation markers (AUC CST6 = 0.88 and AUC LAD1 = 0.90) can be detected. The sensitivity (true positive rate) is plotted against 1-specificity (false positive rate). doi:10.1371/journal.pone.0091440.g002

To the best of our knowledge, these findings are unprecedented beyond the measurement of gene variants in several important in several respects, as previous studies were not tissue based and aspects. First, the potential LAD1 and CST6 DNA methylation- either provided no significant association with therapy response based markers were measured in tumor cells, which directly [12] or reported only limited statistical power [11]. While serum exhibit tumor characteristics that may represent drivers of measurements of CA9 levels revealed no significant differences resistance and biological aggressiveness. Hypermethylation of between therapy responders and non-responders [12] the analysis CST6 and LAD1 exhibited prognostic and predictive value in of genetic variants, possibly interacting with the sunitinib pathway, our study and is a putative biomarker for patient selection. Based identified two VEGFR3 SNPs to be associated with therapy on the clinical outcomes in our study, different therapeutic response and PFS, but not with OS [11]. This study shows that strategies for hypermethylated tumors will be required. After the individual biological variables may affect the response to therapy. network of epigenetic alterations and biological behaviors has been On the other hand, our methylation-based candidate predictors go

PLOS ONE | www.plosone.org 6 March 2014 | Volume 9 | Issue 3 | e91440 Epigenetic Prediction of Therapy Failure in mRCC

Table 4. Sensitivity and specificity.

Sensitivity 95% CI Specificity 95% CI p-value*

CST6 0.818 0.52–0.95 0.857 0.49–0.97 ,0.001 LAD1 0.727 0.43–0.90 1.000 0.65–1.00 0.004 miR-124-3 1.000 0.74–1.00 0.429 0.16–0.75 0.339 miR-9-1 0.636 0.35–0.85 0.571 0.25–0.84 0.319

CI: Confidence interval. *log-rank test. doi:10.1371/journal.pone.0091440.t004 untangled, additional novel targets of therapeutic interventions patients generally face a poor prognosis, and many tumors exhibit may be identified. high methylation for the miR genes as expected for candidate Whether the difference in epigenetic tissue- and genetic blood- prognosticators. based measurements accounts for both epigenetic markers being The independence from clinical or laboratory parameters could related to the PFS and OS of mRCC patients is an interesting not be determined in the present study because the low sample question. Gene variants were only associated with PFS, a surrogate numbers prevented multivariate analysis. Correspondingly, the endpoint for survival measurements in mRCC that has possible relevant questions whether markers could be combined to limitations [31]. To the best of our knowledge, a tissue-based optimize the predictive power or whether markers exhibit molecular marker has not previously been associated with OS. redundant information can only be answered in future studies From a statistical point of view, our epigenetic study delivered by use of enlarged study cohorts. higher HRs in survival analyses and provided a more balanced However, the HRs observed for clinical parameters for patient classification into responders and non-responders than the study outcome were lower with limited accuracy/discriminatory power. by Garcia-Donas et al. [11], and therefore together contributing Our results require confirmation in an independent validation to a higher power of this study. Considering that a much smaller study including the consideration of clinical scoring systems as patient cohort was available for our measurements, our findings confounders. indicate that a strong effect has possibly been identified. Moreover, In conclusion, our study identified LAD1 and CST6 CGI bearing in mind that a growing number of agents can be used for methylation as two epigenetic markers that are associated with the the treatment of mRCC, future identification of an optimum PFS and OS of mRCC patients undergoing antiangiogenic therapy regimen could be facilitated by epigenetic markers that therapy. We have also shown the potential to improve the allow good separation of patients into responders and non- molecular prediction of the response to therapy. Our results responders. further stress the notion that epigenetically altered RCCs exist, Interestingly, the methylation levels of all candidate markers and novel specific strategies may be required to treat patients with clearly decayed into easily distinguishable high and low methyl- such tumors. ation groups, eliminating the need to arbitrarily define cutoff points for dichotomization. Therefore, virtually no overlap existed between the responders and non-responders in the present study. Acknowledgments Thus our LAD1 and CST6 methylation analyses yielded high We thank Margrit Hepke and Christel Reese for technical assistance. specificities of 1.0 and 0.86 for the detection of therapy failure, underlining the possible relevance of these markers in mRCC. Author Contributions This study may also answer whether DNA methylation-based prognosticators represent appropriate predictors of disease. Both Conceived and designed the experiments: IP JS. Performed the miR-9-1 and miR-124-3 [21,22] failed as predictors because no experiments: ND KG. Analyzed the data: JS RS. Contributed reagents/ materials/analysis tools: CS MK MA. Wrote the paper: IP JS. Reviewed association was found with the PFS or OS of patients undergoing the manuscript: ASM MAK VG. therapy. This finding might be explained by the fact that mRCC

References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, et al. (2011) Global cancer Metastatic Renal-Cell Carcinoma Database Consortium prognostic model: a statistics. CA Cancer J Clin. 2011/02/08 ed. population-based study. The lancet oncology 14: 141–148. 2. Motzer RJ, Hutson TE, Tomczak P, Michaelson MD, Bukowski RM, et al. 7. Galsky MD (2013) A prognostic model for metastatic renal-cell carcinoma. (2009) Overall survival and updated results for sunitinib compared with Lancet Oncol 14: 102–103. interferon alfa in patients with metastatic renal cell carcinoma. J Clin Oncol 27: 8. Nickerson P (2008) The impact of immune gene polymorphisms in kidney and 3584–3590. liver transplantation. Clinics in laboratory medicine 28: 455–468, vii. 3. Sternberg CN, Hawkins RE, Wagstaff J, Salman P, Mardiak J, et al. (2013) A 9. Dalgliesh GL, Furge K, Greenman C, Chen L, Bignell G, et al. (2010) randomised, double-blind phase III study of pazopanib in patients with Systematic sequencing of renal carcinoma reveals inactivation of histone advanced and/or metastatic renal cell carcinoma: Final overall survival results modifying genes. Nature 463: 360–363. and safety update. European journal of cancer 49: 1287–1296. 10. Choueiri TK, Vaziri SA, Jaeger E, Elson P, Wood L, et al. (2008) von Hippel- 4. Skolarikos AA, Papatsoris AG, Alivizatos G, Deliveliotis C (2006) Molecular Lindau gene status and response to vascular endothelial growth factor targeted pathogenetics of renal cancer. Am J Nephrol 26: 218–231. therapy for metastatic clear cell renal cell carcinoma. J Urol 180: 860–865; 5. Motzer RJ, Bacik J, Schwartz LH, Reuter V, Russo P, et al. (2004) Prognostic discussion 865–866. factors for survival in previously treated patients with metastatic renal cell 11. Garcia-Donas J, Esteban E, Leandro-Garcia LJ, Castellano DE, del Alba AG, et carcinoma. J Clin Oncol 22: 454–463. al. (2011) Single nucleotide polymorphism associations with response and toxic 6. Heng DY, Xie W, Regan MM, Harshman LC, Bjarnason GA, et al. (2013) effects in patients with advanced renal-cell carcinoma treated with first-line External validation and comparison with other models of the International sunitinib: a multicentre, observational, prospective study. Lancet Oncol 12: 1143–1150.

PLOS ONE | www.plosone.org 7 March 2014 | Volume 9 | Issue 3 | e91440 Epigenetic Prediction of Therapy Failure in mRCC

12. Gigante M, Li G, Ferlay C, Perol D, Blanc E, et al. (2012) Prognostic value of and disease recurrence of patients with clear cell renal cell carcinoma. Br J Cancer serum CA9 in patients with metastatic clear cell renal cell carcinoma under 108: 131–138. targeted therapy. Anticancer Res 32: 5447–5451. 22. Hildebrandt MA, Gu J, Lin J, Ye Y, Tan W, et al. (2010) Hsa-miR-9 13. Sun M, Shariat SF, Cheng C, Ficarra V, Murai M, et al. (2011) Prognostic methylation status is associated with cancer development and metastatic factors and predictive models in renal cell carcinoma: a contemporary review. recurrence in patients with clear cell renal cell carcinoma. Oncogene 29: Eur Urol 60: 644–661. 5724–5728. 14. Varela I, Tarpey P, Raine K, Huang D, Ong CK, et al. (2011) Exome 23. Motoki K, Megahed M, LaForgia S, Uitto J (1997) Cloning and chromosomal sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 mapping of mouse ladinin, a novel basement membrane zone component. in renal carcinoma. Nature 469: 539–542. Genomics 39: 323–330. 15. Baldewijns MM, van Vlodrop IJ, Schouten LJ, Soetekouw PM, de Bruine AP, et 24. Sohaib A (2012) RECIST rules. Cancer imaging : the official publication of the al. (2008) Genetics and epigenetics of renal cell cancer. Biochim Biophys Acta International Cancer Imaging Society 12: 345–346. 1785: 133–155. 25. Sobin LH, Compton CC (2010) TNM seventh edition: what’s new, what’s 16. Vanharanta S, Shu W, Brenet F, Hakimi AA, Heguy A, et al. (2013) Epigenetic changed: communication from the International Union Against Cancer and the expansion of VHL-HIF signal output drives multiorgan metastasis in renal American Joint Committee on Cancer. Cancer 116: 5336–5339. cancer. Nat Med 19: 50–56. 26. NCI National Cancer Institute Dictionary of Cancer Terms. 17. Larkin J, Goh XY, Vetter M, Pickering L, Swanton C (2012) Epigenetic 27. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, et al. (2002) The regulation in RCC: opportunities for therapeutic intervention? Nat Rev Urol 9: 147–155. human genome browser at UCSC. Genome Res 12: 996–1006. 18. Morris MR, Hesson LB, Wagner KJ, Morgan NV, Astuti D, et al. (2003) 28. Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, et al. (2005) Analysis Multigene methylation analysis of Wilms’ tumour and adult renal cell of repetitive element DNA methylation by MethyLight. Nucleic acids research carcinoma. Oncogene 22: 6794–6801. 33: 6823–6836. 19. Morris MR, Ricketts C, Gentle D, Abdulrahman M, Clarke N, et al. (2010) 29. Seidel C, Busch J, Weikert S, Steffens S, Fenner M, et al. (2012) Progression free Identification of candidate tumour suppressor genes frequently methylated in survival of first line vascular endothelial growth factor-targeted therapy is an renal cell carcinoma. Oncogene 30: 1390–1401. important prognostic parameter in patients with metastatic renal cell carcinoma. 20. van Vlodrop IJ, Baldewijns MM, Smits KM, Schouten LJ, van Neste L, et al. Eur J Cancer 48: 1023–1030. (2010) Prognostic significance of Gremlin1 (GREM1) promoter CpG island 30. Team RDC (2011) R: A language and environment for statistical computing R hypermethylation in clear cell renal cell carcinoma. Am J Pathol 176: 575–584. Foundation for Statistical Computing: Vienna, Austria. 21. Gebauer K, Peters I, Dubrowinskaja N, Hennenlotter J, Abbas M, et al. (2013) 31. Knox JJ (2008) Progression-free survival as endpoint in metastatic RCC? Lancet Hsa-mir-124–3 CpG island methylation is associated with advanced tumours 372: 427–429.

PLOS ONE | www.plosone.org 8 March 2014 | Volume 9 | Issue 3 | e91440 Cancer Medicine Open Access ORIGINAL RESEARCH Neuroﬁlament Heavy polypeptide CpG island methylation associates with prognosis of renal cell carcinoma and prediction of antivascular endothelial growth factor therapy response Natalia Dubrowinskaja1, Kai Gebauer1, Inga Peters1,Jo¨ rg Hennenlotter2, Mahmoud Abbas3, Ralph Scherer4, Hossein Tezval1, Axel S. Merseburger1, Arnulf Stenzl2, Viktor Grunwald€ 5, Markus A. Kuczyk1 &Ju¨ rgen Serth1 1Department of Urology, Hannover Medical School, 30625 Hannover, Germany 2Department of Urology, Eberhard-Karls University, 72074 Tu¨bingen, Germany 3Department of Pathology, Hannover Medical School, 30625 Hannover, Germany 4Department of Biometry, Hannover Medical School, 30625 Hannover, Germany 5Department of Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, 30625 Hannover, Germany

Keywords Abstract Methylation, risk assessment, translational research, urological oncology Neurofilament Heavy polypeptid (NEFH) belongs to the group of type IV intermediate filament proteins. DNA methylation of the NEFH promoter and loss of Correspondence expression have previously been shown to activate the AKT/b-catenin pathway Jurgen€ Serth, Department of Urology, in tumor cells. When identifying hypermethylation of the NEFH CpG island OE6247, Carl-Neuberg-Str.1, 30625 (CGI) in renal cell cancer (RCC) we asked whether methylation could provide Hannover, Germany. clinical or prognostic information for RCC and/or predict therapy response in Tel: +49-5115325398; Fax: +49-5115322926; E-mail: [email protected] patients with metastatic RCC (mRCC) undergoing antiangiogenic therapy. Rela- tive methylation of the NEFH CGI was analyzed in 132 RCC samples and 83 Funding Information paired normal tissues using quantitative methylation-specific PCR. Results were This work was supported by internal funding statistically compared with tumor histology, clinicopathological parameters, of the Hannover Medical School. progression-free survival (PFS) as well as with overall survival (OS) in a subset of 18 mRCC patients following antiangiogenic therapy regimens. The NEFH Received: 1 August 2013; Revised: 29 CGI methylation demonstrated a tumor-specific increase (P < 0.001), associa- October 2013; Accepted: 19 November 2013 tion with advanced disease (P < 0.001), and distant metastasis (P = 0.005).

Cancer Medicine 2014; 3(2): 300–309 Higher relative methylation was also signiﬁcantly associated with a poor PFS (HR = 8.6, P < 0.001) independent from the covariates age, gender, diameter of tumors, state of advanced disease, and local and distant metastasis. Median doi: 10.1002/cam4.181 OS following targeted therapy was 29.8 months for patients with low methylation versus 9.8 months for the group with high methylation (P = 0.028). We identiﬁed NEFH methylation as a candidate epigenetic marker for prognosis of RCC patients as well as prediction of anti-vascular endothelial growth factor- based therapy response.

Introduction achieved, patients with advanced disease still have a poor prognosis [2, 3]. Moreover, patient stratiﬁcation for clini- Renal cell cancer (RCC) is among the 10 most frequent cal trials is limited to the use of clinically based scorings causes of cancer death of men in western countries [1]. such as the Memorial Sloane Kettering Cancer Center Surgical treatments as nephrectomy or partial nephron- (MSKCC) or the Heng systems [4, 5]. In view of limita- sparing resection represent the standard therapy of local- tions of current prognostic models it has been argued that ized and locally advanced RCC. Although improvements biologically based markers could improve both the quality of overall survival (OS) due to anti-vascular endothelial of prognostic information and prediction of therapy growth factor (VEGF)-based therapies have been response [6]. Moreover, new biologically based markers

300 ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. N. Dubrowinskaja et al. NEFH Methylation and Renal Cell Cancer Prediction could identify, as a side effect, new molecular targets in renal cancer cell lines we already identified new potential signal transduction of RCC pivotal for progression of the DNA-methylation-based candidate prognosticators for disease [6]. ccRCC [19, 20] and also found the Neurofilament Heavy Clear-cell renal cell carcinoma (ccRCC) is the most fre- polypeptid (NEFH) CGI as a further epigenetic mark of quent histological entity of all RCCs. The question for the potential interest. NEFH is located on chromosome molecular architecture of changes underlying this tumor 22q12.2, encodes for a 200 kDa protein and is classified has not been fully answered at yet. Mutations in the von to the group of type IV intermediate filaments which are Hippel-Lindau (VHL) gene have been found in 52–83% important components of the neuronal cytoskeleton [24]. of ccRCCs patients[7–9] and changes in the polybromo 1 It has been reported that tumor-specific loss of NEFH (PBRM1) gene have been reported as the second most mRNA expression occurs in prostate carcinoma [25]. Fur- common mutation detectable in 23–54% of ccRCCs thermore, higher CGI methylation has been detected in [9–11] while other mutations such as of the SET domain normal esophageal mucosa cells of smokers, indicating containing 2 (SETD2) and BRCA1-associated protein-1 the presence of premalignant epigenetic alterations in pre- (BAP1) genes exhibited only low occurrence [8, 9]. cancerous lesions as a cancer risk factor [26]. Moreover, Therefore, rather few common mutations have been NEFH promoter methylation in esophageal squamous cell revealed to be associated with ccRCC despite exome and carcinoma (ESCC) has been functionally linked with loss genome-wide sequencing analyses of large patient cohorts of expression and activation of the v-akt murine thymo- thus far [9, 10]. ma viral oncogene homolog (AKT)/b-catenin pathway Statistical association of somatic mutations with also leading to increased glycolysis rates and changes in adverse clinical parameters such as higher nuclear grade, mitochondria [27]. necrosis and advanced stage along with evidence for a Here we identified a NEFH methylation marker that relationship with poor survival of patients have only been shows specific hypermethylation in RCC and is signifi- reported for the BAP1 gene [11]. Consistently, The Can- cantly associated with adverse clinicopathological parame- cer Genome Atlas network (TCGA) solely identified ters of the tumor as well as progression-free survival mutations in the BAP1 gene to be associated with a worse (PFS) of RCC patients. Moreover, NEFH methylation survival of patients [9]. Of note, the TCGA study also associates with OS of patients with metastatic disease showed that a great variety of overall rarely observed undergoing targeted therapy regimes. This study suggests genetic alterations including mutations and gains and NEFH methylation both as an independent prognosticator losses of sequences were found to be individually com- and predictor for patients with ccRCC and metastatic bined in tumors thus restraining the identification of sim- disease (mRCC). ple functional conclusions as well as of statistical relationships such as the clinical outcome of patients [9]. Material and Methods On the other hand, many epigenetic DNA-methylation- based alterations have already been reported to occur with Study design and patients a high frequency in ccRCC [9, 12–15] and to show high odds ratios for adverse clinical or pathological parameters Cross-sectional and prognostic analyses were carried out [14, 16–20]. Moreover, a subgroup of these methylation on 114 RCC fresh frozen samples and 83 corresponding markers demonstrated independence from important clin- histologically normal appearing samples (Table 1) as ical parameters, such as stage, grade, diameter of tumor, described previously [20]. Survival analyses for mRCC and status of local or distant metastasis [14, 16, 18, 20, following anti-VEGF-based therapy was done using a 21]. Interestingly, the most frequent common gene muta- cohort of 18 formalin-fixed and paraffin-embedded tions detected so far in ccRCC were either functionally (FFPE) samples (Table 2). Sample collection was appro- related to histone modification and stabilization, thus ved by the local ethics committee and informed consent mechanisms indented with expression states of genes and was obtained from each patient. TNM classification was DNA methylation [22], or, as in case of VHL, were dem- evaluated according to the Union for International Cancer onstrated to cause an increase in epigenetic alterations Control 2002 classification as described before [28]. including DNA methylation [23]. Consequently, epige- Localized and locally advanced RCC describe tumors with netic alterations could be a hallmark of RCC and might pT ≤ 3, lymph node (N) and metastasis (M) negative be an important mean for molecular-based prognosis or (N0, M0). Advanced tumors are pT = 4 and/or lymph prediction of this disease. node positive (N+) and/or positive for distant metastasis In the course of a combined in silico analysis of gene (M+). The histological grading was assessed according to expression and a genome-wide re-expression analysis Thoenes et al. [29]. The time from primary surgery to the using the demethylation agent 5-aza-2′-deoxycytidine in time of the first progressive event including local

ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. 301 NEFH Methylation and Renal Cell Cancer Prediction N. Dubrowinskaja et al.

Table 1. Patient characteristics. Table 2. Characteristics of patients with mRCC undergoing anti-VEGF-based therapy. All RCC survival All RCC % group % n (%)

Total cases 114 50 Total cases 18 (100) Histology Histology ccRCC 82 71.9 39 78 ccRCC 16 (89) papRCC 24 20.2 10 20 Papillary 1 (6) Mixed histology 4 3.5 1 2 Chromophobe 1 (6) Not class. 4 3.5 0 Gender Gender Female 7 (39) Female 40 35 19 38.0 Male 11 (61) Male 74 65 31 62.0 Distant metastasis1 Age, median (years) 65 66.5 M0 0 Distant metastasis M+ 3 (17) M0 88 77.2 41 82.0 Mx 15 (83) M+ 26 22.8 9 18.0 Lymph node metastasis1 Lymph node metastasis N0 2 (11) N0 100 87.7 47 94.0 N1 1 (6) N+ 14 12.3 3 6.0 N2 1 (6) T-classification Nx 14 (78) pT1 11 9.6 1 2.0 T-classification1 pT1a 34 29.8 20 40.0 1 1 (6) pT1b 19 16.7 9 18.0 1a 2 (11) pT2 7 6.1 3 6 1b 2 (11) pT3 5 4.4 2 4 2a 2 (11) pT3a 9 7.9 2 4 3b 6 (33) pT3b/c 24 21.1 12 24 4 2 (11) pT4 1 0.9 0 x 3 (17) NA 4 0.9 1 2.0 Differentiation Differentiation G1 0 (0) G1 22 19.3 6 12.0 G2 12 (67) G1–2 15 13.2 8 16.0 G3 6 (33) G2 58 50.9 28 56.0 TKI-first line G2–G3 8 7.0 3 6.0 Sunitinib 12 (67) G3 11 9.5 5 10.0 Sorafenib 4 (22) State of disease Bevacizumab 1 (6) Loc./Loc.Adv.Disease1 81 71.1 39 78.0 Axitinib 1 (6) Adv. Disease2 32 28.1 11 22.0 Death within follow-up NA 1 0.9 0 No 4 (22) Paired samples Yes 14 (78) All RCC 83 Age, median, min-max (y) 59.5 (48–80) ccRCC 63 OS, median, min-max (m) 12.4 (0.8–59.3) ccRCC, clear cell renal carcinoma; papRCC, papillary renal cell carci- ccRCC, clear cell renal carcinoma; TKI, tyrosine kinase inhibitor; OS, noma. overall survival; mRCC, metastatic renal cell carcinoma. 1pT ≤ 3, N0, M0. 1TNM status refers to the initial histopathological evaluation after 2pT = 4 and/or N+ or M+. nephrectomy. recurrence or a new metastatic site detected by computer identity control by the manufacturer (Cell line services, tomography scan was designated as PFS independent Heidelberg, Germany; Lonza, Basel, Switzerland) exclu- from the initial TNM status. OS was the period of the sively for the purpose of DNA isolation as described pre- first day of systemic therapy until patient’s death or the viously [19, 20]. last day of follow-up. DNA isolation, bisulfite conversion of DNA, Cell lines and quality measurements Human tumor cell lines and primary cells were Isolation of DNA from frozen sections and histopatholog- short-term cultured immediately following purchase and ical evaluation of control sections for estimation of tumor

302 ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. N. Dubrowinskaja et al. NEFH Methylation and Renal Cell Cancer Prediction cell content were performed as described before [20]. transformed into the natural logarithmic scale before con- FFPE samples were punched out as cylinders of approxi- ducting further statistical calculations. Linearity and PCR mately 2 mm height and 1.5 mm diameter by the pathol- efficiency of the qMSP assay were analyzed using linear ogist following examination for histopathology and tumor regression analysis. Mean relative methylation levels cell content. DNA was isolated using an automated observed for paired tumor and adjacent normal appearing MagNA Pure LC system (Roche Diagnostics Deutschland, tissues were compared using the two-sided t-test for Roche Applied Science, Mannheim, Germany). Extracted paired data. Independent tissue sample groups were com- DNA was characterized for yield, purity, and length dis- pared using univariate logistic regression analysis. Dichot- tribution by the use of spectralphotometry and agarose omization of methylation levels for analysis of PFS were gel electrophoresis and then subjected to bisulfite conver- performed using the R package “maxstat” providing sion using the EZ DNA Methylation-GoldTM Kit (Zymo calculation of the optimum threshold. Analysis of PFS Research Corporation, Irvine CA, USA). Yield and degree was carried out using univariate Cox regression analysis. of conversion of converted DNA was controlled by inde- Thus, P-values and hazard ratios (HR) could be calcu- pendent measurements of a repetitive sequence ALU-C4 lated for comparison with results of bivariate analyses (QC1) and a single copy b-Actin (ACTB) sequence considering covariates. OS of patients undergoing targeted (QC-2) as described before.20 Fully methylated and therapy was calculated using log Rank statistics. converted control DNA (M) as well as unmethylated bisulfite-converted control DNA (U) were prepared as Results described previously [20, 30]. Measurement of NEFH CGI methylation in Quantitative methylation-specific PCR technical controls, normal primary cells, and (qMSP) analysis tumor cell lines Quantitation of NEFH methylation was carried out by a The analysis of converted methylated (M), converted quantitative real-time fluorimetric 5′ exonuclease PCR non-methylated (U), and non-converted DNA control assay. The qMSP primers 5′-ACCCGACCGCGACGACTA- samples demonstrated that the NEFH qMSP specifically TA-3′ (forward), 5′-CGTCGAAGTTTATTATGGTTTGAG detects M–DNA while U and non-converted samples â TAGG-3′ (reverse) and the Taqman probe 5′-FAM-CG remained undeterminable both in the QC1 control reac-

CCCTAATACTACCGCAATACCTCCCGC-BHQ-3 were tion as well as in the NEFH-specific PCR (Ct > 45 cycles, created by use of the Beacon DesignerTM software (PRE- Figure 1B). A log2 dilution series adjusted for a constant MIER Biosoft, Palo Alto CA, USA). The qMSP analysis total amount of converted DNA showed good efficiency covered nine CpG sites on chromosome 22 at positions and high linearity of the qMSP assay (Figure 1C). Using 29,876,165, ~169, ~171, ~174, ~206, ~218, ~231, ~263, and the slope of the qMSP calibration line we calculated a ~266 according to the GRCh37/hg19 annotation in the PCR efficiency of 0.95. Measurement of methylation in UCSC genome browser [31, 32]. Real-time PCRs were per- cell lines used as substitutes for frequent human cancers formed as duplicates on an ABI 7900HT (Life technologies, as well as normal primary cells and control DNAs Foster City, CA, USA) in 384-well plates using an auto- demonstrated no methylation neither in normal primary mated liquid handling system as described previously [20]. cells of renal or prostatic origin nor in mock controls The experimenters carried out measurements without (Figure 1D). Low methylation was detected for normal knowledge of type, order, clinicopathological or survival mammary primary cells. Breast and urothelial cancer cell status of samples. Relative methylation levels were calcu- lines overall demonstrated highest relative methylation lated as an analogue of the delta–delta Ct method by nor- values while >25% relative methylation was detected in malizing the difference of NEFH methylation-specific real- two of six RCC and one of three prostate cancer cell lines time detection and methylation independent internal QC1 (Figure 1D). control measurement with the corresponding difference for the fully methylated DNA control samples as described The NEFH CGI shows hypermethylation in previously [20]. RCC The comparison of relative methylation as observed in Statistical analyses paired tumor (TU) and adjacent normal (adN) tissues Explorative statistical data analyses were performed revealed the presence of a tumor-specific hypermethyla- using the statistical software R 2.15 [33]. P<0.05 was tion (P < 0.001). Nearly all of the paired TU versus adN considered as significant. Relative methylation levels were comparisons demonstrated a substantial increase in

ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. 303 NEFH Methylation and Renal Cell Cancer Prediction N. Dubrowinskaja et al.

A TSS Exon 2 Exon 1

CpG island

QMSP

B D ZM 6 NC- DNA Controls 1 21 Mock RPTEC Normal Prec primary cells HMEC 4 2,3,4 Delta Rn NEFH RCC−MF RCC−HS RCC−GS RCC QC1 control 3,4 ACHN 2 A498 786−O 0 10203040 PC−3 LN−cap ProstateCa Cycle number DU−145 T24 RT112 UrothelialCa C HB−CLS2 HB−CLS1 20 EJ28 CLS−439 5637 MDA-MB231 18 MCF-010a MCF-007 HCC-1937 BreastCa 16 HCC-1806 HCC-1599 HCC-1395 Delta Ct value HCC-38 14 SKX msc. HeLa 0 25 50 75 100 0.03 0.1 0.3 1 Relative methylation Relative methylation index

Figure 1. (A) Neurofilament Heavy polypeptid (NEFH)–CpG island (CGI) organization and location of the qMSP assay relative to the transcription start site (TSS) and exon 1 and 2. CpG sites are indicated by vertical lines. (B) qMSP data for control measurements in duplicate of fully methylated DNA (1), unmethylated DNA (2), unconverted DNA (3) and the blank sample (4). (C) Measurement of assay linearity and efficiency using a twofold dilution series of fully methylated in non-methylated control DNA. (D) Relative methylation levels determined by qMSP for control samples, normal primary cells and various cancer cell lines representing human cancers as specified. tumor-specific methylation (Fig. 2A and B). The median related with higher methylation in tumors. No significant relative methylation values found in adN and TU tissues difference was found between tumors of clear cell and corresponded to 0.087% and 0.634% indicating a 7.3-fold papillary histology (P = 0.270, OR 0.87 [0.68–1.11, 95% average increase in highly methylated sequences in CI]). Thus, all subsequent statistical analyses were carried tumors. NEFH methylation of normal and cancer tissues out without consideration of tumor histology. were not correlated (P = 0.65, r = 0.05, Pearson′s correlation analysis). NEFH methylation is independently associated with decreased PFS of patients NEFH methylation and association with Patients measured with methylation higher than the sta- clinicopathological parameters tistically calculated cutoff value of 5.9% relative methyla- We found that higher methylation of the NEFH sub- tion showed a significantly shortened PFS (P < 0.001, region analyzed was statistically associated with the status HR = 8.61 [3.03–24.5, 95%CI], Table 3B). The corre- of distant metastasis, advanced disease, and high-grade sponding Kaplan–Meier plot shows that six out of seven tumors (P = 0.005, OR 1.46 [1.12–1.91, 95%CI]; (86%) patients with higher methylation were found with P < 0.001, OR 1.56 [1.20–2.03, 95%CI]; P = 0.012, OR disease progression within 30 months (Figure 3B). In 1.46 [1.09–1.95, 95%CI], Figure 3A, Table 3A). Age, gen- contrast 80% of tumors identified with low methylation der, diameter of tumors, and lymph node status were not did not show disease progression within 70 months of

304 ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. N. Dubrowinskaja et al. NEFH Methylation and Renal Cell Cancer Prediction

A 0 B 2 − 4 − 6 − RML (ln) RML 8 02 68 − 24 Diff. TU - adN RML (ln) TU - adN RML Diff. 10 − −

adN TU

Figure 2. Tumor-specific hypermethylation of the Neurofilament Heavy polypeptid (NEFH)–CpG island (CGI) sub-region analyzed. (A) Pairwise comparison of relative methylation levels (RML) in natural logarithmic scale for adjacent normal appearing (adN) and paired tumor tissue (TU) samples (B) Assorted differences of RML observed in adN and TU samples. follow-up. High HRs were not only observed in univariate distant metastasis, state of advanced disease as well as poor but also in pairwise bivariate Cox regression analyses con- differentiation of tumor cells. Hence, NEFH appears as a sidering status of distant and lymph node metastasis, state candidate for functional alterations occurring in aggressive of localized or locally advanced and advanced disease, RCCs. This understanding is sustained by our finding that gender, age, tumor diameter, and tumor differentiation as higher NEFH methylation is associated both with PFS of covariates. Bivariate HRs ranged between 5.1 and 17.1 patients of the prognosis cohort as well as OS of the and for the parameter methylation significance was patients undergoing anti-VEGF-based therapy. Interest- observed in all bivariate statistical evaluations (P < 0.001– ingly, we found that methylation appeared as a significant P = 0.022, Table 3C). variable for PFS independent from all available clinical and pathological confounders including state of local and distant metastasis, states of localized/locally advanced or NEFH methylation is associated with OS of advanced disease, tumor diameter, grade of tumors, age, patients undergoing antiangiogenic therapy and gender. NEFH methylation remained as a highly sig- The analysis of NEFH methylation in tumor samples of nificant factor exhibiting high and notable constant HRs in patients undergoing targeted therapy for treatment of all bivariate survival analyses suggesting this marker as mRCC revealed a bimodal distribution of methylation independent prognosticator for RCC. Thus, this study values identifying low- and high-methylated tissue groups shows to the best of our knowledge for the first time, that making a statistical calculation of a cutoff value NEFH methylation is associated with the survival of RCC for dichotomization redundant (Fig. 3C). To evaluate patients and is a candidate predictor for the response of whether both groups differ with respect to the OS of multiple sequential-targeted therapies in advanced RCC. patients we performed a Kaplan–Meier survival analysis Loss of expression of NEFH was first found in prostatic and log-rank statistics (Fig. 3C). We found a median OS tumors [25], while epigenetic alterations of NEFH have ini- of 29.8 (11.7-NE) months for patients with low methyla- tially been reported in the context of “field cancerization” tion in tumor tissues while patients with higher methyla- to occur as premalignant DNA-methylation events detect- tion demonstrated a mean OS of 9.8 (3.0-NE) months able in normal epithelial cells of tissues at higher tumor (P = 0.028). Using a cutoff of 6 months for PFS as risk [26]. Subsequently, NEFH was functionally identified recently suggested as prognosticator of OS [34], analysis to exhibit attributes of a tumor suppressor as knockdown of NEFH methylation allows detection of therapy failure experiments revealed increased tumorigenicity in mice with a sensitivity of 0.91 (0.62–0.98, 95% CI). while expression of NEFH was associated with diminished cell growth and reduced colony formation in vitro [27]. Discussion Moreover, DNA methylation of NEFH associated with high-grade and -stage of ESCC and epigenetic silencing Our methylation analyses showed that higher relative caused activation of the AKT/b-catenin pathway [27], methylation levels of distinct CpG sites within the NEFH therefore, indicating that epigenetic alterations of NEFH CGI are statistically related with unfavorable clinical and could contribute to human carcinogenesis. Considering pathological characteristics of RCCs such as presence of that NEFH methylation or altered expression have not

ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. 305 NEFH Methylation and Renal Cell Cancer Prediction N. Dubrowinskaja et al.

A 20 − 4 − 6 − RML (ln) 8 − 10 −

Kernel distribution cc pap G low G highM0 M1 Loc./ Adv. Loc. Adv.

20 cut off −

4 RML < cut off − 6 − RML (ln) 8 − RML 10

Progression free survival > cut off − 0.0 0.2 0.4 0.6 0.8 1.0 Kernel distribution 0 10203040506070 Time (months)

C 20 −

4 RML < cut off − 6 −

RML (ln) cut off 8 − Overall survival RML

10 > cut off − 0.0 0.2 0.4 0.6 0.8 1.0 Kernel distribution 0 102030405060 Time (months)

Figure 3. (A) Distribution with kernel density estimation for relative methylation levels (RML) in natural logarithmic scale as detected in all tumor samples (left panel). Box plot analyses showing tumor subset-speciﬁc relative methylation levels for clear cell (cc) or papillary (pap) histology, low- (G ≤ 2) or high-grade (G > 2) tumors, negative (M0) or metastasis-positive (M+) patients and localized or locally advanced (pT ≤ 3, N0, M0) or advanced (pT > 3 and/or N+,M+) disease (B) Distribution of relative methylation values in the survival analysis group with kernel distribution and indication of the statistically optimized cutoff value of RML of 5.9% (À2.85 in the natural logarithmic scale, left panel). Kaplan–Meier plot analysis illustrating relative progression-free survival of all renal cell cancer (RCC) patients following dichotomization (right panel). (C) Distribution and cutoff level of relative methylation levels observed for the therapy group (left panel) and Kaplan–Meier plot analysis (right panel) showing relative overall survival of patients under targeted therapies for treatment of advanced disease. been reported for cancers other than ESCC thus far, our The NEFH methylation mark may be also of clinical ﬁndings of hypermethylation taking place both in RCC as interest because it demonstrated independence from both a highly malignant tumor and in tumor cell lines from grade as well as status of distant metastasis, which seems mammary, prostatic, and urothelial cancers further under- remarkable as both factors represent strong classical clini- line the potential relevance of NEFH in carcinogenesis. copathological parameters for the prognosis of disease

306 ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. N. Dubrowinskaja et al. NEFH Methylation and Renal Cell Cancer Prediction

Table 3. Statistical association between NEFH CGI methylation and clinicopathology of patients with RCC; (A) Univariate logistic regression analysis for tumor group comparisons; (B) Univariate analysis of progression-free survival using Cox regression; (C) Bivariate analysis of progression-free survival using Cox regression.

(A) Parameter of dichotomization Median RML1 Median RML1 P-value2 OR 95% CI Adjusted P-value2

Dist. metastasis (M0/M +) À5.5 À4.3 0.005 1.46 1.12–1.91 0.019 Lymph node met. (N0/N +) À5.4 À5.0 0.202 1.22 0.9–1.66 0.362 Grade (low/high) À5.4 À3.6 0.012 1.46 1.09–1.95 0.035 Diameter3 À5.3 À5.5 0.178 1.19 0.93–1.52 0.362 Loc.&loc.adv.dis./adv. dis. À5.6 À4.5 <0.001 1.56 1.20–2.03 0.005

(B) Variate P-value4 HR 95% CI

NEFH methylation <0.001 8.61 3.03–24.5 Metastasis <0.001 5.88 2.17–15.9 Lymph node status 0.283 2.26 0.51–10.0 Loc.&loc.adv.dis./adv. dis. <0.001 5.33 1.99–14.3 Gender 0.187 2.14 0.69–24.5 Age5 0.560 0.75 0.28–2.00 Diameter5 0.200 2.23 0.65–7.63

NEHF methylation <0.001 11 3.53–34.2 Metastasis <0.001 7.27 2.52–20.9 NEHF methylation <0.001 6.96 2.35–20.7 Loc &loc.adv.disease/adv. disease 0.005 4.21 1.53–11.6 NEHF methylation <0.001 7.88 2.75–22.6 Gender 0.357 1.72 0.54–5.42 NEHF methylation <0.001 17.1 4.62–63.3 Age5 0.043 0.28 0.08–0.96 NEHF methylation 0.022 5.07 1.26–20.4 Diameter5 0.368 1.79 0.50–6.38

CI, confidence interval; HR, hazard ratio; OR, odds ratio; RML, Relative methylation level (natural logarithmic scale); CGI, CpG Island; RCC, renal cell carcinoma. 1Median relative methylation level (RML) of dichotomized groups. 2Univariate logistic regression, correction for multiple testing by Holms. 3Localized or locally advanced (T ≤ 3. N0, M0) and advanced disease (pT = 4 and/or N +,M+). 4Univariate Cox regression analysis. 5Dichotomized using median of parameter. 6Bivariate Cox regression analysis. progression. It is known that patients undergoing antian- of 91%. To the best of our knowledge, such a result has giogenic therapy for treatment of metastatic RCC exhibit not been reported for any DNA-based marker at yet, sug- different survival characteristics [34]. We, therefore, inves- gesting this epigenetic mark as a promising candidate tigated in a pilot study whether NEFH methylation is also molecular prognosticator and predictor for RCC. associated with survival of patients treated for advanced This study identified statistical associations between an disease and found high- and low-methylated primary epigenetic alteration and worse patient survival and ther- tumors showing nearly no overlap between both groups. apy response. Thus, this study provides evidence that an Increased methylation associated with a strongly shortened unknown functional relationship of NEFH and cellular median OS and using a cut point of 6 months for PFS as a signaling might exist contributing to the development of prognosticator of OS [34], analysis of NEFH methylation aggressive RCC. Targeted therapeutic intervention in RCC would allow detection of therapy failure with a sensitivity aims at specific components of either the AKT/mamma-

ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. 307 NEFH Methylation and Renal Cell Cancer Prediction N. Dubrowinskaja et al. lian target of rapamycin (mTOR) or the mitogen-acti- 5. Heng, D. Y., W. Xie, M. M. Regan, L. C. Harshman, vated protein kinase (MAPK) pathways, together provid- G. A. Bjarnason, U. N. Vaishampayan, et al. 2013. External ing hypoxia-related signal transduction in RCC [35]. validation and comparison with other models of the Notably, NEFH shows functional interaction with both International Metastatic Renal-Cell Carcinoma Database pathways. First, knockdown experiments in esophageal Consortium prognostic model: a population-based study. cancer cell lines revealed that NEFH is connected with the Lancet Oncol. 14:141–148. AKT pathway via Gsk3ß [27], and, moreover, it has been 6. Galsky, M. D. 2013. A prognostic model for metastatic reported that AKT/mTOR pathway alterations occur in renal-cell carcinoma. Lancet Oncol. 14:102–103. RCC and affect prognosis of patients [36]. Considering 7. Nickerson, M. L., E. Jaeger, Y. Shi, J. A. Durocher, that NEFH has been suggested to be functionally associ- S. Mahurkar, D. Zaridze, et al. 2008. Improved ated with altered metabolism of tumor cells such as identification of von Hippel-Lindau gene alterations in – increased glycolysis via the AKT/b-catenin pathway [27], clear cell renal tumors. Clin. Cancer Res. 14:4726 4734. it can be hypothesized that NEFH alterations could con- 8. Dalgliesh, G. L., K. Furge, C. Greenman, L. Chen, tribute to the recently described metabolic shift of aggres- G. Bignell, A. Butler, et al. 2010. Systematic sequencing of sive renal cancer cells found as a result of integrative renal carcinoma reveals inactivation of histone modifying – genome-wide analyses of molecular alterations in ccRCC genes. Nature 463:360 363. [9]. 9. The Cancer Genome Atlas Research Network. Second, NEFH is a substrate of the MAP kinases Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 2013;499:43–49. MAPK3 and MAPK1 [37] and, phosphorylation of NEFH 10. Varela, I., P. Tarpey, K. Raine, D. Huang, C. K. Ong, by stress-activated p38 protein kinases (MAPK11,-12,-14) P. Stephens, et al. 2011. Exome sequencing identifies has been annotated for the amyotrophic lateral sclerosis frequent mutation of the SWI/SNF complex gene PBRM1 pathway in Kyoto Encyclopedia of Genes and Genomes in renal carcinoma. Nature 469:539–542. indexes [38]. 11. Kapur, P., S. Pena-Llopis, A. Christie, L. Zhrebker, Conclusively, this study identifies increased NEFH A. Pavia-Jimenez, W. K. Rathmell, et al. 2013. Effects on DNA methylation as a candidate marker for independent survival of BAP1 and PBRM1 mutations in sporadic prognosis of PFS of RCC patients. Moreover, our results clear-cell renal-cell carcinoma: a retrospective analysis with suggest that NEFH methylation also predicts OS of met- independent validation. Lancet Oncol. 14:159–167. astatic patients undergoing targeted therapies. Our analy- 12. Baldewijns, M. M., I. J. van Vlodrop, L. J. Schouten, P. M. ses underline the need for a functional analysis of NEFH Soetekouw, A. P. de Bruine, and M. van Engeland. 2008. in RCC and other cancers, possibly contributing for Genetics and epigenetics of renal cell cancer. Biochim. future individualized prognosis and treatment of Biophys. Acta 1785:133–155. patients. 13. Morris, M. R., D. Gentle, M. Abdulrahman, N. Clarke, M. Brown, T. Kishida, et al. 2008. Functional epigenomics Conflict of Interest approach to identify methylated candidate tumour suppressor genes in renal cell carcinoma. Br. J. Cancer None declared. 98:496–501. 14. Morris, M. R., C. Ricketts, D. Gentle, M. Abdulrahman, References N. Clarke, M. Brown, et al. 2010. Identification of 1. Jemal, A., F. Bray, M. M. Center, J. Ferlay, E. Ward, and candidate tumour suppressor genes frequently methylated D. Forman. 2011. Global cancer statistics. CA Cancer J. in renal cell carcinoma. Oncogene 29:2104–2117. Clin. 61:69–90. 15. Morris, M. R., C. J. Ricketts, D. Gentle, F. McRonald, 2. Motzer, R. J., T. E. Hutson, P. Tomczak, M. D. N. Carli, H. Khalili, et al. 2011. Genome-wide methylation Michaelson, R. M. Bukowski, O. Rixe, et al. 2007. analysis identifies epigenetically inactivated candidate Sunitinib versus interferon alfa in metastatic renal-cell tumour suppressor genes in renal cell carcinoma. carcinoma. N. Engl. J. Med. 356:115–124. Oncogene 30:1390–1401. 3. Sternberg, C. N., I. D. Davis, J. Mardiak, C. Szczylik, 16. Yamada, D., S. Kikuchi, Y. N. Williams, M. E. Lee, J. Wagstaff, et al. 2010. Pazopanib in locally Sakurai-Yageta, M. Masuda, T. Maruyama, et al. 2006. advanced or metastatic renal cell carcinoma: results of a Promoter hypermethylation of the potential tumor randomized phase III trial. J. Clin. Oncol. 28:1061–1068. suppressor DAL-1/4.1B gene in renal clear cell carcinoma. 4. Motzer, R. J., J. Bacik, L. H. Schwartz, V. Reuter, P. Russo, Int. J. Cancer 118:916–923. S. Marion, et al. 2004. Prognostic factors for survival in 17. Hildebrandt, M. A., J. Gu, J. Lin, Y. Ye, W. Tan, previously treated patients with metastatic renal cell P. Tamboli, et al. 2010. Hsa-miR-9 methylation status is carcinoma. J. Clin. Oncol. 22:454–463. associated with cancer development and metastatic

308 ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. N. Dubrowinskaja et al. NEFH Methylation and Renal Cell Cancer Prediction

recurrence in patients with clear cell renal cell carcinoma. 27. Kim, M. S., X. Chang, C. LeBron, J. K. Nagpal, J. Lee, Oncogene 29:5724–5728. Y. Huang, et al. 2010. Neurofilament heavy polypeptide 18. van Vlodrop, I. J., M. M. Baldewijns, K. M. Smits, regulates the Akt-beta-catenin pathway in human L. J. Schouten, L. van Neste, W. van Criekinge, et al. 2010. esophageal squamous cell carcinoma. PLoS ONE 5:e9003. Prognostic significance of Gremlin1 (GREM1) promoter 28. Greene, F. L., and L. H. Sobin. 2002. The TNM system: CpG island hypermethylation in clear cell renal cell our language for cancer care. J. Surg. Oncol. 80:119–120. carcinoma. Am. J. Pathol. 176:575–584. 29. Thoenes, W., S. Storkel, and H. J. Rumpelt. 1986. 19. Peters, I., H. Eggers, F. Atschekzei, J. Hennenlotter, Histopathology and classification of renal cell tumors S. Waalkes, W. Trankenschuh, et al. 2012. GATA5 CpG (adenomas, oncocytomas and carcinomas). The basic island methylation in renal cell cancer: a potential cytological and histopathological elements and their use biomarker for metastasis and disease progression. BJU Int. for diagnostics. Pathol. Res. Pract. 181:125–143. 110:E144–E152. 30. Campan, M., D. J. Weisenberger, B. Trinh, and P. W. 20. Gebauer, K., I. Peters, N. Dubrowinskaja, J. Hennenlotter, Laird. 2009. MethyLight. Methods Mol. Biol. 507:325–337. M. Abbas, R. Scherer, et al. 2013. Hsa-mir-124-3 CpG 31. Kent, W. J., C. W. Sugnet, T. S. Furey, K. M. Roskin, island methylation is associated with advanced tumours T. H. Pringle, A. M. Zahler, et al. 2002. The human and disease recurrence of patients with clear cell renal cell genome browser at UCSC. Genome Res. 12:996–1006. carcinoma. Br. J. Cancer 108:131–138. 32. Lander, E. S., L. M. Linton, B. Birren, C. Nusbaum, 21. Atschekzei, F., J. Hennenlotter, S. Janisch, A. Grosshennig, M. C. Zody, J. Baldwin, et al. 2001. Initial sequencing and W. Trankenschuh, S. Waalkes, et al. 2012. SFRP1 CpG analysis of the human genome. Nature 409:860–921. island methylation locus is associated with renal cell cancer 33. R Development Core Team. 2011. R: A language and susceptibility and disease recurrence. Epigenetics environment for statistical computing. R Foundation for 7:447–457. Statistical Computing, Vienna, Austria. Available at http:// 22. Larkin, J., X. Y. Goh, M. Vetter, L. Pickering, and www.R-project.org/. C. Swanton. 2012. Epigenetic regulation in RCC: 34. Seidel, C., J. Busch, S. Weikert, S. Steffens, M. Fenner, opportunities for therapeutic intervention? Nat. Rev. Urol. A. Ganser, et al. 2012. Progression free survival of first line 9:147–155. vascular endothelial growth factor-targeted therapy is an 23. Vanharanta, S., W. Shu, F. Brenet, A. A. Hakimi, important prognostic parameter in patients with metastatic A. Heguy, A. Viale, et al. 2013. Epigenetic expansion of renal cell carcinoma. Eur. J. Cancer 48:1023–1030. VHL-HIF signal output drives multiorgan metastasis in 35. Skolarikos, A. A., A. G. Papatsoris, G. Alivizatos, and renal cancer. Nat. Med. 19:50–56. C. Deliveliotis. 2006. Molecular pathogenetics of renal 24. Al-Chalabi, A., and C. C. Miller. 2003. Neurofilaments and cancer. Am. J. Nephrol. 26:218–231. neurological disease. BioEssays 25:346–355. 36. Pantuck, A. J., D. B. Seligson, T. Klatte, H. Yu, J. T. 25. Schleicher, R. L., S. B. Hunter, M. Zhang, M. Zheng, Leppert, L. Moore, et al. 2007. Prognostic relevance of the W. Tan, C. I. Bandea, et al. 1997. Neurofilament heavy mTOR pathway in renal cell carcinoma: implications for chain-like messenger RNA and protein are present in molecular patient selection for targeted therapy. Cancer benign prostate and down-regulated in prostatic 109:2257–2267. carcinoma. Cancer Res. 57:3532–3536. 37. Perrot, R., R. Berges, A. Bocquet, and J. Eyer. 2008. 26. Oka, D., S. Yamashita, T. Tomioka, Y. Nakanishi, H. Kato, Review of the multiple aspects of neurofilament functions, M. Kaminishi, et al. 2009. The presence of aberrant DNA and their possible contribution to neurodegeneration. Mol. methylation in noncancerous esophageal mucosae in Neurobiol. 38:27–65. association with smoking history: a target for risk 38. Kanehisa, M., S. Goto, S. Kawashima, and A. Nakaya. diagnosis and prevention of esophageal cancers. Cancer 2002. The KEGG databases at GenomeNet. Nucleic Acids 115:3412–3426. Res. 30:42–46.

ª 2014 The Authors. Cancer Medicine published by John Wiley & Sons Ltd. 309 RESEARCH ARTICLE Tumor Specific Epigenetic Silencing of Corticotropin Releasing Hormone -Binding Protein in Renal Cell Carcinoma: Association of Hypermethylation and Metastasis

Hossein Tezval1, Natalia Dubrowinskaja1, Inga Peters1, Christel Reese1, Katrin Serth2, Faranaz Atschekzei1, Jo¨ rg Hennenlotter3, Arnulf Stenzl3, Markus A. Kuczyk1, Ju¨ rgen Serth1*

1 Department of Urology and Urological Oncology, Hannover Medical School, Hannover, Germany, a11111 2 Department of Molecular Biology, Hannover Medical School, Hannover, Germany, 3 Department of Urology, Eberhard Karls University of Tuebingen, Tuebingen, Germany

* [email protected]

Abstract

OPEN ACCESS The relevance of Corticotropin Releasing Hormone (CRH)-system in human malignancies is Citation: Tezval H, Dubrowinskaja N, Peters I, a question of growing interest. Here we investigated hypermethylation and epigenetic silenc- Reese C, Serth K, Atschekzei F, et al. (2016) Tumor ing of the CRH-Binding Protein (CRHBP) gene in clear cell renal cell cancer (ccRCC). Rela- Specific Epigenetic Silencing of Corticotropin Releasing Hormone -Binding Protein in Renal Cell tive methylation of the CRHBP CpG island (CGI) was determined in 17 tumor cell lines as Carcinoma: Association of Hypermethylation and well as 86 ccRCC samples and 66 paired normal tissues using pyrosequencing and quantita- Metastasis. PLoS ONE 11(10): e0163873. tive methylation specific PCR of bisulfite converted DNA. Results were statistically compared doi:10.1371/journal.pone.0163873 with relative mRNA expression levels of CRHBP and clinicopathological parameters of Editor: Javier S Castresana, University of Navarra, patients. Re-expression of CRHBP following 5-aza-2´-deoxycytidine treatment was investi- SPAIN gated by quantitative mRNA expression analysis. Real-time impedance analysis was applied Received: January 14, 2016 for analysis of invasiveness of renal tumor cells following si-RNA knockdown of CRHBP Accepted: September 15, 2016 expression or ectopic expression of CRHBP. We found the CRHBP CGI to be frequently Published: October 3, 2016 methylated in tumor cell lines of renal, prostatic, and bladder cancer. Comparison of methylation in normal and paired renal cancer tissue specimens revealed hypermethylation of the Copyright: © 2016 Tezval et al. This is an open −12 access article distributed under the terms of the CRHBP CGI in tumors (p<1*10 ). DNA methylation and decreased mRNA expression Creative Commons Attribution License, which were correlated (R = 0.83, p<1*10−12). Tumor cell lines showed 5-aza-2´-deoxycytidine permits unrestricted use, distribution, and dependent reduction of methylation and re-expression of CRHBP was associated with reproduction in any medium, provided the original author and source are credited. altered cellular invasiveness of renal cancer cells in real-time impedance invasion assays. Hypermethylation and inverse relationship with mRNA expression were validated in silico Data Availability Statement: Availability of primary patient data is restricted due to ethical reasons as using the TCGA network data. We describe for the first time tumor specific epigenetic silenc- public availability interferes with patient privacy and ing of CRHBP and statistical association with aggressive tumors thus suggesting the CRH is not covered by ethical permissions. Summarized system to contribute to the development of kidney cancer. clinical data as well as all other relevant data are presented within the paper. Please direct further data requests to: [email protected].

Funding: The author(s) received no specific funding for this work.

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 1 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Competing Interests: The authors have declared Introduction that no competing interests exist. The corticotropin releasing hormone (CRH)-family includes the corticotropin releasing hormone (CRH) homologous urocortin proteins (UCN, UCN2, UCN3), their receptors CRHR1 and CRHR2 as well as the cortictropin releasing hormone binding protein CRHRB. Proteins of the CRH-system have been initially identified as hypothalamus directed mediators of neuroen- docrine stress response [1, 2] while recent studies suggest that CRH family members might also play a role in the development of human solid cancers [3]. Moreover, functional studies demonstrated in vitro that migration of tumor cells can be enhanced by CRH protein. This stimulation in turn can be blocked by inhibition of the ERK-pathway [4]. Concurrent alterations in apoptotic behavior and AKT pathway signaling of tumor cells have also been observed following treatment by CRH and UCN2 [5]. In addition, changes in expression both of mRNA as well protein levels have been detected for CRH-family members in a number of human malignancies such as breast, endometrial, lung, prostate and kidney cancer giving further evidence for the relevance of the CRH-system in human cancers [3]. In line we recently identified changes in mRNA and protein expression levels for UCN and CRHR2 in normal and clear cell renal cell carcinoma (ccRCC) tissues [6]. A strong cytoplasmic immunopositivity for UCN was detected in normal proximal renal epithelia whereas tumor cells showed either the combination of nuclear positivity with loss of cytoplasmic signals or solely cytoplasmic immunopositivity. Furthermore, CRHR2 immunopositivity was found to be reduced in endothelia of tumoral microvessels. Interestingly, quantitation of CRHBP mRNA levels revealed a nearly complete loss of mRNA expression in RCC identifying a new member of the CRH-family to be possibly involved in RCC carcinogenesis [7]. The development of ccRCC is associated with loss and/or alteration of chromosome 3q and frequently observed gene mutations in the von Hippel-Lindau (VHL) and PBRM1 genes [8, 9]. Exome wide mutational analyses in substantial number of tumors on the one hand revealed a great number of additional mutations occurring in RCC [10]. On the other hand a pronounced variability of mutations was found, exhibiting nearly individual mutational spectra in tumors thus clearly limiting the clinical usability of mutational information. Interestingly, loss of VHL function was shown to associate with extended epigenetic alterations in RCC [11] and, note- worthy, the most frequent mutations described so far affect genes maintaining the cellular chromatin and histone status, a process that is interrelated with DNA methylation [12]. DNA hypermethylation in RCC has been described for a substantial number of genes, and, moreover to show functional significance in cell lines derived from RCC as well as correlation with histopathological tumor characteristics, clinicopathological parameters and course of the disease [8, 13–18]. In view of our previous findings, demonstrating that mRNA-expression of the CRHBP gene is depleted in tumor tissues, we hypothesized that CRHBP may be epigenetically silenced thus representing a new target of DNA hypermethylation in ccRCC. Our study shows, to our knowledge for the first time, that a member of the CRH-system can undergo epigenetic silencing in a solid human cancer, hence providing new and strong evidence for a significant role of the CRH-system in human tumorigenesis.

Material and Methods Primary cells and tumor cell lines Renal proximal tubular epithelial cells (RPTEC) and primary normal prostatic cells (PreC) were obtained from Lonza (Basel, Switzerland) and renal, urothelial and prostate cancer cell lines ACHN, A498, 786-O, RCC-GS, RCC-HS, RCC-MF, RT112, CLS439, HB-CLS2, EJ28,

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 2 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

5637, T24, DU145, LN-cap and PC3 were purchased from cell line services (CLS, Eppelheim, Germany). Cells were cultured according to the manufactures recommendations and exhibited a total number of 18 passages at the beginning of real-time impedance experiments.

Patients’ characteristics Kidney tumors with clear cell histology of 86 patients (mean age 64 years, 35–90 years) subjected to kidney surgery between 2001 and 2005 collected from the Eberhard Karls University Tuebingen and corresponding 66 tumor free tissues were included in the present study (Table 1). Tissue preparation, storage, pathological evaluation, tumor stage assessment, nuclear grading, and data management have been described previously [19]. The ethics committees "Ethik-Komission an der Medizinischen Fakultät der Eberhard-Karls-Universität und am Uni- versitätsklinikum Tübingen (Head D. Lucht) " and "Ethik-Kommission der Medizinischen Hochschule Hannover (Head H.D. Tröger) " approved the study (ethics votes No. 128/2003V

Table 1. Clinical and histopathological data of patients. Clinico-pathological parameters Number of patients (%)All ccRCC Number of patients (%)Paired ccRCC ccRCC1 ccRCC with pNT2 Total 86 (100) 66 (100) Age (years) Mean 64 65 (minimum-maximum age) (35–90) (35–90) male 53 (62) 40 (61) female 33 (38) 26 (39) pT classiﬁcation pT1 8 (9) 4 (6) pT1a 22 (26) 17 (26) pT1b 14 (16) 12 (18) pT2 6 (7) 4 (6) pT3 2 (2) 1 (1) pT3a 9 (10) 7 (11) pT3b/c 23 (27) 19 (29) pT4 0 (0) 0 (0) others 1 (1) 2 (3) Synchronous lymph nodes metastasis 9 (10) 8 (12) Synchronous distant metastasis 22 (26) 19 (28) Advanced disease (pT3-4 and/or N/M+) 45 (52) 37 (56) Fuhrman grading G1 20 (23) 10 (15) G1-2 9 (10) 7 (10) G2 42 (49) 36 (54) G2-3 5 (6) 5 (7) G3 10 (12) 8 (12)

1ccRCC, clear cell renal cell carcinoma 2pNT paired normal tissue doi:10.1371/journal.pone.0163873.t001

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 3 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

and 1213–2011) and written informed consent was obtained from patients. Organ-confined RCC was defined as pT 2 and N0/M0 and advanced disease as pT 3 and/or N+/M+.

Nucleic acid extraction, DNA bisulfite conversion and DNA methylation analysis Examination of control sections for tumor cell content, DNA isolation from frozen section as well as bisulfite conversion of DNA were performed as reported recently [18]. Genomic DNA from primary cells and the cancer cell lines were isolated using standard proteinase K digestion and phenol-chloroform extraction. CRHBP CGI methylation analysis of cell lines was carried out using pyrosequencing applying the universal reverse primer concept [20]. The forward, 5´-GGAGTTGGTTGGGGAGTA -3´, reverse-universal 5´-GGGACACCGCTGATCGTTTACCCCCRCACAAAATCCCACC TT-3´ and the universal 5’-Biotin- GGGACACCGCTGATCGTTTA-3´ primers were designed using the PyroMark Assay Design 2.0 software (Qiagen, Hilden, Germany). PCR was carried

out in 25 μl consisting of 60 mM Tris-HCl pH 8.5, 15 mM ammonium sulfate, 1.5 mM MgCl2, 0.2 mM dNTP mix, 0.5 U of HotStarTaq DNA Polymerase (Qiagen, Hilden, Germany), 50 ng bisulfite-treated genomic DNA and 0.4 μM of each primer: PCR-cycling conditions were 95°C for 15 minutes, followed by 45 cycles with denaturation at 95°C for 45 seconds, annealing at 60°C for 40 seconds and elongation at 72°C for 40 seconds finished with 1 cycle final elongation at 72°C for 5 minutes. Purification of the biotinylated PCR product, preparation of single strand DNA, annealing of the pyrocequencing primer (5´-TGGTTGGGGAGTAGT-3´) and pyrosequencing in a PyroMark Q24 system (Qiagen, Hilden,Germany) was performed according to the manufacturer´s instruction using PyroGold SQA™ Reagent Kit (Qiagen, Hilden, Ger- many). CpG site quantification was performed using the methylation Software PyroMark Q24. Quantitative methylation specific PCR (qMSP) analysis of tissue specimens was carried out by use of a quantitative real-time fluorimetric 5’ exonuclease PCR assay as described previously [18]. The qMSP primers 5’- AGGGGTTGGTCGGAATCGT -3’ (forward), 5’- AACCTAAAC TACGCTAAATTCCTACG -3’ (reverse) and the Taqman1 probe 5’-FAM- CGCCACCC TCTCCCGCTCCTAACG -BHQ-3 were designed by use of the Beacon Designer™ software (PREMIER Biosoft, Palo Alto CA, USA). We analyzed 7 and 5 CpG sites in the CRHBP CGI (chromosome 5: 76,249,436–76,250,528, CHRBP transcription start site at 76,248,680) at positions 76,249,507, ~510, ~512, ~519, ~526, ~536, ~541 for pyrosequencing and 76,249,622, ~628, ~683, ~696, ~720 for qMSP, respectively, as schematically illustrated in Fig 1A.

Re-expression analysis of CRHBP RCC cells were grown after thawing in normal medium (RPMI1640, 2 mM L-Glutamin, 100 μg/ml Streptomycin, 10% FCS, Biochrom, Berlin, Germany) for 7 days. After passaging cells were treated on day 2–4 with 0,125 μM 5-aza-2`desoxycytidine (Sigma, St. Louis, USA) or a mock solution (control cells) and cultured in normal medium on days 5–7 until harvesting. Isolation of total RNA, reverse transcription and relative quantitation of mRNA levels of internal controls as well CRHBP mRNA were carried out as described before [7]. For ectopic expression of CRHBP 3μg of human cDNA expression vector (NM_001882, Origene Technologies, Rockville, MD, USA) and PerFectin transfection reagent (T303007, Genlantis, San Diego, CA, USA) were applied according to the manufacturer´s instructions in a transfection volume of 1 ml for 6Ã105 cells CHO cells were used as control for transfection and expression efficiency. For SDS-PAGE and Western blot analysis each 0,5 x 106 transfected cells were lysed in 100 μl 2x sample buffer

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 4 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Fig 1. Methylation and re-expression analyses. (A) Structure of the CRHBP CGI and location of the PCR amplicons subjected to qMSP and pryrosequencing analysis relative to the CRHBP transcription start site. Note that only part of CpG sites covered by amplicons is amenable to quantitative methylation analysis as specified in material and methods. Vertical lines represent CpG sites within the CGI. (B) CHRBP pyrosequencing analysis of normal renal proximal tubular cells (RPTEC) and A498 RCC cell lines. (C) Pyrosequencing analysis of CRHBP methylation in controls, primary cells as well as

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 5 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

renal cancer, prostate and bladder cancer cell lines. (D) Re-expression analysis of CRHBP in RCC cell lines following 5-aza- 2´-deoxycytidine treatment (+/-) by the use of qMSP methylation analysis and quantitative real-time PCR of relative CRHBP mRNA expression. doi:10.1371/journal.pone.0163873.g001

(125 mM Tris-HCl pH 6,8, 20% Glycerol, 4% SDS, 5% Mercaptoethanol, 0,025% Bromphenol- blue). Immunoprobing was carried out by use of the anti-Flag-POD antibody (A8592, Sigma- Aldrich, St. Louis, MO, USA).

Real-time impedance measurement of cellular invasiveness Considering that we detected high CRHBP methylation levels and correspondingly no mRNA expression in our RCC cell lines, siRNA knock down experiments for CRHBP could not be directly carried out. Therefore, the RCC-GS and–HS cell lines were first cultured using treatment with 5-aza-2´-deoxycytidine for re-expression of CRHBP mRNA as described above. Instead of harvesting, cells were again sub-cultured, transfected on day 2 with 25 nM Target or TargetPlus control siRNA (Dharmacon, Lafayette, USA) and use of 0.2% DH2 transfection reagent for 24 hours. Cells were then grown for 1 day in normal medium, sub-cultured, culti- vated another day and counted. Each 24,000 cells per well were placed in a nutrient and growth factor deficient medium upon a 2,5% Matrigel (BD, Franklin Lakes, USA) layer in the two chamber CIM-plate system (ACEA, Biosciences, San Diego, CA, USA). Passing through the layer and a supporting membrane into the second chamber containing normal growth medium supplemented with10% FCS and growth upon the microelectrode was monitored by real-time impedance analysis by use of the XCelligence RTCA DP instrument (ACEA Bioscience, San Diego, CA, USA) and taken as a measure of cellular invasiveness. Each experiment was carried out in triplicate. Real time impedance invasion analysis of cells transfected with the plasmid for ectopic expression of CRHBP or the control plasmid was performed as described above for siRNA experiments but using each 30,000 cells per well and a Matrigel concentration of 1.5% in quadruplicate experiments.

Statistical analysis For comparison of kidney tumor tissues and paired tumor adjacent normal tissue samples the paired t-test was applied. Univariate logistic regression models were carried out for independent group comparisons. Means and standard deviations (sd) per group, odds ratios (OR), corresponding 95% confidence intervals (CI) and two-sided p-values are presented. Correlation of relative expression and methylation results were carried using Pearson correlation analysis. The TCGA clear cell kidney carcinoma dataset (KIRC) was applied for validation of statistical results by use of the TCGA KIRC Infinium HumanMethylation450 BeadChip and the TCGA KIRC gene expression by RNAseq (IlluminaHiSeq) level 3 data sets. Hypermethylation was statistically analyzed by use of the paired t-test. The relationship of DNA methylation and mRNA expression was evaluated using the Pearsons correlation analysis and association of methylation and clinico-pathological parameters was analyzed by the use of univariate logistic regression models. P 0.05 was considered to be statistically significant. Adjustments for multiple testing were carried out using the Bonferroni-Hochberg method.

Results CRHBP CGI methylation analyses in cancer cell lines and primary cells Using pyrosequencing analysis (Fig 1A and 1B) for quantitation of relative methylation levels in cells used as models for normal and tumoral human tissues, we found that RPTEC and Prec

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 6 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

cells demonstrated methylation of approx. 20–25% (Fig 1C). In contrast, a relative methylation of 80–100% was detected for six out of six (100%) renal cancer cell lines, five out of six (83%) bladder cancer cell lines and one out of 3 (33%) prostatic carcinoma cell lines (Fig 1C). Besides CLS-439 all cancer cell lines analysed exhibited a degree of relative methylation >50% suggesting CRHBP CGI methylation as a frequent event in urological tumor models (Fig 1C).

Re-expression of CRHBP following 5-aza-2´-deoxycytidine treatment of renal cancer cell lines In view of our previous finding describing substantially reduced mRNA expression in renal tumors [7] we asked whether low mRNA expression of CRHBP in renal cancer cell lines showing a high degree of CRHBP CGI methylation increases due to treatment of cells with the DNA methyltransferase inhibitor 5-aza-2´-deoxycytidine. Re-expression analysis carried out for four cell lines (ACHN, A498, RCC-GS, RCC-HS) showed reduced relative methylation and a concurrent increase in relative mRNA expression levels between 50 and 150 fold following treatment of cells with the inhibitor (Fig 1D).

Analysis of invasion characteristics of renal cancer cell lines following endogenous or ectopic re-expression of CRHBP Real–time impedance analyses of cells grown on top of microelectrodes were applied to measure possible effects of CRHBP on proliferation and invasiveness of the renal cancer cell lines 786-O, RCC-GS and RCC-HS. Considering that all of these cell lines exhibited high CRHBP methylation and undetectable mRNA expression in QPCR analyses, we used both suppression of endogenously re-expressed CRHBP as well as targeted ectopic re-expression of CRHBP. For demethylation and endogenous but unspecific re-expression of CRHBP cells were first grown in the presence of 5-aza-2`desoxycytidine and subsequently subjected to si-RNA treatment to measure the effect of CRHBP suppression upon the cellular behavior in the real-time invasion experiments. Measurements were carried out in triplicate in comparison to onTarget- plus-si-RNA negative controls to monitor the invasion characteristics and indicated that CRHBP-suppression affected the invasiveness of RCC cell lines (Fig 2A and 2B). For example, the si-RNA treated RCC-GS cell line, originally derived from a metastatic primary tumor, showed a significantly increased capability to pass the matrigel-layer used as a measure for invasiveness (Fig 2A). However, changing methyltransferase inhibitor concentrations during pretreatment of cells for re-expression, we also observed reduction of the invasiveness of the cells as a result of si-RNA application (Fig 2B). Corresponding analyses for RCC-HS as a model for a localized cancer demonstrated no significant changes vs. the controls in invasion characteristics (data not shown). Transfection of the native RCC-GS cell line with a CRHBP expression vector showed that ectopic expression of CRHBP is associated with a decreased invasive potential in the Matrigel invasion assay when compared to the RCC-GS cells transfected by use of a mock vector (Fig 2C and 2D).

Analysis of tumor specific hypermethylation of the CRHBP CGI in kidney cancer specimens To answer whether high CGI-methylation of CRHBP associates with tumor tissues and decreased CRHBP expression in RCC, we investigated paired normal and tumoral tissue samples from 66 nephrectomy specimens by the use of qMSP. First, we characterized the qMSP

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 7 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Fig 2. Matrigel invasion assays following CRHBP expression alterations. Real-time impedance measurement of invasiveness of the RCC-GS cell line following si-RNA-knock down of CRHBP in RCC-GS cells treated in advance with 0.125 μM(A) or 0.5 μM(B) 5-aza-2´-deoxycytidine for endogenous re-expression of CRHBP. RCC-GS cells were incubated with 25nM Target (circles) or TargetPlus control (squares) si-RNA and subsequently placed in a nutrient and growth factor deficient medium upon a membrane covered by 2,5% Matrigel layer separating the nutrient and growth factor containing second chamber. Passing through the layer and growth upon the microelectrode (cell index) was monitored against time (seconds) of measurement. Experiments were carried out in triplicate and positive and negative standard deviations of each measurement (bars) are presented. Ectopic re-expression of CRHBP following transfection of CHO control and RCC-GS target cells in western blot analysis (C) and the Matrigel invasion assay (D) using a 1.5% Matrigel layer. Single positive and negative standard deviations from quadruplicate experiments are shown for the CRHBP expression positive and negative RCC-GS cell measurements. doi:10.1371/journal.pone.0163873.g002

assay for linearity and PCR efficiency, finding a coefficient of correlation of R = 0.99 (slope = -3.4, indicating high linearity and efficiency of the assay (Fig 3A)). Analysis of paired normal and tumor tissue samples demonstrated that more than half of the tumor tissues exhibited a substantial increase in methylation thus showing a clear overall hypermethylation of the CRHBP–CGI in RCC tumors (p<1Ã10−12, paired t-test, Fig 3B and 3C). Assessment of median methylation levels for normal and tumor tissues revealed a difference of 4.37 natural logarithmic units corresponding to an approx. 80 fold increase in highly methylated CRHBP CGI sequences in the tumor group.

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 8 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Fig 3. qMSP analysis of CRHBP CGI methylation in renal normal and tumor tissues. (A) Calibration line demonstrating linearity and efficiency of qMSP analysis. (B) Detection of hypermethylation of CRHBP in paired normal (pNT) and tumor samples (TU); (C) Assorted paired difference plot for pairwise methylation differences; (D) Analysis of epigenetic silencing by comparison of relative methylation (qMSP CRHBP) as detected by qMSP and relative mRNA expression (QEXP CRHBP) as detected by quantitative real-time PCR. Paired normal (pNT) and tumor tissue (TU) samples were connected by lines. The overall regression line and 95CI (greyed area) are presented. (E) Analysis of association of relative methylation and metastasis and state of advanced disease. From left to right kernel density estimation of relative methylation values and boxplot presentation of metastasis negative (M0) vs. positive (M1) tumors and localized (Loc.) vs. advanced (Adv.) tumors. lnRML, natural logarithm of relative methylation, ΔlnRML, difference of natural logarithm of relative methylation for tumor (TU) and paired normal tissues (pNT), lnREX, natural logarithm of relative expression of CHRBP mRNA. doi:10.1371/journal.pone.0163873.g003 Inverse relationship of CRHBP CGI methylation and relative mRNA expression levels in renal cancer specimens indicates epigenetic silencing A sample specific comparison of relative methylation detected by qMSP and relative mRNA expression values reported previously [7] showed overall a negative correlation between methylation and mRNA expression in tissue samples of ccRCC specimens (R = -0.55, p<2.7Ã10−12, Fig 3D). Both, hypermethylation data as well as the correlation analysis of methylation vs. expression revealed that the tumor tissues were heterogeneous with respect to CRHRBP–methylation demonstrating clearly separable subgroups of highly and low methylated tumors. A

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 9 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

corresponding correlation analysis considering only tissue pairs with highly methylated tumors demonstrated an even higher coefficient of correlation (R = -0.83, p<1Ã10−12, Fig 3D, upper cohort).

Association of CRHBP methylation and clinicopathological parameters Statistical evaluation of methylation degree and clinicopathological parameters using univariate logistic regression revealed that higher methylation associates with both the state of distant metastasis (p = 0.01, OR 1,18) as well as advanced disease (p = 0.001, OR = 1,2; Fig 3E). Gen- der, age, lymph node metastasis and grade of tumors did not show a relationship with methylation of the CRHBP locus investigated (Table 2).

In silico validation using the TCGA KIRC data set To independently evaluate our findings i.e. tumor specific hypermethylation, the inverse relationship between methylation and mRNA expression as well as the association of methylation with clinicopathological parameters, we interrogated the KIRC data set provided by The Can- cer Genome Atlas (TCGA) research network. We found that eight out of 14 CpG sites annotated to the CRHBP – CGI exhibited data appropriate for statistical evaluation of paired tissue methylation. Seven of the eight loci turned out to demonstrate significant tumor specific hypermethylation (paired t-test, p-values < 1Ã10−10, Bonferroni –Hochberg adjusted for multiple testing, Table 3). Comparison of methylation with mRNA expression revealed for all of these loci a significant inverse correlation of methylation and expression (R = -0.20 to – 0.39; Table 2), Moreover, analysis of statistical associations of methylation with clinicopathological parameters showed significant relationships for five of the loci with high stage, presence of distant metastasis as well as high grade tumors (Table 3).

Discussion The potential relevance of CRH-system alterations in tumor biology has been suggested by a number of reports showing that urocortins as CRH ligands may inhibit tumor growth via effects on vascularization, promote the apoptosis of endothelial cells, and downregulate VEGF expression in vivo [21, 22]. In line, we recently measured a substantial reduction both of CRHBP mRNA and protein immunopositivity in ccRCC [7] thus raising the question whether epigenetic silencing of the gene could occur in tumors. So far, epigenetic alteration i.e. DNA methylation of a member of the CRH system has only been detected for the CRH gene in rats

Table 2. Association of CRHBP methylation and clinicopathology of patients in RCC patients. Parameter p OR(95%CI) Gender 0.081 1.12 (0.99–1.24) Age 0.490 0.96 (0.87–1.07) Lymph node metastasis (N0 vs. N+) 0.085 1.18 (0.98–1.41) Distant metastasis (M0 vs. M+) 0.013 1.18 (1.04–1.34) Grade (low grade vs. high grade tumors) 0.141 1.11 (0.97–1.28) Localized/Advanced 0.001 1.25 (1.08–1.37)

p-values and OR´s refer to statistical analyses using univariate logistic regression. Dichotomization of the parameter age was performed using the median of 64 years. Low grade (G1, G1-2 and G2) and high grade (G2-G3 and G3) deﬁnitions were applied for dichotomization of tumor grade. Localized and advanced tumor states were deﬁned as M0 and N0 and pT<4 (localized) or M+ and/or N+ and/or pT4 (advanced).

doi:10.1371/journal.pone.0163873.t002

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 10 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Table 3. In silico validation of CRHBP methylation results using TCGA KIRC data. Locus (genomic Pos.) 1Hyper-methy-lation 2Gene silencing 3Clinicopathology TN MG p p p p p p cg06495038 (76,247,647) 6.19*10–28 2.37*10−06 2.85*10−05 - 1.04*10−02 1.80*10−04 cg26196496 (76,247,680) 3.29*10−29 1.03*10−02 - --- cg01071966 (76,248,923) 3.01*10−15 4.98*10−04 - --- cg04306063 (76,249,503) 9.61*10−37 2.32*10−11 2.34*10−08 - 7.65*10−07 7.45*10−07 cg07380705 (76,249,898) 1.34*10−30 2.37*10−06 2.12*10−06 - 7.56*10−07 1.14*10−04 cg13157757 (76,250,351) 5.49*10−20 4.42*10−06 1.69*10−05 - 2.96*10−06 1.11*10−04 cg13777717 (76,250,528) 4.21*10−25 2.37*10−06 1.55*10−06 - 9.49*10−06 5.00*10−03

Results are shown exclusively for cg-loci demonstrating tumor-specific hypermethylation. Specification and genomic positions of CpG sites refer to the UCSC Genome Browser on Human Feb. 2009 (GRCh37/hg19) assembly. Only P-values considered as significant are shown for statistical associations obtained in hypermethylation, epigenetic gene silencing and clinicopathological parameter analyses 1paired t-test (Bonferroni-Hochberg adjusted for multiple testing) of 160 paired normal and tumoral tissues using the Methylation450k data set 2Pearson correlation analysis of 297 tumors (Bonferroni-Hochberg adjusted for multiple testing) using TCGA KIRC Infinium HumanMethylation450 BeadChip level 3 data and gene expression by RNAseq (IlluminaHiSeq) level 3 data sets 3univariate logistic regression for methylation comparison of dichotomized subsets of 284 tumors for detection of statistical association with high (> = T3) and low stage (< T3), positive or negative state of lymph node (N) and distant metastasis (M) as well as low (< G3) and high grade (> = G3) tumor subsets. doi:10.1371/journal.pone.0163873.t003

as a result of chronic stress [23]. Here we analyzed whether DNA methylation could be respon- sible for loss of CRHBP expression in ccRCC and associates with clinicopathological parameters of patients. Using cancer cell lines of kidney, prostate and bladder cancers and pyrosequencing for methylation detection we found high relative methylation values in 14 out of 15 cancer cell line tumor models representing three frequent human tumors thus indicating that CGI methylation is a frequent event in human cancers. Furthermore, we found that treatment of RCC cell lines with 5-aza-2´-deoxycytidine leads to a substantial reduction of methylation and concurrent increase in mRNA expression, indicating that methylation substantially contributes in expression regulation and silencing of CRHBP in RCC cell lines. We therefore analyzed whether CRHBP shows hypermethylation in ccRCC tissue samples as well. Our quantitative methylation specific PCR analyses revealed that approximately half the tumors showed high levels of highly methylated CRHBP DNA. Corresponding to cell line analyses, a comparison of mRNA expression levels and CRHBP methylation in the tissue samples elucidated an inverse correlation of methylation and expression i.e. epigenetic silencing in renal tissues as well. Statistical evaluation of tumor methylation levels and clinicopathological parameters of patients exhibited significant associations of CRHBP methylation with the presence of distant metastasis as well as the state of advanced disease therefore indicating a role in the development of more aggressive cancer subtypes. Moreover, in silico validation by use of the KIRC dataset provided by TCGA network study [10] confirmed tumor specific hypermethylation of CRHBP, epigenetic silencing of CRHBP mRNA expression and association of CRHBP methylation with biological aggressive tumor characteristics. Our analysis of the proliferation and invasion characteristics of RCC cancer cell lines following both si-RNA knock down of endogenously re-expressed CRHBP after 5-aza-2´-deoxycytidine treatment as well as ectopic re-expression of CRHBP in an epigenetically silenced RCC cell line interestingly revealed substantial alteration in the matri-gel invasion assay. In case of siRNA knock down the extent and direction of measured effects depended on the

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 11 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

concentration of the methylation inhibitor. However, considering that treatment of cells with 5-aza-2´-deoxycytidine changed expression of an unknown number of other genes and therefore new derived cells with altered microscopic and assumingly also molecular characteristics developed, a specific reaction to CHRBP level changes could be theoretically explainable. On the other hand, ectopic expression of CHRBP also showed alteration, i.e. reduction of cellular invasiveness thus providing further evidence that invasive behavior of renal cancer cells may be affected by CRHBP. Nonetheless, all of our functional analysis are still preliminary and require future functional validation using improved tumor models taking into account the complexity of the CRH signaling network. It has been reported previously that CRH promotes migration of squamous epithelial tumor cells [4]. In prostatic cells it has been found that CRH and Ucn2 affect apoptosis of tumor cells [5]. Moreover it has been demonstrated that the CRH ligand Ucn promotes hepatic cancer cell migration by up-regulating cPLA2 expression via CRHR1 whereas it suppressed tumor cell migration by down-regulating iPLA2 expression via CRHR2 [24]. While these studies identify members of the CRH system to participate in skin and prostatic tumor model cell lines, our preliminary functional analyses provide evidence that the CRH system may also be of relevance for kidney tumors cells. Conclusively, our analysis clearly demonstrated epigenetic silencing of CRHBP hence showing for the first time that a member of the CRH-system is epigenetically silenced in tumor cells and thus providing strong evidence for an involvement of CRH members in human tumorigenesis. Moreover, we found association of DNA methylation with aggressive ccRCC tumor subsets, a finding that appears in nearly perfect concordance with the in silico results provided by the TCGA network. Consequently, both analyses statistically point to CRHBP silencing as a factor contributing to the development of aggressive tumors. Our introductory functional analysis suggests that CRHBP may affect invasive behavior of kidney tumor cells therefore possibly explaining the statistical results. Beside implications for oncological research our results may be also relevant for physiological conditions considering that most genes identified to be epigenetically silenced in tumors already show e.g. age dependent methylation in normal tissues, possibly affecting gene function [23, 25]. To expand our pathophysiological understanding of CRHBP function in tumor biology and ccRCC as well, future studies will rely on identification of improved cell line models to resolve the interplay of CRH peptides, receptors, binding protein and epigenetic alteration in human tissues.

Acknowledgments The results used here for purposes of statistical validation are based upon data generated by the TCGA Research Network: http://cancergenome.nih.gov/. We thank Mrs. M. Hepke and C. Reese for their technical support.

Author Contributions Conceptualization: HT JS MAK. Data curation: JS. Formal analysis: HT IP JS. Funding acquisition: MAK. Investigation: ND CR KS FA JS.

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 12 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

Methodology:ND FA KS CR. Project administration: JS IP HT. Resources: IP JH AS MAK. Supervision: JS MAK. Validation: CR KS. Visualization: JS IP. Writing – original draft: JS HT IP. Writing – review & editing: JS HT IP.

References 1. Vale W, Spiess J, Rivier C, Rivier J. Characterization of a 41-residue ovine hypothalamic peptide that stimulates secretion of corticotropin and beta-endorphin. Science. 1981; 213(4514):1394–7. Epub 1981/09/18. doi: 10.1126/science.6267699 PMID: 6267699. 2. Kageyama K, Bradbury MJ, Zhao L, Blount AL, Vale WW. Urocortin messenger ribonucleic acid: tissue distribution in the rat and regulation in thymus by lipopolysaccharide and glucocorticoids. Endocrinol- ogy. 1999; 140(12):5651–8. Epub 1999/12/01. doi: 10.1210/endo.140.12.7223 PMID: 10579329. 3. Kaprara A, Pazaitou-Panayiotou K, Kortsaris A, Chatzaki E. The corticotropin releasing factor system in cancer: expression and pathophysiological implications. Cellular and molecular life sciences: CMLS. 2010; 67(8):1293–306. Epub 2010/02/10. doi: 10.1007/s00018-010-0265-2 PMID: 20143250. 4. Yang Y, Goldstein BG, Nakagawa H, Katz JP. Kruppel-like factor 5 activates MEK/ERK signaling via EGFR in primary squamous epithelial cells. FASEB journal: official publication of the Federation of American Societies for Experimental Biology. 2007; 21(2):543–50. Epub 2006/12/13. doi: 10.1096/fj. 06-6694com PMID: 17158781. 5. Jin L, Zhang Q, Guo R, Wang L, Wang J, Wan R, et al. Different effects of corticotropin-releasing factor and urocortin 2 on apoptosis of prostate cancer cells in vitro. Journal of molecular endocrinology. 2011; 47(2):219–27. Epub 2011/07/19. doi: 10.1530/JME-11-0048 PMID: 21765100. 6. Tezval H, Jurk S, Atschekzei F, Becker JU, Jahn O, Serth J, et al. Urocortin and corticotropin-releasing factor receptor 2 in human renal cell carcinoma: disruption of an endogenous inhibitor of angiogenesis and proliferation. World journal of urology. 2009. doi: 10.1007/s00345-009-0417-x PMID: 19437022. 7. Tezval H, Atschekzei F, Peters I, Waalkes S, Hennenlotter J, Stenzl A, et al. Reduced mRNA expression level of corticotropin-releasing hormone-binding protein is associated with aggressive human kidney cancer. BMC cancer. 2013; 13:199. Epub 2013/04/24. doi: 10.1186/1471-2407-13-199 PMID: 23607589; PubMed Central PMCID: PMC3653809. 8. Baldewijns MM, van Vlodrop IJ, Schouten LJ, Soetekouw PM, de Bruine AP, van Engeland M. Genet- ics and epigenetics of renal cell cancer. Biochimica et biophysica acta. 2008; 1785(2):133–55. PMID: 18187049. doi: 10.1016/j.bbcan.2007.12.002 9. Varela I, Tarpey P, Raine K, Huang D, Ong CK, Stephens P, et al. Exome sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal carcinoma. Nature. 2011; 469 (7331):539–42. Epub 2011/01/21. doi: 10.1038/nature09639 PMID: 21248752; PubMed Central PMCID: PMC3030920. 10. TCGA. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature. 2013; 499(7456):43–9. Epub 2013/06/25. doi: 10.1038/nature12222 PMID: 23792563. 11. Vanharanta S, Shu W, Brenet F, Hakimi AA, Heguy A, Viale A, et al. Epigenetic expansion of VHL-HIF signal output drives multiorgan metastasis in renal cancer. Nature medicine. 2013; 19(1):50–6. Epub 2012/12/12. doi: 10.1038/nm.3029 PMID: 23223005; PubMed Central PMCID: PMC3540187. 12. Larkin J, Goh XY, Vetter M, Pickering L, Swanton C. Epigenetic regulation in RCC: opportunities for therapeutic intervention? Nature reviews Urology. 2012; 9(3):147–55. Epub 2012/01/18. doi: 10.1038/ nrurol.2011.236 PMID: 22249190. 13. Ibanez de Caceres I, Dulaimi E, Hoffman AM, Al-Saleem T, Uzzo RG, Cairns P. Identification of novel target genes by an epigenetic reactivation screen of renal cancer. Cancer research. 2006; 66 (10):5021–8. doi: 10.1158/0008-5472.CAN-05-3365 PMID: 16707423.

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 13 / 14 CRHBP CGI Hypermethylation in Kidney Cancer

14. Morris MR, Gentle D, Abdulrahman M, Clarke N, Brown M, Kishida T, et al. Functional epigenomics approach to identify methylated candidate tumour suppressor genes in renal cell carcinoma. British journal of cancer. 2008; 98(2):496–501. PMID: 18195710. doi: 10.1038/sj.bjc.6604180 15. Morris MR, Ricketts C, Gentle D, Abdulrahman M, Clarke N, Brown M, et al. Identification of candidate tumour suppressor genes frequently methylated in renal cell carcinoma. Oncogene. 2010. doi: 10. 1038/onc.2009.493 PMID: 20154727. 16. Morris MR, Ricketts CJ, Gentle D, McRonald F, Carli N, Khalili H, et al. Genome-wide methylation analysis identifies epigenetically inactivated candidate tumour suppressor genes in renal cell carcinoma. Oncogene. 2011; 30(12):1390–401. PMID: 21132003. doi: 10.1038/onc.2010.525 17. Peters I, Eggers H, Atschekzei F, Hennenlotter J, Waalkes S, Trankenschuh W, et al. GATA5 CpG island methylation in renal cell cancer: a potential biomarker for metastasis and disease progression. BJU international. 2012; 110(2 Pt 2):E144–52. Epub 2012/02/01. doi: 10.1111/j.1464-410X.2011. 10862.x PMID: 22289415. 18. Gebauer K, Peters I, Dubrowinskaja N, Hennenlotter J, Abbas M, Scherer R, et al. Hsa-mir-124-3 CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell carcinoma. British journal of cancer. 2013; 108(1):131–8. Epub 2013/01/17. doi: 10. 1038/bjc.2012.537 PMID: 23321515; PubMed Central PMCID: PMC3553529. 19. Waalkes S, Atschekzei F, Kramer MW, Hennenlotter J, Vetter G, Becker JU, et al. Fibronectin 1 mRNA expression correlates with advanced disease in renal cancer. BMC cancer. 2010; 10:503. Epub 2010/09/24. doi: 10.1186/1471-2407-10-503 PMID: 20860816; PubMed Central PMCID: PMC2949811. 20. Colella S, Shen L, Baggerly KA, Issa JP, Krahe R. Sensitive and quantitative universal Pyrosequen- cing methylation analysis of CpG sites. BioTechniques. 2003; 35(1):146–50. PMID: 12866414. 21. Hao Z, Huang Y, Cleman J, Jovin IS, Vale WW, Bale TL, et al. Urocortin2 inhibits tumor growth via effects on vascularization and cell proliferation. Proceedings of the National Academy of Sciences of the United States of America. 2008; 105(10):3939–44. PMID: 18308934. doi: 10.1073/pnas. 0712366105 22. Wang J, Xu Y, Xu Y, Zhu H, Zhang R, Zhang G, et al. Urocortin’s inhibition of tumor growth and angiogenesis in hepatocellular carcinoma via corticotrophin-releasing factor receptor 2. Cancer Invest. 2008; 26(4):359–68. PMID: 18443956. doi: 10.1080/07357900701788106 23. Sterrenburg L, Gaszner B, Boerrigter J, Santbergen L, Bramini M, Elliott E, et al. Chronic stress induces sex-specific alterations in methylation and expression of corticotropin-releasing factor gene in the rat. PloS one. 2011; 6(11):e28128. Epub 2011/12/02. doi: 10.1371/journal.pone.0028128 PMID: 22132228; PubMed Central PMCID: PMC3223222. 24. Zhu C, Sun Z, Li C, Guo R, Li L, Jin L, et al. Urocortin affects migration of hepatic cancer cell lines via differential regulation of cPLA2 and iPLA2. Cellular signalling. 2014; 26(5):1125–34. Epub 2014/02/13. doi: 10.1016/j.cellsig.2014.02.002 PMID: 24518041. 25. Issa JP. Aging and epigenetic drift: a vicious cycle. The Journal of clinical investigation. 2014; 124 (1):24–9. Epub 2014/01/03. doi: 10.1172/JCI69735 PMID: 24382386; PubMed Central PMCID: PMC3871228.

PLOS ONE | DOI:10.1371/journal.pone.0163873 October 3, 2016 14 / 14 2.2 Zusammenfassung der Ergebnisse.

2.2.1 Allgemeine Kriterien für die durchgeführten Untersuchungen.

Das Kernstück dieser Dissertation bilden die in den sechs beigefügten Artikeln publizierten Ergebnisse, die die retrospektiven Querschnittsanalysen zusammenfassen. Der Schwerpunkt der Arbeiten wurde unter der Verwendung eines Frischgewebe-Archivs von Tumor- und ggf. gepaarten Normalgewebsproben durchgeführt. Da die zur Verfügung stehenden Kohorten histologisch heterogen waren, wurden genspezifisch, je nach Fragestellung, an Probenanzahl und klinischen Eigenschaften leicht variierende Tumorsubgruppen untersucht (s. Tabelle 2). Im Vorfeld wurden Methylierungs- bzw. Expressionsanalysen der untersuchten Loci an einer Auswahl der Zelllinien, die als repräsentative Modelle für normale Gewebszellen und Tumore des Urogenitaltrakts oder gynäkologische Tumoren eingesetzt wurden, durchgeführt (s. Tabelle 2). Der Nachweis der DNA-Methylierung wurde mit Hilfe der QMSP ausgeführt. Ein für einige Loci insgesamt relativ niedriger gemessener Methylierungslevel in den untersuchten Geweben wäre eventuell auf die Eigenart der QMSP-Methode zurückzuführen. Die niedrige Methylierungswerte können nicht durch die Heterogenität der Zelltypen, die in dem untersuchten Tumorgewebe enthalten waren, erklärt werden, da ein Tumorzellgehalt von ≥ 50 % histopathologisch gesichert wurde. Da methodenbedingt nur einheitlich hoch und dicht methylierte Regionen erfasst werden, würden beispielsweise das Vorliegen der intra-CGI- Heterogenität103 bzw. die Existenz lediglich einer kleineren Subpopulation von Tumorzellen, die an diesem Locus dicht methyliert sind, als Erklärung für niedrige Werte in Frage kommen. Diese Überlegung ist für alle mit QMSP untersuchten Gene gültig. Als zusätzliche Technik zu den QMSP-Analysen wurden gegebenfalls sogenannte Pyrosequenzierungen durchgeführt. Für die Publikationen Nr. 2 und 6 wurden außerdem quantitative Echtzeit-PCR-Analysen (RT-PCR) zum Nachweis der mRNA durchgeführt. Für einen Marker (Publikation Nr. 6) wurden außerdem initiale funktionelle Untersuchungen unter der Verwendung verschiedener Tumorzelllinien der Niere durchgeführt. Bei den statistischen Untersuchungen wurden im Allgemeinen folgende Verfahren angewendet.

Zum Vergleich der mittleren relativen DNA-Methylierungen bzw. mRNA-Expressionslevel gepaarter Tumorgewebsproben und korrespondierender peritumoraler Gewebe wurden gepaarte t-Tests durchgeführt. Die Assoziationen zwischen mittleren relativen DNA- Methylierungen bzw. mRNA-Expressionslevel und klinisch-pathologischen Parametern wurden mit Hilfe univariater logistischer Regressionsanalysen untersucht, die sowohl den statistischen Signifikanzwert (p-Wert) als auch eine Odds Ratio (OR) lieferten, welche als Maß für die beobachteten Effektgrößen diente. Bei vielen klinischen Studien über die Wirksamkeit des zu testenden Therapeutikums definiert der Begriff progressionsfreies Überleben (PFS) den primären oder sekundären Endpunkt und fließt somit in die Auswertung der Studien ein135. In der Praxis ist die Kenntnis über das PFS einer der wichtigen Faktoren bei den Therapieentscheidungen. Die Assoziationen der mittleren relativen DNA-Methylierungen bzw. mRNA-Expressionslevel mit progressionsfreiem Überleben wurden mit Hilfe des univariaten Cox-Regressionsmodells (Cox proportional hazards regression - en.) bzw. Kaplan-Meier-Überlebensanalysen (Kaplan-Meier survival analysis – en.) untersucht. Teilweise wurden zusätzlich paarweise gestaltete bivariate Cox-Regressionsanalysen als Simulation der multivariaten Analysen, die aufgrund der begrenzten Kohortengrößen nicht kalkulierbar waren, durchgeführt. Die Assoziationen der mittleren relativen DNA-Methylierungen mit dem Therapieverlauf wurden mit Hilfe von Kaplan-Meier-Überlebensanalysen und dem Logrank-Test untersucht. Um die mittleren relativen DNA-Methylierungen und mRNA-Expressionslevel zu vergleichen, wurden Pearson-Korrelationsanalysen durchgeführt. Für die statistischen Analysen wurde statistische Software R 2.15.2 verwendet. Mit dem Anliegen der Leserfreundlichkeit und Vermeidung vieler Wiederholungen bei der Beschreibung der Ziele, Gestaltung der Experimenten und Ergebnisse wurden im Folgenden die relevanten Daten über die jeweilige Experiment-Konzeption sowie die Ergebnisse zur besseren Übersicht in vier Tabellen (Nr. 2-5) zusammengefasst. Die schriftliche Kurzbeschreibung der Ergebnisse ist in drei Unterthemen gegliedert: - „Ergebnisse funktioneller Analysen“ - Ergebnisse aus den funktionellen Untersuchungen (Publikations-Nr. 6); - „Deskriptive Ergebnisse“ - Ergebnisse die der beschreibenden Tumorbiologie der Nierenzelltumore zuzuordnen sind (Publikations-Nr. 1-3, 5, 6);

- „Translationale Ergebnisse“ - Ergebnisse aus Untersuchungen von Markern auf deren prognostische oder prädiktive Relevanz (Publikations-Nr. 1-6).

Tabelle 2. Übersicht über das experimentelle Design der zentralen Untersuchungen.

6 5 4 3 2 1 Publikation-Nr. Hsa-mir-124-3 Gen MIR124-3 Therapie CRHBP GATA5 ccRCC GATA3 GATA3 GATA5 MIR9-4 mRCC LAD1 NEFH CST6 RCC 3 1 2 4 keine Messungen / Analysen- durchgeführt metastatic renal cell carcinoma renal cell carcinoma clear cell renal cell carcinoma Subgruppe der Patienten einerdie Erstlinien VEGF-T primär 1 8 7 3 6 3 - - - 6 3 6 2 0 8 4 3 6 2 2 8 7 3 1 6 8 3 7 3 6 3 ------1 - - 4 ------ir rsaaBaeButSonstige Brust Blase Prostata Niere Zelllinien (n) tumoral gesamt herapie unterzogen wurden RCC 3 721 77 135 1 627 86 119 1 025 80 111 3 22 883 18 26 82 132 27 86 119 816 18 816 18 816 18 816 18 1 ccRCC 622 86 2 Gewebskohorten (n) Subgruppen mRCC - - - - 3 Therapie 18 18 18 18 - - - - - 4 peritumorales Gewebe (n) korresp. 66 58 77 87 87 p-ehleugQMSP CpG-Methylierung - p-ehleugQMSP CpG-Methylierung - p-ehleugQMSP CpG-Methylierung - p-ehleugQMSP CpG-Methylierung - mRNA-Expression/ CpG-Methylierung/ mRNA-Expression/ CpG-Methylierung/ RAEpeso qRT-PCR mRNA-Expression CpG-Methylierung p-ehleugQMSP CpG-Methylierung QMSP CpG-Methylierung QMSP CpG-Methylierung p-ehleugQMSP CpG-Methylierung Invasionsverhalten Proliferations- und nlseeeMessverfahren Analyseebene Pyrosequenzierung RT-Impedanz- Messungen qRT-PCR qRT-PCR QMSP / QMSP / QMSP /

2.2.2 Ergebnisse funktioneller Analysen.

In den renalen Tumorzellinien ACHN, A498, RCC-GS und RCC-HS wurde mittels 5-Aza-2`- Deoxycytidin eine unspezifische Reexpression induziert. Anschließend wurden die DNA- Methylierung und mRNA-Level quantifiziert. Es wurde eine inverse Beziehung zwischen DNA-Methylierung und dem mRNA-Level des CRHBP (human corticotrophin releasing factor-binding protein) festgestellt. Die Reduktion der relativen Methylierung ging mit 50- 150-facher Erhöhung des mRNA-Expressionslevel einher. Es wurde eine Charakterisierung des Proliferations- und Invasionsverhaltens der RCC-GS- Zelllinie mittels spezifischer siRNA-Genstillegung (Small interfering RNA) durchgeführt. Abhängig von der angesetzten Konzentration des 5-Aza-2`-Deoxycytidin (Substanz für die dem siRNA-Experiment vorausgehenden unspezifischen Demethylierung) wurden gegenteilige Effekte beobachtet. Verglichen mit den 5-Aza-2`-Deoxycytidin demethylierten und nicht mit der siRNA behandelten Kontrollen haben die mit der höheren 5-Aza2`- Deoxycytidin-Konzentration (0.5µM) behandelten und mit der siRNA-transfizierten Zellen ein invasiveres Verhalten gezeigt. Für die transfizierten Zellen die mit der niedrigeren Konzentration des Demethylierungsreagenz (0.125µM) vorbehandelt wurden ergab sich ein umgekehrtes Bild.

2.2.3 Deskriptive Ergebnisse.

Zelluläre Modelle. Relative DNA-Methylierung wurde für Genloci in MIR124-3, NEFH (Neurofilament Heavy Chain), GATA3 (GATA Binding Protein 3) und GATA5 mittels QMSP und für das CRHBP-Genlocus als globaler Methylierungsgrad mittels Pyrosequenzierung an den zellulären Modellen verschiedener Tumorentitäten und Primärzellen gemessen. Während die untersuchten Genregionen in primären Zellen niedrige oder nicht nachweisbare Methylierung zeigten, wiesen die meisten renalen Tumorzelllinien (je nach Gen in 67-100 % der Zelllinien) eine teilweise hohe relative Methylierung auf.

Gewebskohorten. (s. Tabelle 3.) Die Angaben zu den Probenzahlen finden sich in Tabelle 2 und die berechneten statistischen Werte zu den jeweiligen Aussagen in Tabelle 3. Die relative Methylierung der Genloci MIR124-3, GATA3, GATA5, CRHBP und NEFH und der mRNA-

Expressionslevel der Gene GATA3, GATA5 und CRHBP wurden für die Kohorten der RCC- Tumorgewebsproben (weiter in Text TUs abgekürzt) und korrespondierende peritumorale Geweben (weiter in Text PTs abgekürzt) verglichen. Teilweise wurden Tumorsubgruppen mit den PTs bzw. unter einander verglichen. Für die ausgewählten CpG-Loci in den Genen GATA5 und CRHBP wurden DNA-Methylierungen und mRNA-Expressionen verglichen.

RCC-Gruppe. Eine statistisch signifikante tumorspezifische Hypermethylierung wurde für die Gene MIR124-3, GATA3, GATA5, CRHBP und NEFH beim Vergleich des gesamten RCC- Kollektivs mit dem PTs gezeigt. ccRCC-Subgruppe. Statistisch signifikante tumorspezifische Hypermethylierungen für die Gene MIR124-3, GATA3, und GATA5 wurden beim Vergleich der ccRCC-Subgruppe mit dem PTs gezeigt. Für die Genloci MIR124-3, GATA3, GATA5 und NEFH wurde auch eine signifikant höhere relative Methylierung in den ccRCC-Subgruppe verglichen mit der papRCC-Subgruppe gezeigt. Die mRNA-Expression in den ccRCCs wurde für GATA5 signifikant reduziert, vergleichen mit den Normalgeweben, gezeigt. Ergänzender Vergleich der Methylierung- und Expressionsdaten für GATA5 und CRHBP zeigte inverse Korrelationen zwischen den beiden molekularen Ebenen.

Tabelle 3. Eine Übersicht der wichtigsten beschreibenden Ergebnisse zur Tumorbiologie der Nierenzelltumore. Zusammengefasst nach Genen, Kohorten und Analyseebenen. CRHBP NEFH GATA5 GATA3 MIR124-3 Kohortenvergleich vs. relative Methylierung Analyseebene RAEpeso Vergleich innerhalbmRNA-Expression der ccRCC-Subgruppe papRCC Aussage Aussage Aussage Aussage Aussage ccRCC p-Wert p-Wert p-Wert p-Wert p-Wert RCC PTs 1 3 5 4 2 renal cell carcinoma clear cell renal cell carcinoma papillary renal cell carcinoma yemtyir nRC----inverse Korrelation - - - - - höher methyliert in ccRCC - - Kohorte korrespondierender peritumoraler Gewebe hypermethyliert in - RCC höher hypermethyliert in RCC - höher hypermethylierthypermethyliert in ccRCC in RCC - hypermethylierthypermethyliert in ccRCC in RCC hypermethylierthypermethyliert in ccRCC in RCC RCC <1x10 000*<.01 .2* 001 - - <0.011* 0.026** <0.0001* <0.0001* 001 001 .1*<.0*<0.001*** <0.001* 0.015* - <0.001* <0.001* 001--007 - - 0.027* - - <0.001 .0*001 .0*-- - 0.006* - 0.001* 0.006* 1 vs. PTs -12 * 2 - - - - relative DNA-Methylierung ccRCC 3 vs. PTs schwache Assoziation mit Tumorentwicklung 2 * *** ** Keine Messungen/Analysen- durchgeführt bzw. keine Angabe gepaarter t-Test (gepaarte Proben, wenn nicht anders ke Regressionsanalyse Pearson´s Korrelationsanalyse gepaarter Proben öe ehleti cC - - höher methyliert in ccRCC ccRCC ehleti cC -- methyliert in ccRCC - methyliert in ccRCC 3 vs. papRCC 4 reduzierter mRNA- mRNA -Expression Level in ccRCC ccRCC 3 vs. PTs 2 mRNA-Expression vs. relative Methylierung inverse Korrelation <2.7x 10 nnzeichnet) n -12 *** 5

2.2.4 Translationale Ergebnisse.

Mit der Prognose assoziierte Ergebnisse.

Die Einzelwerte zu den jeweiligen Aussagen über die Signifikanz sind in Tabelle 4 aufgeführt. Relative Methylierungen von Genloci MIR124-3, GATA3, GATA5, NEFH und CRHBP waren signifikant mit klinisch-pathologischen Parametern wie Metastasenstatus (p-Werte variierten zwischen <0.0001 und 0.005) und lokalisierte / fortgeschrittene Krankheit (p-Werte variierten zwischen <0.001-0.024) assoziiert. Des Weiteren wurde für MIR124-3, GATA5 und NEFH eine Assoziation mit dem Tumorgrad (p-Werte: 0.0063-0.012) gezeigt. Nur der GATA5-Genlocus war mit Lymphknotenstatus (p-Wert = 0.03) schwach assoziiert. Für alle weiteren Parameter konnte keine signifikante Assoziation mit den relativen Methylierungen gezeigt werden. Es konnte keine Assoziationen des mRNA-Expressionslevels des GATA5-Gens mit klinisch- pathologischen Parametern, außer mit dem Tumordiameter (P=0.02), gezeigt werden. Die Assoziationen mit progressionsfreiem Überleben wurden in univariaten Cox- Regressionsanalysen mit den relativen Methylierungen für die Genloci MIR124-3, GATA5 und NEFH gezeigt. Die Hazard Ratios betrugen dabei 9.4, 13.0 und 8.6. Bei den gepaarten bivariaten Cox-Regressionsanalysen wurden die relativen Methylierungen dieser drei Genloci als statistisch unabhängige prognostische Faktoren, die mit der PFS stark assoziiert sind, gezeigt. Eine signifikante Assoziation des reduzierten mRNA-Expressionslevel des GATA5-Gens mit verkürztem PFS wurde in univariaten Cox-Regressionsanalysen gezeigt. Eine Assoziation des mRNA-Expressionslevel mit verkürztem PFS wurde bei gepaarten bivariaten Cox- Regressionsanalysen für die Parameter Geschlecht, Alter und Lymphknotenstatus gezeigt.

Tabelle 4. Eine Übersicht der wichtigsten prognose-assoziierten Ergebnisse. Zusammengefasst nach Genen, Analyseparametern und statistischen Verfahren.

Assoziation mit klinikopath. Assoziation mit progressionsfreiem Überleben Parametern Gene * ** *** p-Wert p-Wert HR p-Wert HR Hsa-mir-124-3 relative Methylierung - 0.0005 9.73 -- Geschlecht - 0.2080 2.11 0.0004 10,90 Alter - 0.7058 0,82 0.0005 18,20 Metastasen <0.0010 0.0032 4.86 0.0003 13 Lymphknoten 0.2253 0.8751 1.18 - - Lok/Adv <0.0001 0.0260 4,28 0.0081 5.87 Grad 0.0063 - - - - Diameter 0.0620 0.4278 1.67 0.0006 18.20 GATA3 relative Methylierung - n. s. -- - Metastasen 0.003 - - - - Lymphknoten 0.187 - - - - Lok/Adv 0.024 - - - - Grad 0.658 - - - - GATA5 relative Methylierung - <0.001 13.0 - - Geschlecht - - - - - Alter - 0.362 0.59 <0.001 29.7 Metastasen <0.001 0.012 4.07 <0.001 19.3 Lymphknoten 0.03 - -- - Lok/Adv <0.001 0.061 3.44 0.002 9.55 Grad 0.003 0.001 8.46 0.04 5.35 mRNA-Expression - 0.023 0.25 - - Geschlecht: n. s. 0,489 1.510 0.023 0.25 Alter: n. s. 0.155 0.420 0.035 0.27 Metastasen: n. s. 0.009 4.27 0.113 0.370 Lymphknoten n. s. 0.398 1.20 0.032 0.26 Lok/Adv n. s. 0.03 4.18 0.137 0.390 Grad n. s. <0.001 9.48 0.511 0.640 Diameter 0.02 - - - - NEFH relative Methylierung - <0.001 8.61 - - Geschlecht - 0.187 2.14 <0.001 7.88 Alter - 0.560 0.75 <0.001 17.1 Metastasen 0.005 <0.001 5.88 <0.001 11 Lymphknoten 0.202 0.283 2.26 - - Lok/Adv <0.001 <0.001 5.33 <0.001 6.96 Grad 0.012 - - - - Diameter 0.178 0.200 2.23 0.022 5.07 CRHBP relative Methylierung - ---- Geschlecht 0.081 - - - - Alter 0.490 - - - - Metastasen 0.013 - - - - Lymphknoten 0.085 - - - - Lok/Adv 0.001 - - - - Grad 0.141 - - - - * Univariate logistische Regression ** Cox Regression univariat *** Cox Regression bivariat - Keine Messungen/Analysen durchgeführt Metastasen: Fernmetastasen: M+ vs. M0 Lymphknoten Lymphknotenmetastasen: N+ vs. N0 Adv / Loc fortgeschritten (pT ≥ 3 und / oder pos. Lymphknoten (N+) und / oder pos. Fernmetastasen (M+)) vs. lokalisiert (lokalisiert und lokal fortgeschritten pT ≤ 2, Lymphknoten und Metastasen negativ) Grad Tumorgrad / -differentation: Hoch- (G>2) und Niedrig (G<=2) n. s. keine signifikante Assoziation; keine Einzelwerte in der Publikation gezeigt Methylierungs-Status Status der relativen Methylirerung bzw. globalen Methylierungsgrades HR Hazard Ratio

Mit dem Therapieansprechen assoziierte Ergebnisse.

Die Einzelwerte zu den jeweiligen Aussagen über die Signifikanz sind in Tabelle 5 zusammengestellt. Für die Genloci MIR124-3 und MIR9-1 konnten keine statistischen Assoziationen der relativen Methylierungen mit dem Therapieverlauf festgestellt werden. Bei den statistischen Analysen zeigte die Therapiekohorte bereits anhand der relativen Methylierungen in den Genloci CST6 (cystatin E/M), LAD1 (ladinin 1) und NEFH eine bimodale Verteilung in zwei niedrig und hoch methylierten Gruppen auf. Eine statistische Assoziation der relativen Methylierungen mit dem Therapieansprechen wurde für alle drei Genloci gefunden. Die niedrig und hoch methylierten Gruppen zeigten für CST6 ein medianes progressionsfreies Überleben 11.4 vs. 2.0 (p-Wert=0,009; HR=4.1). Das mediane Gesamtüberleben lag bei 22.9 vs. 3.4 Monate (p-Wert=0.011; HR=4.1). Für den LAD1-Genlocus lagen medianes PFS bei 11.4 vs. 2.0 Monate (p-Wert=0.004; HR=6.4) und OS bei 16.4 vs. 3.4 (p-Wert=0.043; HR=2.9). Die niedrig und hoch methylierten Gruppen zeigten für NEFH ein medianes progressionsfreies Überleben 29.8 vs. 9.8 (p-Wert=0,028). Die Eignung der relativen Methylierung für die Voraussage des Therapieversagens wurde durch die Berechnung der Sensitivität und der Spezifität beurteilt. Der Grenzwert (cutoff - en.) des PFS, welches als ein Parameter für bessere Auftrennung der Therapie-Ansprecher von den Nicht-Ansprechern vorgeschlagen wurde23, wurde auf 6 Monaten gelegt. Bei dichotomisierten Methylierungswerten des Genlocus CST6 ergaben sich die Sensitivität von 0.86 und Spezifität von 0.82. Für den Genlocus LAD1 - 1.0 und 0.73. Für den Genlocus NEFH wurde eine Sensitivität von 0.91 berechnet.

Tabelle 5. Eine Übersicht der wichtigsten mi dem therapieansprechen assoziierten Ergebnisse. Zusammengefasst nach Genen, Analyseparametern und statistischen Verfahren.

Überlebensanalysen* Prädiktion des Therapieversagens** Gen medianes Überleben (Monate, 95% medianes p-Wert CI) Überlebensratio Spezifität Sensitivität niedriegmethyliert hochmethyliert (HR) CST6

PFS0.009 11.4 2.0 4.1 - - OS 0.011 22.9 3.4 4.1 -- Prädiktion des - - - - 0.86 0.82 Therapieversagens LAD1 PFS0.004 11.4 2.0 6.4 - - OS 0.043 16.4 3.4 2.9 -- Prädiktion des - - - - 1.00 0.73 Therapieversagens NEFH OS 0.028 29.8 9.8 --- Prädiktion des - - - - - 0.91 Therapieversagens * Kaplan-Meier-Schätzer und Log-Rank-Test ** Cut Off-Wert = 6 Monate für die Dichtomisierung der Patienten in Responder und Non-Responder (auf die Therapie ansprechende Patienten und nicht auf die Therapie ansprechende Patienten) - Keine Angaben / Keine Messungen bzw. Analysen durchgeführt CI Confidence Interval HR Hazard Ratio PFS progresson-free survival (progressionsfreies Überleben) OS overall survival (Gesamtüberleben)

3 Diskussion

Die Dissertation basiert auf sechs bereits veröffentlichen Publikationen, die dieser Schrift beigefügt sind.

Tabelle 6. Übersicht über die Publikationen, Gene und Zitierungen. Publikationen Untersuchtes Gen Anzahl der Zitierungen Nr. (PubMed (Stand 09.05.2019)) 1 MIR124-3 27 2 GATA5 3 3 GATA3, GATA5 3 4 MIR124-3, MIR9-1, CST6, LAD1 7 5 NEFH 6 6 CRHBP 2

Die publizierten Ergebnisse können grob drei Themenbereichen zugeordnet werden.

Zum dem Bereich funktioneller Analysen gehören die Ergebnisse aus den Reexpressions- experimenten. (Publikation Nr. 6).

Die Ergebnisse aus den DNA-Methylierungs- und mRNA-Analysen in den in vitro- Zellmodellen und in den Kohorten der Tumor- und Normalgewebe des Nierenzellkarzinoms können dem Bereich der deskriptiven Tumorbiologie zugeordnet werden. (Publikationen Nr. 1- 3, 5, 6).

Zu dem Bereich der translationalen medizinischen Forschung zählen die Ergebnisse aus den prognostischen und prädiktiven Fragestellungen der These, die sich mit der Überprüfung der möglichen Bedeutung der untersuchten Marker für die klinische Medizin beschäftigten. (Publikationen Nr. 1 - 6).

Im Folgenden werden diese Aspekte im Kontext des jeweiligen Gens diskutiert.

3.1 MIR124-3.

Die MicroRNAs sind wichtige Faktoren in der Genregulation auf der posttranskriptionellen Ebene, unter anderem auch bei den neoplastischen Prozessen136-140. Die Herunterregulierung der miRNA-Synthese wurde als ein frühes weitverbreitetes Ereignis bei der Tumorgenese beobachtet141,142. Inzwischen wurden einige miRNAs wie zum Beispiel MIR122 für das metastasierende ccRCC143 und für die Wildtyp-BAP1-Subgruppe in einem 11-miRNAs-Panel als prognostische Biomarker vorgeschlagen144. Die miR-124 ist an der Kontrolle der Proliferation, des Zellzyklus und der Migration hämatopoetischer Zellen145 und an Neurogenese beteiligt146-149. Sie wird spezifisch im adulten Nervensystem exprimiert und unterdrückt dort die Expression von hunderten nicht-neuronalen Genen150. Sie wird als vermutlicher Tumorsuppressor angesehen151-157. Es wurde unter anderem über die Herunterregulierung des aggressiven Onkogen EVI1 (Ectopic virus integration 1)158,159 und ebenfalls tumorrelevanten Transkriptionsfaktor Sox9 (SRY-Box 9) durch die miR-124 vermutlich infolge einer Interaktion mit den jeweiligen mRNAs berichten. Anderseits beeinflussen die beiden Proteine ihrerseits die miR-124-Konzentration (hier wäre eine epigenetische Genrepression wahrscheinlich)157,160-166. Des Weiteren wurde sie mit den Prozessen des pathologischen Alterns und dessen Beitrag zu der Tumorgenese in Zusammenhang gebracht152,167. Unsere QMSP-Analysen an den in vitro-Zellmodellen und an den Tumorgewebsproben und korrespondierenden peritumoralen Geweben führten zu der Neuidentifizierung hypermethylierter Loci und sind deskriptiver Tumorbiologie zuzuordnen. Im Gegensatz zur früher berichteten MIR124-3-Hypermethylierung bei den Prostata-Zelllinien LNCaP, DU145 und PC3168 konnten wir in diesen Zelllinien in den von uns untersuchten Loci keine Methylierung bzw. in DU145 lediglich eine niedrige Methylierung detektieren. Als wahrscheinliche Ursache hierfür würde die Erfassung unterschiedlicher Methylierungsstellen mit unterschiedlichen Nachweismethoden (Bisulfitsequenzierung versus QMSP) in Frage kommen. Um die mögliche translationale Relevanz der Ergebnisse einzuschätzen, wurden statistische Evaluationen durchgeführt. Bei den gepaarten Analysen der Tumorgewebe und der korrespondierenden peritumoralen Gewebe wurde für die erhöhte Methylierung eine nominal signifikante tumorspezifische Assoziation gefunden (p=0.026). Beim Vergleich der relativen

Methylierungslevel der gepaarten Gewebe ist eine heterogene Zusammensetzung der Gruppe zu erkennen (Publikation Nr. 1; Fig. 3B). Es ist eine Subgruppe mit etwa gleichbleibenden Methylierungsleveln in Tumoren und peritumoralen Geweben und eine zweite Subgruppe mit starkem Anstieg der Methylierung bei den Tumoren auszumachen. Für gastrisches Gewebe wurde ein Anstieg der MIR124-3-Methylierung bereits mit präneoplastischen Prozessen assoziiert169. In zervikalen Geweben wurde graduell verlaufende Erhöhung der Methylierung von prämalignen zu malignen Geweben gefunden170. Unsere Daten weisen eher daraufhin, dass erst progrediente Tumore in der Niere eine erhöhte Methylierung aufweisen. Entsprechend wurde von uns eine statistische Assoziation zwischen Methylierung und aggressiven Tumoren sowie zwischen Methylierung und Gesamtüberleben gezeigt. Im statistischen Überlebenszeitmodell wurde demonstriert, dass die MIR124-3-Methylierung unabhängig von anderen klinisch-pathologischen Parametern mit dem Gesamtüberleben assoziiert. Insofern kann dieser Locus als unabhängiger Kandidatenprognostikator für die Hochrisiko-Subgruppe vorgeschlagen werden. Die Frage, ob auch weitere Methylierungsänderungen des MIR124-3-Gens bei der RCC-Kanzerogenese als frühes oder spätes Ereignis auftreten, ist noch zu klären. Der MIR124-3-Genlocus wurde ebenfalls in einem prädiktiven Untersuchungsansatz geprüft. Es wurde in untersuchten Therapiesubgruppen keine statistische Assoziation gefunden. Durch andere Arbeitsgruppen wurde, nach der Publikation unserer Ergebnisse, mittlerweile das diagnostische171 und prognostische157,172,173 Potential der MIR-124-3-Methylierung für andere Tumore bestätigt. Für das Ovarialkarzinom wurden MIR124-3- und MIR9-1- Methylierungen als diagnostische und prognostische Kandidaten mit sehr hoher Sensitivität und Spezifität vorgeschlagen174. Inzwischen gibt es Ansätze, die Exo-miR-124 als Krebstherapeutikum einzusetzen. Sie wurde als das effektivste Behandlungsmittel bei dem Gliom gezeigt, und es wird für diesen neuen Therapieansatz aktuell die Option erforscht, die mesenchymalen Stammzellen als natürliche Biofabriken für deren Produktion zu benutzen.175. Es wäre interessant, die Einblicke in die mit MIR124-3 verbundenen Signalwege des EVI1, des Sox9 und der Alterung bei Nierenkarzinomen durch funktionelle Untersuchungen zu vertiefen, auch mit der Aussicht auf Entdeckung neuer therapeutischer Targets.

3.2 MIR9-1.

Die miRNA9-1 wird hochexprimiert in Gehirn und beteiligt sich an der Regulation der neuronalen Differenzierung176. Aus der Literatur geht eine mehrdeutige Rolle der miR-9 bei verschiedenen Tumorentitäten hervor177. Es wurde sowohl über ihre Rolle als Onkogen178-181, als auch als Tumorsuppressor182-186 berichtet. Für ccRCCs wurde sie als tumorsuppressiv88,187 beschrieben und des Weiteren wurde eine Assoziation der MIR9-1-Methylierung mit dem Gesamtüberleben gezeigt88. Die berichteten gegensätzlichen Merkmale dieses Mikromoleküls könnten durch uneinheitliche Signalwege verschiedener Tumorentitäten und intratumoraler Subtypen oder durch das Differieren infolge unterschiedlicher zeitlicher Dynamik der Krankheit oder der Chemotherapie erklärt werden188. Zudem ist anzumerken, dass eine Assoziation der Hypermethylierung bestimmter Loci mit klinisch-pathologischen Parametern (bei der von uns untersuchten MIR9-1-Hypermethylierung war deren Assoziation mit einer schlechteren Prognose der Fall) nicht zwingend auf die genaue Rolle des Gens / der miR bei der Tumorgenese zurückschließen lässt. Es wurde bereits für einige Gene eine locusspezifische Heterogenität des Methylierungsstatus gezeigt, wobei die einzelnen CpG- Stellen in unterschiedlichen Kombinationen / Proportionen methyliert vorliegen und beide Zustände tumorassoziiert sein können59. Es bedarf tiefergreifender funktioneller Untersuchungen, um ein Gen bzw. Protein als Tumorsuppressor, Onkogen oder auch als „Passenger“-Gen für RCCs einzuordnen. Dieses gilt auch für die übrigen von uns untersuchten Gene. Von uns wurde das Potential der MIR9-1-Hypermethylierung als Kandidatenbiomarker für die translationale Medizin untersucht. Die von uns in einem statistischen Modell durchgeführte Analyse des prädiktiven Potenzials einer CGI-Subregion des MIR9-1-Gens hat keine statistischen Zusammenhänge gezeigt. Daher kann die MIR9-1-Subregion nicht unmittelbar als prädiktiver Kandidatenmarker vorgeschlagen werden. Die fehlende Assoziation könnte eventuell auf einen sich dynamisch ändernden MIR9-1-Methylierungsstatus unter Therapieeinfluss zurückgeführt werden. Die Ergebnisse der Arbeitsgruppe um Hildebrandt, welche unter anderem uns als Anstoß für die Therapiestudie dienten, haben 74 neudiagnostizierte Tumore vor Beginn der Therapie untersucht88, in unseren Therapiestudie wurden hingegen die Gewebe bereits behandelter Patienten analysiert. Interessanterweise

86 wurde gezeigt, dass MIR9-1-Methylierung sensitiv gegenüber einem VEGF -Therapeutikums (Erlotonib) war188. Möglicherweise könnte eine Sensitivität, die mit Methylierungsänderungen einhergeht, auch gegenüber weiteren Therapeutika bestehen. Nach unserer Publikation wurde die miR-9 mittlerweile von anderen Arbeitsgruppen als molekularer Marker im 5-Biomarker-Panel für die Identifizierung von Hochrisikopatienten mit kolitisassoziiertem kolorektalem Karzinom189 und in 7-Biomarker-Panel für die Voraussage des Gesamtüberlebens bei dem Leberzellkarzinom190 vorgeschlagen. Im Kontext der Literaturdaten über eine starke Beteiligung der miR-9 an neoplastischen Prozessen wären weitere Untersuchungen der Relevanz dieser miR für die ccRCCs trotz fehlender Signifikanz unserer Ergebnisse denkbar. Neue Anstöße für die funktionellen Untersuchungen der miR-9 in den Nierentumoren könnten interessante in vitro-Experimente an den glatten Muskelzellen der Pulmonalarterie der Ratte geben. Diese Zellart ist die Hauptkomponente, die bei der Hypoxie der Lunge eine vaskuläre Antwort vermittelt. Es wurde gezeigt, dass unter Beteiligung von HIF-1alpha (Hypoxia-inducible factor 1-alpha) eine hypoxieinduzierte gesteigerte miR-9-Expression ausgelöst wird, was das „Umschalten“ der normalen kontraktilen glatten Zellen der Lungenarterie zu einem proliferativen Phänotyp bewirkt191. Dieses Auslösen der Dedifferenzierung der Zellen in die Richtung gesteigerter Proliferation gäbe einen interessanten Forschungsansatz auch in der Tumorgenese der Niere. Auch die Beteiligung an den regulatorischen Vorgängen der Seneszenz wird diskutiert192.

3.3 GATA3.

Namensgebend für die Mitglieder dieser Transkriptionsfaktorenfamilie, die sich an unterschiedlichen Prozessen beteiligen193-198, ist deren Fähigkeit an die DNA-Sequenz „GATA“ zu binden. Nach neueren Erkenntnissen ist GATA3 ein Bestandteil des Foxp3 (Forkhead-Box-Protein P3) -enthaltenden Komplexes in den Treg (regulatory T cells) -Zellen. Es beteiligt sich funktionell als Master-Transkriptionsfaktor an der Zelldifferenzierung der Typ-2 T-Helferzellen199,200 und an der epigenetischen Regulierung der Gene199. Bereits früher wurde von uns eine tumorspezifische verringerte GATA3-mRNA-Expression in den RCC-Geweben201 gezeigt. Ob es einen Zusammenhang zwischen einer veränderten Genexpression und der DNA-Methylierung gibt und ob ferner die eventuell existenten

Methylierungsänderungen ein Potenzial als translationale Kandidatenmarker besitzen, wurde in von unserer Gruppe ausgeführten Analysen geprüft. Dabei wurde eine tumorspezifische Hypermethylierung des GATA3-Genlocus ermittelt. Eine Korrelation der Hypermethylierung mit der fortgeschrittenen Krankheit und dem Metastasenstatus wurden in univariaten, jedoch nicht in bivariaten Analysen gezeigt. Somit konnte für den untersuchten GATA3-Genlocus vorerst kein Potenzial als klinisch relevanter Kandidatenbiomarker festgestellt werden. Vor kurzem wurden bei in silico-Analysen von unterschiedlichen Tumorgeweben, aus der TCGA-Datenbank, unter anderem im GATA3-Gen tumorassoziierte spleißstellen-erzeugende Mutationen gefunden202. Speziell beim Brustkrebs wurden bestimmte Mutationen mit einer schlechteren Prognose für jüngere Brustkrebspatienten assoziiert203, und die immunhistochemische Färbung des GATA3-Proteins wurde als zusätzlicher Marker für den metastasierenden Brustkrebs vorgeschlagen204. Kürzlich wurde ebenfalls über die Beteiligung des GATA3 an den Prozessen der EMT (Epithelial-Mesenchymale Transition), der Störung der Adhäsion und der Entartung in eine aggressive Richtung205-208 berichtet. Diese Signalwege in den Nierenzellkarzinomen genauer zu untersuchen wäre gegebenenfalls von Interesse.

3.4 GATA5.

Zusammen mit den anderen GATA-Familienmitgliedern ist GATA5 in die Differenzierung intestinaler Epithelzellen impliziert209. Es ist beteiligt an der Regulierung der Herzentwicklung10 und interagiert mit weiteren Transkriptionsfaktoren210. Es wurden beispielsweise eine veränderte Expression das GATA5 beim Magenkrebs211-213 und ein Ausfall der Gensequenz durch LOH (loss of heterozygosity) bei Kolorektalkarzinom und Kopf-Hals-Plattenepithelkarzinom berichtet212,213, was auf seine Rolle als Tumorsuppressor zurückschließen ließe. Eine epigenetische Stilllegung des Genes wurde im Kolorektal- und Magenkarzinom beobachtet, wobei allein durch das Wiederherstellen seiner Expression eine Aktivierung stromabwärts liegender Zielgene, trotz dessen bestehender DNA-Methylierung, bewirkt wird214. Gewebebasierte Untersuchungen zeigten eine tumorspezifische Hypermethylierung im Mammakarzinom215 und hepatozellulären Karzinom, bei dem Letzteren auch mit dem Gesamtüberleben assoziiert216.

Der Nachweis der tumorspezifischen Methylierung von GATA5 in RCCs und ihre potenzielle translationale Bedeutung wurden Gegenstand unserer Untersuchungen. Bei den statistischen Analysen der Kohorte der Tumorgewebsproben und der Kohorte korrespondierender peritumoraler Gewebe waren die DNA-Methylierung des GATA5-Genlocus und der GATA5- mRNA-Expressionslevel signifikant assoziiert mit der fortgeschrittenen Krankheit bzw. Metastasierung. Dieses erlaubt, die GATA5-Hypermethylierung als einen unabhängigen Kandidatenprognostikator für verkürztes Gesamtüberleben vorzuschlagen. Der durchschnittlich stark reduzierte GATA5-mRNA-Expressionslevel in ccRCCs korrelierte invers mit der tumorspezifischen Hypermethylierung und könnte eventuell auf die epigenetische Regulierung mit der Beteiligung der DNA-Methylierung zurückgeführt werden. Ferner wäre interessant, die Existenz einer vermeintlichen Subgruppe innerhalb der Kohorte korrespondierender peritumoraler Gewebe zu klären (s. Publikation Nr. 2; Fig. 2). Anhand der relativen Methylierung teilt sich die Kohorte der korrespondierenden peritumoralen Gewebe, bei relativ homogen höheren durchschnittlichen mRNA-Expressionslevel (verglichen mit der Kohorte der Tumorgewebsproben), in hoch und niedrig methylierte Subgruppen auf. Diese frühe Hypermethylierung in der morphologisch und histologisch gesund aussehenden Gewebssubgruppe scheint nicht mit der Stilllegung des Genes einherzugehen, wie es bei den hypermethylierten Tumoren mit inverser Reduktion der mRNA-Expression der Fall ist. Dieses ließe vermuten, dass hier die Expression des GATA5-Genes auf alternativen Wegen erfolgt. Es würde auch im Einklang mit den Beobachtungen von peritumorspezifischen / präneoplastischen Hypermethylierungen einiger Gene bei verschiedenen Tumorentitäten und des Weiteren mit dem Krebsentstehungsmodell der epigenetischen Vorläuferzelle stehen169,217. Um solche alternativen Signalwege und gegebenfalls deren klinisch- pathologische Relevanz aufzudecken, sind umfangreiche funktionelle Untersuchungen erforderlich. Nach unseren Publikationen wurden tumorspezifische GATA5-Hypermethylierungen in den malignen Tumoren des kolorektalen Trakts, des sinosalen Trakts und im Cholangiokarzinom berichtet218-220. Mittlerweile wurde die GATA5-Methylierung (als Kombinationsmarker) für das kolorektale Karzinom für Detektion und Diagnose221 und beim hepatozellulären Karzinom für Prognose216 für non-invasive Tests vorgeschlagen. Interessanterweise wurde die GATA5-

Methylierung kürzlich auch von einer weiteren Gruppe als Kandidatenprognostikator für RCC und ccRCC vorgeschlagen60,222. In funktionellen Tests wurde die Reexpression des GATA5-Gens als zellwachstums- und metastasenhemmend (via Wnt/β-catenin-Signalweg) gezeigt220. Des Weiteren wurde GATA5 unter oxidativem Stress als regulatorischer Partner des transmembranen mitochondrialen Protein Bcl-xL (B-cell lymphoma-extra large) identifiziert223,224. Als Anti-Apoptoseregulator spielt das Bcl-xL auch in der Kanzerogenese eine wichtige Rolle, und die Entwicklung gegen ihn gerichteter inhibitorischer Therapieagenzien ist Gegenstand aktueller Forschung225. Die Untersuchungen dieser Signalwege bzw. Therapieansätze auch an Nierentumoren wären ebenfalls interessant. Die Annahme, dass GATA5 in die Entwicklung des aggressiven Subtyps involviert ist und die Eignung der DNA-Methylierung als klinischer Kandidatenbiomarker wären durch weitere Untersuchungen zu prüfen.

3.5 CST6.

Physiologisch scheint der endogene und sekretorische Proteinaseinhibitor Cystatin E/M226 eine wichtige Rolle bei der Entwicklung der Haut zu spielen227. Bereits die Lokalisation seines Genes in einem LOH-Bereich des 11. Chromosoms lässt auf seine tumorrelevante Eigenschaften schließen228. Es wird mittlerweile beides, seine tumorsuppressive229-237 und seine tumorfördernde Aktivität238, diskutiert. Die CST6-Methylierung wurde als Kandidatenbiomarker für die Klinik zur frühen Detektion des Brustkrebses vorgeschlagen238,239. Von uns wurde eine mögliche translationale Relevanz der DNA-Methylierung im CST6- Locus in einem statistischen Modell für RCC überprüft. Bereits anhand der Methylierungswerte in dem untersuchten CST6-Locus zeigte sich eine deutliche Aufteilung der Patienten (ohne des Überlappens) der Therapiestudie in Ansprecher- und Nichtansprecher-Subgruppen. Darüber hinaus wurden bei den statistischen Analysen eine gute Spezifität (0.86) und Sensitivität (0.82) dieses Markers berechnet. Diese Ergebnisse lassen annehmen, dass der untersuchte CST6-Locus im Zusammenhang mit der Therapieantwort steht und somit die hypermethylierten Tumore eine andere Therapiestrategie

90 benötigen. Die Ergebnisse sind verglichen mit anderen Studien dieser Art einmalig bezüglich ihrer prädiktiven Aussagekraft94,240,241. Insofern kann die CST6-Methylierung als prädiktiver Kandidatenmarker für die RCCs vorgeschlagen werden. Aufgrund der niedrigen Probenzahl war in unserer Studie eine multivariate Analyse der Unabhängigkeit von den klinisch-pathologischen Parametern bzw. die Beantwortung der Frage, ob die Kombinierung der einzelnen Marker eine prädiktive Signatur mit besserer Aussagekraft ergibt, nicht möglich. In der Literatur gibt es jedoch Hinweise darauf, dass dieses Gen auch abhängig von anderen Faktoren individuelle Methylierungsunterschiede aufweisen kann. Es wurde ein signifikanter Unterschied in der CST6-Methylierung bei Rauchern und Nichtrauchern unter Patienten des Lungenadenokarzinoms242 berichtet. Folglich wären in der Zukunft, um das prädiktive Potenzial zu überprüfen, Studien unter Berücksichtigung solche Faktoren nötig. Zum damaligen Zeitpunkt war die uns zur Verfügung stehende an Zahlengröße vergleichsweise kleine Kohorte die größte verfügbare für derartige Studien. Nichtsdestotrotz gibt die gezeigte statistische Signifikanz der Ergebnisse einen Hinweis auf die Relevanz des zugrundeliegenden Effekts. Nach unserer Publikation wurde die Bedeutung des CST6 unter anderem bei Brustkrebs untersucht. Für bestimmte Brustkrebssubgruppen wurde es als potenzieller Tumorpromotor identifiziert243, gleichzeitig wurde eine CST6-Hypermethylierung der freizirkulierenden Tumorzellen als signifikant mit kürzerem progressionsfreiem Überleben gezeigt235. Diese Beispiele zeigen, dass auch für die Nierentumore neben der Überprüfung der klinischen Relevanz des CST6 eine Aufklärung der funktionellen Rolle dieses Proteins noch aussteht.

3.6 LAD1.

Das Gen LAD1 kodiert Ladinin 1, ein in der Basalmembran verankertes Filament, was zu der Stabilität der Dermis-Epithel-Verbindung beiträgt244 und in der Haut und inneren Organen (unter anderem auch der Niere) exprimiert wird245. Bei den linearen IgA-Dermatosen werden die Ladinin1-Proteine mit bestimmten Genvariationen von den IgA-Antikörpern anvisiert, was zu der Störung der Derma-Epidermis-Kohäsion und Hauterkrankungen führt246-248. Es

91 wäre ein Hinweis darauf, dass dieses Protein mitunter für die Integrität der gesunden Zellen im Gewebsverband sorgt. LAD1 wird als ein stromabwärts liegender (in Gegensatz zur miR-9, die in diesem Signalweg stromaufwärts steht) Interaktionspartner des Proto-Onkogens BRAF (Isoform B Rapidly Accelerated Fibrosarcoma) in diversen tumorspezifischen Signalwegen wie der MAPK (mitogen-activated protein kinase) und des AKT/mTOR249 beschrieben. Beim Mammakarzinom agiert LAD1 als Onkogen250 und wurde in den Mamma-Ca.-Zelllinien als Mediator diverser regulatorischer Prozesse251, unter anderem der Mitoseeinleitung252, Proliferation und Migration251,252, identifiziert und mit aggressiven Subtypen assoziiert253. Für Laryngealkarzinom wird ebenfalls seine metastasenfördernde Rolle diskutiert254. Unsere Gruppe hat im Rahmen der Prädiktionsstudie das translationale Potenzial der DNA- Methylierung eines LAD1-Genlocus untersucht. Die von uns gezeigte Hypermethylierung in den primären RCC-Tumorgeweben wurde in statistischen Analysen mit dem progressionsfreien Überleben der Patientengruppe der Erstlinientherapie und mit dem Gesamtüberleben der Sequenztherapiepatienten als signifikant assoziiert gezeigt. Dabei erreicht dieser epigenetische Marker bei der Prädiktion des Therapieversagens eine hohe Spezifität (1.00) und gute Sensitivität (0.73). Auch für LAD1 gelten bezüglich kleiner Anzahl der untersuchten Patienten mit gleichzeitig stark signifikanten Ergebnissen dieselben Überlegungen wie bereits für das CST6. Nach unserer Publikation wurde das LAD1-Protein zusammen mit weiteren Markern als Kandidat für die Molekulardiagnostik des Lungenadenokarzinoms vorgeschlagen255. Des Weiteren wurde die LAD1-Methylierung kürzlich als prognostischer Kandidatenmarker für RCC und ccRCC von einer anderen Arbeitsgruppe bestätigt60,222. Für renale Tumore sind funktionelle Untersuchungen und Einordnung des LAD1 bei den Signalwegen der Tumorgenese und erweiterte Studien bezüglich seiner Eignung als klinischer Biomarker erforderlich.

3.7 NEFH.

Das NF-H (neurofilament heavy chain - en.) wird von den NEFH-Gen kodiert und ist, im Triplet mit zwei anderen Untereinheiten (NF-L (neurofilament light chain - en.) und NF-M (neurofilament medium chain-en.), für den axonalen Durchmesser und intrazellulären Transport als zellstrukturgebendes Protein verantwortlich. Im klassischen Sinne wird dieses Protein als neuronaler Marker bei der Diagnostik verschiedener neuropathologischer Fragestellungen herangezogen256-259. Mittlerweile wurde die Anwesenheit der NF-H- Untereinheit bzw. die Genexpression in verschiedenen Geweben auch außerhalb der Neuronen dokumentiert260-262. Eine veränderte Expression bzw. Methylierung wurde im mit Kanzerogenen exponierten Ösophagus-Plattenepithel und eine erhöhte Methylierung im anliegenden Normalgewebe bei den Ösophaguskarzinom -Patienten gezeigt263. NEFH wurde als Tumorsuppressor, teilweise mit dem aggressiven Subtypen assoziiert, auch in anderen Tumorentitäten beobachtet264-268. Unsere Arbeitsgruppe hat neben der tumorspezifischen Hypermethylierung des NEFH-Locus eine von allen verfügbaren klinisch-pathologischen Parametern unabhängige Assoziation der Methylierung mit dem progressionsfreien Überleben (sowohl in uni- als auch bivariaten Analysen mit konstant hohen HR-Werten) gezeigt. Somit kann der untersuchte epigenetische Marker als unabhängiger Kandidatenprognostikator für ccRCCs vorgeschlagen werden. Bei der statistischen Überprüfung seines prädiktiven Potenzials wurde eine trennscharfe Aufteilung der Therapiepatienten in hoch und niedrig methylierte Subgruppen (stark korrelierend mit dem Gesamtüberleben) gezeigt. Aufgrund dieser Ergebnisse kann die NEFH- Locusmethylierung als prädiktiver Kandidatenmarker ebenfalls vorgeschlagen werden. Mittlerweile wurde die NEFH-Methylierung als ein unabhängiger prognostischer Faktor in Kombination mit weiteren Markern für ein bestimmtes Ovarialkarzinom-Subtyp vorgeschlagen269. Des Weiteren wurde die NEFH-Methylierung kürzlich als Kandidatenmarker für die Prognose bei RCC und ccRCC von einer anderen Arbeitsgruppe bestätigt60,222. Zu den Eigenschaften, speziell der aggressiven Tumorzellen, gehört unter anderem der Verlust der Zellintegrität im Zellverband, mit anschließender Mobilisation und Invasion ins umliegende Gewebe. Beim Eindringen in die dichte extrazelluläre Matrix weisen die Tumorzellen verbesserte Deformationseigenschaften auf, was durch die Veränderungen in der

Zytoskelettstruktur erreicht wird. Während der Tumorneogenese und speziell der EMT finden Dysregulierung und Destabilisierung verschiedener Zytoskelettkomponenten statt, wobei diverse regulatorische Elemente bei der Auflösung der Zellpolarisierung eine Rolle zu spielen scheinen270-273. Es gibt Hinweise darauf, dass unter anderem auch die Neurofilamente, als zytoskelettbildende Strukturen an der Tumorgenese beteiligt sein können. Das als Onkogen bekannte miRNA25 löst die Proliferation und Invasion der Glioblastomzellen aus, und dabei wurde das NF-L als ihr direktes Zielmolekül identifiziert274. Überdies gilt NF-L als Inhibitor des mTOR-Signalweges275. NF-L ist eine der NF-H-Co-Untereinheiten. Inwieweit das NEFH im Einzelnen und Neuroproteine im Allgemeinen in die Signalwege der Mobilisierung der Nierentumorzellen bzw. in die Tumorentstehung und / oder Tumorprogression involviert sind, wäre in weiterführenden Untersuchungen zu klären.

3.8 CRHBP.

Das CRHBP-Gen kodiert ein sekretorisches Glycoprotein des CRH-Systems (Corticotropin- releasing Hormone - en.), welches bei Koordination der Stressantwort und Homöostase eine zentrale Rolle spielt276,277. Das CRHBP-Gen wird physiologisch in den Glomeruli exprimiert278. Es wurde bereits mehrmals über die Beteiligung des CRH-Systems an der physiologischen279- 281 und tumorspezifischen282-286 Modulation der Angiogenese berichtet. Bei den funktionellen Tests wurde gezeigt, dass CRH, CRHR1 (Corticotropin-releasing hormone receptor 1) und CRHR2 (Corticotropin-releasing hormone receptor 2) in die Prozesse der Tumorgenese (unter anderem Tumorangiogenese) involviert sind287-290. Das CRHR1-Protein wurde mittlerweile als prognostischer Marker bei dem Endometriumkarzinom291 und CRH-mRNA als Teil eines mRNA-Panels bei Überwachung des Blasenkarzinoms292 vorgeschlagen. Für das Leberkarzinom wurden in in silico-Analysen eine tumorspezifische Herunterregulierung des CRHBP293 und des Weiteren eine signifikante Assoziation der Expression mit der schlechten Prognose gezeigt294. Von uns wurden initiale funktionelle Untersuchungen an den renalen Tumormodellen durchgeführt. Bei der Charakterisierung des Invasionsverhaltens der mit Hilfe von siRNA- Transfektion CRHBP-reexprimierten Zelllinie RCC-GS wurden abhängig von der angesetzten

Konzentration des 5-Aza-2`-Deoxycytidins (vorgeschaltete unspezifische Demethylierung) gegensätzliche Effekte beobachtet. Die höhere Konzentration (0,5µM) bewirkte ein invasiveres Verhalten der Zellen und die niedrigere (0,125µM) umgekehrt. Eine mögliche Erklärung wäre, dass unspezifische Expressionsänderungen durch die 5-Aza-2`- Deoxycytidin-Behandlung einen heterogenen Einfluss haben und zugleich von der Agenskonzentration abhängig sind. Es wurde berichtet, dass bestimmte Genpromotoren (vermutlich durch das Beibehalten des semi-heterochromatischen Zustandes) nach der Entfernung des Demethylierungsagenzen schnell erneut hypermethyliert und reprimiert werden295. Wenn das gleiche Verhalten auch auf das CRHBP-Gen zuträfe, wäre zusätzlich zu der Optimierung der 5-Aza-2`-Deoxycytidin-Konzentration auch dieser Aspekt bei der Planung weiterer Experimente zu berücksichtigen. Die von unserer Arbeitsgruppe mit immunhistologischen Nachweismethoden und Westernblots bereits früher gezeigte starke Reduktion des CRHBP-Proteins in ccRCC- Geweben, verglichen mit den korrespondierenden peritumoralen Geweben286, steht im Einklang mit den in der Publikation Nr. 6 beschriebenen Ergebnissen unserer Methylierungs- und RNA-Analysen. Die inverse Korrelation des Methylierungsgrades mit der mRNA- Expression286 und die Assoziation der Hypermethylierung mit der Aggressivität der ccRCCs wurden in beiden Fällen durch die in silico-Validierung mit der TCGA KIRC-Datenbank bestätigt (Publikation Nr.6). Unsere statistischen Analysen zeigten, dass dieses Mitglied des CRH-Systems in die Tumorgenese der ccRCCs (und insbesondere des aggressiven Subtyps) involviert ist. Die Eignung der CRHBP-Methylierung / Expression als prognostischer oder prädiktiver Marker sollte in weiteren funktionellen Untersuchungen und Validierungsstudien geprüft werden. Die Beteiligung des CRH-Systems an den Prozessen der Tumorangiogenese macht die beteiligten molekularen Strukturen auch für die Erforschung neuer VEGF- intervenierender Therapieansätze interessant.

3.9 Zusammenfassende Diskussion.

3.9.1 Neue Erkenntnisse.

Funktionelle Untersuchungen. Die neuen Daten der Demethylierungs- und Reexpressionsexperimente an den renalen Tumorzelllinien zeigten eine Beteiligung des CRHBP an dem Invasionsverhaltens der RCC-GS-Zelllinie das durch die Konzentration des Demethylierungsagens beeinflüsst war. Deskriptive Tumorbiologie. Die gewonnenen Daten über die Methylierungsgrade und gegebenfalls Expressionslevel der Genloci MIR124-3, MIR9-1, GATA3, GATA5, CST6, LAD1, NEFH und CRHBP in den untersuchten Zellkulturmodellen und Kohorten der renalen Primärgewebe sind investigativer Natur und tragen zur molekularen Charakterisierung der RCCs bei. Für die Marker GATA5 und CRHBP wurde hier jeweils eine inverse Korrelation zwischen mRNA-Level und Methylierungsgrad gefunden. Translationale Tumorbiologie. Aus den vorgestellten Ergebnissen geht deutlich hervor, dass die DNA-Methylierungsänderungen in MIR124-3, GATA3, GATA5, CST6, LAD1, NEFH und CRHBP mit der Entwicklung der Nierenzellkarzinome signifikant korrelieren. Die DNA- Methylierungen der Genloci MIR124-3, GATA5 und NEFH wurden in der vorgelegten Arbeit als geeignete prognostische Kandidatenmarker für das Rückfallrisiko identifiziert. Unsere Studien zeigten ferner, dass die signifikant erhöhten Methylierungen untersuchter Loci der Gene CST6, LAD1 und NEFH im statistischen Zusammenhang mit der Therapieantwort stehen. Trotz geringer Anzahl der analysierten Proben wurde eine klare, allein durch die Methylierungswerte zustande kommende Dichotomisierung der Proben in therapieantwort- assoziierte Subgruppen beobachtet. Sofern sich die gute Spezifität und Sensitivität dieser epigenetischen Alterationen bei der Therapieprädiktion in weiteren Studien erhärten, lässt sich damit die Erwartung vertiefen, dass anhand des Methylierungsgrades des Tumorgewebes eine Prädiktion bezüglich der Therapiestrategie möglich wäre. Somit können diese Epialterationen als prädiktive Kandidatenmarker vorgeschlagen werden. Inzwischen wurde von unserer Gruppe ein weiterer epigenetischer Kandidatenmarker NELL1 (neural EGFL like 1), der mit fortgeschrittener Krankheit assoziiert, berichtet296. Kürzlich wurden LAD1 und NEFH in einem 4-Methylierungsmarker-Panel, basierend auf den Ergebnissen der Messungen an zwei unabhängigen Kohorten der primären ccRCCs, als Prognosemarker für ccRCCs vorgeschlagen222. GATA5, LAD1 und NEFH gehören auch zu

96 den insgesamt neun für die Prognostik der RCCs von der gleichen Arbeitsgruppe vorgeschlagenen Marker in einem kürzlich erschienenen Review297. Diese Empfehlung ist Ergebnis einer Metaanalysenstudie von 49 einschlägigen Publikationen297. Diese Daten können als erste Validierungen der von uns identifizierten oben genannten Marker angesehen werden. Die Beobachtung, dass die untersuchten epigenetischen Muster aller acht Gene unterschiedliche klinische Relevanz zeigten, ist auf ihre individuellen funktionellen Eigenschaften zurück zu führen. Alle Prozesse im Organismus sind ein Zusammenspiel immenser Vielzahl an internen und externen bio- / chemischen Molekülen und intrinsischen und extrinsischen Faktoren. Es wäre daher nicht zu erwarten, je einen einzelnen robusten und unabhängigen Marker für bestimmte klinische Fragestellungen identifizieren zu können. Sinnvoller erscheint es die Erstellung kombinierter und integrativer fragespezifischer Markerpanels anzustreben.

3.9.2 Aktuelle klinische Anforderungen bei RCCs.

Die Notwendigkeit einer neuen Generation klinischer Marker für die Nierenzellkarzinome ist gegeben. Zwar liefern die aktuellen Modelle eine allgemeine prognostische Aussage, jedoch weniger eine über die Therapiewahl, obwohl den Klinikern eine neue Generation gezielter Krebstherapeutika, die unterschiedliche Therapieregime ermöglichen, bereits zur Verfügung steht. Dabei wären bei der Wahl der Behandlungsmethode unter anderem solche Aspekte wie häufige Resistenzen, schlechte Erfolgschancen bei der späten Diagnose (palliativer Aspekt bei den mRCCs), erhebliche Nebenwirkungen und Kostenfragen zu berücksichtigen. Daher wäre eine Verbesserung bereits vorhandener Entscheidungsmodelle, um die Subgruppe der Therapieansprecher identifizieren zu können und eine weitergehende Individualisierung der Therapieschemata zu erreichen, von großem Interesse.

3.9.3 Aspekte für zukünftige Untersuchungen.

Für die klinische Nutzung molekularer Marker wären funktionelle Untersuchungen und klinische Validierungsstudien notwendig. Bei dem Design zukünftiger Untersuchungen, wie Markerscreening, funktionelle Untersuchungen und klinische Validierungsstudien, um statistisch robuste biomolekulare Signaturen zu entdecken bzw. zu etablieren, sind mehrere allgemeingültige Faktoren zu berücksichtigen. Weiter unten folgen, ohne einen Anspruch auf die Vollständigkeit, einige solche relevante Aspekte. Als Erstes wäre bei der Identifizierung der Methylierungsmarker festzuhalten, dass die zu untersuchenden Loci allein schon aufgrund ihrer Lokalisation im Genom unterschiedlich aussagekräftig bezüglich gesuchter Assoziationen sein können. Beispielsweise weisen einige neuere Untersuchungen111,298-302 daraufhin, dass die funktionell wichtigen Methylierungsänderungen in den Tumoren nicht nur in den „klassischen“ Bereichen der Expressionsteuerung wie Promoter und / oder in den dichtmethylierten CGIs zu finden wären, die bis heute bevorzugt primär untersucht werden. So können je nach Fragestellung sogenannte „CpG island shores“, die bis zu 2 kb die CpG-Inseln flankieren relevant sein, was auf die Kontrollfunktion solcher „shores“ in Bezug auf z. B. Enhancer zurückgeführt wird298,302. Auch die Methylierungen weiterer, von den GpG-Inseln freier, genomischer Strukturen, wären im Stande, als tumorale Biomarker entscheidende klinische Aussagekraft zu liefern. Das wären beispielsweise sogenannte „CpG-Canyons“ (niedrig methylierte Regionen, die oft die Andockstellen für Transkriptionsfaktoren beinhalten)299, repetitive Elemente allgemein111, die innerhalb der Introns liegende Transposons und kryptische bzw. Antisense-Startstellen im einzelnen300. Überdies könnte bereits der Unterschied im Methylierungsniveau einer einzigen CpG-Stelle etwas über die Aktivität eines Genes bzw. über seine Teilnahme an pathologischen Prozessen aussagen303. Die genomweite Sequenzierung und assoziative Analysen der Methylierungsmuster würden für detailliertere Analysen von Nutzen sein. Ein weiterer zu beachtender Faktor wäre, unter anderem auf die sogenannte Feldkanzerisierung und die klonale Evolution der Tumore zurückzuführende, häufig vorhandene intra-/intertumorale epi-/genetische Heterogenität. Mit steigender intratumoraler Heterogenität (unter anderem einhergehend mit durch den therapeutischen Eingriff hervorgerufener sogenannter „bottleneck“ evolution (en.) - genetischer Flaschenhals-

Evolution (dt.)) wäre auch ein steigendes Therapieresistenzrisiko zu erwarten. Eine repräsentative Abbildung der vorhandenen klinisch relevanten Heterogenität, vor dem Hintergrund klinisch bzw. histologisch „konvergenter Phänotypen“ der Tumore, wäre anzustreben. Es wird aktuell diskutiert, ob dafür die Einzelbiopsie-Entnahmen, die nur eine Subregion in dem Tumor erfassen und nicht alle tatsächlich vorhandenen Tumorsubklone (unter anderem klinikrelevante), ausreichen würden304,305. Speziell für die ccRCCs wurde festgestellt, dass die intratumorale Heterogenität der Treibermutationen („driver mutations“) bei den multiregional entnommenen Geweben, verglichen mit den in dem TCGA publizierten Einzelbiopsie-Studien, mit einer höheren Mutationsfrequenz einhergeht51,58. Andererseits wurde gezeigt, dass es möglich wäre die klonale Architektur aus den Einzelbiopsien zu rekonstruieren55. Als Konsequenz aus dieser noch unzureichend erforschten aber eventuell klinisch relevanten Problematik wurden Multiregion-Probenentnahmen und „single-cell“- Analyseverfahren vorgeschlagen. Sie würden die Detektion heterogener molekularer „Passagier-“ und „Treiber“-Veränderungen, unter anderem auch in den prämalignen Läsionen in Bezug auf ihr tumorigenetisches Potenzial ermöglichen306. In unseren Prädiktionsstudien waren multivariate Unabhängigkeitsanalysen wegen der kleinen Probenzahl statistisch nicht durchführbar. Auch die Größe und die Heterogenität des Kollektivs bezüglich unterschiedlicher individueller Faktoren (Geschlecht, Alter, Lebensstil, BMI (body mass index - en.), Diät usw.), die Methylierungsänderungen beeinträchtigen können, wären bei der Planung weiterer Studien zu beachten. Des Weiteren wären beispielsweise die zeitlichen Umstände der Probenentnahme (dynamische Änderungen des molekularen Status, unter anderem durch die therapeutischen Maßnahmen) und die Unterschiede in den Analyseverfahren zu berücksichtigen188,307. Die Vor- und Nachteile der Zelllinien als Tumormodelle für die funktionellen Untersuchungen wären ebenfalls zu berücksichtigen. Die Ergebnisse aus den Experimenten an Zellkulturmodellen, die als komfortable und praktikable „Monokulturen“ der Tumorzellen dienen, geben nur die ersten Anhaltspunkte für weitere Untersuchungen und sind nicht eins zu eins auf die in vivo-Prozesse zu übertragen. Das beruht darauf, dass die Zellen in Kultur der permanenten in vitro-Modelle sich in einer von der natürlichen abweichenden Umgebung befinden und einem abweichenden Selektionsdruck als innerhalb einer sich teilenden Tumorzellpopulation in vivo unterliegen. Daraus resultieren auch entsprechend abweichende molekularbiologische Profile wie Expressions- und Metabolismus-Muster. Die Zellen können

99 sich häufig eine zusätzliche Methylierung (vor allem der gewebsspezifischen Gene) „anlegen“, die teilweise 5-90-fach höher sein kann als in den Ursprungstumoren308. Die Anwendung der Zelllinien bei der Entwicklung neuer Therapeutika könnte auch aufgrund veränderter Methylierungsmuster ihre Limitationen haben. Es wurde bereits gezeigt, dass die renalen Zelllinien veränderte Methylierungsmuster (unter anderem auch der therapierelevanten Gene) im Vergleich zu primären Tumor- und Metastasen-Geweben aufweisen309. Wenngleich der epigenetische Status der Primärtumore im Großen und Ganzen wiedergegeben wird, ist es wichtig die Korrelation der epigenetischen Muster untereinander zu überprüfen121. Es wird mittlerweile die Notwendigkeit „epigenetischer Tumormodelle“ diskutiert306. Die von uns initial gefundenen Tendenzen in Änderungen der DNA- Methylierung in den Zelllinien wurden durch die Messungen in den Gewebskohorten weitgehend bestätigt. Zu den starken Hilfsmitteln bei dem Screening und der Validierung von Kandidatenmarkern gehören seit einigen Jahren in silico-Analysen, bei denen bioinformatische Verfahren für die Extraktion klinisch nützlicher Informationen verwendet werden. Dabei wird eine gewaltige Menge an externen Daten (unter anderem aus dem TCGA- Projekt) alleine oder in Kombination mit eigenen experimentell generierten Daten ausgewertet310,311. Als effizienter und präziser, verglichen mit konventionellen „Eine-Ebene- Datensätzen“, werden aktuell integrative Analysen sogenannter „cross-omics“ Daten (molekulare Profile auf beispielsweise genetischer, epigenetischer und Expressionsebenen) diskutiert59,312,313. Durch die Definierung molekularer Subtypen der ccRCCs mit Hilfe von Kombination genomischer und epigenomischer Daten, konnte beispielsweise eine Aussage über den Erfolg von Sunitinibtherapie bei metastasierenden Patienten gemacht werden314. Die Durchführung der in silico-Analysen ist ein sinnvolles Werkzeug bei den Biomarkeruntersuchungen.

100

3.9.4 Perspektiven für die klinikorientierte RCC-Forschung.

Neben den prognostischen und prädiktiven Fragestellungen wäre der Einsatz biomolekularer Marker in weiteren translationalen Bereichen der medizinischen Forschung denkbar. Sinnvoll wären beispielsweise die Einsatzmöglichkeiten der biomolekularen Marker für die Diagnostik bzw. Kontrolle prä-neoplastischer Läsionen, für die Überwachung von Patienten mit nachteiligen biomolekularen Merkmalen bzw. die Therapieüberwachung zwecks rechtzeitiger Resistenzindikation. Überdies würde die Suche nach prognostischen und prädiktiven Biomarkern die Kandidaten für die neuartigen therapeutischen Interventionen wie beispielsweise miRNA-Inhibitoren bzw. miRNA-Mimics315 oder auch für epigenetische Modulatoren, wie DNA-Methyltransferasen- Inhibitoren und Histon-Deacetylasen-Inhibitoren316, liefern. Auch für den neuen therapeutischen Ansatz, bei dem die kompetitiven Fähigkeiten therapiesensitiver Klone ausgenutzt werden, wäre das Erstellen individueller epigenetischer molekularer Profile des Patienten vom Nutzen54,317.

101

4 Zusammenfassung.

Identifizierung epigenetischer Alterationen als Biomarker für Genese, Progression und Therapieansprechen des Nierenzellkarzinoms. Natalia Dubrowinskaja

Fragestellung. Die aktuellen prognostischen Modelle, die zwecks Risikostratifizierung und Therapieauswahl bei der Behandlung der RCCs herangezogen werden, weisen einige Limitationen auf. Vor dem Hintergrund, häufig vorhandener Therapeutikaresistenzen, aktueller gezielter Therapieansätze und einer mutmaßlichen ausgeprägten inter- und intratumoralen Heterogenität ist eine Verbesserung bereits vorhandener Entscheidungsmodelle erforderlich. In dieser Dissertation wurde der Frage nachgegangen, ob die untersuchten epigenetischen Alterationen das Potenzial besitzen, brauchbare Informationen zur Verbesserung der Prognose bzw. Prädiktion des Nierenzellkarzinoms zu liefern. Erkenntnisse. 1) Es wurde eine Beteiligung des CRHBP-Gens an dem Invasionsverhalten der tumorrelevanten Nierenzellkarzinom-Zelllinie RCC-HS gezeigt, was zum Verständnis und Aufklärung der hinter der Tumorentwicklung bei den Nieren stehender molekularer Mechanismen beiträgt. 2) Es wurden neue Informationen über DNA-Methylierungssignaturen der in vitro- Zellmodelle und renaler Primärgewebe der Gene MIR124-3, MIR9-1, GATA3, GATA5, CST6, LAD1, NEFH und CRHBP gewonnen. 3) Hinsichtlich einer potenziellen Rolle als prognostische bzw. prädiktive Marker für RCC wurden statistisch signifikante Ergebnisse für die Methylierungssignaturen der Gene MIR124-3, GATA5, CST6, LAD1, NEFH und CRHBP erzielt. Fazit. Die Ergebnisse dieser Arbeit weisen klar daraufhin, dass DNA- Methylierungsalterationen für prognostische oder prädiktive Fragestellungen bei RCCs aussichtsreiche Kandidaten darstellen. Offene Fragen. Zukünftige Untersuchungen müssen bei der Ausweitung bzw. Optimierung methylierungsbasierter Signaturen einige relevante Faktoren berücksichtigen. Des Weiteren sind prospektive, unabhängige, multizentrische Evaluierungen notwendig.

102

Ausblick. Ferner ist zu erwarten, dass die Durchführung funktioneller Untersuchungen an identifizierten Kandidaten zu einem verbesserten Verständnis der Tumorbiologie beitragen könnte bzw. Grundlage zur Identifizierung neuer therapeutischer Zielstrukturen darstellen könnte.

103

5 Summary.

Identification of epigenetic alterations as a biomarker of genesis, progression and therapy response of renal cell carcinoma. Natalia Dubrowinskaja

Background. The current prognostic models used for risk stratification and therapy selection in the treatment of RCCs have some limitations regarding their prognostic and predictive value. Based on settings of frequently existing therapeutic resistance, currently used targeted therapies and a putative pronounced inter- and intratumoral heterogeneity, an improvement of already existing decision models is necessary. The aim of this thesis was to examine whether the investigated epigenetic alterations have the potential to provide useful information for improving the prognosis or prediction of renal cell carcinoma. Findings. 1) Involvement of the CRHBP gene in the invasive behavior of the renal tumor relevant cell culture line RCC-HS has been demonstrated, which contributes to the understanding and elucidation of the molecular mechanisms behind the tumor development of the kidney. 2) New information was obtained on DNA methylation signatures of in vitro cell culture models and renal primary tissues of the genes MIR124-3, MIR9-1, GATA3, GATA 5, CST6, LAD1, NEFH and CRHBP. 3) Regarding a potential role as prognostic or predictive markers of RCCs, statistically significant results were obtained for the methylation signatures of the genes MIR124-3, GATA5, CST6, LAD1, NEFH, and CRHBP. Conclusions. The results of this work clearly indicate that DNA methylation alterations are promising candidates for prognostic or predictive Markers in RCCs. Open questions. Future investigations should consider some relevant aspects in extending or optimizing methylation-based signatures. Furthermore, prospective, independently, multicenter evaluations will be necessary. Outlook. It can be further expected that identified candidates after functional examinations could contribute to an improved understanding of tumor biology or could represent a basis for the identification of new therapeutic target structures.

104

6 Literaturverzeichnis.

(1) Bray, F.; Ferlay, J.; Soerjomataram, I.; Siegel, R. L.; Torre, L. A.; Jemal, A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer. J. Clin. 2018, 68, 394-424.

(2) Ferlay, J.; Colombet, M.; Soerjomataram, I.; Mathers, C.; Parkin, D. M.; Pineros, M.; Znaor, A.; Bray, F. Estimating the global cancer incidence and mortality in 2018: GLOBOCAN sources and methods. Int. J. Cancer 2019, 144, 1941-1953.

(3) Robert-Koch-Institit Krebs in Deutschland 2011/2012; 2015; (abgerufen am: 23.03.2019).

(4) McGuire, S. World Cancer Report 2014. Geneva, Switzerland: World Health Organization, International Agency for Research on Cancer, WHO Press, 2015. Adv. Nutr. 2016, 7, 418-419.

(5) Al-Bayati, O.; Hasan, A.; Pruthi, D.; Kaushik, D.; Liss, M. A. Systematic review of modifiable risk factors for kidney cancer. Urol. Oncol. 2019, 37, 359-371.

(6) Fisher, R.; Gore, M.; Larkin, J. Current and future systemic treatments for renal cell carcinoma. Semin. Cancer Biol. 2013, 23, 38-45.

(7) NCI, N. C. I. SEER, Cancer Stat Facts. 2016; (abgerufen am: 31.05.2019).

(8) Cohen, N. M.; Kenigsberg, E.; Tanay, A. Primate CpG islands are maintained by heterogeneous evolutionary regimes involving minimal selection. Cell 2011, 145, 773-786.

(9) Curtis, S. A.; Cohen, J. V.; Kluger, H. M. Evolving Immunotherapy Approaches for Renal Cell Carcinoma. Curr. Oncol. Rep. 2016, 18, 57.

(10) Reiter, I.; Krammer, B.; Schwamberger, G. Cutting edge: differential effect of apoptotic versus necrotic tumor cells on macrophage antitumor activities. J. Immunol. 1999, 163, 1730-1732.

(11) Cao, Y. Endogenous angiogenesis inhibitors and their therapeutic implications. Int. J. Biochem. Cell Biol. 2001, 33, 357-369.

(12) Folkman, J. Endogenous angiogenesis inhibitors. APMIS 2004, 112, 496-507.

(13) Faivre, S.; Kroemer, G.; Raymond, E. Current development of mTOR inhibitors as anticancer agents. Nat. Rev. Drug Discov. 2006, 5, 671-688.

(14) Vignot, S.; Faivre, S.; Aguirre, D.; Raymond, E. mTOR-targeted therapy of cancer with rapamycin derivatives. Ann. Oncol. 2005, 16, 525-537.

(15) Tanneeru, K.; Guruprasad, L. Ligand-based 3-D pharmacophore generation and molecular docking of mTOR kinase inhibitors. J. Mol. Model. 2012, 18, 1611-1624.

(16) Keir, M. E.; Butte, M. J.; Freeman, G. J.; Sharpe, A. H. PD-1 and its ligands in tolerance and immunity. Annu. Rev. Immunol. 2008, 26, 677-704.

105

(17) Zarour, H. M. Reversing T-cell Dysfunction and Exhaustion in Cancer. Clin. Cancer Res. 2016, 22, 1856-1864.

(18) Reiter, M. A. Nierenzellkarzinom Medikamentöse Therapie und prognostische Modelle. Onkologe 2014, 12, 1241-1251.

(19) Albiges, L.; Oudard, S.; Negrier, S.; Caty, A.; Gravis, G.; Joly, F.; Duclos, B.; Geoffrois, L.; Rolland, F.; Guillot, A.; Laguerre, B.; Legouffe, E.; Kohser, F.; Dietrich, P. Y.; Theodore, C. A.; Escudier, B. Complete remission with tyrosine kinase inhibitors in renal cell carcinoma. J. Clin. Oncol. 2012, 30, 482-487.

(20) Procopio, G.; Verzoni, E.; Bracarda, S.; Ricci, S.; Sacco, C.; Ridolfi, L.; Porta, C.; Miceli, R.; Zilembo, N.; Bajetta, E.; Italian Trials in Medical Oncology (ITMO) group Overall survival for sorafenib plus interleukin-2 compared with sorafenib alone in metastatic renal cell carcinoma (mRCC): final results of the ROSORC trial. Ann. Oncol. 2013, 24, 2967-2971.

(21) Motzer, R. J.; Basch, E. Targeted drugs for metastatic renal cell carcinoma. Lancet 2007, 370, 2071-2073.

(22) Porta, C.; Sabbatini, R.; Procopio, G.; Paglino, C.; Galligioni, E.; Ortega, C. Primary resistance to tyrosine kinase inhibitors in patients with advanced renal cell carcinoma: state-of-the-science. Expert Rev. Anticancer Ther. 2012, 12, 1571-1577.

(23) Seidel, C.; Busch, J.; Weikert, S.; Steffens, S.; Fenner, M.; Ganser, A.; Grunwald, V. Progression free survival of first line vascular endothelial growth factor-targeted therapy is an important prognostic parameter in patients with metastatic renal cell carcinoma. Eur. J. Cancer 2012, 48, 1023-1030.

(24) Arbeitsgemeinschaft der Wissenschaftlichen Medizinischen Fachgesellschaften e.V. (AWMF) , der Deutschen Krebsgesellschaft e.V. (DKG) und der Deutschen Krebshilfe S3-Leitlinie Diagnostik, Therapie und Nachsorge des Nierenzellkarzinoms. (abgerufen am: 31.05.2019)

(25) Pichler, M.; Hutterer, G. C.; Chromecki, T. F.; Jesche, J.; Kampel-Kettner, K.; Groselj-Strele, A.; Hoefler, G.; Pummer, K.; Zigeuner, R. Comparison of the 2002 and 2010 TNM classification systems regarding outcome prediction in clear cell and papillary renal cell carcinoma. Histopathology 2013, 62, 237-246.

(26) Kim, H. L.; Han, K. R.; Zisman, A.; Figlin, R. A.; Belldegrun, A. S. Cachexia-like symptoms predict a worse prognosis in localized t1 renal cell carcinoma. J. Urol. 2004, 171, 1810-1813.

(27) Bensalah, K.; Leray, E.; Fergelot, P.; Rioux-Leclercq, N.; Tostain, J.; Guille, F.; Patard, J. J. Prognostic value of thrombocytosis in renal cell carcinoma. J. Urol. 2006, 175, 859-863.

(28) Flanigan, R. C.; Polcari, A. J.; Hugen, C. M. Prognostic variables and nomograms for renal cell carcinoma. Int. J. Urol. 2011, 18, 20-31.

(29) Leibovich, B. C.; Lohse, C. M.; Crispen, P. L.; Boorjian, S. A.; Thompson, R. H.; Blute, M. L.; Cheville, J. C. Histological subtype is an independent predictor of outcome for patients with renal cell carcinoma. J. Urol. 2010, 183, 1309-1315.

106

(30) Mekhail, T. M.; Abou-Jawde, R. M.; Boumerhi, G.; Malhi, S.; Wood, L.; Elson, P.; Bukowski, R. Validation and extension of the Memorial Sloan-Kettering prognostic factors model for survival in patients with previously untreated metastatic renal cell carcinoma. J. Clin. Oncol. 2005, 23, 832-841.

(31) Heng, D. Y.; Xie, W.; Regan, M. M.; Warren, M. A.; Golshayan, A. R.; Sahi, C.; Eigl, B. J.; Ruether, J. D.; Cheng, T.; North, S.; Venner, P.; Knox, J. J.; Chi, K. N.; Kollmannsberger, C.; McDermott, D. F.; Oh, W. K.; Atkins, M. B.; Bukowski, R. M.; Rini, B. I.; Choueiri, T. K. Prognostic factors for overall survival in patients with metastatic renal cell carcinoma treated with vascular endothelial growth factor-targeted agents: results from a large, multicenter study. J. Clin. Oncol. 2009, 27, 5794-5799.

(32) Heng, D. Y.; Xie, W.; Regan, M. M.; Harshman, L. C.; Bjarnason, G. A.; Vaishampayan, U. N.; Mackenzie, M.; Wood, L.; Donskov, F.; Tan, M. H.; Rha, S. Y.; Agarwal, N.; Kollmannsberger, C.; Rini, B. I.; Choueiri, T. K. External validation and comparison with other models of the International Metastatic Renal-Cell Carcinoma Database Consortium prognostic model: a population-based study. Lancet Oncol. 2013, 14, 141-148.

(33) Galsky, M. D. A prognostic model for metastatic renal-cell carcinoma. Lancet Oncol. 2013, 14, 102-103.

(34) Linehan, W. M.; Ricketts, C. J. The metabolic basis of kidney cancer. Semin. Cancer Biol. 2013, 23, 46-55.

(35) Li, B.; Qiu, B.; Lee, D. S.; Walton, Z. E.; Ochocki, J. D.; Mathew, L. K.; Mancuso, A.; Gade, T. P.; Keith, B.; Nissim, I.; Simon, M. C. Fructose-1,6-bisphosphatase opposes renal carcinoma progression. Nature 2014, 513, 251-255.

(36) Kelly, P. M.; Davison, R. S.; Bliss, E.; McGee, J. O. Macrophages in human breast disease: a quantitative immunohistochemical study. Br. J. Cancer 1988, 57, 174-177.

(37) Seliger, B.; Schlaf, G. Structure, expression and function of HLA-G in renal cell carcinoma. Semin. Cancer Biol. 2007, 17, 444-450.

(38) Dunker, K.; Schlaf, G.; Bukur, J.; Altermann, W. W.; Handke, D.; Seliger, B. Expression and regulation of non-classical HLA-G in renal cell carcinoma. Tissue Antigens 2008, 72, 137-148.

(39) Noessner, E.; Brech, D.; Mendler, A. N.; Masouris, I.; Schlenker, R.; Prinz, P. U. Intratumoral alterations of dendritic-cell differentiation and CD8(+) T-cell anergy are immune escape mechanisms of clear cell renal cell carcinoma. Oncoimmunology 2012, 1, 1451-1453.

(40) Daurkin, I.; Eruslanov, E.; Stoffs, T.; Perrin, G. Q.; Algood, C.; Gilbert, S. M.; Rosser, C. J.; Su, L. M.; Vieweg, J.; Kusmartsev, S. Tumor-associated macrophages mediate immunosuppression in the renal cancer microenvironment by activating the 15-lipoxygenase-2 pathway. Cancer Res. 2011, 71, 6400-6409.

(41) Eruslanov, E.; Stoffs, T.; Kim, W. J.; Daurkin, I.; Gilbert, S. M.; Su, L. M.; Vieweg, J.; Daaka, Y.; Kusmartsev, S. Expansion of CCR8(+) inflammatory myeloid cells in cancer patients with urothelial and renal carcinomas. Clin. Cancer Res. 2013, 19, 1670-1680.

107

(42) Xu, L.; Zhu, Y.; Chen, L.; An, H.; Zhang, W.; Wang, G.; Lin, Z.; Xu, J. Prognostic value of diametrically polarized tumor-associated macrophages in renal cell carcinoma. Ann. Surg. Oncol. 2014, 21, 3142-3150.

(43) Adeniran, A. J.; Shuch, B.; Humphrey, P. A. Hereditary Renal Cell Carcinoma Syndromes: Clinical, Pathologic, and Genetic Features. Am. J. Surg. Pathol. 2015, 39, e1-e18.

(44) Al-Bayati, O.; Hasan, A.; Pruthi, D.; Kaushik, D.; Liss, M. A. Systematic review of modifiable risk factors for kidney cancer. Urol. Oncol. 2019.

(45) Desimone, M. C.; Rathmell, W. K.; Threadgill, D. W. Pleiotropic effects of the trichloroethylene- associated P81S VHL mutation on metabolism, apoptosis, and ATM-mediated DNA damage response. J. Natl. Cancer Inst. 2013, 105, 1355-1364.

(46) Frew, I. J.; Moch, H. A clearer view of the molecular complexity of clear cell renal cell carcinoma. Annu. Rev. Pathol. 2015, 10, 263-289.

(47) Dalgliesh, G. L.; Furge, K.; Greenman, C.; Chen, L.; Bignell, G.; Butler, A.; Davies, H.; Edkins, S.; Hardy, C.; Latimer, C.; Teague, J.; Andrews, J.; Barthorpe, S.; Beare, D.; Buck, G.; Campbell, P. J.; Forbes, S.; Jia, M.; Jones, D.; Knott, H.; Kok, C. Y.; Lau, K. W.; Leroy, C.; Lin, M. L.; McBride, D. J.; Maddison, M.; Maguire, S.; McLay, K.; Menzies, A.; Mironenko, T.; Mulderrig, L.; Mudie, L.; O'Meara, S.; Pleasance, E.; Rajasingham, A.; Shepherd, R.; Smith, R.; Stebbings, L.; Stephens, P.; Tang, G.; Tarpey, P. S.; Turrell, K.; Dykema, K. J.; Khoo, S. K.; Petillo, D.; Wondergem, B.; Anema, J.; Kahnoski, R. J.; Teh, B. T.; Stratton, M. R.; Futreal, P. A. Systematic sequencing of renal carcinoma reveals inactivation of histone modifying genes. Nature 2010, 463, 360-363.

(48) Bihr, S.; Ohashi, R.; Moore, A. L.; Ruschoff, J. H.; Beisel, C.; Hermanns, T.; Mischo, A.; Corro, C.; Beyer, J.; Beerenwinkel, N.; Moch, H.; Schraml, P. Expression and Mutation Patterns of PBRM1, BAP1 and SETD2 Mirror Specific Evolutionary Subtypes in Clear Cell Renal Cell Carcinoma. Neoplasia 2019, 21, 247-256.

(49) Turajlic, S.; Xu, H.; Litchfield, K.; Rowan, A.; Horswell, S.; Chambers, T.; O'Brien, T.; Lopez, J. I.; Watkins, T. B. K.; Nicol, D.; Stares, M.; Challacombe, B.; Hazell, S.; Chandra, A.; Mitchell, T. J.; Au, L.; Eichler-Jonsson, C.; Jabbar, F.; Soultati, A.; Chowdhury, S.; Rudman, S.; Lynch, J.; Fernando, A.; Stamp, G.; Nye, E.; Stewart, A.; Xing, W.; Smith, J. C.; Escudero, M.; Huffman, A.; Matthews, N.; Elgar, G.; Phillimore, B.; Costa, M.; Begum, S.; Ward, S.; Salm, M.; Boeing, S.; Fisher, R.; Spain, L.; Navas, C.; Gronroos, E.; Hobor, S.; Sharma, S.; Aurangzeb, I.; Lall, S.; Polson, A.; Varia, M.; Horsfield, C.; Fotiadis, N.; Pickering, L.; Schwarz, R. F.; Silva, B.; Herrero, J.; Luscombe, N. M.; Jamal-Hanjani, M.; Rosenthal, R.; Birkbak, N. J.; Wilson, G. A.; Pipek, O.; Ribli, D.; Krzystanek, M.; Csabai, I.; Szallasi, Z.; Gore, M.; McGranahan, N.; Van Loo, P.; Campbell, P.; Larkin, J.; Swanton, C.; TRACERx Renal Consortium Deterministic Evolutionary Trajectories Influence Primary Tumor Growth: TRACERx Renal. Cell 2018, 173, 595-610.e11.

(50) Brugarolas, J. Molecular genetics of clear-cell renal cell carcinoma. J. Clin. Oncol. 2014, 32, 1968-1976.

(51) Cancer Genome Atlas Research Network Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 2013, 499, 43-49.

108

(52) Varela, I.; Tarpey, P.; Raine, K.; Huang, D.; Ong, C. K.; Stephens, P.; Davies, H.; Jones, D.; Lin, M. L.; Teague, J.; Bignell, G.; Butler, A.; Cho, J.; Dalgliesh, G. L.; Galappaththige, D.; Greenman, C.; Hardy, C.; Jia, M.; Latimer, C.; Lau, K. W.; Marshall, J.; McLaren, S.; Menzies, A.; Mudie, L.; Stebbings, L.; Largaespada, D. A.; Wessels, L. F.; Richard, S.; Kahnoski, R. J.; Anema, J.; Tuveson, D. A.; Perez-Mancera, P. A.; Mustonen, V.; Fischer, A.; Adams, D. J.; Rust, A.; Chan-on, W.; Subimerb, C.; Dykema, K.; Furge, K.; Campbell, P. J.; Teh, B. T.; Stratton, M. R.; Futreal, P. A. Exome sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal carcinoma. Nature 2011, 469, 539-542.

(53) Domazet-Loso, T.; Tautz, D. Phylostratigraphic tracking of cancer genes suggests a link to the emergence of multicellularity in metazoa. BMC Biol. 2010, 8:66.

(54) Gerlinger, M.; Horswell, S.; Larkin, J.; Rowan, A. J.; Salm, M. P.; Varela, I.; Fisher, R.; McGranahan, N.; Matthews, N.; Santos, C. R.; Martinez, P.; Phillimore, B.; Begum, S.; Rabinowitz, A.; Spencer-Dene, B.; Gulati, S.; Bates, P. A.; Stamp, G.; Pickering, L.; Gore, M.; Nicol, D. L.; Hazell, S.; Futreal, P. A.; Stewart, A.; Swanton, C. Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing. Nat. Genet. 2014, 46, 225-233.

(55) Huang, Y.; Wang, J.; Jia, P.; Li, X.; Pei, G.; Wang, C.; Fang, X.; Zhao, Z.; Cai, Z.; Yi, X.; Wu, S.; Zhang, B. Clonal architectures predict clinical outcome in clear cell renal cell carcinoma. Nat. Commun. 2019, 10(1):1245.

(56) Gerlinger, M.; Rowan, A. J.; Horswell, S.; Math, M.; Larkin, J.; Endesfelder, D.; Gronroos, E.; Martinez, P.; Matthews, N.; Stewart, A.; Tarpey, P.; Varela, I.; Phillimore, B.; Begum, S.; McDonald, N. Q.; Butler, A.; Jones, D.; Raine, K.; Latimer, C.; Santos, C. R.; Nohadani, M.; Eklund, A. C.; Spencer-Dene, B.; Clark, G.; Pickering, L.; Stamp, G.; Gore, M.; Szallasi, Z.; Downward, J.; Futreal, P. A.; Swanton, C. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N. Engl. J. Med. 2012, 366, 883-892.

(57) Sanfrancesco, J. M.; Cheng, L. Complexity of the genomic landscape of renal cell carcinoma: Implications for targeted therapy and precision immuno-oncology. Crit. Rev. Oncol. Hematol. 2017, 119, 23-28.

(58) Sato, Y.; Yoshizato, T.; Shiraishi, Y.; Maekawa, S.; Okuno, Y.; Kamura, T.; Shimamura, T.; Sato-Otsubo, A.; Nagae, G.; Suzuki, H.; Nagata, Y.; Yoshida, K.; Kon, A.; Suzuki, Y.; Chiba, K.; Tanaka, H.; Niida, A.; Fujimoto, A.; Tsunoda, T.; Morikawa, T.; Maeda, D.; Kume, H.; Sugano, S.; Fukayama, M.; Aburatani, H.; Sanada, M.; Miyano, S.; Homma, Y.; Ogawa, S. Integrated molecular analysis of clear-cell renal cell carcinoma. Nat. Genet. 2013, 45, 860-867.

(59) Cancer Genome Atlas Research Network; Weinstein, J. N.; Collisson, E. A.; Mills, G. B.; Shaw, K. R.; Ozenberger, B. A.; Ellrott, K.; Shmulevich, I.; Sander, C.; Stuart, J. M. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 2013, 45, 1113-1120.

(60) Joosten, S. C.; Smits, K. M.; Aarts, M. J.; Melotte, V.; Koch, A.; Tjan-Heijnen, V. C.; van Engeland, M. Epigenetics in renal cell cancer: mechanisms and clinical applications. Nat. Rev. Urol. 2018, 15, 430-451.

109

(61) Schimansky, I. Subtypspezifische DNA-Methylierung von CDH3 (P-Cadeherin) in humanen Nierenzellkarzinomen[dissertation]. Hannover: Medizinische Hochshule Hannover; 2011.

(62) Ding, L.; Ellis, M. J.; Li, S.; Larson, D. E.; Chen, K.; Wallis, J. W.; Harris, C. C.; McLellan, M. D.; Fulton, R. S.; Fulton, L. L.; Abbott, R. M.; Hoog, J.; Dooling, D. J.; Koboldt, D. C.; Schmidt, H.; Kalicki, J.; Zhang, Q.; Chen, L.; Lin, L.; Wendl, M. C.; McMichael, J. F.; Magrini, V. J.; Cook, L.; McGrath, S. D.; Vickery, T. L.; Appelbaum, E.; Deschryver, K.; Davies, S.; Guintoli, T.; Lin, L.; Crowder, R.; Tao, Y.; Snider, J. E.; Smith, S. M.; Dukes, A. F.; Sanderson, G. E.; Pohl, C. S.; Delehaunty, K. D.; Fronick, C. C.; Pape, K. A.; Reed, J. S.; Robinson, J. S.; Hodges, J. S.; Schierding, W.; Dees, N. D.; Shen, D.; Locke, D. P.; Wiechert, M. E.; Eldred, J. M.; Peck, J. B.; Oberkfell, B. J.; Lolofie, J. T.; Du, F.; Hawkins, A. E.; O'Laughlin, M. D.; Bernard, K. E.; Cunningham, M.; Elliott, G.; Mason, M. D.; Thompson, D. M.,Jr; Ivanovich, J. L.; Goodfellow, P. J.; Perou, C. M.; Weinstock, G. M.; Aft, R.; Watson, M.; Ley, T. J.; Wilson, R. K.; Mardis, E. R. Genome remodelling in a basal- like breast cancer metastasis and xenograft. Nature 2010, 464, 999-1005.

(63) Yachida, S.; Jones, S.; Bozic, I.; Antal, T.; Leary, R.; Fu, B.; Kamiyama, M.; Hruban, R. H.; Eshleman, J. R.; Nowak, M. A.; Velculescu, V. E.; Kinzler, K. W.; Vogelstein, B.; Iacobuzio- Donahue, C. A. Distant metastasis occurs late during the genetic evolution of pancreatic cancer. Nature 2010, 467, 1114-1117.

(64) Tao, Y.; Ruan, J.; Yeh, S. H.; Lu, X.; Wang, Y.; Zhai, W.; Cai, J.; Ling, S.; Gong, Q.; Chong, Z.; Qu, Z.; Li, Q.; Liu, J.; Yang, J.; Zheng, C.; Zeng, C.; Wang, H. Y.; Zhang, J.; Wang, S. H.; Hao, L.; Dong, L.; Li, W.; Sun, M.; Zou, W.; Yu, C.; Li, C.; Liu, G.; Jiang, L.; Xu, J.; Huang, H.; Li, C.; Mi, S.; Zhang, B.; Chen, B.; Zhao, W.; Hu, S.; Zhuang, S. M.; Shen, Y.; Shi, S.; Brown, C.; White, K. P.; Chen, D. S.; Chen, P. J.; Wu, C. I. Rapid growth of a hepatocellular carcinoma and the driving mutations revealed by cell-population genetic analysis of whole-genome data. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 12042-12047.

(65) Hofflin, R.; Roth, W.; Sultmann, H.; Grullich, C.; Hatiboglu, G.; Nyarangi-Dix, J.; Schonberg, G.; Teber, D.; Hadaschik, B.; Pahernik, S.; Hohenfellner, M.; Duensing, S. Intratumoral heterogeneity in renal cell carcinoma. Molecular basis and translational implications. Urologe A. 2015, 54, 800-803.

(66) Klatte, T.; Rao, P. N.; de Martino, M.; LaRochelle, J.; Shuch, B.; Zomorodian, N.; Said, J.; Kabbinavar, F. F.; Belldegrun, A. S.; Pantuck, A. J. Cytogenetic profile predicts prognosis of patients with clear cell renal cell carcinoma. J. Clin. Oncol. 2009, 27, 746-753.

(67) Parker, A. S.; Leibovich, B. C.; Lohse, C. M.; Sheinin, Y.; Kuntz, S. M.; Eckel-Passow, J. E.; Blute, M. L.; Kwon, E. D. Development and evaluation of BioScore: a biomarker panel to enhance prognostic algorithms for clear cell renal cell carcinoma. Cancer 2009, 115, 2092-2103.

(68) Monzon, F. A.; Alvarez, K.; Peterson, L.; Truong, L.; Amato, R. J.; Hernandez-McClain, J.; Tannir, N.; Parwani, A. V.; Jonasch, E. Chromosome 14q loss defines a molecular subtype of clear- cell renal cell carcinoma associated with poor prognosis. Mod. Pathol. 2011, 24, 1470-1479.

(69) Sanjmyatav, J.; Junker, K.; Matthes, S.; Muehr, M.; Sava, D.; Sternal, M.; Wessendorf, S.; Kreuz, M.; Gajda, M.; Wunderlich, H.; Schwaenen, C. Identification of genomic alterations associated with

110 metastasis and cancer specific survival in clear cell renal cell carcinoma. J. Urol. 2011, 186, 2078- 2083.

(70) Eichelberg, C.; Chun, F. K.; Bedke, J.; Heuer, R.; Adam, M.; Moch, H.; Terracciano, L.; Hinrichs, K.; Dahlem, R.; Fisch, M.; Schlomm, T.; Sauter, G.; Minner, S. Epithelial cell adhesion molecule is an independent prognostic marker in clear cell renal carcinoma. Int. J. Cancer 2013, 132, 2948-2955.

(71) Divgi, C. R.; Uzzo, R. G.; Gatsonis, C.; Bartz, R.; Treutner, S.; Yu, J. Q.; Chen, D.; Carrasquillo, J. A.; Larson, S.; Bevan, P.; Russo, P. Positron emission tomography/computed tomography identification of clear cell renal cell carcinoma: results from the REDECT trial. J. Clin. Oncol. 2013, 31, 187-194.

(72) Farber, N. J.; Wu, Y.; Zou, L.; Belani, P.; Singer, E. A. Challenges in RCC Imaging: Renal Insufficiency, Post-Operative Surveillance, and the Role of Radiomics. Kidney Cancer. J. 2015, 13, 84-90.

(73) Rowe, S. P.; Gorin, M. A.; Hammers, H. J.; Som Javadi, M.; Hawasli, H.; Szabo, Z.; Cho, S. Y.; Pomper, M. G.; Allaf, M. E. Imaging of metastatic clear cell renal cell carcinoma with PSMA-targeted (1)(8)F-DCFPyL PET/CT. Ann. Nucl. Med. 2015, 29, 877-882.

(74) Morrissey, J. J.; Mellnick, V. M.; Luo, J.; Siegel, M. J.; Figenshau, R. S.; Bhayani, S.; Kharasch, E. D. Evaluation of Urine Aquaporin-1 and Perilipin-2 Concentrations as Biomarkers to Screen for Renal Cell Carcinoma: A Prospective Cohort Study. JAMA Oncol. 2015, 1, 204-212.

(75) Ball, M. W.; Gorin, M. A.; Guner, G.; Pierorazio, P. M.; Netto, G.; Paller, C. J.; Hammers, H. J.; Diaz, L. A.; Allaf, M. E. Circulating Tumor DNA as a Marker of Therapeutic Response in Patients With Renal Cell Carcinoma: A Pilot Study. Clin. Genitourin. Cancer. 2016, 14, e515-e520.

(76) Gatto, F.; Maruzzo, M.; Magro, C.; Basso, U.; Nielsen, J. Prognostic Value of Plasma and Urine Glycosaminoglycan Scores in Clear Cell Renal Cell Carcinoma. Front. Oncol. 2016, 6, 253.

(77) Pena, C.; Lathia, C.; Shan, M.; Escudier, B.; Bukowski, R. M. Biomarkers predicting outcome in patients with advanced renal cell carcinoma: Results from sorafenib phase III Treatment Approaches in Renal Cancer Global Evaluation Trial. Clin. Cancer Res. 2010, 16, 4853-4863.

(78) Armstrong, A. J.; George, D. J.; Halabi, S. Serum lactate dehydrogenase predicts for overall survival benefit in patients with metastatic renal cell carcinoma treated with inhibition of mammalian target of rapamycin. J. Clin. Oncol. 2012, 30, 3402-3407.

(79) Tran, H. T.; Liu, Y.; Zurita, A. J.; Lin, Y.; Baker-Neblett, K. L.; Martin, A. M.; Figlin, R. A.; Hutson, T. E.; Sternberg, C. N.; Amado, R. G.; Pandite, L. N.; Heymach, J. V. Prognostic or predictive plasma cytokines and angiogenic factors for patients treated with pazopanib for metastatic renal-cell cancer: a retrospective analysis of phase 2 and phase 3 trials. Lancet Oncol. 2012, 13, 827-837.

(80) Zurita, A. J.; Jonasch, E.; Wang, X.; Khajavi, M.; Yan, S.; Du, D. Z.; Xu, L.; Herynk, M. H.; McKee, K. S.; Tran, H. T.; Logothetis, C. J.; Tannir, N. M.; Heymach, J. V. A cytokine and

111 angiogenic factor (CAF) analysis in plasma for selection of sorafenib therapy in patients with metastatic renal cell carcinoma. Ann. Oncol. 2012, 23, 46-52.

(81) Hegde, P. S.; Jubb, A. M.; Chen, D.; Li, N. F.; Meng, Y. G.; Bernaards, C.; Elliott, R.; Scherer, S. J.; Chen, D. S. Predictive impact of circulating vascular endothelial growth factor in four phase III trials evaluating bevacizumab. Clin. Cancer Res. 2013, 19, 929-937.

(82) Funakoshi, T.; Lee, C. H.; Hsieh, J. J. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat. Rev. 2014, 40, 533-547.

(83) Pawlowski, R.; Muhl, S. M.; Sulser, T.; Krek, W.; Moch, H.; Schraml, P. Loss of PBRM1 expression is associated with renal cell carcinoma progression. Int. J. Cancer 2013, 132, E11-17.

(84) Heinzelmann, J.; Henning, B.; Sanjmyatav, J.; Posorski, N.; Steiner, T.; Wunderlich, H.; Gajda, M. R.; Junker, K. Specific miRNA signatures are associated with metastasis and poor prognosis in clear cell renal cell carcinoma. World J. Urol. 2011, 29, 367-373.

(85) Slaby, O.; Redova, M.; Poprach, A.; Nekvindova, J.; Iliev, R.; Radova, L.; Lakomy, R.; Svoboda, M.; Vyzula, R. Identification of MicroRNAs associated with early relapse after nephrectomy in renal cell carcinoma patients. Genes Chromosomes Cancer 2012, 51, 707-716.

(86) Fritz, H. K.; Lindgren, D.; Ljungberg, B.; Axelson, H.; Dahlback, B. The miR(21/10b) ratio as a prognostic marker in clear cell renal cell carcinoma. Eur. J. Cancer 2014, 50, 1758-1765.

(87) Nofech-Mozes, R.; Khella, H. W.; Scorilas, A.; Youssef, L.; Krylov, S. N.; Lianidou, E.; Sidiropoulos, K. G.; Gabril, M.; Evans, A.; Yousef, G. M. MicroRNA-194 is a Marker for Good Prognosis in Clear Cell Renal Cell Carcinoma. Cancer. Med. 2016, 5, 656-664.

(88) Hildebrandt, M. A.; Gu, J.; Lin, J.; Ye, Y.; Tan, W.; Tamboli, P.; Wood, C. G.; Wu, X. Hsa-miR- 9 methylation status is associated with cancer development and metastatic recurrence in patients with clear cell renal cell carcinoma. Oncogene 2010, 29, 5724-5728.

(89) Fabrizio, F. P.; Costantini, M.; Copetti, M.; la Torre, A.; Sparaneo, A.; Fontana, A.; Poeta, L.; Gallucci, M.; Sentinelli, S.; Graziano, P.; Parente, P.; Pompeo, V.; De Salvo, L.; Simone, G.; Papalia, R.; Picardo, F.; Balsamo, T.; Flammia, G. P.; Trombetta, D.; Pantalone, A.; Kok, K.; Paranita, F.; Muscarella, L. A.; Fazio, V. M. Keap1/Nrf2 pathway in kidney cancer: frequent methylation of KEAP1 gene promoter in clear renal cell carcinoma. Oncotarget 2017, 8, 11187-11198.

(90) Kapur, P.; Pena-Llopis, S.; Christie, A.; Zhrebker, L.; Pavia-Jimenez, A.; Rathmell, W. K.; Xie, X. J.; Brugarolas, J. Effects on survival of BAP1 and PBRM1 mutations in sporadic clear-cell renal- cell carcinoma: a retrospective analysis with independent validation. Lancet Oncol. 2013, 14, 159-167.

(91) Tian, Y.; Arai, E.; Gotoh, M.; Komiyama, M.; Fujimoto, H.; Kanai, Y. Prognostication of patients with clear cell renal cell carcinomas based on quantification of DNA methylation levels of CpG island methylator phenotype marker genes. BMC Cancer 2014, 14, 772.

(92) Lin, Y. L.; Wang, Y. P.; Li, H. Z.; Zhang, X. Aberrant Promoter Methylation of PCDH17 (Protocadherin 17) in Serum and its Clinical Significance in Renal Cell Carcinoma. Med. Sci. Monit. 2017, 23, 3318-3323.

112

(93) Ito, N.; Eto, M.; Nakamura, E.; Takahashi, A.; Tsukamoto, T.; Toma, H.; Nakazawa, H.; Hirao, Y.; Uemura, H.; Kagawa, S.; Kanayama, H.; Nose, Y.; Kinukawa, N.; Nakamura, T.; Jinnai, N.; Seki, T.; Takamatsu, M.; Masui, Y.; Naito, S.; Ogawa, O. STAT3 polymorphism predicts interferon-alfa response in patients with metastatic renal cell carcinoma. J. Clin. Oncol. 2007, 25, 2785-2791.

(94) Garcia-Donas, J.; Esteban, E.; Leandro-Garcia, L. J.; Castellano, D. E.; del Alba, A. G.; Climent, M. A.; Arranz, J. A.; Gallardo, E.; Puente, J.; Bellmunt, J.; Mellado, B.; Martinez, E.; Moreno, F.; Font, A.; Robledo, M.; Rodriguez-Antona, C. Single nucleotide polymorphism associations with response and toxic effects in patients with advanced renal-cell carcinoma treated with first-line sunitinib: a multicentre, observational, prospective study. Lancet Oncol. 2011, 12, 1143-1150.

(95) Jonasch, E.; McCutcheon, I. E.; Waguespack, S. G.; Wen, S.; Davis, D. W.; Smith, L. A.; Tannir, N. M.; Gombos, D. S.; Fuller, G. N.; Matin, S. F. Pilot trial of sunitinib therapy in patients with von Hippel-Lindau disease. Ann. Oncol. 2011, 22, 2661-2666.

(96) van der Veldt, A. A.; Eechoute, K.; Gelderblom, H.; Gietema, J.; Guchelaar, H. J.; van Erp, N. P.; van den Eertwegh, A. J.; Haanen, J. B.; Mathijssen, R. H.; Wessels, J. A. Genetic polymorphisms associated with a prolonged progression-free survival in patients with metastatic renal cell cancer treated with sunitinib. Clin. Cancer Res. 2011, 17, 620-629.

(97) Xu, C. F.; Bing, N. X.; Ball, H. A.; Rajagopalan, D.; Sternberg, C. N.; Hutson, T. E.; de Souza, P.; Xue, Z. G.; McCann, L.; King, K. S.; Ragone, L. J.; Whittaker, J. C.; Spraggs, C. F.; Cardon, L. R.; Mooser, V. E.; Pandite, L. N. Pazopanib efficacy in renal cell carcinoma: evidence for predictive genetic markers in angiogenesis-related and exposure-related genes. J. Clin. Oncol. 2011, 29, 2557- 2564.

(98) Kim, H. C.; Lee, J. S.; Kim, S. H.; So, H. S.; Woo, C. Y.; Lee, J. L. Sunitinib treatment for metastatic renal cell carcinoma in patients with von hippel-lindau disease. Cancer. Res. Treat. 2013, 45, 349-353.

(99) Cho, D.; Signoretti, S.; Dabora, S.; Regan, M.; Seeley, A.; Mariotti, M.; Youmans, A.; Polivy, A.; Mandato, L.; McDermott, D.; Stanbridge, E.; Atkins, M. Potential histologic and molecular predictors of response to temsirolimus in patients with advanced renal cell carcinoma. Clin. Genitourin. Cancer. 2007, 5, 379-385.

(100) Nishikawa, M.; Miyake, H.; Harada, K.; Fujisawa, M. Expression level of phosphorylated-4E- binding protein 1 in radical nephrectomy specimens as a prognostic predictor in patients with metastatic renal cell carcinoma treated with mammalian target of rapamycin inhibitors. Med. Oncol. 2014, 31, 792. Epub 2013 Dec 4.

(101) Morris, M. R.; Ricketts, C.; Gentle, D.; Abdulrahman, M.; Clarke, N.; Brown, M.; Kishida, T.; Yao, M.; Latif, F.; Maher, E. R. Identification of candidate tumour suppressor genes frequently methylated in renal cell carcinoma. Oncogene 2010, 29, 2104-2117.

(102) Deckers, I. A.; Schouten, L. J.; Van Neste, L.; van Vlodrop, I. J.; Soetekouw, P. M.; Baldewijns, M. M.; Jeschke, J.; Ahuja, N.; Herman, J. G.; van den Brandt, P. A.; van Engeland, M. Promoter Methylation of CDO1 Identifies Clear-Cell Renal Cell Cancer Patients with Poor Survival Outcome. Clin. Cancer Res. 2015, 21, 3492-3500.

113

(103) van Vlodrop, I. J.; Baldewijns, M. M.; Smits, K. M.; Schouten, L. J.; van Neste, L.; van Criekinge, W.; van Poppel, H.; Lerut, E.; Schuebel, K. E.; Ahuja, N.; Herman, J. G.; de Bruine, A. P.; van Engeland, M. Prognostic significance of Gremlin1 (GREM1) promoter CpG island hypermethylation in clear cell renal cell carcinoma. Am. J. Pathol. 2010, 176, 575-584.

(104) Peters, I.; Eggers, H.; Atschekzei, F.; Hennenlotter, J.; Waalkes, S.; Trankenschuh, W.; Grosshennig, A.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. GATA5 CpG island methylation in renal cell cancer: a potential biomarker for metastasis and disease progression. BJU Int. 2012, 110, E144-52.

(105) Agathanggelou, A.; Cooper, W. N.; Latif, F. Role of the Ras-association domain family 1 tumor suppressor gene in human cancers. Cancer Res. 2005, 65, 3497-3508.

(106) Serth, J.; Tezval, H.; Peters, I.; Atschekzei, F.; Rehmet, K.; Jurk, S.; Albrecht, K.; Kuczyk, M. A.; Merseburger, A. S. Methylation of the RASSF1A tumor suppressor gene promoter. Risk factor for carcinogenesis of urological tumors. Urologe A. 2008, 47, 1117-1120.

(107) Tezval, H.; Merseburger, A. S.; Matuschek, I.; Machtens, S.; Kuczyk, M. A.; Serth, J. RASSF1A protein expression and correlation with clinicopathological parameters in renal cell carcinoma. BMC Urol. 2008, 8:12.

(108) Atschekzei, F.; Hennenlotter, J.; Janisch, S.; Grosshennig, A.; Trankenschuh, W.; Waalkes, S.; Peters, I.; Dork, T.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. SFRP1 CpG island methylation locus is associated with renal cell cancer susceptibility and disease recurrence. Epigenetics 2012, 7, 447-457.

(109) Evelonn, E. A.; Landfors, M.; Haider, Z.; Kohn, L.; Ljungberg, B.; Roos, G.; Degerman, S. DNA methylation associates with survival in non-metastatic clear cell renal cell carcinoma. BMC Cancer 2019, 19(1):65.

(110) Bird, A. P. DNA methylation and the frequency of CpG in animal DNA. Nucleic Acids Res. 1980, 8, 1499-1504.

(111) Fazzari, M. J.; Greally, J. M. Epigenomics: beyond CpG islands. Nat. Rev. Genet. 2004, 5, 446- 455.

(112) Wood, L. D.; Parsons, D. W.; Jones, S.; Lin, J.; Sjoblom, T.; Leary, R. J.; Shen, D.; Boca, S. M.; Barber, T.; Ptak, J.; Silliman, N.; Szabo, S.; Dezso, Z.; Ustyanksky, V.; Nikolskaya, T.; Nikolsky, Y.; Karchin, R.; Wilson, P. A.; Kaminker, J. S.; Zhang, Z.; Croshaw, R.; Willis, J.; Dawson, D.; Shipitsin, M.; Willson, J. K.; Sukumar, S.; Polyak, K.; Park, B. H.; Pethiyagoda, C. L.; Pant, P. V.; Ballinger, D. G.; Sparks, A. B.; Hartigan, J.; Smith, D. R.; Suh, E.; Papadopoulos, N.; Buckhaults, P.; Markowitz, S. D.; Parmigiani, G.; Kinzler, K. W.; Velculescu, V. E.; Vogelstein, B. The genomic landscapes of human breast and colorectal cancers. Science 2007, 318, 1108-1113.

(113) Chen, R. Z.; Pettersson, U.; Beard, C.; Jackson-Grusby, L.; Jaenisch, R. DNA hypomethylation leads to elevated mutation rates. Nature 1998, 395, 89-93.

114

(114) Schuebel, K. E.; Chen, W.; Cope, L.; Glockner, S. C.; Suzuki, H.; Yi, J. M.; Chan, T. A.; Van Neste, L.; Van Criekinge, W.; van den Bosch, S.; van Engeland, M.; Ting, A. H.; Jair, K.; Yu, W.; Toyota, M.; Imai, K.; Ahuja, N.; Herman, J. G.; Baylin, S. B. Comparing the DNA hypermethylome with gene mutations in human colorectal cancer. PLoS Genet. 2007, 3, 1709-1723.

(115) Chan, T. A.; Glockner, S.; Yi, J. M.; Chen, W.; Van Neste, L.; Cope, L.; Herman, J. G.; Velculescu, V.; Schuebel, K. E.; Ahuja, N.; Baylin, S. B. Convergence of mutation and epigenetic alterations identifies common genes in cancer that predict for poor prognosis. PLoS Med. 2008,5(5): e114.

(116) Yi, J. M.; Dhir, M.; Van Neste, L.; Downing, S. R.; Jeschke, J.; Glockner, S. C.; de Freitas Calmon, M.; Hooker, C. M.; Funes, J. M.; Boshoff, C.; Smits, K. M.; van Engeland, M.; Weijenberg, M. P.; Iacobuzio-Donahue, C. A.; Herman, J. G.; Schuebel, K. E.; Baylin, S. B.; Ahuja, N. Genomic and epigenomic integration identifies a prognostic signature in colon cancer. Clin. Cancer Res. 2011, 17, 1535-1545.

(117) Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev. 2002, 16, 6-21.

(118) Frigola, J.; Song, J.; Stirzaker, C.; Hinshelwood, R. A.; Peinado, M. A.; Clark, S. J. Epigenetic remodeling in colorectal cancer results in coordinate gene suppression across an entire chromosome band. Nat. Genet. 2006, 38, 540-549.

(119) Coolen, M. W.; Stirzaker, C.; Song, J. Z.; Statham, A. L.; Kassir, Z.; Moreno, C. S.; Young, A. N.; Varma, V.; Speed, T. P.; Cowley, M.; Lacaze, P.; Kaplan, W.; Robinson, M. D.; Clark, S. J. Consolidation of the cancer genome into domains of repressive chromatin by long-range epigenetic silencing (LRES) reduces transcriptional plasticity. Nat. Cell Biol. 2010, 12, 235-246.

(120) Easwaran, H. P.; Van Neste, L.; Cope, L.; Sen, S.; Mohammad, H. P.; Pageau, G. J.; Lawrence, J. B.; Herman, J. G.; Schuebel, K. E.; Baylin, S. B. Aberrant silencing of cancer-related genes by CpG hypermethylation occurs independently of their spatial organization in the nucleus. Cancer Res. 2010, 70, 8015-8024.

(121) Baylin, S.; Bestor, T. H. Altered methylation patterns in cancer cell genomes: cause or consequence? Cancer. Cell. 2002, 1, 299-305.

(122) Coulondre, C.; Miller, J. H. Analysis of base substitutions induced by ultraviolet radiation in Escherichia coli. Natl. Cancer Inst. Monogr. 1978, (50), 115-119.

(123) Ohm, J. E.; Baylin, S. B. Stem cell chromatin patterns: an instructive mechanism for DNA hypermethylation? Cell. Cycle 2007, 6, 1040-1043.

(124) Tsai, H. C.; Baylin, S. B. Cancer epigenetics: linking basic biology to clinical medicine. Cell Res. 2011, 21, 502-517.

(125) Bedford, M. T.; van Helden, P. D. Hypomethylation of DNA in pathological conditions of the human prostate. Cancer Res. 1987, 47, 5274-5276.

115

(126) Kim, Y. I.; Giuliano, A.; Hatch, K. D.; Schneider, A.; Nour, M. A.; Dallal, G. E.; Selhub, J.; Mason, J. B. Global DNA hypomethylation increases progressively in cervical dysplasia and carcinoma. Cancer 1994, 74, 893-899.

(127) Lin, C. H.; Hsieh, S. Y.; Sheen, I. S.; Lee, W. C.; Chen, T. C.; Shyu, W. C.; Liaw, Y. F. Genome-wide hypomethylation in hepatocellular carcinogenesis. Cancer Res. 2001, 61, 4238-4243.

(128) Cadieux, B.; Ching, T. T.; VandenBerg, S. R.; Costello, J. F. Genome-wide hypomethylation in human glioblastomas associated with specific copy number alteration, methylenetetrahydrofolate reductase allele status, and increased proliferation. Cancer Res. 2006, 66, 8469-8476.

(129) Gebauer, K.; Peters, I.; Dubrowinskaja, N.; Hennenlotter, J.; Abbas, M.; Scherer, R.; Tezval, H.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. Hsa-mir-124-3 CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell carcinoma. Br. J. Cancer 2013, 108, 131-138.

(130) Peters, I.; Dubrowinskaja, N.; Kogosov, M.; Abbas, M.; Hennenlotter, J.; von Klot, C.; Merseburger, A. S.; Stenzl, A.; Scherer, R.; Kuczyk, M. A.; Serth, J. Decreased GATA5 mRNA expression associates with CpG island methylation and shortened recurrence-free survival in clear cell renal cell carcinoma. BMC Cancer 2014, 14:101.

(131) Peters, I.; Gebauer, K.; Dubrowinskaja, N.; Atschekzei, F.; Kramer, M. W.; Hennenlotter, J.; Tezval, H.; Abbas, M.; Scherer, R.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. GATA5 CpG island hypermethylation is an independent predictor for poor clinical outcome in renal cell carcinoma. Oncol. Rep. 2014, 31, 1523-1530.

(132) Peters, I.; Dubrowinskaja, N.; Abbas, M.; Seidel, C.; Kogosov, M.; Scherer, R.; Gebauer, K.; Merseburger, A. S.; Kuczyk, M. A.; Grunwald, V.; Serth, J. DNA methylation biomarkers predict progression-free and overall survival of metastatic renal cell cancer (mRCC) treated with antiangiogenic therapies. PLoS One 2014, 9, e91440.

(133) Dubrowinskaja, N.; Gebauer, K.; Peters, I.; Hennenlotter, J.; Abbas, M.; Scherer, R.; Tezval, H.; Merseburger, A. S.; Stenzl, A.; Grunwald, V.; Kuczyk, M. A.; Serth, J. Neurofilament Heavy polypeptide CpG island methylation associates with prognosis of renal cell carcinoma and prediction of antivascular endothelial growth factor therapy response. Cancer. Med. 2014, 3, 300-309.

(134) Tezval, H.; Dubrowinskaja, N.; Peters, I.; Reese, C.; Serth, K.; Atschekzei, F.; Hennenlotter, J.; Stenzl, A.; Kuczyk, M. A.; Serth, J. Tumor Specific Epigenetic Silencing of Corticotropin Releasing Hormone -Binding Protein in Renal Cell Carcinoma: Association of Hypermethylation and Metastasis. PLoS One 2016, 11, e0163873.

(135) Green, S.; at al. Clinical trials in oncology. In Chapman&Hall, Ed.; Chapman&Hall: 2002 in Boca Raton (Florida): CRC Press .

(136) Iorio, M. V.; Croce, C. M. MicroRNA dysregulation in cancer: diagnostics, monitoring and therapeutics. A comprehensive review. EMBO Mol. Med. 2012, 4, 143-159.

116

(137) Zhang, H.; Li, H. miR-137 inhibits renal cell carcinoma growth in vitro and in vivo. Oncol. Lett. 2016, 12, 715-720.

(138) Chanudet, E.; Wozniak, M. B.; Bouaoun, L.; Byrnes, G.; Mukeriya, A.; Zaridze, D.; Brennan, P.; Muller, D. C.; Scelo, G. Large-scale genome-wide screening of circulating microRNAs in clear cell renal cell carcinoma reveals specific signatures in late-stage disease. Int. J. Cancer 2017, 141, 1730- 1740.

(139) He, L.; Hannon, G. J. MicroRNAs: small RNAs with a big role in gene regulation. Nat. Rev. Genet. 2004, 5, 522-531.

(140) Li, Y.; Zhang, D.; Wang, J. MicroRNA373 promotes tumorigenesis of renal cell carcinoma in vitro and in vivo. Mol. Med. Rep. 2017, 16, 7048-7055.

(141) Croce, C. M. Causes and consequences of microRNA dysregulation in cancer. Nat. Rev. Genet. 2009, 10, 704-714.

(142) Melo, S. A.; Esteller, M. A precursor microRNA in a cancer cell nucleus: get me out of here! Cell. Cycle 2011, 10, 922-925.

(143) Fan, Y.; Ma, X.; Li, H.; Gao, Y.; Huang, Q.; Zhang, Y.; Bao, X.; Du, Q.; Luo, G.; Liu, K.; Meng, Q.; Zhao, C.; Zhang, X. miR-122 promotes metastasis of clear-cell renal cell carcinoma by downregulating Dicer. Int. J. Cancer 2018, 142, 547-560.

(144) Ge, Y. Z.; Xu, L. W.; Zhou, C. C.; Lu, T. Z.; Yao, W. T.; Wu, R.; Zhao, Y. C.; Xu, X.; Hu, Z. K.; Wang, M.; Yang, X. B.; Zhou, L. H.; Zhong, B.; Xu, Z.; Li, W. C.; Zhu, J. G.; Jia, R. P. A BAP1 Mutation-specific MicroRNA Signature Predicts Clinical Outcomes in Clear Cell Renal Cell Carcinoma Patients with Wild-type BAP1. J. Cancer. 2017, 8, 2643-2652.

(145) Dickstein, J.; Senyuk, V.; Premanand, K.; Laricchia-Robbio, L.; Xu, P.; Cattaneo, F.; Fazzina, R.; Nucifora, G. Methylation and silencing of miRNA-124 by EVI1 and self-renewal exhaustion of hematopoietic stem cells in murine myelodysplastic syndrome. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 9783-9788.

(146) Krichevsky, A. M.; Sonntag, K. C.; Isacson, O.; Kosik, K. S. Specific microRNAs modulate embryonic stem cell-derived neurogenesis. Stem Cells 2006, 24, 857-864.

(147) Cao, X.; Pfaff, S. L.; Gage, F. H. A functional study of miR-124 in the developing neural tube. Genes Dev. 2007, 21, 531-536.

(148) Makeyev, E. V.; Zhang, J.; Carrasco, M. A.; Maniatis, T. The MicroRNA miR-124 promotes neuronal differentiation by triggering brain-specific alternative pre-mRNA splicing. Mol. Cell 2007, 27, 435-448.

(149) Visvanathan, J.; Lee, S.; Lee, B.; Lee, J. W.; Lee, S. K. The microRNA miR-124 antagonizes the anti-neural REST/SCP1 pathway during embryonic CNS development. Genes Dev. 2007, 21, 744- 749.

117

(150) Conaco, C.; Otto, S.; Han, J. J.; Mandel, G. Reciprocal actions of REST and a microRNA promote neuronal identity. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 2422-2427.

(151) Zhang, Z.; Gong, Q.; Li, M.; Xu, J.; Zheng, Y.; Ge, P.; Chi, G. MicroRNA-124 inhibits the proliferation of C6 glioma cells by targeting Smad4. Int. J. Mol. Med. 2017, 40, 1226-1234.

(152) Kim, B. C.; Jeong, H. O.; Park, D.; Kim, C. H.; Lee, E. K.; Kim, D. H.; Im, E.; Kim, N. D.; Lee, S.; Yu, B. P.; Bhak, J.; Chung, H. Y. Profiling age-related epigenetic markers of stomach adenocarcinoma in young and old subjects. Cancer. Inform. 2015, 14, 47-54.

(153) Gong, M.; Liang, T.; Jin, S.; Dai, X.; Zhou, Z.; Gao, M.; Huang, S.; Luo, J.; Zou, L.; Zou, X. Methylation-mediated silencing of miR-124 facilitates chondrogenesis by targeting NFATc1 under hypoxic conditions. Am. J. Transl. Res. 2017, 9, 4111-4124.

(154) Dallaire, A.; Garand, C.; Paquel, E. R.; Mitchell, S. J.; de Cabo, R.; Simard, M. J.; Lebel, M. Down regulation of miR-124 in both Werner syndrome DNA helicase mutant mice and mutant Caenorhabditis elegans wrn-1 reveals the importance of this microRNA in accelerated aging. Aging (Albany NY) 2012, 4, 636-647.

(155) Yuan, Q.; Sun, T.; Ye, F.; Kong, W.; Jin, H. MicroRNA-124-3p affects proliferation, migration and apoptosis of bladder cancer cells through targeting AURKA. Cancer. Biomark 2017, 19, 93-101.

(156) Zhang, J.; Lu, Y.; Yue, X.; Li, H.; Luo, X.; Wang, Y.; Wang, K.; Wan, J. MiR-124 suppresses growth of human colorectal cancer by inhibiting STAT3. PLoS One 2013, 8, e70300.

(157) Wang, H.; Zhang, T. T.; Jin, S.; Liu, H.; Zhang, X.; Ruan, C. G.; Wu, D. P.; Han, Y.; Wang, X. Q. Pyrosequencing quantified methylation level of miR-124 predicts shorter survival for patients with myelodysplastic syndrome. Clin. Epigenetics 2017, 9:91.

(158) Laricchia-Robbio, L.; Fazzina, R.; Li, D.; Rinaldi, C. R.; Sinha, K. K.; Chakraborty, S.; Nucifora, G. Point mutations in two EVI1 Zn fingers abolish EVI1-GATA1 interaction and allow erythroid differentiation of murine bone marrow cells. Mol. Cell. Biol. 2006, 26, 7658-7666.

(159) Senyuk, V.; Premanand, K.; Xu, P.; Qian, Z.; Nucifora, G. The oncoprotein EVI1 and the DNA methyltransferase Dnmt3 co-operate in binding and de novo methylation of target DNA. PLoS One 2011, 6, e20793.

(160) Grandjean, V.; Gounon, P.; Wagner, N.; Martin, L.; Wagner, K. D.; Bernex, F.; Cuzin, F.; Rassoulzadegan, M. The miR-124-Sox9 paramutation: RNA-mediated epigenetic control of embryonic and adult growth. Development 2009, 136, 3647-3655.

(161) Sekido, R.; Lovell-Badge, R. Sex determination and SRY: down to a wink and a nudge? Trends Genet. 2009, 25, 19-29.

(162) Matheu, A.; Collado, M.; Wise, C.; Manterola, L.; Cekaite, L.; Tye, A. J.; Canamero, M.; Bujanda, L.; Schedl, A.; Cheah, K. S.; Skotheim, R. I.; Lothe, R. A.; Lopez de Munain, A.; Briscoe, J.; Serrano, M.; Lovell-Badge, R. Oncogenicity of the developmental transcription factor Sox9. Cancer Res. 2012, 72, 1301-1315.

118

(163) Jo, A.; Denduluri, S.; Zhang, B.; Wang, Z.; Yin, L.; Yan, Z.; Kang, R.; Shi, L. L.; Mok, J.; Lee, M. J.; Haydon, R. C. The versatile functions of Sox9 in development, stem cells, and human diseases. Genes & Diseases 2014, 1, 149-161.

(164) Larsimont, J. C.; Youssef, K. K.; Sanchez-Danes, A.; Sukumaran, V.; Defrance, M.; Delatte, B.; Liagre, M.; Baatsen, P.; Marine, J. C.; Lippens, S.; Guerin, C.; Del Marmol, V.; Vanderwinden, J. M.; Fuks, F.; Blanpain, C. Sox9 Controls Self-Renewal of Oncogene Targeted Cells and Links Tumor Initiation and Invasion. Cell. Stem Cell. 2015, 17, 60-73.

(165) Wang, X.; Liu, Y.; Liu, X.; Yang, J.; Teng, G.; Zhang, L.; Zhou, C. MiR-124 inhibits cell proliferation, migration and invasion by directly targeting SOX9 in lung adenocarcinoma. Oncol. Rep. 2016, 35, 3115-3121.

(166) Zhou, F.; Zhang, C.; Guan, Y.; Chen, Y.; Lu, Q.; Jie, L.; Gao, H.; Du, H.; Zhang, H.; Liu, Y.; Wang, X. Screening the expression characteristics of several miRNAs in G93A-SOD1 transgenic mouse: altered expression of miRNA-124 is associated with astrocyte differentiation by targeting Sox2 and Sox9. J. Neurochem. 2018, 145, 51-67.

(167) Bernstein, C.; Bernstein, H. Epigenetic reduction of DNA repair in progression to gastrointestinal cancer. World J. Gastrointest. Oncol. 2015, 7, 30-46.

(168) Chu, M.; Chang, Y.; Guo, Y.; Wang, N.; Cui, J.; Gao, W. Q. Regulation and methylation of tumor suppressor miR-124 by androgen receptor in prostate cancer cells. PLoS One 2015, 10, e0116197.

(169) Ando, T.; Yoshida, T.; Enomoto, S.; Asada, K.; Tatematsu, M.; Ichinose, M.; Sugiyama, T.; Ushijima, T. DNA methylation of microRNA genes in gastric mucosae of gastric cancer patients: its possible involvement in the formation of epigenetic field defect. Int. J. Cancer 2009, 124, 2367-2374.

(170) Wilting, S. M.; van Boerdonk, R. A.; Henken, F. E.; Meijer, C. J.; Diosdado, B.; Meijer, G. A.; le Sage, C.; Agami, R.; Snijders, P. J.; Steenbergen, R. D. Methylation-mediated silencing and tumour suppressive function of hsa-miR-124 in cervical cancer. Mol. Cancer. 2010, 9:167.

(171) Harada, T.; Yamamoto, E.; Yamano, H. O.; Nojima, M.; Maruyama, R.; Kumegawa, K.; Ashida, M.; Yoshikawa, K.; Kimura, T.; Harada, E.; Takagi, R.; Tanaka, Y.; Aoki, H.; Nishizono, M.; Nakaoka, M.; Tsuyada, A.; Niinuma, T.; Kai, M.; Shimoda, K.; Shinomura, Y.; Sugai, T.; Imai, K.; Suzuki, H. Analysis of DNA methylation in bowel lavage fluid for detection of colorectal cancer. Cancer. Prev. Res. (Phila) 2014, 7, 1002-1010.

(172) Shindo, T.; Shimizu, T.; Nojima, M.; Niinuma, T.; Maruyama, R.; Kitajima, H.; Kai, M.; Itoh, N.; Suzuki, H.; Masumori, N. Evaluation of Urinary DNA Methylation as a Marker for Recurrent Bladder Cancer: A 2-Center Prospective Study. Urology 2018, 113, 71-78.

(173) Kim, Y. H.; Lee, W. K.; Lee, E. B.; Son, J. W.; Kim, D. S.; Park, J. Y. Combined Effect of Metastasis-Related MicroRNA, miR-34 and miR-124 Family, Methylation on Prognosis of Non- Small-Cell Lung Cancer. Clin. Lung Cancer. 2017, 18, e13-e20.

119

(174) Braga, E. A.; Loginov, V. I.; Filippova, E. A.; Burdennyi, A. M.; Pronina, I. V.; Kazubskaya, T. P.; Khodyrev, D. S.; Utkin, D. O.; Kushlinskii, D. N.; Adamyan, L. V.; Kuslinskii, N. E. Diagnostic Value of a Group of MicroRNA Genes Hypermethylated in Ovarian Carcinoma. Bull. Exp. Biol. Med. 2018, 166, 253-256.

(175) Lang, F. M.; Hossain, A.; Gumin, J.; Momin, E. N.; Shimizu, Y.; Ledbetter, D.; Shahar, T.; Yamashita, S.; Parker Kerrigan, B.; Fueyo, J.; Sawaya, R.; Lang, F. F. Mesenchymal stem cells as natural biofactories for exosomes carrying miR-124a in the treatment of gliomas. Neuro Oncol. 2018, 20, 380-390.

(176) Delaloy, C.; Liu, L.; Lee, J. A.; Su, H.; Shen, F.; Yang, G. Y.; Young, W. L.; Ivey, K. N.; Gao, F. B. MicroRNA-9 coordinates proliferation and migration of human embryonic stem cell-derived neural progenitors. Cell. Stem Cell. 2010, 6, 323-335.

(177) Fu, L.; Shi, J.; Liu, A.; Zhou, L.; Jiang, M.; Fu, H.; Xu, K.; Li, D.; Deng, A.; Zhang, Q.; Pang, Y.; Guo, Y.; Hu, K.; Zhou, J.; Wang, Y.; Huang, W.; Jing, Y.; Dou, L.; Wang, L.; Xu, K.; Ke, X.; Nervi, C.; Li, Y.; Yu, L. A minicircuitry of microRNA-9-1 and RUNX1-RUNX1T1 contributes to leukemogenesis in t(8;21) acute myeloid leukemia. Int. J. Cancer 2017, 140, 653-661.

(178) Chen, P.; Price, C.; Li, Z.; Li, Y.; Cao, D.; Wiley, A.; He, C.; Gurbuxani, S.; Kunjamma, R. B.; Huang, H.; Jiang, X.; Arnovitz, S.; Xu, M.; Hong, G. M.; Elkahloun, A. G.; Neilly, M. B.; Wunderlich, M.; Larson, R. A.; Le Beau, M. M.; Mulloy, J. C.; Liu, P. P.; Rowley, J. D.; Chen, J. miR-9 is an essential oncogenic microRNA specifically overexpressed in mixed lineage leukemia- rearranged leukemia. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 11511-11516.

(179) Wang, Y.; Zhou, Y.; Graves, D. T. FOXO transcription factors: their clinical significance and regulation. Biomed. Res. Int. 2014, 2014, 925350.

(180) Zhou, B.; Xu, H.; Xia, M.; Sun, C.; Li, N.; Guo, E.; Guo, L.; Shan, W.; Lu, H.; Wu, Y.; Li, Y.; Yang, D.; Weng, D.; Meng, L.; Hu, J.; Ma, D.; Chen, G.; Li, K. Overexpressed miR-9 promotes tumor metastasis via targeting E-cadherin in serous ovarian cancer. Front. Med. 2017, 11, 214-222.

(181) Dong, Z. C.; Zhang, D.; Wang, S. B.; Lin, Z. Q. Target inhibition on GSK-3beta by miR-9 to modulate proliferation and apoptosis of bladder cancer cells. Eur. Rev. Med. Pharmacol. Sci. 2018, 22, 3018-3026.

(182) Lehmann, U.; Hasemeier, B.; Christgen, M.; Muller, M.; Romermann, D.; Langer, F.; Kreipe, H. Epigenetic inactivation of microRNA gene hsa-mir-9-1 in human breast cancer. J. Pathol. 2008, 214, 17-24.

(183) Yu, Y.; Wan, Y.; Huang, C. The biological functions of NF-kappaB1 (p50) and its potential as an anti-cancer target. Curr. Cancer. Drug Targets 2009, 9, 566-571.

(184) Rodriguez-Otero, P.; Roman-Gomez, J.; Vilas-Zornoza, A.; Jose-Eneriz, E. S.; Martin-Palanco, V.; Rifon, J.; Torres, A.; Calasanz, M. J.; Agirre, X.; Prosper, F. Deregulation of FGFR1 and CDK6 oncogenic pathways in acute lymphoblastic leukaemia harbouring epigenetic modifications of the MIR9 family. Br. J. Haematol. 2011, 155, 73-83.

120

(185) Tsai, K. W.; Liao, Y. L.; Wu, C. W.; Hu, L. Y.; Li, S. C.; Chan, W. C.; Ho, M. R.; Lai, C. H.; Kao, H. W.; Fang, W. L.; Huang, K. H.; Lin, W. C. Aberrant hypermethylation of miR-9 genes in gastric cancer. Epigenetics 2011, 6, 1189-1197.

(186) Wang, L. Q.; Kwong, Y. L.; Kho, C. S.; Wong, K. F.; Wong, K. Y.; Ferracin, M.; Calin, G. A.; Chim, C. S. Epigenetic inactivation of miR-9 family microRNAs in chronic lymphocytic leukemia-- implications on constitutive activation of NFkappaB pathway. Mol. Cancer. 2013;12:173.

(187) Pronina, I. V.; Klimov, E. A.; Burdennyy, A. M.; Beresneva, E. V.; Fridman, M. V.; Ermilova, V. D.; Kazubskaya, T. P.; Karpukhin, A. V.; Braga, E. A.; Loginov, V. I. Methylation of the genes for the microRNAs miR-129-2 and miR-9-1, changes in their expression, and activation of their potential target genes in clear cell renal cell carcinoma. Mol. Biol. (Mosk) 2017, 51, 73-84.

(188) Chen, X.; Zhu, L.; Ma, Z.; Sun, G.; Luo, X.; Li, M.; Zhai, S.; Li, P.; Wang, X. Oncogenic miR-9 is a target of erlotinib in NSCLCs. Sci. Rep. 2015, 5, 17031.

(189) Toiyama, Y.; Okugawa, Y.; Tanaka, K.; Araki, T.; Uchida, K.; Hishida, A.; Uchino, M.; Ikeuchi, H.; Hirota, S.; Kusunoki, M.; Boland, C. R.; Goel, A. A Panel of Methylated MicroRNA Biomarkers for Identifying High-Risk Patients With Ulcerative Colitis-Associated Colorectal Cancer. Gastroenterology 2017, 153, 1634-1646.e8.

(190) Zhang, J.; Chong, C. C.; Chen, G. G.; Lai, P. B. A Seven-microRNA Expression Signature Predicts Survival in Hepatocellular Carcinoma. PLoS One 2015, 10, e0128628.

(191) Shan, F.; Li, J.; Huang, Q. Y. HIF-1 alpha-induced up-regulation of miR-9 contributes to phenotypic modulation in pulmonary artery smooth muscle cells during hypoxia. J. Cell. Physiol. 2014, 229, 1511-1520.

(192) O'Loghlen, A.; Brookes, S.; Martin, N.; Rapisarda, V.; Peters, G.; Gil, J. CBX7 and miR-9 are part of an autoregulatory loop controlling p16(INK) (4a). Aging Cell. 2015, 14, 1113-1121.

(193) Ko, L. J.; Engel, J. D. DNA-binding specificities of the GATA transcription factor family. Mol. Cell. Biol. 1993, 13, 4011-4022.

(194) Bossard, P.; Zaret, K. S. GATA transcription factors as potentiators of gut endoderm differentiation. Development 1998, 125, 4909-4917.

(195) Crispino, J. D. GATA1 in normal and malignant hematopoiesis. Semin. Cell Dev. Biol. 2005, 16, 137-147.

(196) Bresnick, E. H.; Lee, H. Y.; Fujiwara, T.; Johnson, K. D.; Keles, S. GATA switches as developmental drivers. J. Biol. Chem. 2010, 285, 31087-31093.

(197) Wan, Y. Y. GATA3: a master of many trades in immune regulation. Trends Immunol. 2014, 35, 233-242.

(198) Fujiwara, T. GATA Transcription Factors: Basic Principles and Related Human Disorders. Tohoku J. Exp. Med. 2017, 242, 83-91.

121

(199) Fang, D.; Cui, K.; Hu, G.; Gurram, R. K.; Zhong, C.; Oler, A. J.; Yagi, R.; Zhao, M.; Sharma, S.; Liu, P.; Sun, B.; Zhao, K.; Zhu, J. Bcl11b, a novel GATA3-interacting protein, suppresses Th1 while limiting Th2 cell differentiation. J. Exp. Med. 2018, 215, 1449-1462.

(200) Rudra, D.; deRoos, P.; Chaudhry, A.; Niec, R. E.; Arvey, A.; Samstein, R. M.; Leslie, C.; Shaffer, S. A.; Goodlett, D. R.; Rudensky, A. Y. Transcription factor Foxp3 and its protein partners form a complex regulatory network. Nat. Immunol. 2012, 13, 1010-1019.

(201) Tavares, T. S.; Nanus, D.; Yang, X. J.; Gudas, L. J. Gene microarray analysis of human renal cell carcinoma: the effects of HDAC inhibition and retinoid treatment. Cancer. Biol. Ther. 2008, 7, 1607-1618.

(202) Jayasinghe, R. G.; Cao, S.; Gao, Q.; Wendl, M. C.; Vo, N. S.; Reynolds, S. M.; Zhao, Y.; Climente-Gonzalez, H.; Chai, S.; Wang, F.; Varghese, R.; Huang, M.; Liang, W. W.; Wyczalkowski, M. A.; Sengupta, S.; Li, Z.; Payne, S. H.; Fenyo, D.; Miner, J. H.; Walter, M. J.; Cancer Genome Atlas Research Network; Vincent, B.; Eyras, E.; Chen, K.; Shmulevich, I.; Chen, F.; Ding, L. Systematic Analysis of Splice-Site-Creating Mutations in Cancer. Cell. Rep. 2018, 23, 270-281.e3.

(203) Wang, M. X.; Ren, J. T.; Tang, L. Y.; Ren, Z. F. Molecular features in young vs elderly breast cancer patients and the impacts on survival disparities by age at diagnosis. Cancer. Med. 2018; 7(7):3269-3277.

(204) Shield, P. W.; Crouch, S. J.; Papadimos, D. J.; Walsh, M. D. Gata3 Immunohistochemical Staining is A Useful Marker for Metastatic Breast Carcinoma in Fine Needle Aspiration Specimens. J. Cytol. 2018, 35, 90-93.

(205) Li, Y.; Ishiguro, H.; Kawahara, T.; Kashiwagi, E.; Izumi, K.; Miyamoto, H. Loss of GATA3 in bladder cancer promotes cell migration and invasion. Cancer. Biol. Ther. 2014, 15, 428-435.

(206) Si, W.; Huang, W.; Zheng, Y.; Yang, Y.; Liu, X.; Shan, L.; Zhou, X.; Wang, Y.; Su, D.; Gao, J.; Yan, R.; Han, X.; Li, W.; He, L.; Shi, L.; Xuan, C.; Liang, J.; Sun, L.; Wang, Y.; Shang, Y. Dysfunction of the Reciprocal Feedback Loop between GATA3- and ZEB2-Nucleated Repression Programs Contributes to Breast Cancer Metastasis. Cancer. Cell. 2015, 27, 822-836.

(207) Kim, K. S.; Kim, J.; Oh, N.; Kim, M. Y.; Park, K. S. ELK3-GATA3 axis modulates MDA-MB- 231 metastasis by regulating cell-cell adhesion-related genes. Biochem. Biophys. Res. Commun. 2018, 498, 509-515.

(208) Li, X.; Jin, J.; Yang, S.; Xu, W.; Meng, X.; Deng, H.; Zhan, J.; Gao, S.; Zhang, H. GATA3 acetylation at K119 by CBP inhibits cell migration and invasion in lung adenocarcinoma. Biochem. Biophys. Res. Commun. 2018, 497, 633-638.

(209) Gao, X.; Sedgwick, T.; Shi, Y. B.; Evans, T. Distinct functions are implicated for the GATA-4, - 5, and -6 transcription factors in the regulation of intestine epithelial cell differentiation. Mol. Cell. Biol. 1998, 18, 2901-2911.

122

(210) Krasinski, S. D.; Van Wering, H. M.; Tannemaat, M. R.; Grand, R. J. Differential activation of intestinal gene promoters: functional interactions between GATA-5 and HNF-1 alpha. Am. J. Physiol. Gastrointest. Liver Physiol. 2001, 281, G69-84.

(211) Bai, Y.; Akiyama, Y.; Nagasaki, H.; Yagi, O. K.; Kikuchi, Y.; Saito, N.; Takeshita, K.; Iwai, T.; Yuasa, Y. Distinct expression of CDX2 and GATA4/5, development-related genes, in human gastric cancer cell lines. Mol. Carcinog. 2000, 28, 184-188.

(212) Fujiwara, Y.; Emi, M.; Ohata, H.; Kato, Y.; Nakajima, T.; Mori, T.; Nakamura, Y. Evidence for the presence of two tumor suppressor genes on chromosome 8p for colorectal carcinoma. Cancer Res. 1993, 53, 1172-1174.

(213) Jones, J. W.; Raval, J. R.; Beals, T. F.; Worsham, M. J.; Van Dyke, D. L.; Esclamado, R. M.; Wolf, G. T.; Bradford, C. R.; Miller, T.; Carey, T. E. Frequent loss of heterozygosity on chromosome arm 18q in squamous cell carcinomas. Identification of 2 regions of loss--18q11.1-q12.3 and 18q21.1- q23. Arch. Otolaryngol. Head. Neck. Surg. 1997, 123, 610-614.

(214) Akiyama, Y.; Watkins, N.; Suzuki, H.; Jair, K. W.; van Engeland, M.; Esteller, M.; Sakai, H.; Ren, C. Y.; Yuasa, Y.; Herman, J. G.; Baylin, S. B. GATA-4 and GATA-5 transcription factor genes and potential downstream antitumor target genes are epigenetically silenced in colorectal and gastric cancer. Mol. Cell. Biol. 2003, 23, 8429-8439.

(215) Wang, D.; Yang, P. N.; Chen, J.; Zhou, X. Y.; Liu, Q. J.; Li, H. J.; Li, C. L. Promoter hypermethylation may be an important mechanism of the transcriptional inactivation of ARRDC3, GATA5, and ELP3 in invasive ductal breast carcinoma. Mol. Cell. Biochem. 2014, 396, 67-77.

(216) Mzik, M.; Chmelarova, M.; John, S.; Laco, J.; Slaby, O.; Kiss, I.; Bohovicova, L.; Palicka, V.; Nekvindova, J. Aberrant methylation of tumour suppressor genes WT1, GATA5 and PAX5 in hepatocellular carcinoma. Clin. Chem. Lab. Med. 2016, 54, 1971-1980.

(217) Feinberg, A. P.; Ohlsson, R.; Henikoff, S. The epigenetic progenitor origin of human cancer. Nat. Rev. Genet. 2006, 7, 21-33.

(218) Nishioka, Y.; Ueki, T.; Hokazono, K.; Nagayoshi, K.; Tanaka, M. Comparative detection of aberrantly methylated DNA in preoperative and postoperative stool from patients with colorectal cancers. Int. J. Biol. Markers 2015, 30, e81-7.

(219) Chmelarova, M.; Laco, J.; Kovarikova, H.; Baranova, I.; Dundr, P.; Nemejcova, K.; Michalek, J.; Vosmik, M.; Palicka, V. Elevated DNA methylation in malignant tumors of the sinonasal tract and its association with patient survival. Biomed. Pap. Med. Fac. Univ. Palacky Olomouc Czech. Repub. 2018; 162(3):232-238.

(220) Liu, P.; Zhou, T. F.; Qiu, B. A.; Yang, Y. X.; Zhu, Y. J.; An, Y.; Zhao, W. C.; Wu, Y. T.; Ma, P. F.; Li, J. B.; Xia, N. X. Methylation-Mediated Silencing of GATA5 Gene Suppresses Cholangiocarcinoma Cell Proliferation and Metastasis. Transl. Oncol. 2018, 11, 585-592.

123

(221) Zhang, X.; Song, Y. F.; Lu, H. N.; Wang, D. P.; Zhang, X. S.; Huang, S. L.; Sun, B. L.; Huang, Z. G. Combined detection of plasma GATA5 and SFRP2 methylation is a valid noninvasive biomarker for colorectal cancer and adenomas. World J. Gastroenterol. 2015, 21, 2629-2637.

(222) van Vlodrop, I. J. H.; Joosten, S. C.; De Meyer, T.; Smits, K. M.; Van Neste, L.; Melotte, V.; Baldewijns, M. M. L. L.; Schouten, L. J.; van den Brandt, P. A.; Jeschke, J.; Yi, J. M.; Schuebel, K. E.; Ahuja, N.; Herman, J. G.; Aarts, M. J.; Bosman, F. T.; Van Criekinge, W.; van Engeland, M. A Four-Gene Promoter Methylation Marker Panel Consisting of GREM1, NEURL, LAD1, and NEFH Predicts Survival of Clear Cell Renal Cell Cancer Patients. Clin. Cancer Res. 2017, 23, 2006-2018.

(223) Silva, M.; Richard, C.; Benito, A.; Sanz, C.; Olalla, I.; Fernandez-Luna, J. L. Expression of Bcl- x in erythroid precursors from patients with polycythemia vera. N. Engl. J. Med. 1998, 338, 564-571.

(224) Wu, G. J.; Wang, W.; Lin, Y. L.; Liu, S. H.; Chen, R. M. Oxidative stress-induced apoptotic insults to rat osteoblasts are attenuated by nitric oxide pretreatment via GATA-5-involved regulation of Bcl-X L gene expression and protein translocation. Arch. Toxicol. 2016, 90, 905-916.

(225) Timucin, A. C.; Basaga, H.; Kutuk, O. Selective targeting of antiapoptotic BCL-2 proteins in cancer. Med. Res. Rev. 2018;39(1):146-175.

(226) Ni, J.; Abrahamson, M.; Zhang, M.; Fernandez, M. A.; Grubb, A.; Su, J.; Yu, G. L.; Li, Y.; Parmelee, D.; Xing, L.; Coleman, T. A.; Gentz, S.; Thotakura, R.; Nguyen, N.; Hesselberg, M.; Gentz, R. Cystatin E is a novel human cysteine proteinase inhibitor with structural resemblance to family 2 cystatins. J. Biol. Chem. 1997, 272, 10853-10858.

(227) Zeeuwen, P. L.; van Vlijmen-Willems, I. M.; Hendriks, W.; Merkx, G. F.; Schalkwijk, J. A null mutation in the cystatin M/E gene of ichq mice causes juvenile lethality and defects in epidermal cornification. Hum. Mol. Genet. 2002, 11, 2867-2875.

(228) Stenman, G.; Astrom, A. K.; Roijer, E.; Sotiropoulou, G.; Zhang, M.; Sager, R. Assignment of a novel cysteine proteinase inhibitor (CST6) to 11q13 by fluorescence in situ hybridization. Cytogenet. Cell Genet. 1997, 76, 45-46.

(229) Sotiropoulou, G.; Anisowicz, A.; Sager, R. Identification, cloning, and characterization of cystatin M, a novel cysteine proteinase inhibitor, down-regulated in breast cancer. J. Biol. Chem. 1997, 272, 903-910.

(230) Murthy, R. V.; Arbman, G.; Gao, J.; Roodman, G. D.; Sun, X. F. Legumain expression in relation to clinicopathologic and biological variables in colorectal cancer. Clin. Cancer Res. 2005, 11, 2293-2299.

(231) Ai, L.; Kim, W. J.; Kim, T. Y.; Fields, C. R.; Massoll, N. A.; Robertson, K. D.; Brown, K. D. Epigenetic silencing of the tumor suppressor cystatin M occurs during breast cancer progression. Cancer Res. 2006, 66, 7899-7909.

(232) Cheng, T.; Hitomi, K.; van Vlijmen-Willems, I. M.; de Jongh, G. J.; Yamamoto, K.; Nishi, K.; Watts, C.; Reinheckel, T.; Schalkwijk, J.; Zeeuwen, P. L. Cystatin M/E is a high affinity inhibitor of

124 cathepsin V and cathepsin L by a reactive site that is distinct from the legumain-binding site. A novel clue for the role of cystatin M/E in epidermal cornification. J. Biol. Chem. 2006, 281, 15893-15899.

(233) Gawenda, J.; Traub, F.; Luck, H. J.; Kreipe, H.; von Wasielewski, R. Legumain expression as a prognostic factor in breast cancer patients. Breast Cancer Res. Treat. 2007, 102, 1-6.

(234) Zeeuwen, P. L.; Cheng, T.; Schalkwijk, J. The biology of cystatin M/E and its cognate target proteases. J. Invest. Dermatol. 2009, 129, 1327-1338.

(235) Benezeder, T.; Tiran, V.; Treitler, A. A. N.; Suppan, C.; Rossmann, C.; Stoeger, H.; Cote, R. J.; Datar, R. H.; Balic, M.; Dandachi, N. Multigene methylation analysis of enriched circulating tumor cells associates with poor progression-free survival in metastatic breast cancer patients. Oncotarget 2017, 8, 92483-92496.

(236) Johnstone, C. N.; Pattison, A. D.; Gorringe, K. L.; Harrison, P. F.; Powell, D. R.; Lock, P.; Baloyan, D.; Ernst, M.; Stewart, A. G.; Beilharz, T. H.; Anderson, R. L. Functional and genomic characterisation of a xenograft model system for the study of metastasis in triple-negative breast cancer. Dis. Model. Mech. 2018;11(5)dmm.032250.

(237) Wallin, H.; Apelqvist, J.; Andersson, F.; Ekstrom, U.; Abrahamson, M. Low-level internalization of cystatin E/M affects legumain activity and migration of melanoma cells. J. Biol. Chem. 2017, 292, 14413-14424.

(238) Chimonidou, M.; Tzitzira, A.; Strati, A.; Sotiropoulou, G.; Sfikas, C.; Malamos, N.; Georgoulias, V.; Lianidou, E. CST6 promoter methylation in circulating cell-free DNA of breast cancer patients. Clin. Biochem. 2013, 46, 235-240.

(239) Li, Z.; Guo, X.; Wu, Y.; Li, S.; Yan, J.; Peng, L.; Xiao, Z.; Wang, S.; Deng, Z.; Dai, L.; Yi, W.; Xia, K.; Tang, L.; Wang, J. Methylation profiling of 48 candidate genes in tumor and matched normal tissues from breast cancer patients. Breast Cancer Res. Treat. 2015, 149, 767-779.

(240) Knox, J. J. Progression-free survival as endpoint in metastatic RCC? Lancet 2008, 372, 427-429.

(241) Gigante, M.; Li, G.; Ferlay, C.; Perol, D.; Blanc, E.; Paul, S.; Zhao, A.; Tostain, J.; Escudier, B.; Negrier, S.; Genin, C. Prognostic value of serum CA9 in patients with metastatic clear cell renal cell carcinoma under targeted therapy. Anticancer Res. 2012, 32, 5447-5451.

(242) Tessema, M.; Yingling, C. M.; Liu, Y.; Tellez, C. S.; Van Neste, L.; Baylin, S. S.; Belinsky, S. A. Genome-wide unmasking of epigenetically silenced genes in lung adenocarcinoma from smokers and never smokers. Carcinogenesis 2014, 35, 1248-1257.

(243) Li, Q.; Zheng, Z. C.; Ni, C. J.; Jin, W. X.; Jin, Y. X.; Chen, Y.; Zhang, X. H.; Chen, E. D.; Cai, Y. F. Correlation of Cystatin E/M with Clinicopathological Features and Prognosis in Triple-Negative Breast Cancer. Ann. Clin. Lab. Sci. 2018, 48, 40-44.

(244) Uitto, J.; Pulkkinen, L. Molecular complexity of the cutaneous basement membrane zone. Mol. Biol. Rep. 1996, 23, 35-46.

125

(245) Kim, M. S.; Pinto, S. M.; Getnet, D.; Nirujogi, R. S.; Manda, S. S.; Chaerkady, R.; Madugundu, A. K.; Kelkar, D. S.; Isserlin, R.; Jain, S.; Thomas, J. K.; Muthusamy, B.; Leal-Rojas, P.; Kumar, P.; Sahasrabuddhe, N. A.; Balakrishnan, L.; Advani, J.; George, B.; Renuse, S.; Selvan, L. D.; Patil, A. H.; Nanjappa, V.; Radhakrishnan, A.; Prasad, S.; Subbannayya, T.; Raju, R.; Kumar, M.; Sreenivasamurthy, S. K.; Marimuthu, A.; Sathe, G. J.; Chavan, S.; Datta, K. K.; Subbannayya, Y.; Sahu, A.; Yelamanchi, S. D.; Jayaram, S.; Rajagopalan, P.; Sharma, J.; Murthy, K. R.; Syed, N.; Goel, R.; Khan, A. A.; Ahmad, S.; Dey, G.; Mudgal, K.; Chatterjee, A.; Huang, T. C.; Zhong, J.; Wu, X.; Shaw, P. G.; Freed, D.; Zahari, M. S.; Mukherjee, K. K.; Shankar, S.; Mahadevan, A.; Lam, H.; Mitchell, C. J.; Shankar, S. K.; Satishchandra, P.; Schroeder, J. T.; Sirdeshmukh, R.; Maitra, A.; Leach, S. D.; Drake, C. G.; Halushka, M. K.; Prasad, T. S.; Hruban, R. H.; Kerr, C. L.; Bader, G. D.; Iacobuzio-Donahue, C. A.; Gowda, H.; Pandey, A. A draft map of the human proteome. Nature 2014, 509, 575-581.

(246) Ishiko, A.; Shimizu, H.; Masunaga, T.; Hashimoto, T.; Dmochowski, M.; Wojnarowska, F.; Bhogal, B. S.; Black, M. M.; Nishikawa, T. 97-kDa linear IgA bullous dermatosis (LAD) antigen localizes to the lamina lucida of the epidermal basement membrane. J. Invest. Dermatol. 1996, 106, 739-743.

(247) Marinkovich, M. P.; Taylor, T. B.; Keene, D. R.; Burgeson, R. E.; Zone, J. J. LAD-1, the linear IgA bullous dermatosis autoantigen, is a novel 120-kDa anchoring filament protein synthesized by epidermal cells. J. Invest. Dermatol. 1996, 106, 734-738.

(248) Motoki, K.; Megahed, M.; LaForgia, S.; Uitto, J. Cloning and chromosomal mapping of mouse ladinin, a novel basement membrane zone component. Genomics 1997, 39, 323-330.

(249) Rusinek, D.; Swierniak, M.; Chmielik, E.; Kowal, M.; Kowalska, M.; Cyplinska, R.; Czarniecka, A.; Piglowski, W.; Korfanty, J.; Chekan, M.; Krajewska, J.; Szpak-Ulczok, S.; Jarzab, M.; Widlak, W.; Jarzab, B. BRAFV600E-Associated Gene Expression Profile: Early Changes in the Transcriptome, Based on a Transgenic Mouse Model of Papillary Thyroid Carcinoma. PLoS One 2015, 10(12):e0143688.

(250) Wang, X.; Guda, C. Integrative exploration of genomic profiles for triple negative breast cancer identifies potential drug targets. Medicine (Baltimore) 2016, 95, e4321.

(251) Sun, Y.; Liu, W. Z.; Liu, T.; Feng, X.; Yang, N.; Zhou, H. F. Signaling pathway of MAPK/ERK in cell proliferation, differentiation, migration, senescence and apoptosis. J. Recept. Signal Transduct. Res. 2015, 35, 600-604.

(252) Herbst, R. S. Review of epidermal growth factor receptor biology. Int. J. Radiat. Oncol. Biol. Phys. 2004, 59, 21-26.

(253) Roth, L.; Srivastava, S.; Lindzen, M.; Sas-Chen, A.; Sheffer, M.; Lauriola, M.; Enuka, Y.; Noronha, A.; Mancini, M.; Lavi, S.; Tarcic, G.; Pines, G.; Nevo, N.; Heyman, O.; Ziv, T.; Rueda, O. M.; Gnocchi, D.; Pikarsky, E.; Admon, A.; Caldas, C.; Yarden, Y. SILAC identifies LAD1 as a filamin-binding regulator of actin dynamics in response to EGF and a marker of aggressive breast tumors. Sci. Signal. 2018, 11, 10.1126.

126

(254) Klobucar, M.; Sedic, M.; Gehrig, P.; Grossmann, J.; Bilic, M.; Kovac-Bilic, L.; Pavelic, K.; Kraljevic Pavelic, S. Basement membrane protein ladinin-1 and the MIF-CD44-beta1 integrin signaling axis are implicated in laryngeal cancer metastasis. Biochim. Biophys. Acta 2016, 1862, 1938-1954.

(255) Codreanu, S. G.; Hoeksema, M. D.; Slebos, R. J. C.; Zimmerman, L. J.; Rahman, S. M. J.; Li, M.; Chen, S. C.; Chen, H.; Eisenberg, R.; Liebler, D. C.; Massion, P. P. Identification of Proteomic Features To Distinguish Benign Pulmonary Nodules from Lung Adenocarcinoma. J. Proteome Res. 2017, 16, 3266-3276.

(256) Rosengren, L. E.; Karlsson, J. E.; Karlsson, J. O.; Persson, L. I.; Wikkelso, C. Patients with amyotrophic lateral sclerosis and other neurodegenerative diseases have increased levels of neurofilament protein in CSF. J. Neurochem. 1996, 67, 2013-2018.

(257) Teunissen, C. E.; Iacobaeus, E.; Khademi, M.; Brundin, L.; Norgren, N.; Koel-Simmelink, M. J.; Schepens, M.; Bouwman, F.; Twaalfhoven, H. A.; Blom, H. J.; Jakobs, C.; Dijkstra, C. D. Combination of CSF N-acetylaspartate and neurofilaments in multiple sclerosis. Neurology 2009, 72, 1322-1329.

(258) Jonsson, M.; Zetterberg, H.; van Straaten, E.; Lind, K.; Syversen, S.; Edman, A.; Blennow, K.; Rosengren, L.; Pantoni, L.; Inzitari, D.; Wallin, A. Cerebrospinal fluid biomarkers of white matter lesions - cross-sectional results from the LADIS study. Eur. J. Neurol. 2010, 17, 377-382.

(259) Niemela, V.; Landtblom, A. M.; Blennow, K.; Sundblom, J. Tau or neurofilament light-Which is the more suitable biomarker for Huntington's disease? PLoS One 2017, 12, e0172762.

(260) Luciano, L.; Groos, S.; Reale, E. Brush cells of rodent gallbladder and stomach epithelia express neurofilaments. J. Histochem. Cytochem. 2003, 51, 187-198.

(261) O'Shaughnessy, P. J.; Morris, I. D.; Baker, P. J. Leydig cell re-generation and expression of cell signaling molecules in the germ cell-free testis. Reproduction 2008, 135, 851-858.

(262) Takahashi, N.; Ishizuka, B. The involvement of neurofilament heavy chain phosphorylation in the maturation and degeneration of rat oocytes. Endocrinology 2012, 153, 1990-1998.

(263) Oka, D.; Yamashita, S.; Tomioka, T.; Nakanishi, Y.; Kato, H.; Kaminishi, M.; Ushijima, T. The presence of aberrant DNA methylation in noncancerous esophageal mucosae in association with smoking history: a target for risk diagnosis and prevention of esophageal cancers. Cancer 2009, 115, 3412-3426.

(264) Schleicher, R. L.; Hunter, S. B.; Zhang, M.; Zheng, M.; Tan, W.; Bandea, C. I.; Fallon, M. T.; Bostwick, D. G.; Varma, V. A. Neurofilament heavy chain-like messenger RNA and protein are present in benign prostate and down-regulated in prostatic carcinoma. Cancer Res. 1997, 57, 3532- 3536.

(265) Kim, M. S.; Chang, X.; LeBron, C.; Nagpal, J. K.; Lee, J.; Huang, Y.; Yamashita, K.; Trink, B.; Ratovitski, E. A.; Sidransky, D. Neurofilament heavy polypeptide regulates the Akt-beta-catenin pathway in human esophageal squamous cell carcinoma. PLoS One 2010, 5, e9003.

127

(266) Hummon, A. B.; Pitt, J. J.; Camps, J.; Emons, G.; Skube, S. B.; Huppi, K.; Jones, T. L.; Beissbarth, T.; Kramer, F.; Grade, M.; Difilippantonio, M. J.; Ried, T.; Caplen, N. J. Systems-wide RNAi analysis of CASP8AP2/FLASH shows transcriptional deregulation of the replication-dependent histone genes and extensive effects on the transcriptome of colorectal cancer cells. Mol. Cancer. 2012; 11:1.

(267) Calmon, M. F.; Jeschke, J.; Zhang, W.; Dhir, M.; Siebenkas, C.; Herrera, A.; Tsai, H. C.; O'Hagan, H. M.; Pappou, E. P.; Hooker, C. M.; Fu, T.; Schuebel, K. E.; Gabrielson, E.; Rahal, P.; Herman, J. G.; Baylin, S. B.; Ahuja, N. Epigenetic silencing of neurofilament genes promotes an aggressive phenotype in breast cancer. Epigenetics 2015, 10, 622-632.

(268) Revill, K.; Wang, T.; Lachenmayer, A.; Kojima, K.; Harrington, A.; Li, J.; Hoshida, Y.; Llovet, J. M.; Powers, S. Genome-wide methylation analysis and epigenetic unmasking identify tumor suppressor genes in hepatocellular carcinoma. Gastroenterology 2013, 145, 1424-35.e1-25.

(269) Huang, R. L.; Chen, H. J.; Chen, L. Y.; Chao, T. K.; Lin, W. Y.; Liew, P. L.; Su, P. H.; Weng, Y. C.; Wang, Y. C.; Liao, C. C.; Hsu, Y. W.; Chen Wang, H.; Lai, H. C. Epigenetic loss of heparan sulfate 3-O-sulfation sensitizes ovarian carcinoma to oncogenic signals and predicts prognosis. Int. J. Cancer 2018;143(8):1943-1953.

(270) Kalluri, R.; Weinberg, R. A. The basics of epithelial-mesenchymal transition. J. Clin. Invest. 2009, 119, 1420-1428.

(271) Gu, X.; Jiang, R.; Fowler, M. R. Acantholytic squamous cell carcinoma in upper aerodigestive tract: histopathology, immunohistochemical profile and epithelial mesenchymal transition phenotype change. Head. Neck. Pathol. 2012, 6, 438-444.

(272) Yuan, W.; Yuan, Y.; Zhang, T.; Wu, S. Role of Bmi-1 in regulation of ionizing irradiation- induced epithelial-mesenchymal transition and migration of breast cancer cells. PLoS One 2015, 10, e0118799.

(273) Tang, Y.; He, Y.; Zhang, P.; Wang, J.; Fan, C.; Yang, L.; Xiong, F.; Zhang, S.; Gong, Z.; Nie, S.; Liao, Q.; Li, X.; Li, X.; Li, Y.; Li, G.; Zeng, Z.; Xiong, W.; Guo, C. LncRNAs regulate the cytoskeleton and related Rho/ROCK signaling in cancer metastasis. Mol. Cancer. 2018, 17, 77.

(274) Peng, G.; Yuan, X.; Yuan, J.; Liu, Q.; Dai, M.; Shen, C.; Ma, J.; Liao, Y.; Jiang, W. miR-25 promotes glioblastoma cell proliferation and invasion by directly targeting NEFL. Mol. Cell. Biochem. 2015, 409, 103-111.

(275) Haddad, L. A.; Smith, N.; Bowser, M.; Niida, Y.; Murthy, V.; Gonzalez-Agosti, C.; Ramesh, V. The TSC1 tumor suppressor hamartin interacts with neurofilament-L and possibly functions as a novel integrator of the neuronal cytoskeleton. J. Biol. Chem. 2002, 277, 44180-44186.

(276) Vamvakopoulos, N. C.; Sioutopoulou, T. O.; Durkin, S. A.; Nierman, W. C.; Wasmuth, J. J.; McPherson, J. D. Mapping the human corticotropin releasing hormone binding protein gene (CRHBP) to the long arm of chromosome 5 (5q11.2-q13.3). Genomics 1995, 25, 325-327.

128

(277) Sterrenburg, L.; Gaszner, B.; Boerrigter, J.; Santbergen, L.; Bramini, M.; Elliott, E.; Chen, A.; Peeters, B. W.; Roubos, E. W.; Kozicz, T. Chronic stress induces sex-specific alterations in methylation and expression of corticotropin-releasing factor gene in the rat. PLoS One 2011, 6, e28128.

(278) Cuellar, L. M.; Fujinaka, H.; Yamamoto, K.; Miyamoto, M.; Tasaki, M.; Zhao, L.; Tamer, I.; Yaoita, E.; Yoshida, Y.; Yamamoto, T. Identification and localization of novel genes preferentially expressed in human kidney glomerulus. Nephrology (Carlton) 2009, 14, 94-104.

(279) Bale, T. L.; Giordano, F. J.; Vale, W. W. A new role for corticotropin-releasing factor receptor- 2: suppression of vascularization. Trends Cardiovasc. Med. 2003, 13, 68-71.

(280) Tsurumi, Y.; Murohara, T.; Krasinski, K.; Chen, D.; Witzenbichler, B.; Kearney, M.; Couffinhal, T.; Isner, J. M. Reciprocal relation between VEGF and NO in the regulation of endothelial integrity. Nat. Med. 1997, 3, 879-886.

(281) Bale, T. L.; Giordano, F. J.; Hickey, R. P.; Huang, Y.; Nath, A. K.; Peterson, K. L.; Vale, W. W.; Lee, K. F. Corticotropin-releasing factor receptor 2 is a tonic suppressor of vascularization. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 7734-7739.

(282) Arbiser, J. L.; Karalis, K.; Viswanathan, A.; Koike, C.; Anand-Apte, B.; Flynn, E.; Zetter, B.; Majzoub, J. A. Corticotropin-releasing hormone stimulates angiogenesis and epithelial tumor growth in the skin. J. Invest. Dermatol. 1999, 113, 838-842.

(283) Wang, J.; Xu, Y.; Xu, Y.; Zhu, H.; Zhang, R.; Zhang, G.; Li, S. Urocortin's inhibition of tumor growth and angiogenesis in hepatocellular carcinoma via corticotrophin-releasing factor receptor 2. Cancer Invest. 2008, 26, 359-368.

(284) Hao, Z.; Huang, Y.; Cleman, J.; Jovin, I. S.; Vale, W. W.; Bale, T. L.; Giordano, F. J. Urocortin2 inhibits tumor growth via effects on vascularization and cell proliferation. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 3939-3944.

(285) Tezval, H.; Jurk, S.; Atschekzei, F.; Becker, J. U.; Jahn, O.; Serth, J.; Kuczyk, M. A. Urocortin and corticotropin-releasing factor receptor 2 in human renal cell carcinoma: disruption of an endogenous inhibitor of angiogenesis and proliferation. World J. Urol. 2009, 27, 825-830.

(286) Tezval, H.; Jurk, S.; Atschekzei, F.; Serth, J.; Kuczyk, M. A.; Merseburger, A. S. The involvement of altered corticotropin releasing factor receptor 2 expression in prostate cancer due to alteration of anti-angiogenic signaling pathways. Prostate 2009, 69, 443-448.

(287) Jin, L.; Chen, J.; Li, L.; Li, C.; Chen, C.; Li, S. CRH suppressed TGFbeta1-induced Epithelial- Mesenchymal Transition via induction of E-cadherin in breast cancer cells. Cell. Signal. 2014, 26, 757-765.

(288) Jin, L.; Li, C.; Li, R.; Sun, Z.; Fang, X.; Li, S. Corticotropin-releasing hormone receptors mediate apoptosis via cytosolic calcium-dependent phospholipase A(2) and migration in prostate cancer cell RM-1. J. Mol. Endocrinol. 2014, 52, 255-267.

129

(289) Giannogonas, P.; Apostolou, A.; Manousopoulou, A.; Theocharis, S.; Macari, S. A.; Psarras, S.; Garbis, S. D.; Pothoulakis, C.; Karalis, K. P. Identification of a novel interaction between corticotropin releasing hormone (Crh) and macroautophagy. Sci. Rep. 2016, 6, 23342.

(290) Fang, X.; Hong, Y.; Dai, L.; Qian, Y.; Zhu, C.; Wu, B.; Li, S. CRH promotes human colon cancer cell proliferation via IL-6/JAK2/STAT3 signaling pathway and VEGF-induced tumor angiogenesis. Mol. Carcinog. 2017, 56, 2434-2445.

(291) Sato, N.; Takagi, K.; Suzuki, T.; Miki, Y.; Tanaka, S.; Nagase, S.; Warita, H.; Fukudo, S.; Sato, F.; Sasano, H.; Ito, K. Immunolocalization of corticotropin-releasing hormone (CRH) and its receptors (CRHR1 and CRHR2) in human endometrial carcinoma: CRHR1 as a potent prognostic factor. Int. J. Gynecol. Cancer 2014, 24, 1549-1557.

(292) Pichler, R.; Fritz, J.; Tulchiner, G.; Klinglmair, G.; Soleiman, A.; Horninger, W.; Klocker, H.; Heidegger, I. Increased Accuracy of a novel mRNA-based Urine Test for Bladder Cancer Surveillance. BJU Int. 2017;121(1):29-37.

(293) Ho, D. W.; Kai, A. K.; Ng, I. O. TCGA whole-transcriptome sequencing data reveals significantly dysregulated genes and signaling pathways in hepatocellular carcinoma. Front. Med. 2015, 9, 322-330.

(294) Xia, H. B.; Wang, H. J.; Fu, L. Q.; Wang, S. B.; Li, L.; Ru, G. Q.; He, X. L.; Tong, X. M.; Mou, X. Z.; Huang, D. S. Decreased CRHBP expression is predictive of poor prognosis in patients with hepatocellular carcinoma. Oncol. Lett. 2018, 16, 3681-3689.

(295) McGarvey, K. M.; Fahrner, J. A.; Greene, E.; Martens, J.; Jenuwein, T.; Baylin, S. B. Silenced tumor suppressor genes reactivated by DNA demethylation do not return to a fully euchromatic chromatin state. Cancer Res. 2006, 66, 3541-3549.

(296) Peters, I.; Dubrowinskaja, N.; Hennenlotter, J.; Antonopoulos, W. I.; Von Klot, C. A. J.; Tezval, H.; Stenzl, A.; Kuczyk, M. A.; Serth, J. DNA methylation of neural EGFL like 1 (NELL1) is associated with advanced disease and the metastatic state of renal cell cancer patients. Oncol. Rep. 2018, 40, 3861-3868.

(297) Joosten, S. C.; Deckers, I. A.; Aarts, M. J.; Hoeben, A.; van Roermund, J. G.; Smits, K. M.; Melotte, V.; van Engeland, M.; Tjan-Heijnen, V. C. Prognostic DNA methylation markers for renal cell carcinoma: a systematic review. Epigenomics 2017, 9, 1243-1257.

(298) Irizarry, R. A.; Ladd-Acosta, C.; Wen, B.; Wu, Z.; Montano, C.; Onyango, P.; Cui, H.; Gabo, K.; Rongione, M.; Webster, M.; Ji, H.; Potash, J.; Sabunciyan, S.; Feinberg, A. P. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat. Genet. 2009, 41, 178-186.

(299) Asmar, F.; Søgaard, A.; Grønbæk, K. Chapter 2 - DNA Methylation and Hydroxymethylation in Cancer. 2015, 9-30.

(300) Flanagan, J. M.; Wild, L. An epigenetic role for noncoding RNAs and intragenic DNA methylation. Genome Biol. 2007, 8(6):307.

130

(301) Kitazawa, S.; Haraguchi, R.; Kitazawa, R. Morphology-oriented epigenetic research. Histochem. Cell Biol. 2018, 150, 3-12.

(302) Meier, K.; Recillas-Targa, F. New insights on the role of DNA methylation from a global view. Front. Biosci. (Landmark Ed) 2017, 22, 644-668.

(303) Nile, C. J.; Read, R. C.; Akil, M.; Duff, G. W.; Wilson, A. G. Methylation status of a single CpG site in the IL6 promoter is related to IL6 messenger RNA levels and rheumatoid arthritis. Arthritis Rheum. 2008, 58, 2686-2693.

(304) Gerlinger, M.; Swanton, C. How Darwinian models inform therapeutic failure initiated by clonal heterogeneity in cancer medicine. Br. J. Cancer 2010, 103, 1139-1143.

(305) Sankin, A.; Hakimi, A. A.; Mikkilineni, N.; Ostrovnaya, I.; Silk, M. T.; Liang, Y.; Mano, R.; Chevinsky, M.; Motzer, R. J.; Solomon, S. B.; Cheng, E. H.; Durack, J. C.; Coleman, J. A.; Russo, P.; Hsieh, J. J. The impact of genetic heterogeneity on biomarker development in kidney cancer assessed by multiregional sampling. Cancer. Med. 2014, 3, 1485-1492.

(306) Flavahan, W. A.; Gaskell, E.; Bernstein, B. E. Epigenetic plasticity and the hallmarks of cancer. Science 2017;357(6348).

(307) Yap, T. A.; Gerlinger, M.; Futreal, P. A.; Pusztai, L.; Swanton, C. Intratumor heterogeneity: seeing the wood for the trees. Sci. Transl. Med. 2012, 4, 127ps10.

(308) Smiraglia, D. J.; Rush, L. J.; Fruhwald, M. C.; Dai, Z.; Held, W. A.; Costello, J. F.; Lang, J. C.; Eng, C.; Li, B.; Wright, F. A.; Caligiuri, M. A.; Plass, C. Excessive CpG island hypermethylation in cancer cell lines versus primary human malignancies. Hum. Mol. Genet. 2001, 10, 1413-1419.

(309) Winter, S.; Fisel, P.; Buttner, F.; Rausch, S.; D'Amico, D.; Hennenlotter, J.; Kruck, S.; Nies, A. T.; Stenzl, A.; Junker, K.; Scharpf, M.; Hofmann, U.; van der Kuip, H.; Fend, F.; Ott, G.; Agaimy, A.; Hartmann, A.; Bedke, J.; Schwab, M.; Schaeffeler, E. Methylomes of renal cell lines and tumors or metastases differ significantly with impact on pharmacogenes. Sci. Rep. 2016, 6, 29930.

(310) Lasseigne, B. N.; Burwell, T. C.; Patil, M. A.; Absher, D. M.; Brooks, J. D.; Myers, R. M. DNA methylation profiling reveals novel diagnostic biomarkers in renal cell carcinoma. BMC Med. 2014;12:235.

(311) Chen, G.; Wang, Y.; Wang, L.; Xu, W. Identifying prognostic biomarkers based on aberrant DNA methylation in kidney renal clear cell carcinoma. Oncotarget 2017, 8, 5268-5280.

(312) Dimitrieva, S.; Schlapbach, R.; Rehrauer, H. Prognostic value of cross-omics screening for kidney clear cell renal cancer survival. Biol. Direct 2016, 11, 68.

(313) Moon, M.; Nakai, K. Integrative analysis of gene expression and DNA methylation using unsupervised feature extraction for detecting candidate cancer biomarkers. J. Bioinform Comput. Biol. 2018, 16, 1850006.

(314) Beuselinck, B.; Job, S.; Becht, E.; Karadimou, A.; Verkarre, V.; Couchy, G.; Giraldo, N.; Rioux-Leclercq, N.; Molinie, V.; Sibony, M.; Elaidi, R.; Teghom, C.; Patard, J. J.; Mejean, A.;

131

Fridman, W. H.; Sautes-Fridman, C.; de Reynies, A.; Oudard, S.; Zucman-Rossi, J. Molecular subtypes of clear cell renal cell carcinoma are associated with sunitinib response in the metastatic setting. Clin. Cancer Res. 2015, 21, 1329-1339.

(315) Titze-de-Almeida, R.; David, C.; Titze-de-Almeida, S. S. The Race of 10 Synthetic RNAi-Based Drugs to the Pharmaceutical Market. Pharm. Res. 2017, 34, 1339-1363.

(316) Perri, F.; Longo, F.; Giuliano, M.; Sabbatino, F.; Favia, G.; Ionna, F.; Addeo, R.; Della Vittoria Scarpati, G.; Di Lorenzo, G.; Pisconti, S. Epigenetic control of gene expression: Potential implications for cancer treatment. Crit. Rev. Oncol. Hematol. 2017, 111, 166-172.

(317) Gatenby, R. A.; Silva, A. S.; Gillies, R. J.; Frieden, B. R. Adaptive therapy. Cancer Res. 2009, 69, 4894-4903.

132

Danksagung.

Ich möchte mich bei allen Personen bedanken, die mir diese Dissertation ermöglicht haben.

Herrn Prof. Dr. Markus A. Kuczyk, Leiter der der Klinik für Urologie und Urologische Onkologie der Medizinischen Hochschule Hannover, danke ich für die freundliche Überlassung des Dissertationsthemas und Ermöglichung der Durchführung dieser Arbeit in seiner Klinik.

Herrn Dr. rer. nat. Jürgen Serth danke ich herzlich für die Themastellung, die hervorragende Betreuung und seine ständige Diskussions- und Hilfsbereitschaft bei allen meinen Anliegen, unter anderem auch für die Mühen bei dem Korrekturlesen.

Herrn Prof. Dr. Jürgen Alves möchte ich ebenfalls einen großen herzlichen Dank für seine professionelle Begleitung meiner Arbeit mit vielen wertvollen Anregungen und steten Hilfsbereitschaft aussprechen.

Des Weiteren gilt mein ganz besonderer herzlicher Dank Herrn Dr. rer. nat. Thilo Dörk- Bousset auf dessen vollste freundschaftliche Unterstützung und Anregung bei allen meinen Anliegen ich glücklicherweise immer zählen konnte.

Bei allen meinen Kollegen aus der Abteilungen der Urologischen Forschung, der Molekularen Gynäkologie und der Molekularen Immunologie möchte ich mich ebenfalls für die freundschaftliche Arbeitsatmosphäre und stete Hilfsbereitschaft herzlich bedanken. Namentlich (alphabetisch geordnet) möchte ich vor allem folgende Personen nennen – Ute Adenaw, Dr. Faranaz Atschekzei, Dr. Natalia Bogdanova, Dr. Kristine Bousset, Margrit Hepke, Christel Reese, Peter Schürmann, Britta Wieland.

Zum Schluss bedanke ich mich bei meiner Familie für die bedingungslose und uneingeschränkte Unterstützung.

133

Eidesstattliche Erklärung.

Hiermit erkläre ich, dass ich die Dissertation „Identifizierung epigenetischer Alterationen als Biomarker für Genese, Progression und Therapieansprechen des Nierenzellkarzinoms.“ selbstständig verfasst habe. Bei der Anfertigung wurden folgende Hilfen Dritter in Anspruch genommen: Prof. Dr. M. A. Kuczyk - Betreuer intern, Dr. J. Serth - Betreuer intern, Prof. Dr. Alves - Zweitbetreuer intern, Dr. Dörk-Bousset - stellvertretender Zweitbetreuer intern.

Ich habe keine entgeltliche Hilfe von Vermittlungs- bzw. Beratungsdiensten (Promotionsberater oder anderer Personen) in Anspruch genommen. Niemand hat von mir unmittelbar oder mittelbar entgeltliche Leistungen für Arbeiten erhalten, die im Zusammenhang mit dem Inhalt der vorgelegten Dissertation stehen. Ich habe die Dissertation an folgenden Institutionen angefertigt: Medizinische Hochschule Hannover.

Die Dissertation wurde bisher nicht für eine Prüfung oder Promotion oder für einen ähnlichen Zweck zur Beurteilung eingereicht. Ich versichere, dass ich die vorstehenden Angaben nach bestem Wissen vollständig und der Wahrheit entsprechend gemacht habe.

Ort, Datum______Unterschrift:______

134

Liste etwaiger wissenschaftlicher Veröffentlichungen

(1) Delahanty, R. J.; Xiang, Y. B.; Spurdle, A.; Beeghly-Fadiel, A.; Long, J.; Thompson, D.; Tomlinson, I.; Yu, H.; Lambrechts, D.; Dork, T.; Goodman, M. T.; Zheng, Y.; Salvesen, H. B.; Bao, P. P.; Amant, F.; Beckmann, M. W.; Coenegrachts, L.; Coosemans, A.; Dubrowinskaja, N.; Dunning, A.; Runnebaum, I. B.; Easton, D.; Ekici, A. B.; Fasching, P. A.; Halle, M. K.; Hein, A.; Howarth, K.; Gorman, M.; Kaydarova, D.; Krakstad, C.; Lose, F.; Lu, L.; Lurie, G.; O'Mara, T.; Matsuno, R. K.; Pharoah, P.; Risch, H.; Corssen, M.; Trovik, J.; Turmanov, N.; Wen, W.; Lu, W.; Cai, Q.; Zheng, W.; Shu, X. O. Polymorphisms in inflammation pathway genes and endometrial cancer risk. Cancer Epidemiol. Biomarkers Prev. 2013, 22, 216-223.

(2) Dubrowinskaja, N.; Gebauer, K.; Peters, I.; Hennenlotter, J.; Abbas, M.; Scherer, R.; Tezval, H.; Merseburger, A. S.; Stenzl, A.; Grunwald, V.; Kuczyk, M. A.; Serth, J. Neurofilament Heavy polypeptide CpG island methylation associates with prognosis of renal cell carcinoma and prediction of antivascular endothelial growth factor therapy response. Cancer. Med. 2014, 3, 300-309.

(3) Eeles, R. A.; Kote-Jarai, Z.; Al Olama, A. A.; Giles, G. G.; Guy, M.; Severi, G.; Muir, K.; Hopper, J. L.; Henderson, B. E.; Haiman, C. A.; Schleutker, J.; Hamdy, F. C.; Neal, D. E.; Donovan, J. L.; Stanford, J. L.; Ostrander, E. A.; Ingles, S. A.; John, E. M.; Thibodeau, S. N.; Schaid, D.; Park, J. Y.; Spurdle, A.; Clements, J.; Dickinson, J. L.; Maier, C.; Vogel, W.; Dork, T.; Rebbeck, T. R.; Cooney, K. A.; Cannon-Albright, L.; Chappuis, P. O.; Hutter, P.; Zeegers, M.; Kaneva, R.; Zhang, H. W.; Lu, Y. J.; Foulkes, W. D.; English, D. R.; Leongamornlert, D. A.; Tymrakiewicz, M.; Morrison, J.; Ardern-Jones, A. T.; Hall, A. L.; O'Brien, L. T.; Wilkinson, R. A.; Saunders, E. J.; Page, E. C.; Sawyer, E. J.; Edwards, S. M.; Dearnaley, D. P.; Horwich, A.; Huddart, R. A.; Khoo, V. S.; Parker, C. C.; Van As, N.; Woodhouse, C. J.; Thompson, A.; Christmas, T.; Ogden, C.; Cooper, C. S.; Southey, M. C.; Lophatananon, A.; Liu, J. F.; Kolonel, L. N.; Le Marchand, L.; Wahlfors, T.; Tammela, T. L.; Auvinen, A.; Lewis, S. J.; Cox, A.; FitzGerald, L. M.; Koopmeiners, J. S.; Karyadi, D. M.; Kwon, E. M.; Stern, M. C.; Corral, R.; Joshi, A. D.; Shahabi, A.; McDonnell, S. K.; Sellers, T. A.; Pow-Sang, J.; Chambers, S.; Aitken, J.; Gardiner, R. A.; Batra, J.; Kedda, M. A.; Lose, F.; Polanowski, A.; Patterson, B.; Serth, J.; Meyer, A.; Luedeke, M.; Stefflova, K.; Ray, A. M.; Lange, E. M.; Farnham, J.; Khan, H.; Slavov, C.; Mitkova, A.; Cao, G.; … Dubrowinskaja, N.; … ; UK Genetic Prostate Cancer Study Collaborators/British Association of Urological Surgeons' Section of Oncology; UK ProtecT Study Collaborators; PRACTICAL Consortium; Easton, D. F. Identification of seven new prostate cancer susceptibility loci through a genome-wide association study. Nat. Genet. 2009, 41, 1116-1121.

(4) Gebauer, K.; Peters, I.; Dubrowinskaja, N.; Hennenlotter, J.; Abbas, M.; Scherer, R.; Tezval, H.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. Hsa-mir-124-3 CpG island methylation is associated with advanced tumours and disease recurrence of patients with clear cell renal cell carcinoma. Br. J. Cancer 2013, 108, 131-138.

(5) Lin, W. Y.; Camp, N. J.; Cannon-Albright, L. A.; Allen-Brady, K.; Balasubramanian, S.; Reed, M. W.; Hopper, J. L.; Apicella, C.; Giles, G. G.; Southey, M. C.; Milne, R. L.; Arias-Perez, J. I.; Menendez-Rodriguez, P.; Benitez, J.; Grundmann, M.; Dubrowinskaja, N.; Park-Simon, T. W.; Dork, T.; Garcia-Closas, M.; Figueroa, J.; Sherman, M.; Lissowska, J.; Easton, D. F.; Dunning, A. M.; Rajaraman, P.; Sigurdson, A. J.; Doody, M. M.; Linet, M. S.; Pharoah, P. D.; Schmidt, M. K.; Cox, A. A role for XRCC2 gene polymorphisms in breast cancer risk and survival. J. Med. Genet. 2011, 48, 477-484.

(6) Long, J.; Zheng, W.; Xiang, Y. B.; Lose, F.; Thompson, D.; Tomlinson, I.; Yu, H.; Wentzensen, N.; Lambrechts, D.; Dork, T.; Dubrowinskaja, N.; Goodman, M. T.; Salvesen, H. B.; Fasching, P. A.; Scott, R. J.; Delahanty, R.; Zheng, Y.; O'Mara, T.; Healey, C. S.; Hodgson, S.; Risch, H.; Yang, H. P.; Amant, F.; Turmanov, N.; Schwake, A.; Lurie, G.; Trovik, J.; Beckmann, M. W.; Ashton, K.; Ji, B. T.; Bao, P. P.; Howarth, K.; Lu, L.; Lissowska, J.; Coenegrachts, L.; Kaidarova, D.; Durst, M.; Thompson, P. J.; Krakstad, C.; Ekici, A. B.; Otton, G.; Shi, J.; Zhang, B.; Gorman, M.; Brinton, L.; Coosemans, A.; Matsuno, R. K.; Halle, M. K.; Hein, A.; Proietto, A.; Cai, H.; Lu, W.; Dunning, A.; Easton, D.; Gao, Y. T.; Cai, Q.; Spurdle, A. B.; Shu, X. O. Genome-wide association study identifies a possible susceptibility locus for endometrial cancer. Cancer Epidemiol. Biomarkers Prev. 2012, 21, 980-987.

(7) Meyer, A.; Coinac, I.; Bogdanova, N.; Dubrowinskaja, N.; Turmanov, N.; Haubold, S.; Schurmann, P.; Imkamp, F.; von Klot, C.; Merseburger, A. S.; Machtens, S.; Bremer, M.; Hillemanns, P.; Kuczyk, M. A.; Karstens, J. H.; Serth, J.; Dork, T. Apoptosis gene polymorphisms and risk of prostate cancer: a hospital-based study of German patients treated with brachytherapy. Urol. Oncol. 2013, 31, 74-81.

(8) Muranen, T. A.; Blomqvist, C.; Dork, T.; Jakubowska, A.; Heikkila, P.; Fagerholm, R.; Greco, D.; Aittomaki, K.; Bojesen, S. E.; Shah, M.; Dunning, A. M.; Rhenius, V.; Hall, P.; Czene, K.; Brand, J. S.; Darabi, H.; Chang-Claude, J.; Rudolph, A.; Nordestgaard, B. G.; Couch, F. J.; Hart, S. N.; Figueroa, J.; Garcia-Closas, M.; Fasching, P. A.; Beckmann, M. W.; Li, J.; Liu, J.; Andrulis, I. L.; Winqvist, R.; Pylkas, K.; Mannermaa, A.; Kataja, V.; Lindblom, A.; Margolin, S.; Lubinski, J.; Dubrowinskaja, N.; Bolla, M. K.; Dennis, J.; Michailidou, K.; Wang, Q.; Easton, D. F.; Pharoah, P. D.; Schmidt, M. K.; Nevanlinna, H.; … Dubrowinskaja, N.; … Patient survival and tumor characteristics associated with CHEK2:p.I157T - findings from the Breast Cancer Association Consortium. Breast Cancer Res. 2016, 18, 98-016-0758-5.

(9) Peters, I.; Dubrowinskaja, N.; Abbas, M.; Seidel, C.; Kogosov, M.; Scherer, R.; Gebauer, K.; Merseburger, A. S.; Kuczyk, M. A.; Grunwald, V.; Serth, J. DNA methylation biomarkers predict progression-free and overall survival of metastatic renal cell cancer (mRCC) treated with antiangiogenic therapies. PLoS One 2014, 9, e91440.

(10) Peters, I.; Dubrowinskaja, N.; Hennenlotter, J.; Antonopoulos, W. I.; Von Klot, C. A. J.; Tezval, H.; Stenzl, A.; Kuczyk, M. A.; Serth, J. DNA methylation of neural EGFL like 1 (NELL1) is associated with advanced disease and the metastatic state of renal cell cancer patients. Oncol. Rep. 2018, 40, 3861-3868.

(11) Peters, I.; Dubrowinskaja, N.; Kogosov, M.; Abbas, M.; Hennenlotter, J.; von Klot, C.; Merseburger, A. S.; Stenzl, A.; Scherer, R.; Kuczyk, M. A.; Serth, J. Decreased GATA5 mRNA expression associates with CpG island methylation and shortened recurrence-free survival in clear cell renal cell carcinoma. BMC Cancer 2014, 14, 101-2407-14-101.

(12) Peters, I.; Dubrowinskaja, N.; Tezval, H.; Kramer, M. W.; von Klot, C. A.; Hennenlotter, J.; Stenzl, A.; Scherer, R.; Kuczyk, M. A.; Serth, J. Decreased mRNA expression of GATA1 and GATA2 is associated with tumor aggressiveness and poor outcome in clear cell renal cell carcinoma. Target Oncol. 2015, 10, 267-275.

(13) Peters, I.; Gebauer, K.; Dubrowinskaja, N.; Atschekzei, F.; Kramer, M. W.; Hennenlotter, J.; Tezval, H.; Abbas, M.; Scherer, R.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. GATA5 CpG island hypermethylation is an independent predictor for poor clinical outcome in renal cell carcinoma. Oncol. Rep. 2014, 31, 1523-1530.

(14) Prokofyeva, D.; Bogdanova, N.; Dubrowinskaja, N.; Bermisheva, M.; Takhirova, Z.; Antonenkova, N.; Turmanov, N.; Datsyuk, I.; Gantsev, S.; Christiansen, H.; Park-Simon, T. W.; Hillemanns, P.; Khusnutdinova, E.; Dork, T. Nonsense mutation p.Q548X in BLM, the gene mutated in Bloom's syndrome, is associated with breast cancer in Slavic populations. Breast Cancer Res. Treat. 2013, 137, 533-539.

(15) Sogkas, G.; Dubrowinskaja, N.; Bergmann, A. K.; Lentes, J.; Ripperger, T.; Fedchenko, M.; Ernst, D.; Jablonka, A.; Geffers, R.; Baumann, U.; Schmidt, R. E.; Atschekzei, F. Progressive Immunodeficiency with Gradual Depletion of B and CD4(+) T Cells in Immunodeficiency, Centromeric Instability and Facial Anomalies Syndrome 2 (ICF2). Diseases 2019, 7, 10.3390/diseases7020034.

(16) Tezval, H.; Dubrowinskaja, N.; Peters, I.; Reese, C.; Serth, K.; Atschekzei, F.; Hennenlotter, J.; Stenzl, A.; Kuczyk, M. A.; Serth, J. Tumor Specific Epigenetic Silencing of Corticotropin Releasing Hormone -Binding Protein in Renal Cell Carcinoma: Association of Hypermethylation and Metastasis. PLoS One 2016, 11, e0163873.

(17) von Klot, C. A.; Dubrowinskaja, N.; Peters, I.; Hennenlotter, J.; Merseburger, A. S.; Stenzl, A.; Kuczyk, M. A.; Serth, J. Rho GDP dissociation inhibitor-beta in renal cell carcinoma. Oncol. Lett. 2017, 14, 8190-8196.