UNIVERSIDAD DE LA FRONTERA FACULTAD DE CIENCIAS AGROPECUARIAS Y FORESTALES

Identificación de de expresión diferencial en cáncer pancreático mediante hibridación sustractiva por supresión/microarray y su evaluación como potenciales biomarcadores

Tesis presentada a la Dirección Académica de Postgrado de la Universidad de La Frontera para obtener el grado de Doctor en Ciencias mención Biología Celular y Molecular Aplicada.

JAIME ANDRES ESPINOZA RUIZ

GUIA: DR. MANUEL GIDEKEL

CO GUIA: OSVALDO PODHAJC :K RECiBa DO Programa Form<:~clón Capital Humano Avanzado

TEMUCO-CH 1LE 2O D IC 2012 2012 lflgr~sª Nfl;, ...... !:;¿_.;?._,J!.;~.... . -······················· OiJ&tfn~ '"'"""''''""lt''~tnu"""''"'"•n•• - -- ························-············· UNIVERSIDAD DE LA FRONTERA

DE CIENCIAS AGROPECUARIAS Y FOREST . ~~~"'\

Q~.(:.C\(;)N -·\ .· ,-:.; . AV.,;,).lSM't.:.;A r;. i ~ N' :;:1 . :;¡ ?O>:rr~iO j INFORME DE APROBACIÓN . /l \:. .. - ' ·.:~.-.:-~ TESIS DE DOCTORADO

Se informa a la Dirección Académica de Postgrado de la Universidad de La Frontera que la Tesis de Doctorado presentada por la candidata

JAIME ANDRES ESPINOZA RUIZ ha sido aprobada por la Comisión Informante de Tesis' como requisito para el Grado de Doctor en Ciencias mención Biología Celular y Molecular Aplicada, en el examen de defensa de Tesis rendido el ... .::f .... de ... ~~~~~ ...... de 2012.

Guía de Tesis

Dr. Manuel Gidekel

Ca-Guía

Dr. Osvaldo Podhajcer

Evaluador Interno

Dra. Ana Gutiérrez Moraga

Evaluador Externo

Dr. Guillermo Mazzolini

Dr. lván Roa ¿. /¡¡. COMISION EXAMINADORA

PROFESOR GUIA

Dr. Manuel Gidekel Vice Rectoría Investigación y postgrado. Universidad de La Frontera, Temuco-Chile

PROFESOR CO-GUIA

Dr. Osvaldo Podhajcer. Laboratorio de Terapia Molecular y Celular, Fundación Instituto Leloir. Buenos Aires, Argentina.

PROFESOR EVALUADOR INTERNO

Dra. Ana Gutiérrez Moraga. Departamento de producción agropecuaria. Facultad de Ciencias Agropecuarias y Forestales. Universidad de La Frontera, Temuco- Chile

PROFESORES EVALUADORES EXTERNOS

Dr. Guillermo Mazzolini Laboratorio de Terapia Génica Facultad de Medicina, Universidad Austral. Buenos Aires-Argentina

Dr.lván Roa Servicio de Anatomía Patológica de la Clínica Alemana Santiago-Chile Esta tesis fue realizada en el Laboratorio CTI Salud - VentureL@b. Escuela de Negocios. Universidad Adolfo lbáñez. Santiago - Chile y contó con el financiamiento de:

1. Consorcio de Tecnología e Innovación para la Salud S.A. CONICYT PIA-06 2. Beca Doctorado en Chile CONICYT N° 21080240. 3. Beca CHILE Pasantía Doctoral en el Extranjero. N° 75110023. 4. Beca de Apoyo de Tesis Doctoral CONICYT N° 24100124. 5. Beca de Exención de Arancel UFRO 2008-2012. INDICE DE CONTENIDOS

IN DICE DE CONTENIDOS ...... O AGRADECIMIENTOS ...... 3 PRODUCTIVIDAD CIENTÍFICA ...... 5 RESUMEN ...... S ABSTRACT ...... 1 O 1.1NTRODUCCIÓN ...... 12 1.1 El páncreas ...... 12 1.1.1 Ubicación, morfología y tipos celulares del páncreas ...... 12 1.1.2 Antecedentes epidemiológicos ...... 14 1.1.3 Diagnóstico y tratamiento ...... 15 1.1.4 Patofisiología ...... 16 1.1.5 Alteraciones genéticas del adenocarcinoma ductal pancreático ...... 18 1.2 Transcriptómica y Cáncer ...... 22 1.2.1 El transcriptoma ...... 22 1.2.2 El transcriptoma del cáncer ...... 24 1.2.3 El transcriptoma del adenocarcinoma ductal pancreático ...... 25 1.3 Microarray de expresión génica ...... 26 1.4 Hibridación sustractiva por supresión ...... 28 2. PRESENTACION DEL PROBLEMA ...... 33 2.1 Justificación ...... 33 2.2 Hipótesis ...... 33 2.3 Objetivo general de la tesis ...... 34 2.4 Objetivos específicos ...... 34 3. MATERIAL Y MÉTODOS ...... 35 3.1. Muestras clínicas ...... 35 3.2 Extracción de RNA ...... 35 3.2.1 Extracción de RNA de tejidos tumorales y no neoplásicos ...... 35 3.2.2 Determinación de la integridad del RNA...... 36 3.3 Hibridación supresiva sustractiva (SSH) ...... 37 3.3.1 Síntesis de cONA mediante tecnología SMART ...... 37 3.3.2 Sustracción génica mediante PCR-Select ...... 41 3.4 Amplificación y marcaje de las muestras ...... 43 3.5 Hibridación de los microarrays ...... 47 3.6 Extracción de la fluorescencia y análisis bioinformático ...... 48 3.7 Síntesis de cONA y PCR cuantitativo en tiempo real ...... 50 3.8 lnmunohistoquímica ...... 53 3.8.1. Anticuerpos ...... 53 4. RESULTADOS ...... 57 4.1 Diseño experimental ...... 57 4.2 Estrategia Directa ...... 59 4.3 Estrategia SSH ...... 59 4.3.1 Síntesis de cONA mediante tecnología Super SMART ...... 62 4.3.2 Ligación de adaptadores ...... 62 4.3.3 Sustracción génica por PCR-Se/ect ...... 64 4.4 Integración de los perfiles de expresión génica analizados mediante microarray de las estrategias Directa y SSH ...... 67 4.4.1 Secuencias de expresión diferencial entre tejidos tumorales y no neoplásicos, identificados mediante estrategia Directa y SSH ...... 67 4.4.2 Análisis de enriquecimiento de ontología de genes ...... 70 4.4.3 Ambas estrategias identifican genes que codifican para proteínas de secreción.

··································································································································· 75 4.5 Validación de los genes diferenciales obtenidos por la estrategia directa y SSH por PCR en tiempo real...... 81 4.6 Validación de genes diferenciales mediante tinción inmunohistoquímica ...... 83 4.6.1. Genes candidatos de la estrategia Directa ...... 83 4.6.1.1. PMEPA 1 prostate transmembrane , androgen induced 1 ...... 83 4.6.1.2 BHLHE40 basic helix-loop-helix fa mi/y, member e40 ...... 86 4.6.1.3. CALU calumenin ...... 88 4.6.2. Genes identificados en la estrategia Directa contrastados con la base de datos del Human Atlas Protein ...... 91 4.6.3. Genes candidatos de la estrategia SSH ...... 93 4.6.3.1. PLEKHM1 pleckstrin homo/ogy domain containing, family M (with RUN doma in) member 1...... 93 4.6.3.2. WNT9A wingless-type MMTV integration site family, member 9A ...... 99 4.6.3.3. MMP? matrix metallopeptidase 7 (matrilysin, uterine) y STAT1 signa/ transducer and activator of transcription 1, 91 kDa ...... 102 4.6.3.4. DPYSL4 dihydropyrimidinase-like 4 ...... 104 4.6.4. Genes identificados en la estrategia SSH contrastados con la base de datos del Human Atlas Protein ...... 105 5. DISCUSION ...... 107 5.1 La estrategia Directa permite la identificación de transcritos relacionados con la reacción desmoplástica del ADP ...... 108 5.2 La estrategia SSH permite enriquecer grupos de genes distintos de los identificados mediante la estrategia Directa ...... 111 5.3 Ambas estrategias permiten identificar genes que codifican para proteínas de secreción ...... 115 5.4 Validación de genes candidatos de la estrategia Directa ...... 117 5.5 Validación de genes candidatos de la estrategia SSH ...... 120 5.6 Realidad actual del diagnóstico de adenocarcinoma ductal pancreático ...... 122 6. CONCLUSIONES GENERALES ...... 127 7. BIBLIOGRAFÍA ...... 129 8. ANEXOS ...... 148 PATENTE PROVISORIA 2010 ...... 150 PATENTE PROVISORIA 2012 ...... 154 ARTICULO ENVIADO 1...... 156 ARTÍCULO ENVIADO 2 ...... 224 ARTÍCULO PUBLICADO (Ce-autoría) ...... 263 "El hecho de ver un tumor «pequeño» y extraerlo del cuerpo no nos garantiza que estemos libres del cáncer, algo que todavía nos cuesta creer. En definitiva, una mamografía o un frotis de Pap no son un retrato del cáncer en su infancia. Como cualquier retrato, lo hacemos con la esperanza de que capte algo esencial del sujeto: su psique, su ser interno, su futuro, su comportamiento. «Todas las fotografías son fieles -se complacía en decir el artista Richard Avedon-, pero ninguna es la verdad»".

El emperador de todos los males: Una biografía del cáncer Siddhartha Mukherjee [1]

" ... los genes son como las teclas de un piano; Aunque son esenciales, es el contexto el que hace la música" Nelson C.M. & Bissell M.J. [2]

2 AGRADECIMIENTOS

Quiero agradecer a mi padre, madre y hermana, quienes me dieron en primer lugar la oportunidad de escoger este camino y me han apoyado constantemente durante este proceso. Un agradecimiento no es suficiente para demostrar el amor y admiración que tengo por ellos.

Al Dr. Manuel Gidekel y la Dra. Ana Gutierrez por darme la oportunidad de desarrollar mi proyecto en su laboratorio y apoyarme en el término de la tesis.

A Carolina Bizama, Felipe Benavente y Edgardo Salvatierra, mis más cercanos compañeros de laboratorio con quienes compartí las alegrías del trabajo en el laboratorio y me apoyaron cuando lo necesité. Gracias por compartir sus experiencias conmigo y brindarme su amistad.

A mis compañeros de laboratorio de hoy y de siempre: Jennifer Osorio, Graciela Berríos, Gustavo Cabrera, Mónica Ramírez, Claudia Calderón, Ismael Riquelme, Daniel Weinecker, Claudia Calderón, Yamilé Bernardo, Alejandra Sandoval, Helga Weber, Eduardo Sagredo y Francisco Gamín, por todos los gratos momentos, el apoyo brindado y sus contribuciones para implementar, ejecutar, interpretar y discutir los resultados obtenidos en este proyecto.

A los integrantes del laboratorio de Proteomics of Cancer del Danish Cancer Society, Copenhagen, Dinamarca. A José Moreira, Sofia Svensson, Anni Handesten y Lene Brogaard por su apoyo, entusiamo y constante preocupación por mi. Quiero agradecer en especial a Julio Celis y Teresa Cabezón, por recibirme en su laboratorio, apoyarme y compartir sus conocimientos conmigo. Me permitieron recordar las razones del por que escogí este camino tanto tiempo atrás.

Al Dr. Osvaldo Podhajcer, Dr. Guillermo Mazzolini, Dr. Elmer Fernández y Dr. lván Roa por sus valiosos aportes al desarrollo de este proyecto. 3 A Soledad Lantadilla, Gabriela Bascuñan, Tamara Sánchez del Servicio de Anatomía Patológica de Clínica Alemana de Santiago, por la buena voluntad y la enorme ayuda que me dieron durante este proceso.

A Jorge Dinamarca y Nicolás Saavedra, quienes tuvieron la voluntad y paciencia de enviarme decenas (cientos quizás) de papers científicos que les solicité durante los últimos años.

4 PRODUCTIVIDAD CIENTÍFICA

La presente tesis de Doctor originó hasta la fecha la siguiente productividad científica (ver anexos):

Propiedad intelectual • GIDEKEL Manuel, PODHAJCER Osvaldo, ROA lván, BIZAMA Carolina, BENAVENTE Felipe, ESPINOZA Jaime, SALVATIERRA Edgardo, FERNANDEZ Elmer, GUTIERREZ Ana, ROA Juan Carlos, MAZZOLINI Guillermo: Novel genes and uses thereof, expression profile of colon, gastric and pancreatic cancer. Patent USPTO non-provisional and PCT. Número de presentación 61/404,141, fecha 09/28/201 O.

• GIDEKEL Manuel, ESPINOZA Jaime, CABEZON Teresa, MAZZOLINI Guillermo. Tumor biomarkers for pancreatic cancer. Patent USPTO provisional. Filling number 61/634832.

Publicaciones

• JA Espinoza, C Bizama, F Benavente, EA Fernández, E Salvatierra, HA Gutierrez, 1 Roa, G Mazzolini, O Podhajcer, M Gidekel. Low-abundance transcriptome of pancreatic ductal adenocarcinoma as a source of potential molecular markers. Artículo sometido a Molecular Oncology.

• C Bizama, F Benavente, JA Espinoza, E Salvatierra, HA Gutierrez, EA Fernández, EA Sagredo, 1 Roa, G Mazzolini, M Gidekel & OL Podhjcer. Low abundance trancriptome analyses identifies novel markers, specific intracelular pathways and target genes in advanced human gastrointestinal cancer. Artículo sometido a Cancer Research.

Otras publicaciones generadas durante el período 2008-2012

• M Malvicini, M lngolotti, F Piccioni, M García, J Bayo, C Atorrasagasti, L Alaniz, J Aquino, JA Espinoza, M Gidekel, OG Scharovsky, P Matar & G Mazzolini. Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and transfer of IL-12. Molecular Oncology 2011, 5(3):242-55. 5 • JA Espinoza, U Paasch & JV Villegas. Mitochondrial membrane potential disruption pattern in human sperm. Human Reproduction 2009 24(9):2079- 2085.

• JA Espinoza, MA Schulz, R Sánchez & JV Villegas. lntegrity of Mitochondrial Membrane Potential Reflects Human Sperm Quality. Andrologia 2009,41:51- 54.

• P Navarrete, JA Espinoza, J Parodi, JG Álvarez & R Sánchez. Protective effect of Fallopian tubal fluid against activated leukocyte-induced sperm DNA fragmentation: preliminary results. Andrologia 2009, 41 (3): 196-198.

Cursos

• Functional Genomics and Systems Bio/ogy. Wellcome Trust Genome Campus, Hinxton, Cambridge, UK. 16-25 Junio 201 O.

• FEBS Advanced Lecture Course on Translational Cancer Research. Hotel Porto Bay Falesia, Algarve, Portugal September 27 - October 4, 2011.

Pasantías de investigación

• Laboratory of Cancer Proteomic, Danish Cancer Society Research Center, Copenhagen, Denmark. Agosto-Diciembre 2011. Investigador responsable: Dr. Julio Celis.

Presentaciones en congresos y cursos

• Bizama C., Benavente F., Espinoza JA., Salvatierra E., Fernández E., Sagredo E. A., Roa 1., Mazzolini G., Gidekel M., Podhajcer OL. Transcriptome analyses identifies specific intracellular pathways and target genes in gastric cancer. CTI-Salud, Universidad de La Frontera. XXXV Reunión Anual Sociedad de Bioquímica y Biología Molecular de Chile. 2-5 de Octubre 2012. Puerto Varas. Premio mención honrosa en categoría mejor trabajo de incorporación.

• Espinoza JA, Salvatierra E, Freeman T, Bizama C & Gidekel M. Network analysis reveals a strong desmoplastic signature of pancreatic ductal adenocarcinoma. FEBS Advanced Course on Translational Cancer Research. Algarve, Portugal. September 27- October 4, 2011. 6 • Sagredo E, Espinoza JA, Brito C, Bizama C, Cabrera G y Gutiérrez-Moraga A. Transcriptomic network analysis of Arabidopsis thaliana under UVB-stress particular transcriptions factors involved in the response to stress. VI Encuentro de Estudiantes de Ingeniería en Biotecnología (ENEIB). Universidad Nacional Andrés Bello (UNAB) Viña del Mar, Chile. 23 a 25 de Agosto de 2011.

• Sagredo E, Espinoza JA, Brito C, Bizama C, Cabrera G, Gutiérrez-Moraga A y Gidekel M. Transcriptomic network analysis of Arabidopsis thaliana reveals specific gene regulation under cold, salt and UV-B stress conditions. XXXIV Reunión Anual Sociedad de Bioquímica y Biología Molecular de Chile, Valdivia, Chile. 27 al 30 de Septiembre de 2011.

• Sagredo E, Espinoza JA, Brito C, Bizama C, Cabrera G, Gutiérrez-Moraga A y Gidekel M. Transcriptomic network analysis of Arabidopsis thaliana under UVB-stress reveals time-dependent changes of biological processes involved in the adaptation to stress. Workshop de Genómica Vegetal y Biología de Sistemas. Santiago, Chile. 1O y 11 de Noviembre de 2011

7 RESUMEN

El adenocarcinoma ductal pancreático (ADP) representa el decimotercer cáncer más frecuente a nivel mundial y es la causa de muerte de alrededor de 260,000 personas cada año. Esta enfermedad es más común en la tercera edad y solo alrededor del 20% de los pacientes presentan tumores localizados y potencialmente curables mediante la extracción quirúrgica del órgano. La tasa de sobrevivencia en pacientes a los 5 años es menor al 5%, lo que convierte al ADP en uno de los tumores sólidos más mortales. El pésimo pronóstico del ADP está directamente vinculado a la falta de metodologías de diagnóstico temprano y de tratamientos antineoplásicos efectivos. La sobrevida de los pacientes mejora sustancialmente si el tumor es diagnosticado en etapas tempranas de la enfermedad. Recientemente se demostró que la progresión del ADP, desde la generación de la célula tumoral de origen hasta la aparición del primer clan metastásico, ocurre en un tiempo promedio de 15 años, permitiendo una hipotética ventana de tiempo para el diagnóstico temprano de la enfermedad. En este contexto, es de importancia identificar biomarcadores moleculares que permitan detectar la lesión tumoral y cuya implementación pueda ser transferible al uso clínico rutinario. En este estudio empleamos la metodología de hibridación sustractiva por supresión (SSH, de su sigla en inglés) acoplada a microarray para estudiar el transcriptoma de baja abundancia de tumores de adenocarcinoma ductal pancreático, con el objetivo de identificar moléculas que se expresen diferencialmente con respecto de tejidos no neoplásicos y que posean especificidad por el tejido tumoral. La estrategia SSH permite enriquecer diferentes genes a los identificados mediante la estrategia directa (microarray convencional), permitiendo una reducción de genes de alta abundancia relacionados con matrix extracelular y favoreciendo la identificación de genes de función predominantemente intracelular. Por otra parte, la estrategia SSH enriqueció genes relacionados con la vía de señalización de Wnt asociada a {3-catenin. Identificamos incremento de los niveles del transcrito y la proteína de los genes WNT9A y PLEKHM1 en tejidos de ADP. La proteína WNT9A se expresa a bajos niveles en las células ductales de tejidos pancreáticos no neoplásicos, sin embargo, mayores niveles de expresión se observaron en tejidos tumorales de ADP. La proteína codificada por el gen PLEKHM1 se expresa de forma exclusiva en células tumorales y no en otros tipos celulares del páncreas. Ambas proteínas se expresan de forma diferencial y presentan gran especificidad por el tejido tumoral. En conclusión, con esta investigación hemos identificado nuevos potenciales biomarcadores de cáncer pancreático a partir del transcriptoma de baja abundancia del ADP, analizado mediante hibridación sustractiva por supresión. Nuevos marcadores de alta especificidad pueden contribuir a desarrollar nuevas metodologías de diagnóstico molecular para el cáncer pancreático. ABSTRACT

Pancreatic ductal adenocarcinoma (PDA) is the thirteenth most common cancer worldwide and is the cause of death of about 260,000 people each year. This disease is more common in old age and only about 20% of patients have tumors potentially curable by surgical removal of the organ. The survival rate in patients at 5 years is less than 5%, making the ADP in one of the deadliest salid tumors. The poor prognosis of ADP is directly linked to the lack of methods for early diagnosis and effective cancer treatments. The patient survival improves substantially if the tumor is diagnosed in early stages of the disease. Recently it was demonstrated that the progression of ADP from the generation of the tumor cell of origin to the first appearance of metastatic clone, occurs in an average of 15 years, allowing a hypothetical time window for early diagnosis of the disease. In this context it is important to identify molecular biomarkers to detect the tumor and whose implementation can be transferable to routine clinical use. In this study we used a suppressive subtractive hybridization (SSH) coupled to microarray to study the low abundance transcriptome of pancreatic ductal adenocarcinoma tumors, with the aim of identify molecules that are differentially expressed with respect to non­ neoplastic tissues. The SSH strategy enrich different genes to those identified by the direct strategy (conventional microarray), allowing a reduction of genes of high abundance extracellular matrix-related and favoring the identification of genes predominantly intracellular function. Moreover, the strategy SSH enriched genes related to Wnt signaling pathway associated with ~-catenin, a signaling pathway of increasing importance in cancer. We identified higher transcript and protein levels of PLEKHM1 and WNT9A in ADP tissues. WNT9A protein is expressed at low levels in the pancreatic ductal cells of non-neoplastic tissue, however, higher levels of expression were observed in tumor tissues ADP. The protein encoded by PLEKHM1 is expressed exclusively in tumor cells and not in other cell types of the páncreas. Both are expressed differentially and are highly specific for tumoral tissue. In conclusion, we found new potential biomarkers of pancreatic cancer from the transcriptome of low abundance of ADP, analyzed by suppression subtractive hybridization. Highly specific markers can help to develop new molecular diagnostic methods for pancreatic cancer. 1. INTRODUCCIÓN

1.1 El páncreas

1.1.1 Ubicación, morfología y tipos celulares del páncreas

El páncreas es un órgano con funciones endocrinas y exocrinas, que se ubica en la región retroperitoneal de la parte superior de la cavidad abdominal, junto al bazo y detrás del hígado y el estomágo (Figura 1A y 1B). En humanos, el órgano posee distintivamente una cabeza, cuerpo y cola, con la cabeza del páncreas contactando directamente la regional duodenal del intestino. Las células del páncreas están organizadas en distintos lóbulos compuestos principalmente por células acinares, las cuales se ordenan estructuralmente en acinos (Figura 1C). Las células acinares representan el 90-95% de la masa total del páncreas y son las encargadas de producir las enzimas digestivas que son vertidas en el duodeno para la digestión de los alimentos. Entre las enzimas que producen se encuentra la carboxipeptidasa 1(Figura 1H), ribonucleasa A (Figura 11), proteasa, elastasa, fosfolipasa, amilasa y tri psi na. Otro tipo celular del acino es la célula centroacinar progenitora que expresa SOX9, proteína reguladora de la transcripción (Figura 1J). Las enzimas son transportadas a través de los duetos hacia el duodeno. Los duetos están compuestos de células de origen epitelial, que expresan, entre otros marcadores, queratina 19 (Figura 1K). El componente endocrino del páncreas lo constituyen los islotes de Langerhans donde se encuentran las células alfa (productoras de la hormona glucagón) (Figura 1D), las células beta (productoras de la hormona insulina) (Figura 1E) y las células delta (productoras de la hormona somatostatina) (Figura 1F). Otro tipo celular caracterizado recientemente es la célula estelar pancreática [3], la cual reside entre los acinos pancreáticos normales, extendiendo proyecciones celulares. Se caracteriza por expresar marcadores moleculares asociados a fibroblastos y células de musculatura lisa, como la actina de músculo liso (Figura 1G). Otros tipo celulares productores de hormonas son las denominadas células PP (PP ce//), que producen el polipéptido pancreático y las células épsilon, que producen la hormona grelina.

12 A Vista posterior B Vista anterior

Cola

Cuerpo

Islote de ...-~-t--:­ Langerhans

.___--+-- Acinos . ':... ;.;. <,• ·, ••

Dueto

Islotes de Langerhans Células estelares Células Delta ( Ste/late ce/ls) ...... ' f ... • • ~ .. \·.'·.·,...... :-,•-"'··- . . ' ~. ·.~· . ~·. . . \1:\ . .. . . ·~ ~; t ,. •J/.' ,.• : ._· •r .."' , _.. , .•• • .. r ..... ;,· ·~,"' .. • Glucagón Insulina Somatostatina Actina de músculo liso Células acinares

Carboxipeptidasa A 1 Ribonucleasa, RNasaA SRY (sex determining Keratina 19 familia, 1 region Y)-box 9

Figura 1. Ubicación, morfología y tipos celulares del páncreas. A) Vista posterior y B) Vista anterior de la ubicación retroperiteoneal del páncreas, oculto tras el hígado y el estomágo y posicionado al lado del bazo. El dueto biliar atraviesa en páncreas en su camino al duodeno. C) Los tres principales compartimentos del páncreas: los acinos productores de enzimas digestivas, los duetos que transportan las enzimas hacia el duodeno y los islotes de Langerhans, componente endocrino del páncreas. D) 13 Tinciones inmunohistoquímicas de cortes histológicos de tejidos pancreáticos normales marcados con diferentes anticuerpos primarios dirigidos contra proteína de específicas de tipo celular. Imágenes de figura 1A y 18 fueron obtenidas del recurso online open­ access Science Photo Library (http://www.sciencephoto.com/). Las imágenes de las figuras 1D-K fueron obtenidas desde el Human Atlas Protein [4] respetando su política de uso de imágenes (http://www.proteinatlas.org/about/datausage).

1.1.2 Antecedentes epidemiológicos

El adenocarcinoma ductal pancreático (ADP) es la neoplasia más común del páncreas, constituyendo cerca del 85% de todos los tumores de este órgano. Esta neoplasia representa el decimotercer cáncer más frecuente del mundo, con aproximadamente 232,000 nuevos casos diagnosticados anualmente. La tasa de sobrevivencia a los 5 años oscila entre el 3 y 5%, la más baja dentro de los cánceres importantes [5]. En Chile, el ADP posee el índice de mortalidad más elevado de todas las neoplasias, donde prácticamente el 99% de los pacientes fallecen al corto o mediano plazo [6]. En Estados Unidos, el ADP constituye la cuarta causa de muerte por neoplasias, con una cifra estimada para el año 2008 de 37,000 nuevos casos y más de 34,000 muertes [7]. La mortalidad es dos a tres veces superior en países desarrollados en comparación con países menos desarrolldos [8]. Las razones para la baja sobrevivencia del ADP incluyen la típica naturaleza agresiva de este tipo de tumores, el diagnóstico tardío, la baja tasa de resección quirúrgica y la falta de terapias efectivas [5]. El ADP es predominantemente una enfermedad de la tercera edad, siendo los 73 años la mediana al momento del diagnóstico y rara vez se presenta antes de los 40 años. El tabaquismo es el principal factor de riesgo modificable, donde se estima que al menos uno de cada cuatro casos es producido por la adicción al tabaco. Otros factores de riesgo establecidos para esta neoplasia incluyen dietas ricas en carnes y grasas, bajos niveles de folato sérico, diabetes mellitus crónica y pancreatitis crónica [9]. El ADP es responsable de una cantidad sustancial de carcinomas con tumor primario desconocido debido a que con frecuencia se encuentra ampliamente diseminado al momento del diagnóstico y el sitio primario de origen no es detectado a menos que el estudio sea cuidadoso [1 0]. Debido a su patrón infiltrativo y 14 diseminación temprana, rara vez el tumor primario forma una lesión compacta de más de 5 cm, contribuyendo a que la opción de extraer quirúrgicamente el órgano tan solo pueda ser considerada en menos del 20% de los casos [11]. El ADP produce una masa tumoral firme y altamente esclerótica. Los bordes del tumor suelen estar mal definidos y con largas extensiones de carcinoma más allá de la masa tumoral principal. A nivel microscópico, el tejido tumoral se compone de un epitelio neoplásico infiltrante de formas glandulares con una intensa reacción desmoplástica. La invasión vascular y perineural se encuentran presentes en la mayoría de los cánceres extraídos quirúrgicamente y son muy comunes las metástasis a los ganglios linfáticos regionales y al hígado [9].

1.1.3 Diagnóstico y tratamiento

Al momento de la presentación de la enfermedad, los pacientes frecuentemente sufren de dolores abdominales profundos y difusos, que se localizan ampliamente alrededor del área del tumor. La mayoría de los pacientes manifiestan síntomas sistémicos como astenia (sensación de fatiga), anorexia (falta de apetito) y pérdida de peso. Otras manifestaciones menos frecuentes abarcan trombosis venosas profundas y superficiales, paniculitis (inflamación del tejido adiposo subcutáneo), alteración de las funciones hepáticas, obstrucción de la salida gástrica, aumento de la circunferencia abdominal y depresión [12]. Más del 40% de los pacientes con cáncer pancreático asisten tres o más veces a la consulta médica antes de ser diagnosticados, situación que solo es superada por el mieloma múltiple con más del 50% de los pacientes [13]. La presencia de síntomas inespecíficos y la carencia de un sistema de screening sensible y específico para esta enfermedad, contribuyen enormemente a la ineficacia del diagnóstico del cáncer pancreático, que se ve reflejada con el alto número de visitas al médico que deben realizar los pacientes antes de ser diagnosticado correctamente. La evaluación de un paciente se centra en el diagnóstico y la estadificación de la enfermedad, así como en la evaluación de la resecabilidad y la paliación de los

15 síntomas. La tomografía computarizada helicoidal multicorte junto con la administración intravenosa de material de contraste es el procedimiento de elección para la . evaluación inicial. La molécula CA 19-9 es el único biomarcador que posee una utilidad clínica demostrada, siendo empleado para monitorear la terapia y la detección temprana de recurrencia de la enfermedad después del tratamiento, sin embargo, posee escasa utilidad como metodología de diagnóstico temprano de lesiones tumorales y no es específico de cáncer pancreático [12, 14]. Actualmente existe una urgente necesidad de identificar nuevas estrategias moleculares de diagnóstico temprano de ADP, sin embargo, la carencia de biomarcadores específicos no es solo un problema del ADP. El campo de investigación de biomarcadores para biomedicina a proporcionado no más de 100 biomarcadores con utilidad clínica hasta la fecha, a pesar de las más de 150,000 publicaciones que han reportado el hallazgo de biomarcadores, dejando al descubierto la baja eficiencia de trasladar biomarcadores candidatos al ámbito clínico [15]. Actualmente, la droga escogida para el tratamiento del cáncer pancreático es gemcitabine, un análogo de nucleósido que interfiere con la replicación del DNA. La sobrevivencia al cabo de un año en pacientes tratados con esta droga es tan sólo del 18%, representando un avance significativo pero modesto en la calidad de vida del paciente [16, 17]. En ensayos clínicos fase 3, la combinación de gemcitabine y erlotinib, inhibidor del receptor de crecimiendo epidérmico, ha demostrado ser modestamente superior al tratamiento individual con gemcitabine. Erlotinib fue aprobado por la Food and Drug Administration (FDA) de EE.UU. para el tratamiento de cáncer pancreático. A pesar de la gran cantidad de investigaciones publicadas en el campo de la terapia del cáncer pancreático, actualmente no hay disponibilidad de terapias eficaces que extiendan la vida de los pacientes más allá de unos pocos meses [12].

1.1.4 Patofisiología

Los adenocarcinomas ductales pancreáticos evolucionan a partir de lesiones precursoras no invasivas como las neoplasias intraepiteliales pancreáticas (Pancreatic

16 intraepithelial neoplasia; PaniNs, de su sigla en idioma inglés), las cuales adquieren clonalmente una serie de alteraciones genéticas y epigenéticas. Además, otros tumores pancreáticos invasivos pueden originarse a partir de neoplasias papilares mucinosas intraductales o neoplasias quísticas mucinosas [14]. Las PaniNs se dividen en tres grados en función del grado de atipia epitelial. Las lesiones con atipia mínima se designan como PaniN-1, aquellos con atipia moderada son PaniN-2, y aquellos con marcada atipia son PaniN-3 (carcinoma in situ). Además, las lesiones PaniN-1 se subdividen en los tipos plano (PaniN-1A) y papilares (PaniN- 1B) (Figura 2).

;~:f.{."~~ ... -/~: ··n. ~ ·; -~,~~~' .~: ·.. ·. ·~,11~ . . . .. • . {:' ~ . . . .; .. - ... .:.-. . . )\ r. . , 1 • • , • .;~ ~ • • '\" •• t-~:-,.. ,.--;.J' • .• · • '! , <". . .'. .} \ \r t ~-- ·.- ~: ~~ .. !.'. ·.: ...... {' ' .'(...... • ~· .. t., '~~·' ..,._ ~ •. . ~ .:.: . ··¡, . . .•.. 1• • ~ -¡. ' •• . ' •• ' ' " ,, ' ' r,. ,. '· ~·, ; , ·:"· . . .' 1 "''(~ • .\. , ~,_:·; ' .. ..: 't' l..... ft '''-' v, ~~J'i ': ·. ~::;·-~ ~~-~-- ~- ··:.·~ •.,'. ,-T.~:;·~,;¡, ;~~~?~-:V::· .._:

•• 11 • • P~llf"ile.: ·. ~'.f .~IN:W' ~-~ .. · -: • Pa~lN;_~·. '·· ·,/ :'PaniN-3 Figura 2. Progresión morfológica de las lesiones PaniN desde el tejido normal hasta la lesión PaniN-3. PaniN=pancreatic intraepithelial neoplasia. Imagen adaptada sin modificaciones desde referencia [18].

Cuando las lesiones preneoplásicas (PaniN) destruyen la lámina basal, el tumor adquiere un fenotipo invasivo. A nivel macroscópico, el tumor presenta bordes mal definidos que son difíciles de distinguir de otras lesiones benignas que forman cicatrices, como la pancreatitis crónica. Microscópicamente, el tumor se presenta a menudo "bien diferenciado", con elementos glandulares similares a los duetos pancreáticos normales [19]. A pesar de que el ADP se encuentra con frecuencia bien diferenciado, tiene una gran propensión a la infiltración insidiosa y se caracteriza por presentar invasión perineural en la totalidad de los casos, siendo el ADP el cáncer donde más frecuentemente se identifica este fenómeno y que se asocia al fuerte dolor y a un mal pronóstico [20].

17 Una de las principales características del ADP es la extensa reacción desmoplástica que se encuentra en la mayoría de los casos y que constituye, probablemente, la más intensa desmoplasia observada entre todos los tumores epiteliales [21]. Esta desmoplasia se caracteriza por la acumulación de tejido fibrótico (principalmente por acumulación de colágeno), células inflamatorias y formación de vasos sanguíneos. Actualmente sabemos que la célula pancreatica estelar (stel/ate ce//) participa activamente en la generación de esta reacción desmoplástica mediante la sobreproducción de colágenos y otras moléculas de matriz extracelular [22], fenómeno que contribuye a la quimioresistencia del ADP a terapias con agentes antineoplásicos [23] y nanopartículas [24].

1.1.5 Alteraciones genéticas del adenocarcinoma ductal pancreático

El ADP es una enfermedad causada por mutaciones heredadas y somáticas, caracterizándose por tener una firma mutacional que la diferencia de otras neoplasias pancreáticas. La mutación del oncogén KRAS2 es la alteración genética más común, presente en -90-95% de los casos a nivel mundial [25-27], reportándose en Chile una frecuencia de -60% [28]. La elevada presencia de mutaciones de KRAS en las lesiones preneoplásicas del ADP humano apoya su rol como gen precursor del proceso neoplásico, lo que ha sido corroborado reiteradas veces en modelos animales, donde la expresión de KRAS mutante ha mostrado ser un requisito previo para la aparición del ADP [29, 30]. El gen p16/CDKN2A (también conocido como INK4A) es el gen supresor de tumores más comúnmente inactivado en ADP (80-95%). Otros genes con alta frecuencia de mutación en el ADP son TP53 (50-75%), DPC4 (45-55%) y BRAC2 (7-10%) [18]. El proyecto del Atlas del Genoma del Cáncer (The Cancer Genome Atlas; TCGA http://cancergenome.nih.gov/) publicó el año 2008 un análisis genético integral del exorna de 24 tumores de ADP estudiados mediante next generation sequencing [31]. El estudió demostró que al menos 12 procesos se encuentran alterados genéticamente en la gran mayoría de los ADPs, estableciendo genes de las vías de señalización de

KRAS, Wnt/Notch, Hedgehog, TGF-~ y regulación de la transición entre G1/S que se

18 encontraron mutados en todos los casos estudiados y explican las principales características de la tumorogénesis pancreática (Figura 3). La intervención terapéutica de las vías o procesos de señalización, en vez de alteraciones genéticas individuales, provee un nuevo paradigma para el tratamiento de esta enfermedad [31].

M'rC. WNT9A. CASPlO. VCP. GATA6. TCF.:. CAD. liiPl MAP2. TSC2

TGF BR 2 __--· .,, CRCC4. BMPR2 fRCC6. CP300. S\IAf~. •, RANBP2. S\IAD3 .. · WnVNotc:h ApopCOSis IICnalinC '-., TPS3 DNA ' TGFf damate IICMiinC .f ¡~./ . control AGIIGU7. / / - t 1 CDKN2A. CCXA2BPA. ¡ [ Pancr~vtic-. , FBX\\17. DEPOC2. Smiol GTPase· ·. /' ' CHDl. C..'IOC~r R~&ulatlon of . APC PL CR l dependen! 2 G1!S phaM ': u,n.Hng• transltlon 1 i ADA.\111 Replaton TB\'> OPP6. of lnwaslon MfPJA so o PCSK6 LPR2. APG-lA. GL/1. BOC PRSS23 KRAS Homophltlc: CRfBBP llenallng c:eU C·Jun ed~ N·lermln<~l lntecrtn I

Figura 3. Principales vías de señalización alteradas en el adenocarcinoma ductal pancreático. 12 vías y procesos cuyos genes que los componen presentan alteraciones genéticas en la mayoría de los casos de ADP. Genes identificados en al menos 16 de 24 casos de ADP son señalados en el círculo más externo. GTPase= guanosine triphosphatase; TGF-13= transforming growth factor {3. Imagen adaptada sin modificaciones desde referencia [32]

KRAS mutado posee un rol activo en el origen de la lesión neoplásica pancreática, donde su expresión permite a la célula conservar su viabilidad durante el desarrollo tumoral temprano [33]. Este proceso está asociado al concepto de "adicción oncogénica", donde los tumores requieren la expresión y actividad sostenida de un solo gen aberrante para su desarrollo, a pesar de la acumulación de otras lesiones oncogénicas [34]. Las mutaciones de KRAS son detectables en etapas tan tempranas como la lesión Pan 1N-1 A [35] 19 El modelo actual de progresión tumoral para el ADP (Figura 4) propone que la mutación de KRAS y el acortamiento de los telómeros son pasos importantes para el comienzo del proceso carcinogénico, así como la mutaciones en los genes TP53 y DPC4 (SMAD4; SMAD fa mi/y member 4) son relevantes para desencadenar el fenotipo invasivo del adenocarcinoma. En etapas iniciales, el ADP adquiere alteraciones genómicas relacionadas con disfunción del telómero y desregulación del ciclo celular. La elevada inestabilidad genómica frecuentemente persiste después de la diseminación del cáncer, resultando así en una evolución constante, paralela e incluso convergente entre las distintas metástasis [36], sin embargo, la mayoría de los re­ arreglos genómicos ocurren tempranamente en el tumor primario y antes de la enfermedad metastásica. El año 2012, Yachida et al evaluaron cuantitativamente los tiempos requeridos para la evolución genética del ADP, estableciendo un tiempo promedio de alrededor de 1O años entre la adquisión de la primera mutación y el nacimiento de la primera célula parental tumoral no metastásica (Figura 5). Al menos otros cinco años más son necesarios para la adquisición de la capacidad metastásica y, en promedio, los pacientes fallecen al cabo de dos años a partir de ese momento [37]. Este estudio establece que la enfermedad transcurre en más de dos décadas desde su inicio hasta la muerte del paciente. Lamentablemente, la mayoría de los casos son diagnosticados en la etapa T3 (Figura 5), donde las células metastásicas ya se encuentran diseminadas y creciendo en órganos distantes. Sin embargo, desde el inicio de la enfermedad hasta antes de la metástasis, existe una ventana de tiempo teórica de -18 años, donde el cáncer puede ser diagnosticado. Este hallazgo estimula el desarrollo de nuevas alternativas de diagnóstico para el ADP.

20 Figura 4. Modelo PaniN de progresión tumoral. Las alteraciones genéticas y moleculares más comunes de cada estadía morfológico son mostradas. PaniN=pancreatic intraepithelial neoplasia. Imagen adaptada sin modificaciones desde referencia [38].

1 Normal 1-+ ~a ni~-+ 1 PaniN2J-+ 1 PaniN31-+

T1=11.7 yrs T2=6.8 yrs T3=2.7 yrs

lnitiation Gain of Gain of Death invasive metastatic ability ability Figura 5. Línea de tiempo estimada para la progresión genética del ADP. Sucesivas ondas de progresión clonal ocurren con la adquisión de mutaciones adicionales y que corresponden con el modelo de progresión Pan IN (T1 ). En T2 el ADP adquiere un fenotipo invasivo y durante T3, el tumor comienza la diseminación metastásica. En promedio, 21.2 años transcurren desde el inicio del tumor hasta el fallecimiento del paciente debido a la enfermedad metastásica. Imagen adaptada sin modificaciones desde referencia [39].

21 1.2 Transcriptómica y Cáncer

1.2.1 El transcriptoma

El transcriptoma es el conjunto completo de transcritos y sus niveles de expresión en una célula, en un determinado momento del desarrollo o estado fisiológico. Los transcritos incluyen a diversas clases de moléculas de RNA, tales como los RNAs mensajeros (mRNA), RNAs de transferencia (tRNA), RNAs ribosómicos (rRNA) y a otras clases de RNAs no codificantes [40]. A diferencia del genoma, el cual es relativamente estable, el transcriptoma es dinámico y varía enormemente dependiendo del tipo celular, tejidos y condiciones ambientales. Como los mRNAs transportan la información codificante del genoma (exorna) para la síntesis proteíca, se espera que los niveles de expresión de estas moléculas reflejen sus respectivos niveles de proteínas. Ha sido una práctica común en investigación usar las concentraciones de los mRNAs como un reflejo de las concentraciones y actividades de sus correspondientes proteínas, asumiendo así que la abundancia de los transcritos es el principal determinante de las concentraciones de las proteínas [41 ]. Estudios previos donde se han comparado los niveles de mRNA y proteínas han concluido que la correlación es pobre, sin embargo, estas conclusiones fueron obtenidas con una baja cantidad de genes debido, principalmente, a limitaciones técnicas [42, 43]. Recientemente, un estudio publicado por Schwanhausser et al [44] midió simultáneamente los niveles de mRNA y proteína de más de 5,000 genes en células mamíferas. El estudio concluyó que los niveles de mRNA explican alrededor del 40% de la variabilidad observada en los niveles de proteínas (en células en división y no sincronizadas), destacando la importancia del control traduccional del ribosoma en el proceso de la regulación de la expresión génica. Este porcentaje es similar al observado en otros estudios [41]. Futuras investigaciones aún deben dilucidar el aporte de otros procesos a la regulación global de la expresión génica, tales como son la transcripción, degradación de mRNA y degradación de proteínas [41, 44].

22 Determinados mRNAs y proteínas pueden tener diferentes abundancias y tasas de degradación según sus funciones. Por ejemplo, mRNAs y proteínas de genes metabólicos tienden a ser muy estables y tener elevadas razones de proteína/mRNA. Por el contrario, proteínas involucradas en la organización de cromatina y regulación de la transcripción tienden a ser rápidamente degradadas. Por lo tanto, la regulación de las proteínas y sus mRNAs reflejan roles biológicos específicos: proteínas regulatorias pueden ser producidas y degradadas muy rápido para reaccionar ante estímulos, mientras que proteínas estructurales o "housekeeping" pueden tener vidas más largas [41]. Las investigaciones transcriptómicas que han empleado las tecnologías de microarray y secuenciación de RNA (RNA-seq) han revelado que los niveles de expresión de mRNAs se desplegan a través de un amplio rango, observándose trancritos altamente representados y otros de muy baja expresión. En promedio, la mediana de expresión de transcritos es de solo 17 copias por célula, como se ha observado en fibroblastos murinos [44]. Sin embargo, al analizar transcritos individualmente las diferencias son dramáticas. En poblaciones purificadas de células Th2, los transcritos de GATA3 se encuentran en promedio entre 25 a 150 copias por célula, en cambio, los transcritos de TBX21 se encuentran en menos de una copia por célula. Estas variaciones en los niveles de expresión, estudiadas ya sea mediante microarray o RNA-seq, han revelado dos componentes sobrepuestos de distribución de niveles de expresión génica: genes de alta y de baja expresión. Los genes de alta expresión participan activamente de las funcionales celulares, sin embargo, la relevancia funcional de los genes de baja abundancia ha sido puesta en duda [45]. Recientemente, un estudio proteómico y transcriptómico realizado en tres líneas celulares tumorales de distintos tipos de cáncer e histología (U-2 OS, osteosarcoma humano; U-251 MG, human glioma; A-431, carcinoma epidermoide), reveló que los transcritos expresados únicamente en un tipo celular y no en los otros dos, poseen una baja abundancia relativa, comparados con genes de expresión moderada o alta que se expresan de forma similar en las tres líneas celulares [45]. Esta característica sugiere que los genes de baja expresión pueden tener potencial como marcadores de

23 identidad y por lo tanto, pueden ser empleados como biomarcadores para la identificación, por ejemplo, de células tumorales. La baja expresión de muchos transcritos implica que hay diferencias transcriptómicas sustanciales entre las células, incluso en aquellas cultivadas clonalmente, sugiriendo que cada célula posee una firma transcriptómica individual, si es que no única. Esta característica reta el concepto de que exista un transcriptoma individual y estable mediante el cual se pueda caracterizar una célula. La convergencia de conclusiones obtenidas por estudios recientes como la caracterización del transcriptóma de célula única [46] y un modelo transcripcional dinámico y de modificación veloz [47], permiten inferir que el transcriptoma humano posee una ontogenia altamente específica, con identidad posicional, dinamismo, plasticidad y diversidad [48].

1.2.2 El transcriptoma del cáncer

La secuencia y estructura del genoma cáncer, junto a sus variaciones epigenéticas, influyen en varios fenotipos mesurables dentro los cuales encontramos la progresión tumoral, la metástasis, la respuesta a drogas y la sobrevivencia de los pacientes. Estas variaciones genómicas y su influencia en el tumor pueden ser conducidas a través de la alteración de vías de señalización oncogénicas, que a su vez pueden ser evaluadas a través del análisis de sus perfiles de expresión génica. En el ámbito clínico, el uso de los perfiles de expresión génica aplicados al estudio del cáncer ha servido para identificar diversos subtipos moleculares [49-51] y para predecir sobrevivencia y recurrencia de la enfermedad [52-54]. Por otra parte, estas tecnologías han sido aplicadas además para disectar molecularmente las vías de señalización involucradas en la carcinogénesis y evaluar así su potencial como alternativa terapeútica [55-57]. El cáncer de mama representa una de las neoplasias más intensamente estudiadas a nivel de expresión génica desde los inicios del uso masivo de las tecnologías de microarray [50-52]. Luego de más de una década de estudios, se han

24 obtenido importantes conclusiones: los tumores que expresan el receptor de estrógeno representan una grupo molecularmente muy distinto de los tumores que no expresan el receptor, principalmente debido a la fuerte influencia hormonal en la transcripción; por otra parte, los niveles de expresión de genes relacionados con proliferación tumoral representan un factor pronóstico en tumores positivos para el receptor de estrógeno [58]. El estudio del transcriptoma del cáncer de mama ha contribuido enormemente a comprender la biología de este tumor, sin embargo, la medición de expresión génica no ha sido introducida a ningún nivel de la práctica clínica [58, 59].

1.2.3 El transcriptoma del adenocarcinoma ductal pancreático

El ADP es un tumor caracterizado por tener una firma genética característica [9], que le confiere un comportamiento agresivo y difícil de intervenir terapéuticamente. Además, este tumor cursa paralelamente con una intensa reacción estroma! caracterizada por una gran acumulación de material fibrótico e infiltración de células inflamatorias de origen inmune. La tecnología de microarray ha sido empleada para entender las complejidades de la heterogenidad del ADP y su regulación transcripcional. A nivel de expresión génica, las masas tumorales del ADP pueden ser divididas en al menos cuatro tipos de compartimentos con arquitecturas definidas: 1) epitelio neoplástico; 2) estroma juxtatumoral (estroma inmediatamente adyacentes al tumor); 3) panestroma (el resto del estroma no juxtatumoral) y 4) el tejido angio­ endotelial [60]. Estos estudios han identificado una serie de moléculas cuya expresión se encuentra mas elevada en tejidos tumorales en comparación a los tejidos no neoplásticos [61-63]. El ejemplo de la molécula mesotelina (mesothelin) destaca la aplicación versátil que puede tener el descubrimiento de un solo gen up-regulated. Mesotelina codifica para una proteína localizada en la superficie de la célula y que se encuentra up-regulated ubicuamente en la mayoría de los casos de ADP, observándose infrecuentemente en lesiones preneoplásicas no invasivas y totalmente ausente de los tejidos normales pancreáticos [64, 65]. El marcaje con anticuerpo anti­ mesotelina es útil como marcador auxiliar para el diagnóstico de ADP en muestras de

25 biopsia o citologías dudosas [66] y, debido a que mesotelina también se secreta, la detección de niveles séricos elevados se ha propuesto como un posible biomarcador de ADP [67]. Además, anticuerpos monoclonales humanizados anti-mesotelina e inmunotoxinas de Pseudomonas conjugadas a anti-mesotelina han sido desarrolladas como estrategia terapéuticas basada en esta proteína [64]. El caso de la mesotelina demuestra que los estudios de expresión génica tienen el potencial de impactar el manejo clínico actual del ADP. Luego de la masificación el uso de la tecnología de microarray, surgieron varios estudios transcriptómicos aplicados a diferentes aspectos de a biología del ADP: identificación de biomarcadores [62, 63, 68-71], estudio de la metástasis [72, 73], caracterización molecular de la reacción desmoplástica [60, 74] y de la pancreatitis crónica [75]. Si bien estos trabajos han contribuido enormemente a la comprensión de la biología del cáncer pancreático, a la fecha, pocos descubrimientos han sido transferidos al ámbito clínico en materia de diagnóstico molecular [14, 76]. En este contexto, el transcriptoma de baja expresión del ADP representa una fuente de moléculas con potencial de convertirse en biomarcadores. Una forma eficiente de evaluar un transcriptoma de baja abundancia es mediante la tecnología de hibridación sustractiva por supresión.

1.3 Microarray de expresión génica

Los microarrays son plataformas de alto rendimiento, diseñadas para medir la expresión de miles de genes simultáneamente. Esta tecnología se desarrolló rápidamente y su uso en investigación explotó entre los años 1999 y 2002, permitiendo asi la caracterización de los transcriptomas de múltiples especies, tanto a nivel fisiológico como en enfermedad, siendo el cáncer una de las áreas más extensamente estudiadas con esta tecnología [77]. Diferentes tipos de microarray han sido diseñados, pero todos se basan en el principio de detectar transcritos a través de hibridación de ácidos nucleicos. Una gran cantidad de sondas (spotted cDNAs u oligonucleótidos impresos), cada una correspondiente a una parte específica de un transcrito, son

26 adheridas a un substrato sólido, usualmente portaobjetos de vidrio. Los mRNAs extraídos de cada muestra experimental, controles o referencias son convertidas en cDNAs o cRNAs de hebra simple, marcadas con una molécula fluorescente y posteriormente hibridadas con las sondas complementarias impresas en el array (Figura VA) [78]. Las diferencias de expresión génica son calculadas en base a la cuantificación y análisis estadístico de la señal de fluorescencia emitida por los cDNAs o cRNAs marcados [79] . Los mícroarrays de dos colores (two-colour mícroarrays) permiten que dos muestras biológicas diferentes sean hibridadas en el mismo mícroarray, debido a que dos fluróforos distintos son usados para marcar cada muestra de cONA. Esta estrategia permite eliminar parcialmente las diferencias originadas en los distintos arrays, tales como la eficiencia de la hibridación y las variaciones en el tamaño del spot [80]. Dos muestras biológicas son hibridadas en un solo mícroarray de dos colores y varios mícroarrays son usualmente necesarios para analizar las muestras de un determinado experimento. Debido a esto, múltiples forma de combinaciones posibles emergen al momento de decidir como serán combinadas las muestras en estos mícroarrays. Estas combinaciones son denominadas "diseño de hibridación" o "diseño experimental", e influyen enormemente en la eficiencia y el poder de posteriores análisis estadísticos. Un diseño experimental popular es el denominado diseño de referencia (reference desígn) y representa un tipo de comparación indirecta (Figura 6). Este diseño emplea un cONA de referencia común en todos los arrays contra el que todas las otras muestras experimentales serán comparadas. Este tipo de diseño experimental permite ser extendido para incluir muestras adicionales en diferentes momentos, siempre y cuando la misma muestra de referencia esté disponible; es relativamente fácil de analizar estadísticamente y como las muestras son procesadas de la misma manera, reduce la posibilidad de errores en la manipulación [81, 82].

27 A TR+ 8 "'"' Extracción Marcaje ~ ~ ~::""1.. ...1\N 1\N IVV ..N\/~ EXP1 C1 mRNAen cONA Células estudio \ Tejidos Hibridación 00000 IV\/ /VV NV ... 0000 &EXP2 REF BC2 NVIVV IVV 0000 1 00000 Gen "'"" M:~c:je rep~i:do "' activado ""...... ,.1\N NV ../\IV NV Escanear 8 8 ~.,_ ...,. 1\N N\/ Análisis EXP3 C3 mRNA cONA Células referencia Tejidos Figura 6. Diseño de referencia. A) Las muestras experimentales y controles procesadas para extraer el RNA en estudio y sintetizar el cONA, el cual será marcado con un fluoróforo (ej: Alexa 647) e hibridado en el microarray junto a la muestra de referencia, que ha sido procesada de la misma manera pero conjugada a un fluoróforo distinto (ej: Alexa 555). B) Esquema del diseño de referencia. La muestra de referencia es hibridada competitivamente junto con cada muestra en estudio. REF= muestra de referencia; EXP1-3= replicados biológicos experimentales; C1-3: replicados biológicos de los controles. Figura adaptada con modificaciones desde referencias [78] y [81].

1.4 Hibridación sustractiva por supresión

La hibridación sustractiva por supresión ( Suppression subtractive hybridization; SSH, de su sigla en inglés) ha sido ampliamente usada para identificar transcritos (cONAs) que discriminan dos muestras complejas muy relacionadas entre sí [83]. La metodología de SSH normaliza (ecualiza) la abundancia de transcritos entre los transcriptomas en estudio, permitiendo enriquecer más de 1000 veces los transcritos de expresión diferencial, independientemente de su abundancia relativa original [84]. La metodología de SSH está basada globalmente en el efecto de supresión de la reacción en cada de la polimerasa (PCR; polimerase chain rection) [85, 86], combinando la normalización y sustracción en un solo paso. Primero, el cONA es sintetizado a partir del RNA extraído de las muestras en estudio. La población de cONA donde serán identificados los transcritos de interés se denomina tester y la población de cONA referencia utilizada para sustraer se denomina driver. Si la cantidad de RNA inicial no es suficiente, la tecnología de amplificación SMART (Switching Mechanism At 28 5'end of RNA Iranscript) es adecuada para amplificar cDNAs de alta calidad. A continuación, los cDNAs de doble cadena son sintetizados independientemente a partir de las muestras tester y driver, y posteriormente digeridas con enzimas de restricción que generen extremos romos, tales como Rsa 1 o Alu l. La muestra tester es luego dividida en dos porciones, cada una es ligada a dos diferentes adaptadores de doble cadena (1 y 2R). Los extremos de los adaptadores no están fosforilados, por lo tanto solo una hebra de cada adaptador es ligada covalentemente a los extremos 5'de los cDNAs. Los pasos moleculares que ocurren durante la SSH son descritos en la figura 7. En la primera hibridación, se añade un exceso de cONA driver a cada muestra tester (1 y 2R). Las muestran son posteriormente desnaturalizadas mediante calor y luego alineadas. Luego de este paso, en cada muestra se generan los tipos de moléculas A, B, C y D (Figura 7). Durante esta primera hibridación el grupo de moléculas de hebras simple de las muestras tester es normalizado, equiparando asi las concentraciones de los cDNAs de alta y baja abundancia. La normalización ocurre porque el paso de alineamiento genera homohíbridos (B) y heterohíbridos (C) más rápidamente para las moléculas de mayor abundancia, debido a la cinética de segundo orden de las hibridaciones de ácidos nucleicos. Las hebras menos abundantes permanecen en hebra simple. Controlando la duración de la hibridación, las hebras simples de los cDNAs de alta abundancia son reducidas al mismo nivel de aquellas menos abundantes, normalizando así la representatividad de los cDNAs de la muestra tester. Al mismo tiempo, la población de moléculas tipo A son significativamente enriquecidas para transcritos de expresión diferencial, debido a que los cDNAs comunes para las muestras tester y driver forman moléculas tipo C. Durante la segunda hibridación, las dos muestras de la primera hibridación son combinadas y alineadas en presencia de cONA driver adicional, desnaturalizado recientemente. Bajo estas condiciones, solo las hebras simples tester tipo A pueden reasociarse y formar nuevos híbridos tipo B, C y E (Figura 7). Los híbridos tipo E son moléculas tester de doble cadena con diferentes extremos, uno de los cuales corresponde al adaptador 1 y el otro al adaptador 2R. Posteriormente, cONA driver desnaturalizado es luego añadido para enriquecer la fracción E. Toda la población de moléculas es luego sometida a dos rondas de PCR para amplificar selectivamente las secuencias de expresión diferencial. Antes de la 29 primera PCR, los extremos de las hebras son rellenados, creando el sitio complementario para la unión de los primers necesarios para la amplificación. Las moléculas tipo A y D carecen de sitios de unión a primers y no pueden ser amplificados. Las moléculas tipo B forman estructuras de bucle plegadas sobre si mismas, suprimiendo así el alineamiento de primers y la posterior amplificación. Solo las moléculas tipo E son amplificados exponencialmente debido que tienen distintos tipos de adaptadores en sus extremos, y por lo tanto, diferentes sitios de alineamiento para primers. Estas moléculas son el principal producto enriquecido en las muestras sustraídas y representan transcritos de baja abundancia y de expresión diferencial en la muestra tester [87]. La metodología SSH puede ser acoplada a una plataforma de microarray para incrementar el rendimento y la cobertura de la identificación de transcritos [88-92]. Esta estrategia ha sido validada metodológicamente para identificar transcritos de relativa especificidad en tejidos tumorales tales como carcinoma hepatocelular [88-90], cáncer de mama [88, 93], cáncer nasofaríngeo [88, 92] y carcinoma pulmonar escamoso [91]. La estrategia SSH-microarray es de particular utilidad en el enriquecimiento de transcritos de baja abundancia, que como se ha descrito previamente, suelen tener una expresión específica de tipo celular [94] y también pueden ser empleados para identificar tipos específicos de tumores [88]. Sin embargo, un hecho importante de destacar, es que ningún estudio realizado en tumores utilizando la metodología de SSH-microarray ha desarrollado una evaluación exhaustiva de expresión de las respectivas proteínas codificadas por esos transcritos específicos. Esto es de importancia debido a que, como sabemos, solo un -40% de la abundancia de mRNAs tiene una correlación con sus respectivas proteínas [41] y estas poseen además un rango dinámico de concentración 900 veces superior a los mRNAs [44], convirtiendo a las proteínas en buenos candidatos para ser medidos en los fluidos y tejidos. Estudios previos han empleado la tecnología de hibridación sustractiva por supresión acoplada a microarray con el objetivo de identificar potenciales marcadores de cáncer [88-92]

30 e= Tester cONA+ Adaptador 1 Driver cONA Tester cONA+ Adaptador 2R 1 1 l Primera l Hibridación &:J A A----­B-1::~- e•~=~ ==- e Segunda Hibridación • Mezclar muestras - • Añadir Driver desnaturalizado - • Alinear (Anneal) l E

A -11111--- A s•ll---11- &:J--r= B ..:J-oEII -=:1-­ e e••-- &:J - ! A

No amplificación

-~~---->~ B o B.

e } Amplificación lineal

E Amplificación exponencial 3'~5' e!>

Figura 7. Fundamento molecular de la hibridación sustractiva por supres1on. Moléculas tipo E se forman sólo si la secuencia se encuentra up-regulated en el cONA tester. Cajas negras representan la parte exterior del adaptador 1 y 2R y cebador de correspondiente secuencia para la PCR primaria. Cajas rojas representan la parte interna del adaptador 1 y la correspondiente secuencia de cebador de PCR nested 1. Las cajas verdes representan la parte interna del adaptador 2R y la correspondiente secuencia de cebador de PCR nested 2R

31 Considerando los antecedentes presentados, el presente proyecto de investigación tuvo como propósito aplicar tecnologías de transcriptómica para identificar nuevos genes candidatos y evaluar su aplicación como potenciales herramientas de diagnóstico del adenocarcinoma ductal pancreático.

32 2. PRESENTACION DEL PROBLEMA

2.1 Justificación

El ADP es la causa de muerte de alrededor de 260,000 personas cada año a nivel mundial. Esta enfermedad no posee terapias efectivas y carece de metodologías de diagnóstico temprano. Actualemtne, es de urgencia identificar nuevas estrategias que permitan el desarrollo de metodologías de diagnóstico eficientes para el ADP. El presente trabajo de tesis tiene por objetivo investigar la biología del ADP en busca de moléculas que puedan servir como marcadores de la enfermedad, con el propósito final de desarrollar nuevas metodologías de diagnóstico. Basados en estos antecedentes se ha planteado la siguiente hipótesis y objetivos de identificación:

2.2 Hipótesis

La hipótesis de trabajo sostiene que mediante la aplicación de la metodología de hibridación sustractiva por supresión, acoplada a análisis de microarray y bioinformática, es posible identificar genes de expresión diferencial entre muestras neoplásicas y no neoplásicas de pacientes con ADP. Esta información permite la identificación de genes up-regulated asociados a carcinogénesis pancreática que podrían servir como potenciales marcadores génicos de la enfermedad.

33 2.3 Objetivo general de la tesis

El objetivo general es identificar nuevos marcadores de adenocarcinoma ductal pancreático (ADP), mediante análisis del transcriptoma de alta y baja abundancia en tejido neoplásico y no neoplásico, provenientes de pacientes con ADP avanzando. Abordamos esta pregunta mediante dos estrategias de genómica funcional: hibridación sustractiva supresiva (SSH) acoplada a detección por microarray.

2.4 Objetivos específicos

Determinación de genes de expresión diferencial en tejidos de adenocarcinoma ductal pancreático y tejidos no neoplásicos de páncreas mediante tecnología de microarray (estrategia directa). Determinación de genes de expresión diferencial en tejidos de adenocarcinoma ductal pancreático y tejidos no neoplásicos de páncreas mediante hibridación sustractiva por supresión acoplada a tecnología de microarray (estrategia SSH). Comparación de los genes de expresión diferencial obtenidos por ambas estrategias. Enriquecimiento de ontologías de genes y pathways. Selección de genes candidatos. Identificación de potenciales marcadores moleculares mediante análisis inmunohistoquímico sistemático en tejidos de ADP empleando tissue microarrays.

34 3. MATERIAL Y MÉTODOS

3.1. Muestras clínicas

El análisis de expresión génica mediante microarray y SSH-microarray fue realizado empleando muestras del Banco Nacional de Tumores. Los tejidos de cáncer pancreático y su correspondiente tejido adyacente no neoplásico fueron colectados de 6 pacientes con diagnóstico histológico de adenocarcinoma ductal pancreático avanzado sometidos a resección quirúrgica sin terapia adyuvante previa a la cirugía. Las muestras fueron procesadas e incubadas en RNAiater (Ambion lnc, Austin Tx, USA) durante 12 horas a 4 oc. Posteriormente, las muestras fueron introducidas en criotubos, sometidas a congelación en isopentano a -50 oc por 2 min y almacenadas a -80 oc hasta su utilización. Se consideró como tiempo máximo aceptable 30 minutos desde la extracción del tejido hasta su congelación. Para la validación inmunohistoquímica se emplearon tissue microarrays {TMAs) de cáncer pancreático que contienen tejidos de 30 pacientes, incluyendo 2 cores de tejido tumoral por cada core de tejido no neoplásico del mismo paciente (AccuMax Array A307). El análisis de expresión en tejidos normales fue realizado usando un TMA Pantomic Normal Tissues MN0661, construido con 33 tejidos normales por duplicado (Pantomics lnc.) y donado amablemente por el Dr. Julio Celis.

3.2 Extracción de RNA

3.2.1 Extracción de RNA de tejidos tumorales y no neoplásicos

Para la extracción de RNA (Ribonucleic acid) total de los tejidos se utilizó el reactivo de Trizo!, (lnvitrogen, Carlsbad, CA, USA) siguiendo las instrucciones del proveedor. Los tejidos (1 00-150 mg) fueron pulverizados mediante congelación en nitrógeno líquido y homogenizados con 1 mi de reactivo Trizo!. La suspensión fue incubada durante 5 mina temperatura ambiente {TA) y centrifugada a 12.000 g durante 1O m in a 4 oc. Luego de la separación, la fase acuosa fue recuperada en un tubo 35 nuevo al cual se adicionaron 200 ~-ti de cloroformo. El tubo fue agitado vigorosamente e incubado en hielo durante 3 m in y centrifugado a 12000 g por 1O m in. El sobrenadante fue recuperado en un tubo y el RNA fue precipitado con 500 ~-ti de isopropanol. El tubo fue incubado por 10 min en hielo y centrifugado a 12000 g por 10 min. El RNA precipitado fue lavado una vez con 500 ~-ti etanol 75% y centrifugado a 7500 g por 5 min. Luego, el RNA precipitado fue secado durante 5 mina temperatura ambiente (TA) y resuspendido en 30- 50 ~-ti de agua desionizada estéril tratada con DEPC O, 1 %. A continuación, los RNAs fueron sometidos a digestión enzimática con DNAsa 1 (Ambion lnc, Austin Tx, USA) para degradar trazas de DNA genómico, seguido de una purificación mediante columnas de afinidad RNeasy (Qiagen, Hilden, Germany). La concentración y pureza del RNA extraído fue determinada por espectrofometría UV (NanoDrop Technologies, USA), seguida de una evaluación de la integridad del RNA mediante electroforesis en gel de agarosa. Finalmente, las muestras de RNA fueron almacenadas a -80 oc hasta su utilización. El RNA proveniente de tejido normal pancreático fue obtenido comercialmente (Ciontech, Palo Alto, CA, USA).

3.2.2 Determinación de la integridad del RNA

Para determinar la integridad del ARN, se realizó una electroforesis horizontal en gel de agarosa al 1.5 % teñido con bromuro de etidio (0,5 ¡Jg 1 mi) con cada una de las muestras extraídas. Durante este procedimiento, 500 ng de cada muestra de RNA fue nivelada a un volumen total de 5 !JI y mezclada con 1 !JI de buffer azul celeste 6X (Winkler, Santiago, Chile). Las muestras fueron cargadas en el gel y la electroforesis se realizó en buffer TAE 1X (Tris-Acetato-EDTA) a 50 voltios durante 30 min. Para visualizar el RNA, el gel fue expuesto a luz ultravioleta en un transluminador (Aipha lnnotech Corporation, CA, USA).

36 3.3 Hibridación supresiva sustractiva (SSH)

La hibridación sustractiva supresiva requiere como material inicial un cONA de alta calidad, sintetizado a partir de las muestras de RNA extraídas de los casos tumorales y sus respectivos tejidos no neoplásicos. Para la síntesis de cONA se empleó el kit Super SMART PCR cONA Synthesis (Ciontech, Terra Ave, CA, USA), siguiendo las indicaciones del manufacturador. Posteriormente, los procedimientos de la SSH fueron realizados empleando el kit PCR-Select cONA Subtraction (Ciontech, Terra Ave, CA, USA), con algunas modificaciones.

3.3.1 Síntesis de cONA mediante tecnología SMART

Los métodos comúnmente usados para la síntesis de cONA se basan en la habilidad de la enzima transcriptasa reversa (TR) de transcribir una hebra simple de cONA a partir de mRNA en una sola reacción. Sin embargo, debido a que la TR no siempre transcribe la secuencia completa del mRNA, los extremos 5'de los genes tienden a estar pobremente representados en las poblaciones de cONAs. Este es a menudo el caso para los mRNAs largos, especialmente si la única estrategia de cebado (priming) es el uso de primers oligo(dT) o si el mRNA tiene estructuras secundarias persistentes. En ausencia de degradación de RNA, las moléculas de cONA truncadas se deben a la tendencia de la TR a detenerse antes de que se complete la transcripción. La tecnología SMART fue desarrollada con el propósito de enriquecer preferentemente cONAs de longitud completa. La síntesis de cONA mediante SMART emplea nanogramos de RNA inicial. Como primer de la primera reacción se emplea un oligo(dT) modificado (3' SMART COS Primer 11 A). Cuando la TR alcanza el extremo 5', la actividad de transferasa terminal de la enzima añade unos pocos nucleótidos adicionales, principalmente desoxicitosinas, al extremo 3'de la hebra de cONA. El oligonucleótido SMART, que posee una secuencia oligo(G) en su extremo 3', se une complementariamente a la secuencia de desoxicitosinas sintetizada previamente, creando un templado extendido.

37 A continuación, la TR cambia de templado y continua extendiendo hasta el final del templado. El cONA de longitud completa y de hebra simple recién sintetizado, contiene la secuencia completa del extremo 5'del mRNA, así como también, las secuencias que son complementarias al oligonucleótido SMART. La secuencia de anclaje SMART y la secuencia poli(A) sirven como sitios de cebado para la amplificación de extremo a extremo del cONA [95] (Figura 8). El procedimiento realizado se describe a continuación: a) Síntesis de la primera hebra de cONA

Por cada reacción se utilizó 1 IJg de RNA de cada muestra en un volumen de 50 IJI. El RNA fue incubado con 71JI de 3' SMART COS Primer IIA (12 IJM) y 7 IJI SMART IIA oligonucleotide (12 IJM) a 65 °C por 2 min. Luego se redujo la temperatura a 46 °C y se adicionó a la reacción la siguiente mezcla: 20 IJI de 5X First-Strand Buffer, 2 IJI OTT (1 00 mM), 1O IJI de 50X dNTP (1 O mM), 2,5 IJI RNaseOUT (40 U 1 IJI), 2 IJI SuperScript 111 Reverse Transcriptase 200 U 1 IJI y 5,5 IJI de agua desionizada estéril (AOESI). La reacción fue mezclada gentilmente e incubada a 46 oc por 90 min y luego a 70 oc por 15 min en un termociclador (Thermal Cycler 2720, Applied Biosystem). Como control de síntesis se utilizó 1O ng de RNA de placenta humana. La purificación del cONA fue realizada utilizando el kit NucleoSpin Extract 11 (Macherey-Nagel, Oüren, Alemania). A los 100 IJI de cada reacción de síntesis fueron añadidos 212 IJI de Buffer NT y se mezclaron con la pipeta. La solución fue traspasada a una columna NucleoSpin Extract 11 (Macherey-Nagel, Oüren, Alemania) y centrifugada a 14.000 rpm por 1 min y así descartar el eluído. Luego, la columna fue lavada con 600 IJI de Wash Buffer NT3 y centrifugada a 14.000 rpm por 1 min, descartando nuevamente el eluído. Para remover cualquier residuo de buffer la columna fue centrifugada a 14.000 rpm por 2 min. Posteriormente, la columna fue puesta en un tubo plástico nuevo y el cONA fue eluído con 50 IJI de agua Milli-Q, incubándose por 2 min a TA y centrifugando a 14.000 rpm por 1 min. Este paso fue repetido nuevamente con 35 IJI de agua Milli- Q, obteniendo así un volumen total de aproximadamente 85 IJI. 38 poli(A) RNA 0 5' · ... . ;0G poli(~)3'. primer CDS Smtes1s de pnmera oligonucleótido Hebra mediante TR SMARTIIA ! 0 5' poli(A} 3' ;0G Adición de dC adicionales ! Un solo 0 5'. poli{A) 3' paso ;0G CCC Cambio de templado ! y extensión mediante TR -GGG poli(A) 3' -ccc Amplificación de cONA ! mediante LD-PCR con primer PCR

cONA de doble cadena Figura 8. Fundamento molecular de la amplificación de cONA mediante tecnología SMART. La figura es descrita en el texto (punto 3.5.1 ). b) Amplificación del cONA por reacción de polimerasa en cadena (PCR).

Para identificar el número de ciclos óptimos necesarios para comenzar con la SSH , el cONA fue amplificado mediante PCR de larga distancia (LD-PCR) con 15, 18, 21, 24, y 27 ciclos separadamente y el producto de PCR fue chequeado mediante electroforesis en gel de agarosa 1, 2 % [96]. Para cada reacción el cONA fue mezclado con 172 1-11 de ADESI, 30 !JI de 10X Advantage 2 PCR Buffer, 6 !JI de 50X dNTP (10 mM), 6 1-11 de 5' PCR Primer 11 A (12 1-JM), 6 1-11 50X Advantage 2 Polymerase Mix. El volumen total de la PCR fue de 300 !JI, el que se dividió en 3 alícuotas de 100 !JI cada una (etiquetadas A, By C). Las tres alícuotas fueron sometidas al siguiente ciclo termal: 95 °C por 1 min y 15 ciclos de 95 °C por 5 seg, 65 oc por 5 seg y 68 °C por 6 min. Para determinar el ciclo óptimo de la reacción, a partir del tubo C se obtuvo una alícuota de 30 1-11 al cual se denominó tubo experimental, el cual fue incubado 15 ciclos adicionales de PCR, extrayendo 5 !JI de la reacción cada 3 ciclos. Todas las alícuotas obtenidas para cada caso fueron chequeadas por electroforesis en gel de agarosa al

39 1,2 % y se seleccionó como óptimo el ciclo anterior a la saturación del PCR. Una vez determinado el ciclo óptimo, se procedió a realizar los ciclos faltantes a los tubos A, B y C. Finalmente, se detuvo la reacción con 2 ¡JI de EOTA 0,5 M y se chequeó la amplificación de cada tubo mediante electroforesis en gel de agarosa 1,2%. Luego de la reacción de PCR, los tubos A, B y C fueron mezclados y purificados por extracción con igual volumen de fenol: cloroformo: alcohol isoamílico (25:24: 1 ). La mezcla fue agitada vigorosamente y centrifugada a 14.000 rpm por 1 O min. Luego, con la finalidad de concentrar el cONA, la fase superior acuosa fue traspasada a un nuevo tubo, mezclada con 700 ¡JI de n-butano! y se centrifugada a 14.000 rpm durante 1 min. Este último paso fue necesario repetirlo con 100 ¡JI de n-butano! hasta conseguir un volumen final de 40-70 ¡JI. Como última etapa de la purificación, se traspasó todo el cONA concentrado a una columna Chroma Spin - 1000 (Ciontech, Terra Ave, CA, USA) y se adicionó buffer TNE 1X hasta completar un volumen final de 320 ¡JI. Finalmente, se recuperó el cONA de la columna por centrifugación a 700 g por 3 min. Adicionalmente, cONA residual fue recuperado de la columna por elución con 75 ¡JI de TNE 1X.

e) Digestión con Rsal

Esta etapa fue realizada para generar, fragmentos de cONA más cortos y con extremos romos que son necesarios para la posterior ligación de adaptadores y sustracción. Antes de proceder con la digestión con Rsa 1, se guardó una alícuota de 1O ¡JI de cONA purificado para chequear el efecto de la digestión. La digestión fue realizada utilizando la fracción purificada de cONA (-320 !JI), 36 ¡JI de buffer de digestión Rsal 1 OX y 1,5 ¡JI de enzima Rsal (1 O U /¡JI). La reacción fue incubada a 3rC por 4 horas. Para confirmar la digestión, se realizó una electroforesis en gel agarosa al 1,2% de 1O ¡JI de cONA no cortados y 1O ¡JI de cONA digerido con Rsal (Figura xxx). Para terminar la reacción se adicionó al tubo 8 ¡JI de 0,5 M de EOTA. El cONA digerido posteriormente fue purificado, para esto se utilizó el kit E.Z.N.A. Cycle-Pure (Omega bio-tek, Norcross, GA, USA). A cada reacción le fueron adicionados 6 volúmenes de buffer CP, mezclando la solución con pipeta y la totalidad

40 de volumen fue transferido a una columna HiBind DNA. Posteriormente, fue centrifugado a 10.000 g por 1 m in a TA y se descartó el eluído. Luego, la columna fue lavada dos veces con 700 ¡JI y 500 ¡JI de DNA Wash Buffer respectivamente y centrifugación a 10.000 g por 1 m in. Finalmente, el cONA fue recuperado desde la desde la columna con 30 ¡JI de agua MilliQ y centrifugación a alta velocidad.

3.3.2 Sustracción génica mediante PCR-Select

Este método permite la detección de secuencias génicas diferencialmente expresadas en una población y no en otra. Aunque existe diferentes variaciones de este método, la teoría básica es simple: Primero ambas poblaciones de RNA son convertidas a cONA. El cONA que contiene los transcritos de interés se denomina tester y la muestra de referencia, driver. Tester y driver son hibridados y las secuencias híbridas son removidas. Los fragmentos de cONA del tester que no hibridan, contienen genes potencialmente presentes en el tester y no en el driver (Figura 7). La metodología SSH fue realizada con el protocolo del kit de Clontech con algunas modificaciones. Durante la PCR secundaria el partidor PCR Nested primer 1 fue reemplazado por la secuencia 5'- CTAATACGACTCACTATAGGGCTCGAGCGGCC-3' que posee un sitio promotor T7 que permite la transcripción in Vitro del amplicón sustractivo, para la posterior hibridación en el microarray. El procedimiento realizado para la SSH se describe a continuación:

a) Ligación de adaptadores

Cada muestra tester es separada en dos alícuotas de 120 ng y sometidas a una ligación independiente con los adaptadores 1 y 2R, resultando dos poblaciones de cONA tester. Como control de cONA tester no sustraído, en un tubo nuevo fueron mezclados 2 ¡JI de Tester 1 y 2 ¡JI de Tester 2R. Posteriormente, las reacciones de ligación fueron incubadas a 16 oc durante 14 h. Finalizada la incubación, la reacción es detenida añadiendo 1 ¡JI de EDTA/Giicógeno a cada tubo e incubada durante 5 m in

41 a 72 oc. Posteriormente, cada cONA tester y cONA control fue diluido 1:200 en agua para la determinación de eficiencia de la ligación de adaptadores mediante PCR, empleando los partidores PCR primer 1 y 3' GAPDH para amplificar fragmentos desde cDNAs correctamente ligados y los partidores GAPDH 3'y GAPDH 5', para determinar los niveles totales del cONA de GAPDH. Para completar la extensión de los adaptadores la reacción fue incubada a 75 oc por 5 min, seguida de un programa de termociclado de 1 ciclo a 94 oc por 30 seg y 25 ciclos de 94 oc por 1O seg, 65 oc por 30 seg y 68 oc por 150 seg. Finalmente, 5 !JI de cada tubo fueron analizados en gel de agarosa 1,2%. La eficiencia de la reacción se relaciona presencia de la banda de 750 pb con los primer G3PDH 3' y PCR primer 1, que debe ser de una intensidad similar a la banda de 500 pb obtenida con ambos primer para G3PDH.

b) Primera y segunda hibridación

Durante la primera hibridación fue mezclada separadamente una pequeña cantidad de cONA tester ligado con cada adaptador y un exceso de cONA driver. Luego, se añadieron 2 gotas de aceite mineral a cada tubo, fueron centrifugados brevemente, incubados a 98 oc por 90 seg y finalmente a 68 oc por 8 horas. En la segunda hibridación, las dos muestras de la primera hibridación se mezclaron e hibridaron con un exceso de cONA driver desnaturalizado. Durante esta reacción 1 !JI de cONA driver digerido (300 ng/IJI) fue diluído, con 1 1-11 de buffer de hibridación 4X y 2 1-11 de agua estéril. Luego, 1 !JI de esta dilución fue desnaturalizado a 98 oc por 90 seg y mezclado simultáneamente con las muestras de hibridación 1 y 2R. Esta reacción fue incubada a 68 oc por 12 h. Durante esta etapa se enriquecerán los genes diferencialmente expresados mediante la formación de moléculas hibridas que contendrán diferentes adaptadores en cada extremo.

e) Amplificación por PCR

Luego de la segunda hibridación, los cONA diferencialmente expresados son amplificados selectivamente mediante dos amplificaciones por PCR. La primera

42 amplificación de las secuencias tester específicas se realizó utilizando el partidor PCR 1. Posteriormente, para la PCR secundaria (nested) se emplearon los partidores PCR nested primer 1 modificado y el PCR nested 2R para reducir los niveles de amplificación inespecífica y enriquecer las secuencias diferencialmente expresadas. Las condiciones del PCR primario fueron las siguientes: una incubación de 75°C por 5 m in, seguida por un total de 27 ciclos a 94 oc por 30 seg, 66 oc por 30 seg y 72oC por 90 seg. Para la PCR secundaria se utilizaron las mismas condiciones de termociclado, pero solo durante 12 ciclos a una temperatura de annealing de 68° C. Finalmente, ambos productos de PCR fueron chequeados en un gel de agarosa al 1,2%.

d) Evaluación de la eficiencia de la sustracción.

La eficiencia de la sustracción fue analiada mediante el monitoreo por PCR cuantitativa en tiempo real de genes controles en muestras sustraídas y no sustraídas. La reacción de PCR fue preparada con el reactivo Brillant 111 de Stratagene y cargada con 2 ng de amplicón sustractivo y amplificada en un equipo Stratagene Mx3000p (Stratagene, Cedar Creek, TX, USA) con el siguiente programa de termociclado: 1 ciclo de 3 min a 95° C; 55 ciclos de 1O seg a 95° C, 15 seg a 60° C y 15 seg a 72. El retraso de la amplificación en más de cinco ciclos amplificación de la muestra sustraída comparada con la muestra sin sustraer es considerado como eficiente.

3.4 Amplificación y marcaje de las muestras

La amplificación y marcajde las muestras se realizó mediante transcripción In Vitro empleando kit comercial SuperScript lndirect RNA Amplification System (lnvitrogen, Carlsbad, CA, USA). El fundamento molecular se describe en la figura 9.

43 ~ - . mRNA ... --;_ • • 1 • -,..-,... • ...... ~ AAAAAAA TTTTTTITT-Promotor T7 ¡Transcriptasa Reversa Síntesis primera hebra cONA 1 hebracDNAmRNA:::::::::::::::::::::::::::::::AAAAAAA TTTTTTITT-PromotorT? !Rnasa HE. coli •

TTTTTTITT-P ! cONA doble hebra 12 hebra cDNA- AAAAAAA-PromotorT7 1 hebra cDNA TTTTTTITT-PromotorT7 !NTPs + Aminoallyl Uracilos + RNA Polimerasa T7 Amplificación de mRNA: Transcripción In Vitro mediada porT7 .:::::::::::::::~uuuuuuuuuuuuuuuu aRNA • aminoallyl ...... ~ ...... uuuuuuuu ...... uuuuuuuu amplificado

Figura 9. Transcripción In Vitro conducida por promotor T7. La primera hebra de cONA es sintetizada a partir de un mRNA, empleado un primer politimina-promotor T7 y la enzima SuperScript 11/. La segunda hebra es sintetizada con una polimerasa de E. co/i que emplea como primers pequeñas secuencias de RNA, resultantes de la degradación incompleta producida por la RNasa H de E. coli. La enzima ligasa de E. coli une los puentes restantes entre los nucleótidos. La doble hebra de cONA es incubada con nucleótidos normales y uracilos modificados con cadenas alifáticas (Uracilo aminoal/yl) para la amplificación de aRNA. A continuación, el aRNA generado es conjugado covalentemente con los fluoróforos correspondientes.

El protocolo es descrito a continuación:

a) Síntesis de la primera hebra de cONA

La reacción para la síntesis de la primera hebra de cONA se preparó a partir de 1 IJg de RNA, en el caso de estrategia Directa, y 500 ng de cONA sustraído en el caso de la estrategia SSG. A cada reacción se le añadió 1 J.JI de primer T7-oligo(dT) completando a un volumen de 1O 1-11 con agua desionizada estéril. La solución fue

44 incubada a 70 oc por 1O m in y luego en hielo por 1 m in. Terminada la incubación, fue adicionada a la reacción una mezcla con 4 IJI de 5X First-Strand Buffer, 2 IJI de OTT O, 1 M, 1 IJI de dNTP mix 1O mM, 1 IJI de RNaseOUT (40 U 1 IJI) y 2 IJI de SuperScript 111 RT (200 U 1 IJI). Posteriormente, la reacción fue incubada a 46 oc por 2 horas y a 70 oc por 10 min.

b) Síntesis de la segunda hebra de cONA y purificación

Luego de terminada la síntesis de la primera hebra, los 20 IJI de la reacción fueron adicionados a una mezcla constituida por 91 IJI de AOESI-OEPC, 30 IJI de 5X

Second-Strand Buffer, 3 IJI de dNTP mix 1O mM, 4 IJI de E. coli DNA Polymerase 1 (1 O U/ul), 1 IJI de E. coli ONA Ligase (1 O U 1 IJI) y 1 IJI E. coli RNase H (2 U 1 IJI). Gentilmente, se homogeinizó la solución con la pipeta y se incubó a 16 oc por 2 horas (Figura 8). Finalmente, la reacción fue colocada en hielo. Para la purificación del cONA, los 150 IJI de la reacción fueron mezclados con 500 IJI de cONA Loading Buffer y transferidos a la columna Low-Eiution Volume Spin Catridge. El eluído fue descartado por centrifugación a 6.000 g a TA por 1 min. Posteriormente, la columna fue lavada con 700 IJI de cONA Wash-Buffer y centrifugada a 6.000 g por 1 min a TA. Para eliminar el buffer residual la columna fue centrifugada nuevamente a 6.000 g por 2 min. Finalmente, se agregaron 24 IJI de AOESI-OEPC al centro de la columna incubándose por 2 mina T.A. El cONA fue eluído por centrifugación a 10.000 g por 1 mina TA.

e) Transcripción In Vitro y purificación de aRNA.

Para proceder a la transcripción in vitro del cONA de doble cadena, al cONA eluído en el paso anterior se le adicionan los siguientes componentes: 1,5 IJI de ATP (100 mM), 1,51JI de CTP (100 mM), 1,51JI de GTP (100 mM), 0,751JI de UTP (100 mM), 2 ¡.JI de aa-UTP (50 mM), 4 IJI de 1OX T7 Reaction Buffer y 7 IJI de T7 Enzyme Mix. La solución fue mezclada e incubada a 37 °C por 14 horas aproximadamente. Terminada la incubación se eliminaron las trazas de cONA mediante la adición de 2 IJI de ONase 1 (1 U 1 IJI) e incubación a 37 °C por 30 min. Para la purificación de aRNA, se adicionó a

45 la reacción 160 IJI de aRNA Binding Buffer y 100 IJI de Etanol 100%. La solución fue mezclada con la pipeta y luego transferida a la columna Purelink Spin Column. El eluído fue descartado por centrifugación a 12.000 gaTA por 15 seg. La columna fue lavada dos veces con 500 j.JI de aRNA Wash-Buffer y centrifugada a 12.000 g por 15 seg a T.A. El buffer residual fue eliminado de la columna por centrifugación a 12.000 g durante 2 m in. Finalmente, se agregó 100 IJI de ADESI-DEPC al centro de la columna e incubó por 1 mina T.A. El aRNA fue eluído por centrifugación a 12.000 g por 2 mina TA.

d) Cuantificación y marcaje del aRNA

Una vez terminada la amplificación, el aRNA fue cuantificado y se procedió al marcaje con el fluoróforo correspondiente. El aRNA proveniente de las muestras normales y tumorales sustraídas y sin sustraer fue marcado con Alexa Fluor 647, mientras que el aRNA de la referencia fue marcado con Alexa 555 (lnvitrogen, Carlsbad, CA, USA). Para realizar la incorporación del fluoróforo se redujo el volumen inicial de 5 j.Jg de aRNA a 3 IJI utilizando una centrífuga al vacío (Speed vac, Savant, NY, USA). Una vez alcanzado el volumen de reacción, el fluoróforo correspondiente fue resuspendido en 46 j.JI de DMSO (volumen para 4 reacciones). Posteriormente, el aRNA concentrado fue combinado con 8 j.JI de 2X Coupling Buffer y 11 IJI del fluoróforo resuspendido, la mezcla fue mezclada con vórtex, centrifugada brevemente e incubada por 30 min aTA en oscuridad. Finalmente, el aRNA marcado fue purificado como se describió anteriormente. Luego de la purificación del aRNA, se determinó la eficiencia de incorporación de picomoles de fluoróforo empleando para ello un espectrofotómetro Nanodrop ND 1000 que mide simultáneamente la cantidad de RNA y el fluoróforo incorporado.

46 3.5 Hibridación de los microarrays

Los vidrios utilizados para la hibridación fueron 48,5K The Human Exonic Evidence Based 0/igonuc/eotide (HEEBO) (Microarray lnc, Huntsville, USA). Este microarray está constituidos por un set 48.958 sondas de oligonucleótidos de 70 Mer, de las cuales 40.604 representan sondas específicas de genes y 4.650 sondas de control. Este microarray fue desarrollado por una colaboración entre la Universidad de Stanford e !Ilumina. Para la hibridación de los vidrios se siguieron las instrucciones del proveedor que se describen a continuación.

a) Prehibridación de los vidrios

La pre-hibridización en un paso que se realiza para aumentar la sensibilidad de la señal. Durante este procedimiento se incuban dos slides de microarray en un tubo de 50 mi con la solución de pre-hibridización (SSC 5X, SDS O, 1% y O, 1% BSA) a 50 oc durante 30 min. Luego, los slides son lavados 5 veces durante intervalos de 1 min con agua desionizada. Finalmente, los vidrios fueron secados mediante centrifugación. Los cubreobjetos de vidrio para los microarrays (25 x 60 mm) fueron lavados con etanol 70% y secados cuidadosamente. Por último, los bordes de los slides fueron marcados con un lápiz hidrófobo (Pen, Burlingame, CA, USA) para evitar la pérdida de la solución durante la hibridación.

b) Hibridación de los vidrios

La sonda de hibridación contuvo 60 pmoles de aRNA de referencia universal marcado con Alexa Fluor 555 y 60 pmoles de aRNA de la condición experimental marcado con Alexa Fluor 647 en un volumen final de 25 ¡.JI. Esta solución fue combinada con 25 ¡.JI del buffer de hibridación 2X (formamida desionizada 50%, SSC 1OX, SDS 0,2%, 0.02% DNA esperma de salmón) en un tubo ámbar de 1 ,5 mi y fue incubada a 95°C por 2 min. El volumen total de reacción fue distribuido sobre la superficie del arreglo, cuidando de no formar burbujas. Finalmente, los vidrios fueron 47 incubados en un horno de hibridación (lnSiide Out, Boekel Scientific, Pennsylvania, PA, USA) a 42°C durante 16 horas.

e) Lavados post-hibridación

Luego de hibridar la sonda es necesario remover el material excedente que no fue hibridado en la matriz, para esto todos los vidrios fueron lavados 4 veces en agitación, cuidando de no secar el vidrio entre lavados. El primer lavado se realizó durante 5 min utilizando la solución de lavado 1 (SSC 2X, SOS O, 1%) previamente calentada a 42 °C. El segundo lavado se realizó durante 5 min con solución de lavado 2 (SSC 0,1X, SOS 0,1%) aTA y los dos últimos lavados durante 1 min utilizando solución 3 (SSC 0,1X). Finalmente, los vidrios fueron secados por centrifugación y escaneados en un ScanArray Gx Microarray Scanner (Perkin Elmer, Waltham, MA).

3.6 Extracción de !a fluorescencia y análisis bioinformático

La intensidad de señal de los vidrios fue cuantificada con el Software SpotReader (Niles Scientific, USA). Para el grillado de la imagen se utilizó el patrón de "orange pack 1" y círculo elíptico adaptativo. Los artefactos dentro del slide fueron eliminados de forma manual. El análisis de los datos y la obtención de los genes diferenciales fueron realizados en colaboración con el Dr. Elmer Fernández del Grupo de Minería de Datos, Facultad de Ingeniería de la Universidad Católica de Córdoba (UCC), Argentina. El análisis de los chips se realizó mediante el programa estadístico "R" (www.r­ proyect.org) y el paquete de análisis LIMMA junto con aplicaciones propias de modelos mixtos. Los datos fueron procesados usando los métodos de sustracción de fondo standard y normexp [97]. El método standard es la sustracción clásica donde a cada spot se le resta su fondo directamente. El método normexp el fondo como una distribución mezcla entre normal y exponencial. Para la normalización dentro del vidrio se utilizó el método printipploess y para normalización de intensidades entre vidrios se utilizaron dos métodos de normalización: GQuantile y SCALE utilizando la función

48 mediana de sus desvíos absolutos (MAD) [98]. La normalización de GQuantile supone que dado que todos los vidrios están hibridados con la referencia universal marcada con fluoróforo verde (Aiexa 555), esta debería tener el mismo comportamiento en todos los slides, por lo tanto los datos son forzados a seguir este comportamiento y se genera una distribución única basada en los cuantiles del fluróforo Alexa 555, la cual es posteriormente utilizada sobre todos los vidrios. La normalización "SCALE" supone que dado que los valores a comparar tumor versus adyacente son los cocientes entre muestra/referencia universal, el logaritmo de este cociente (LogFC) debe tener un comportamiento "normal". Esta normalización tiende a hacer que las distribuciones del LogFC tengan una distribución simétrica semejante para cada chip. Para la obtención de genes de expresión diferencial entre las muestras tumorales y las adyacentes al tumor, se utilizó el siguiente modelo lineal: LogFCtgp = J-19 + f3t+ pp + f.tgp, donde J-19 es un efecto global del gen "g", f3t es el efecto del tumor (parámetro de interés), Pp,..,.N(O. v;) es el efecto aleatorio que modela el apareamiento de las muestras de un mismo paciente y Et:gp""·N(O,a:) es el error. El parámetro J-19 contiene la expresión global del gen más el efecto del tejido adyacente, por lo que el parámetro f3t nos indica si el tumor produce un efecto en exceso o en depreciación de la expresión. Para la identificación

Alexa 555 Alexa 647 Imagen compuesta de los genes que se diferencian con respecto a Tumor versus Adyacente se evalúa si e Figura 10. Imagen representativa de s/ide de microarray HEEBO hibridado. El escaneo del slide fue realizado con una resolución de 5 1Jm. La imagen compuesta es generada por la sobreposición de la imagen Alexa 555 y Alexa 647.

49 este parámetro f3t es estadísticamente distinto de cero. Por otro lado, dadas las características de este tipo de experimento y la baja cantidad de muestras también imponemos como condición que este valor supere un determinado umbral. El parámetro Jlg contiene la expresión global del gen más el efecto del tejido no neoplásico, por lo que el parámetro f3t nos indica si el tumor produce un efecto en exceso o en depreciación de la expresión. Para la identificación de los genes que se diferencian con respecto a Tumor versus Adyacente se evalúa si este parámetro f3teS estadísticamente distinto de cero. Por otro lado, dadas las características de este tipo de experimento y la baja cantidad de muestras también imponemos como condición que este valor supere un determinado umbral. La determinación de estos umbrales (tanto para la significancia estadística, como para el umbral del nivel de expresión) no es trivial y no existe un consenso específico, lo cual implica que su determinación sea característica de cada experimento y basada en la premisa de encontrar información biológicamente relevante.· La visualización de los datos mediante heatmaps fueron realizados con el programa MeV: multiExperimentis Viewer que es parte de TM4 Microarray Software Suite (http://www.tm4.org/mev/) [99]. Los análisis de enriquecimiento de funciones y de vías de señalización se realizaron utilizando la base de datos DAVID Bioinformatic Resources 6.7. La significancia estadística (valor p <0.05) es calculada con un test exacto de Fisher modificado (EASE score) [100].

3.7 Síntesis de cONA y PCR cuantitativo en tiempo real

Para la síntesis de cONA se utilizó el kit comercial AffinityScript qRT-PCR (Stratagene, Cedar Creek, TX, USA), siguiendo las especificaciones del fabricante. La transcripción reversa se realizó a partir de 1 ¡.Jg de RNA y como primers se usó una mezcla de oligo(dT) (170 ng) y random primer (30 ng) en un volumen total de 20 !JI. La combinación de estos dos tipos de primers permite enriquecer los extremos 3' y 5'del cONA, optimizando la calidad del templado. La reacción fue incubada durante 5 min a

50 25°C, 45 min a 42 oc y 5 min a 95°C. La reacción de qPCR fue realizada en un volumen total de 20 ¡JI, utilizando como templado 2 ¡JI de cONA previamente diluido 1/1 O y una mezcla de 0,5 ¡JI de partidores (1 O !JM), 1O ¡JI Brilliant 11 SYBR Green (Stratagene, Cedar Creek, TX, USA) y 7,5 ¡JI de agua desionizada. La reacción de qPCR se realizó en un termociclador de tiempo real Stratagene Mx3000p (Stratagene, Cedar Cree k, TX, USA), utilizando el siguiente ciclo termal: 1O m in a 95°C y 45 ciclos de 15 sega 95 oc, 15 sega 60 oc, 15 sega 72 oc. El programa AmplifX v1.5.4 fue empleado para el diseño de los primers y posteriormente contrastados con la herramienta web PrimerBLAST para confirmar especificidad (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). Las secuencias de los primers empleados son descritas en la tabla 2. Luego de las reacciones de PCR, los amplicones fueron analizados mediante gel de electroforesis en agarosa 1% para confirmar la amplificación de un fragmento único y corroborar peso molecular correspondiente. El fragmento único fue contrastado a su vez con la curva de melting obtenida durante la fase final de la qPCR. Las eficiencias de las reacciones de PCR fueron evaluadas empleando el programa LinReg con el objetivo de establecer que las eficiencias estuvieran dentro de rangos aceptables de 90 a 110%, sin embargo, tres pares de primers tuvieron porcentajes menores a 90% [101]. La identificación de genes normalizadores fue realizada usando el software GeNorm (http://medgen.ugent.be/-jvdesomp/genorm/) empleando los datos de todos los genes en estudio más un panel de genes normalizadores candidatos. El propósito de la normalización es remover las diferencias en las muestras (como cantidad o calidad de RNA) con el objetivo de identificar la variación real específica del gen. Los genes normalizadores o controles internos deben poseer una variación mínima. Para validar la estabilidad de los genes el algoritmo del software Genorm mide un valor M comparando el promedio de variación pareada entre cada gen y todos los demás. A menor valor M menor es la variabilidad, permitiendo así indentificar los mejores genes normalizadores [1 02]. Escogimos desde la literatura un panel de cinco genes candidatos [1 03, 104], los cuales fueron analizados en muestras pancreáticas provenientes de pacientes y líneas celulares. Los genes QARS y TBP fueron

51 identificados como los genes más estables y fueron usados como normalizadores para los posteriores experimentos de qRT-PCR (Figura 10).

Símbolo Sentido Secuencia 5'-->3' Amplicón Eficiencia QARS Forward ACCTGAACCTGGCATCACTACA 100 pb 101% Reverse CCAAGACGCTCAAACTGGAACT TBP Forward ACCAGGTGATGCCCTTCTGTAA 180 pb 99% Reverse CCTCAAACCAACTTGTCAACAGC Forward CAGCAACTGAACCTGCCATT CALU 66 pb 93.50% Reverse TTGGGCCAAGCTTTCCTAGA BHLHE40 Forward CACGATCAGCAATCAGGCATAGT 150 pb 102% Reverse CAACGGCATATGGAGTGTCCTT Forward TTCAGCTCTTGACCTGTCCC TOP2A 108 pb 93.50% Reverse CAAATGTTGTCCCCGAGTCT Forward CGTAAGCACATTCGGGCCATTT COL4A1 108pb 102% Reverse TCAGGCCTAGTGGTCCGAATCT Forward TTTCCCTGACAGCTGTGTACTC CTSK 109 pb 99% Reverse TGTGAAAATCTCCAGCCTGTACC Forward ACTTTTGGGAGCACGGACTGT SPARC 143 pb 99% Reverse TTTTGGCCTTCCTGGCTGAAAC Forward AAGGCTTGAACCAACCTACGGA FN1 128 pb 91.50% Reverse AAAGCCTAAGCACTGGCACAAC Forward TGTGTACCTGCCCAATTGTGAC IGFBPS 116 pb 97% Reverse GCAGCTTCATCCCGTACTTGT Forward TGCGTAGGTGAAAAGGCAGAAC PMEPA1 149 pb 91.50% Reverse AGCTTGTGCATTCAGACCAGAC Forward ACCCTTGAAATCCGTGAAACCG SBN02 170 pb 86.30% Reverse AGAGGTGAGCGGGCAATAACT CD55 Forward GCTTTGGAAGGCCGTACAAGTT 196 pb 99.50% Re verse AAGGCTGTTTGAGGGATGCAGA OASL Forward AGCAGGTGCTCCTTAGCCAAAT 146 pb 88.50% Reverse AGGATGAAGCTGTTGGGGTTGA KNG1 Forward ATTGACTTCGTGGCCAGGGAAA 168 pb 97.3 Re verse TCCCAGTGGTTGACAGTTGACA SYT12 Forward TTCCCCATCGCAAATGCAGTTC 191 pb 77% Re verse TTGGACCTCTGGTTCTTGCCAT MMP7 Forward TGGGACATTCCTCTGATCCT 165 pb 88% Reverse TGAATGGATGTTCTGCCTGA Forward TGCAGATGCAAGCCCAGAATCA LAMA3 109 pb 99.40% Reverse GCTCAGTCCCATCTCTGTTGCAAT SYNE2 Forward CTGATGCTCGCTGGAAAGAGTT 171 pb 96.80% Reverse GCCGCAGTGCTGTCTCTAAA PLEKHM1 Forward AAGTGACCTGGTCCTAAGCTGT 105 pb 94.50% Reverse GGCAAGTTCAGTGATGCTTTGG DPYSL4 Forward AGCCCTCATCCAGGGAAGTTTT 175 pb 90% Reverse TTCACATCAGAGGCAGGATGAC STAT1 Forward TCCGTGGCACTGCATACAATCT 184 pb 93.70% Reverse TGCCGAATTCCCAAAGGGCAAA

52 Tabla 2. Secuencia de primers usados para la verificación de ambas estrategias mediante qRT-PCR. Símbolo azul=genes normalizadores; Símbolo verde= genes de la estrategia Directa; Símbolo rojo=genes de a estrategia SSH.

Valores de estabilidad de expresión promedio de genes normalizadores

POL2R TFCP2 UBE2D

Genes menos estables Genes más estables 111( • Figura 10. Genes de mayor estabilidad identificados mediante el programa Genorm. QARS= glutaminyl-tRNA synthetase; TBP= TATA box binding protein; UBE2D2= ubiquitin-conjugating enzyme E2D 2; TFCP2= transcription factor CP2; POL2R= polymerase (RNA) 11 (DNA directed) polypeptide L, 7.6kDa.

El procesamiento de los datos de qPCR fue realizado con el software de Stratagene MxPro, siguiendo las intrucciones del proveedor. El análisis estadístico de los datos fue realizado con el Software GraphPad Prism version 5.0 (GraphPad Software lnc., San Diego, USA), utilizando análisis de tipo t-test de student y los valores de p value < 0.05 fueron considerados estadísticamente significativos.

3.8 lnmunohistoquímica

3.8.1. Anticuerpos

Los siguientes anticuerpos fueron empleados en este estudio: PLEKHM1 (Human Atlas Protein HPA021558; dilución 1:75); MMP7 (R&D Systems; dilución 1:500); STAT1 (Affinity antibodies; dilución 1:400); DPYSL4 (Abcam; dilución 1 :600); WNT9A (Human Atlas Protein HPA011223; dilución 1:25); BHLHE40 (Human Atlas Protein HPA028921; dilución 1:300), PMEPA1 (Abnova H00056937-M01; 1 :100), 53 CALU (Human Atlas Protein HPA006018; dilución 1:300). Adicionalmente las muestras fueron procesadas con anticuerpos anti-p53 (Dako clone D0-7; dilución 1:250) para detección de p53 mutado; anti-Ki67 (Dako clone MIB-1; dilución 1:200) para evaluar proliferación celular y anti-keratina 19 (Thermo scientific; dilución 1:1 000) para identificar células de origen epitelial. Las diluciones de trabajo de todos los anticuerpos fueron testeadas en controles positivos para establecer concentración óptima. Las tinciones inmunohistoquímicas de los anticuerpos anti-PMEPA 1, anti­ BHLHE40 y anti-CALU fueron realizados en colaboración con el Dr. lván Roa y la TM Soledad Lantadilla del Laboratorio de Patología y Biología Molecular de la Clínica Alemana de Santiago. Los tejidos tumorales y adyacentes fueron fijados por inmersión en formalina tamponada neutra al 10 % (pH 6.8-7.2), deshidratadas en diluciones seriadas de alcohol, clarificadas con xileno y embebidos en bloques de parafina. Luego, se cortaron secciones de 4 ¡Jm usando un microtomo de rotación (RM 2245, Leica Nussloch, Germany), las cuales fueron posteriormente montadas sobre un portaobjetos silanizado. Antes de proceder con la inmunohistoquímica, las secciones fueron incubadas a 60 oc durante 30 min, desparafinadas en xileno y rehidratados en diluciones seriadas de etanol. La recuperación antigénica se realizó en buffer citrato a 10 mM (pH 6.0) en una vaporera durante 10 min. Para eliminar la actividad peroxidasa endógena, los portaobjetos fueron incubados con una solución acuosa de peróxido de hidrogeno al 3% durante 1O m in a temperatura ambiente y lavados con buffer TBS­ Tween O, 1% (TBS-T). El bloqueo de cargas inespecíficas se realizó con una solución bloqueadora de proteínas durante 15 min a temperatura ambiente (DAKO, #X0909). Luego los cortes fueron incubados con 100 !JI de cada anticuerpo primario diluído adecuadamente en solución diluyente de anticuerpo (DAKO, #S3022) durante 40 min a temperatura ambiente en cámara húmeda. Luego de la incubación, los cortes fueron lavados con una buffer TBS-T e incubados con 50 !JI de anticuerpo secundario HRP polímero conjugado por 15 min a temperatura ambiente en cámara húmeda (Super Picture Polymer, lnvitrogen Cat. N°. 87-8963). Posteriormente, los cortes fueron lavados con TBS-T e incubados con 50 !JI de sustrato cromógeno (Nova Red, Vector Lab. # SK-4800), monitoreando la reacción bajo microscopio. Para diferenciar las estructuras celulares las secciones fueron contrateñidas con hematoxilina de Mayer y 54 alcalinizadas con solución de carbonato de litio saturada. Finalmente, fueron deshidratadas con diluciones de alcoholes, aclaradas en xileno y montadas. El análisis de los tissue arrays de los anticuerpos anti-PMEPA 1, anti-BHLHE40 y anti-CALU fue analizado semicuantitativamente asignando scores a la intensidad y celularidad. La intensidad de la tinción fue evaluada en una escala de 0-3 (O=negativo; 1=1eve; 2=moderada y 3=fuerte). El porcentaje de células con inmunoreactividad fue evaluada en una escala de 1-4 (1=0-25% de células positivas; 26-50% de células positivas; 51-75% de células positivas y 76-100% de células positivas). El staining score fue calculado como el múltiplo de la intensidad y a celularidad, generando una escala de 0-12. Las tinciones inmunohistoquímicas de los anticuerpos anti-PLEKHM1, anti­ WNT9A, anti-MMP7, anti-STAT1, anti-DPYSL4, anti-Ki67, anti-Keratin 19 y anti-p53 fueron realizadas en colaboración con el Dr. Julio Celis y la Dra. Teresa Cabezón del Group Cancer Proteomics del Danish Cancer Society Research Center, Copenhagen, Denmark. Cortes histológicos de 4 ¡.Jm fueron montados en slides Super Frost, incubados a 65°C durante 60 min, desparafinados, bloqueadas con peróxido de hidrógeno 1% durante 1O m in y rehidratados mediante sucesivas incubaciones en alcoholes graduados. La recuperación antigénica fue realizada incubando los slides en un buffer de Tris/EDTA pH 9.0 (10mM Tris, 1mM EDTA) y calentados en microondas a 750W durante 1O m in. Posteriormente, las muestras fueron incubadas en suero 1% en buffer TBS para bloquear tinción inespecífica. La presencia del antígeno fue evaluada incubando las muestras durante 45 min con el anticuerpo primario seguido de una detección con anticuerpo secundario conjugado a un complejo peroxidasa EnVision poly-HRP system (DAKO) durante 45 min. Luego las muestras fueron incubadas con DAB+Chromogen (DAKO) durante 15 min para el desarrollo del color y finalmente contrastadas con hematoxilina. El análisis de las tinciones inmunohistoquímicas para los anticuerpos anti­ PLEKHM1, anti-WNT9A, anti-MMP7 y anti-STAT1 fue realizado usando el sistema automático ACIS 111 (DAKO) para digitalizar y cuantificar la tinción IHQ. El software ACIS 111 permite reconocer de forma separada pixels de color café (tinción positiva) y pixels de color azul (contratinción de hematoxilina). Para el análisis de los TMAs

55 generamos un staining score en función de la intensidad de la tinción y el porcentaje de células que presentan inmunoreactividad dentro del core (staining score).

56 4. RESULTADOS

4.1 Diseño experimental

Las muestras de PDAC y sus respectivos tejidos no neoplásicos fueron analizadas mediante dos estrategias. La primera, denominada estrategia Directa, consiste en amplificar el RNA mediante transcripción In Vitre e hibridar el material directamente en el microarray. Esta estrategia busca evaluar el transcriptoma nativo sin pasos previos de intervención. La segunda estrategia, denominada SSH, consiste en aplicar la hibridación sustractiva por supresión a los tejidos tumorales y no neoplásicos (testers), empleando un tejido normal de páncreas como muestra sustractora (driver). Esta estrategia permite detectar genes diferenciales entre las muestras en estudio y enriquece transcritos de expresión específica. Estos genes no son detectados mediante la estrategia Directa de microarray [88, 90]. Tanto para la estregia Directa como para la estrategia SSH, las muestras tumorales y no neoplásicas fueron comparadas entre sí, empleando un diseño de referencia (Figura 6) [81]. Las muestras objetivo fueron hibridadas con el fluoróforo Alexa 647 que emite fluorescencia color rojo. De la misma manera, se obtuvo aRNA a partir de una muestra Referencia Universal marcada con el fluoroforo Alexa 555, que emite fluorescencia de color verde. Cada muestra tumoral y no neoplásica (en ambas estrategias) fue incubada competitivamente con la muestra Referencia Universal. De esta manera, la muestra de referencia se encuentra hibridada de manera común en todos los microarrays. En la figura 11 se describen las etapas del diseño experimental.

57 Seis tejidos de ADP y sus correspondientes tejidos adyacentes no neoplásticos 1 l Extracción de RNA y J análisis de integridad

1 1 [ Estrategia Directa J [ Estrategia SSH l t • Síntesis de cONA • Amplificación de cONA mediante tecnología SMART • Digestión con RSAI ...... _

1 TESTER 1 DRIVER • Síntesis cONA -- Tejido • Transcripción In vitro y marcaje [ Ligación de J normal adaptadores J + • Hibridaciones • PCR primario y anidado • Purificación y eficiencia de SSH • Hibridación en microarrnys HEEBO • Escaneo y extracción de la fluorescencia + • Transcripción In vitro y marcaje • Hibridación en microarrays HEEBO • Escaneo y extracción de la fluorescencia .. • Análisis bioinformático • Validación de datos mediante qRT-PCR • Análisis de ontología de genes y pathways

• Validación de genes candidatos a nivel de proteína en tejidos de ADP, tejidos no neoplásicos y tejidos normales mediante IHQ.

Figura 11. Diseño experimental. Seis tejidos de ADP y sus respectivos tejidos no neoplásicos fueron empleados como testers, mientras que un tejido normal de páncreas fue empleado como driver. SSH= Subtractive supressive hybridization; ADP=Adecarcinoma ductal pancreático; HEEBO= Human Exonic Evidence Based 0/igonuc/eotide; qRT-PCR= quantitative reverse transcription - polymerase chain reaction; IHQ= inmunohistoquímica.

58 4.2 Estrategia Directa

Para el análisis del transcriptoma de los tejidos tumorales y adyacentes fueron empleados microarrays de oligonucleótidos HEEBO provistos por Microarray lnc (http://www.microarrays.com/). Este microarray permite la medición de alrededor de 40.600 transcritos mediante sondas específicas. Luego de la extracción de RNA desde los tejidos tumorales y no neoplásicos, cada muestra fue sometida a una transcripción In Vitro con el objetivo de amplificar material suficiente para ser hibridado sobre los slides de microarray. La transcripción In Vitre mediada por promotor T7 es una de las metodologías más populares y eficientes, permitiendo obtener aRNA amplificado que conserva la abundancia relativa de transcritos observada en la muestra original [1 05]. Las muestras objetivo fueron posteriormente marcadas con el fluoróforo Alexa 647 (color rojo) e hibridado de forma comparativa frente a un mismo aRNA de referencia Universal (Ciontech) marcado con Alexa 555 (color verde). Los fluoróforos Alexa 555 y 647 son más fotoestables y presentan una mayor correlación de señal que los colorantes clásicos Cy3 y Cy5 [1 06]. La descripción de los resultados de la estrategia directa será abordada en capítulos posteriores en conjunto con los resultados de la estrategia SSH.

4.3 Estrategia SSH

La hibridación sustractiva por supresión (SSH) es una técnica descrita por Diatchenko et al, 1996 y que facilita la detección de genes diferencialmente expresados entre dos sistemas biológicos complejos [96]. Este procedimiento, consiste en dos pasos sucesivos de hibridación entre una muestra prueba que presenta los transcritos de interés (tester) y un exceso de RNA mensajero proveniente de una muestra control (driver), que conduce a la normalización de moléculas específicas en la muestra tester, con lo que se favorece el enriquecimiento de los transcritos de baja abundancia (Figura 7). Sin embargo, uno de los principales inconvenientes de esta técnica es que para la detección de los transcritos se requiere de la creación de una biblioteca sustractiva que conlleva muchas veces a la pérdida de información y con esto se ve limitada la sensibilidad de la técnica [90]. Por otro lado, la tecnología de 59 microarray ofrece una extraordinaria plataforma para la detección de una gran cantidad de transcritos, siendo una limitante importante la baja sensibilidad que presenta esta metodología para la detección de transcritos de bajo nivel expresión [1 07]. Por lo tanto, la combinación de ambas tecnologías sería una herramienta de alto rendimiento y sensibilidad para la detección de transcritos raros y de baja abundancia [90]. La estrategia SSH fue desarrollada en paralelo con la estrategia Directa, empleando los mismos seis tejidos tumorales con su respectivo tejido adyacente no neoplásico (Figura Diseño). Las muestras tester tumorales y no neoplásicas fueron sustraídas de forma independiente contra una muestra pool de RNA normal pancreático como driver. Debido a la baja cantidad de RNA extraída a partir de las muestras en estudio, amplificamos los RNA poli-A mediante la tecnología SMART, basada en la técnica de PCR (ver Material y Métodos 3.5.1 ). De esta manera fue obtenido el material suficiente para comenzar el protocolo de SSH. Para acoplar la última etapa de la SSH con la transcripción In Vitro (paso previo de la hibridación sobre el microarray), el partidor PCR nested primer 1 fue modificado mediante la adición de la secuencia del promotor T7. En la PCR secundaria (nested) se emplea el partidor PCR nested primer 1 modificado en conjunto con el PCR nested primer 2, para de esta forma amplificar mediante PCR las secuencias diferencialmente expresadas y a la vez, incorporar en cada transcrito el promotor necesario para la posterior síntesis de aRNA (Figura 12).

60 PCR Primaria

Promotorl7 Adaptador 1 S' -CTAATACGACTCACTATAGGGCTCGAGCGGCCGCCCGGGCAGGT-3' 1 1 5'-CTAATACGACTCACTATAGGGC-3' PCR primer 1 Promotorl7 Adaptador 2R S' -CTAATACGACTCACTATAGGGCAGCGTGGTCGCGGCCGAGGT-3'

PCR Secundaria (Nested}

Promotorl7 Adaptador 1 5'-CTAATACGACTCACTATAGGGCTCGAGCGGCCGCCCGGGCAGGT-3' 1 1 5'-CTAATACGACTCACTATAGGGCTCGAGCGGCC-3' Nested PCR primer 1 modificado

Promotorl7 Adaptador 2R 5'-CTAATACGACTCACTATAGGGCAGCGTGGTCGCGGCCGAGGT-3' 1 1 5'-AGCGTGGTCGCGGCCGAGGT-3' Nested PCR primer 2R

Figura 12. Primers empleados para la PCR primaria y secundaria de la SSH. PCR primer 1 es empleado para amplificar los transcritos enriquecidos durante la SSH. La PCR secundaria (nested) fue modificada mediante la adición de la secuencia del promotor T7 en el Nested PCR primer 1(originalmente ausente). De esta manera, el promotor es conservado en los amplicones y permite la posterior transcripción In vitre y marcaje de las muestras procesadas mediante la estrategia SSH.

61 4.3.1 Síntesis de cONA mediante tecnología Super SMART

Para obtener la cantidad de cONA necesaria para el primer paso de la SSH, las muestras fueron amplificadas mediante una variante de la PCR basada en la tecnología SMART (ver Material y Métodos 3.5.1 ). El principio de esta tecnología es la utilización de un primer oligo-dT modificado (3'SMART COS Primer 11 A) y de un oligonucleótido SMART para la síntesis de la primera cadena de cONA, los cuales posteriormente servirán como partidores universales para la amplificación del cONA por reacción de polimerasa en cadena de larga distancia (LO-PCR) [95]. En la figura 13A se observa se muestra una figura representativa de la determinación de ciclo óptimo de la amplificación SMART, donde la cantidad de cONA amplificado alcanza una meseta. Comúnmente, es posible observar un barrido entre 100-2000 pb a partir de los RNAs aislados de las muestras en estudio. El cONA de doble hebra fue purificado usando las columnas Chroma Spin - 1000 y sometido a una digestión enzimática con la enzima RSAI, la cual reconoce y cliva el sitio 5' ... GTtAC ... 3', generando moléculas con extremos romos, necesarios para la posterior ligación de los adaptadores. En la figura 138 se observa un ejemplo de digestión enzimática con RSAI, donde la muestra sin digerir (RSAI-) presenta cONAs de entre 0.5 a 1 .5 kb de peso, los cuales disminuyen su peso luego de la digestión (RSAI+).

4.3.2 Ligación de adaptadores

Luego de la digestión enzimática, cada muestra es dividida en dos alícuotas, las cuales serán ligadas cada una con un adaptador distinto, llamados 1 y 2R. La eficiencia de la ligación de adaptadores es verificada mediante PCR, utilizando como parámetro la detección de la ligación de los adaptadores en el gen GAPOH. Para este objetivo, se empleó el PCR Primer 1 (que une la secuencia adaptadora) y el partidor GAPOH 3' para detectar el transcrito correctamente ligado, mientras que los partidores 3'y 5' de GAPOH fueron usados para medir los niveles totales de GAPOH. Para 62 confirmar la correcta ligación de los adaptadores deberíamos observar la presencia de una banda de 750 pb, la cual no debe ser menos de cuatro veces la intensidad de la banda de 500 pb, que representa el nivel total de GAP OH en la muestra (Figura 14 ).

A Ciclos Ciclos J 17- 'O MP .lí 15, 18 •- - -~~-+1kb - ...... , . -~

Caso 56 Adyacente Caso 56 Tumor

B MP 'RSAI- RSA~+ 3kb+ - 1kb+-- -... -" m

Pan creas normal Figura 13. Análisis de la amplificación del cONA por PCR de larga distancia usando la tecnología Super SMART y digestión enzimática con RSAI. A) El ciclo óptimo fue determinado de forma independiente para cada muestra y consistió en un ciclo antes de la meseta de la PCR. Para el caso 56, el ciclo óptimo para la muestra adyacente y tumoral fue de 23 ciclos. B) Digestión enzimática del RNA normal de páncreas antes (RSA-) y después (RSA+). MP= marcador de peso molecular.

63 A 750 pb 1 Adaptador GAPDH, 500' pb B Adaptador 1 Adaptador 2R MP l l .. ,...---..---...,----• GAPDH ligado con !!!!!! • J.. adaptadoffis ~~~ ~:::- ---.- 1 1 • GAPDH total Caso 636 No neoplástico

Figura 14. Determinación de eficiencia de la ligación de adaptadores. A) El transcrito de GAPOH ligado debe generar un fragmento de 750 pb, mientras que el transcrito de GAPOH generará una banda de 500 pb. B) Ligación eficiente de la muestra 636 No neoplásica. MP= marcador de peso molecular.

4.3.3 Sustracción génica por PCR-Se/ect

Luego de ligación de adaptadores, las muestras son sometidas a dos rondas de hibridaciones en presencia de cONA driver desnaturalizado (Ver Material y Métodos 3.5.2b). Al término de la segunda hibridación, las muestras son posteriormente amplificadas mediante dos reacciones de PCR consecutivas. El PCR primario consiste en la amplificación de secuencias específicas del tester con el PCR Primer 1. El PCR secundario (nested) es realizado con los primers PCR nested 1 modificado y nested 2R y tiene como objetivo reducir los niveles de amplificación inespecífica y enriquecer las secuencias diferencialmente expresadas. A continuación, el amplicón sustractivo es purificado y sometido a transcripción in vitre y marcado con Alexa 647 para efectuar la posterior hibridación y detección con el microarray HEEBO (ver Material y Métodos 3.7 y 3.8). La eficiencia de la sustracción se determina mediante la comparación de la abundancia de cONA de genes control en las muestras sustraídas y sin sustraer (amplicón de la PCR secundaria). El protocolo convencional de análisis de eficiencia se 64 realiza midiendo semicuantitativamente los niveles de transcritos del gen GAPDH, donde el retardo en la amplificación en más de 5 ciclos es considerado una sustracción exitosa. Para medir la eficiencia de la sustracción reemplazamos el protocolo convencional de punto final por una reacción de PCR cuantitativa en tiempo real. Para la comparación entres muestras sustraídas (SSH) y no sustraídas (No SSH) las muestras fueron normalizadas por carga, añadiendo 2 ng de amplicón sustractivo a cada reacción de PCR. Las curvas de amplificación (Figura 15) muestran que el cONA de GAPDH fue sustraído eficientemente en 11 de 12 muestras.

65 Cidos Ciclos Ciclos : 56T : A ciclos• 6.34

..... NoSSH ...... SSH Figura 15. Análisis de eficiencia de la sustracción. Curvas de amplificación del cONA de GAPDH de las muestras sustraídas (SSH) y no sustraídas (No SSH) en cada caso. Las muestras sustraídas (círculos azules) presentan un retraso de más de cinco ciclos en su amplificación comparados con las muestras no sustraídas (cuadrados cafés), con excepción de la muestras 687 A. Cada muestra fue normalizada por carga con 2 ng de amplicón por reacción de PCR, empleando el mismo cyc/e threshold (Ct) para todos los análisis. T=tejido tumoral; A=tejido adyacente no neoplásico.

66 4.4 Integración de los perfiles de expresión génica analizados mediante microarray de las estrategias Directa y SSH

4.4.1 Secuencias de expresión diferencial entre tejidos tumorales y no neoplásicos, identificados mediante estrategia Directa y SSH

Luego de la hibridación y extracción de la fluorescencia de los slides de microarray de ambas estrategias (ver Materiales y Métodos 3. 7 y 3.8) se procedió con la sustracción del ruido (background), la normalización de las intensidades y un paso de filtración de datos. Las sondas que reportaron una baja intensidad y/o un valor de ruido superior a la intensidad del spot fueron retiradas del análisis. De esta forma se descartaron los datos que aportan al ruido global del análisis, logrando una considerable reducción de los falsos positivos dentro del grupo de genes diferenciales. El análisis estadístico de genes de expresión diferencial fue realizado utilizando el paquete LIMMA del entorno "R", junto con aplicaciones propias de modelos mixtos. Este tipo de modelos es apropiado para analizar datos pareados, por ejemplo, diferentes muestras o mediciones provenientes del mismo paciente. En cada estrategia fueron comparados los mismos grupos: Tejido tumoral y tejidos no neoplásicos adyacentes a la lesión tumoral. Para determinar la eficiencia de la sustracción a nivel de la matriz de microarray, escogimos 100 genes housekeeping, cuyos niveles de expresión fueron comparados entre ambas estrategias. En promedio se observa una disminución de señal en la estragia de SSH de un 20% en -49 genes, 50% en -37 genes y 80% en -14 genes (Tabla 2). Los genes que presentaron una reducción de señal mayor al 50% en el 75% (al menos 9 de 12) de las muestras son: NACA (11 genes); RPL 13A (9 genes); RPL29 (12 genes); TAF2 (9 genes); LDHA (9 genes); RPL 19 (10 genes); TKT (9 genes); FAU (10 genes); ZFP36L 1 (11 genes) y RPS11 (10 genes), de los cuales, 4 genes participan como componentes estructurales de ribosomas. Establecimos criterios de selección de secuencias diferenciales en función de su significancia estadística y fold change de expresión. Para la estrategia Directa

67 escogimos los secuencias diferenciales con valores p < 0.01 y con fold changes mayores a 0.4 y menores a -0.4, obteniendo un total de 736 secuencias, de las cuales 505 se encuentran up-regu/ated y 231 down-regulated en los tejidos de adenocarcinoma ductal. Para la estrategia SSH escogimos las secuencias diferenciales con valores p < 0.05 y con fold changes mayores a 0.35 y menores a - 0.35, obteniendo un total de 616 secuencias, de las cuales se encuentran 312 up­ regulated y 304 down-regu/ated en tejidos de adenocarcinoma ductal. La cantidad de secuencias obtenidas mediante estos criterios permiten realizar análisis de enriquecimiento de términos ontológicos para determinar que procesos, funciones y compatimentos celulares pueden ser identificados en ambas estrategias. Las variaciones de expresión génica de ambas estrategias son mostradas mediante heatmaps en las figuras 16A y 168.

68 A Estrategia Directa B Estrategia SSH No neoplásico ADP

-2 o 2 log2 fold change

-2 o 2 log2 fold change Figura 16. Heatmaps de los valores de expresión (fo/d changes) de los genes diferenciales entre tejidos tumorales y no neoplásicos identicados mediante ambas estrategias. A) 736 secuencias diferenciales identificados mediante la estrategia Directa. B) 616 secuencias diferenciales identificadas mediante la estrategia SSH. Valores expresados como log2 del fold change obtenido de la razón entre la intensidad de fluorescencia de canales (rojo/verde)

69 Tabla 2. Resumen de sustracción de genes 100 constitutivos exónicos en cada muestra.

Reducción 56T 56 A 271T 271A 558T 558A 617T 617A 636T 636A 687T 687A Promedio de señal 20% 42 39 37 53 54 60 64 59 58 42 49 42 49.9 50% 28 22 25 44 49 39 49 44 42 32 39 32 37.1 80% 10 5 9 14 16 21 26 19 13 18 16 6 14.4

4.4.2 Análisis de enriquecimiento de ontología de genes

Las listas de genes de expresión diferencial obtenidas mediante ambas estrategias fueron sometidas a la base de datos web DAVID Bioinformatics Resources 6.7. Esta herramienta permite la extracción de características de relevancia biológica asociada a largas listas de genes obtenidas mediante diversas plataformas de análisis, facilitando la exploración e interpretación de experimentos. El concepto de análisis de enriquecimiento se basa en que si los procesos biológicos son anormales en un estudio determinado, los genes desregulados pueden tener un mayor potencial (enriquecimiento) de ser seleccionados como grupos relevantes mediante un tecnología de screening de alto rendimiento, como la tecnología de microarray [108]. Cuando las listas de genes a comparar poseen distintos tamaños y diferente límites, el enriquecimiento absoluto de los valores p pueden variar de lista a lista. Sin embargo, el orden relativo de los términos se mantiene muy estable, permitiendo obtener conclusiones globales consistentes a través de listas de distintos tamaños obtenidas a partir del mismo conjunto de datos. El análisis de ontología (; GO) cubre tres dominios: componente celular (cellular component), las partes de una célula o su ambiente extracelular; función molecular (molecular function), la actividad elemental de un producto génico a nivel molecular, tales como catálisis o capacidad de unión; y proceso biológico (biological process), operaciones o conjuntos de eventos moleculares con inicios y términos definidos, pertinentes para el funcionamiento de unidades vivientes integradas: células, tejidos, órganos y organismos [1 09]. Para obtener el grado de enriquecimiento, un determinado fondo (background) debe ser establecido para realizar la comparación. El genoma de Horno sapies fue 70 empleado como fondo para la comparación de las listas de genes obtenidas mediante ambas estrategias. El enriquecimiento de los términos GO es mostrado en la figura 17 y es representado por el Log10 del valor p. Los procesos biológicos de mayor enriquecimiento identificados por la estrategia Directa comprenden genes cuyos productos están relacionados con comunicación celular, respuesta al stress y al daño, organización de estructura extracelular, adhesión celular, migración celular, metabolismo del coágeno y respuesta inflamatoria (Figura 17A). Estos procesos celulares no fueron enriquecidos en la estrategia SSH, la cual permitió el enriquecimiento de procesos relacionados con morfogénesis de órganos, metabolismo del calcio, diferenciación de stem ce//s y vía de señalización de Wnt. A nivel de componente celular, se observa en la estrategia Directa un fuerte enriquecimiento de genes cuya expresión es de ubicación preferentemente extracelular, hallazgo contrario al obervado en la estrategia SSH, donde el enriquecimiento se debe a genes de ubicación predominantemente intracelular aunque no es significativamente estadístico (Figura 178). Los genes relacionados "Wnt receptor signaling pathway through beta-catenin" son APC, CCND1, TCF7L2 y WNT9A. Los genes relacionados a "Stem ce// differentiation" ERCC2, MSI2, ACE, APC, RIF1, TFC7L2. Empleando la base de datos KEGG (http://www.genome.jp/kegg/) realizamos un análisis complemetario con el obejtivo de identificar vías de señalización presentes en la lista de genes de la estrategia SSH. La vía Wnt signaling pathway (hsa04310-Homo sapiens) posee 9 genes representados en la lista, lo cual se encuentra en concordancia con lo identificado mediante el análisis de procesos biológicos (Figura 18). Estos genes son WNT9A, NFAT5, TCF7L2, CCND1 (cycD1), MMP7 (Uterine), PPP2R 1B (PP2A), APC Y TBL 1Y (TBL 1 ). La vía de señalización posee una creciente importante en la investigación biomédica del cáncer.

71 Biological Process

Organ morphogenesis-¡.•..¡.-• -SSH Response to calcium ion WDirecta Cellular Component Stem cell differentiation Cell projection -SSH 'Vnt receptor signaling pathway through beta-catenin Synapse O Directa lntracellular signaling casca de Plasma membrana part Glycoprotein metabolic process Cell junction Regulation of catalytic activity-1...----+-' Adherens junction lnflammatory response..¡,---¡--' Nuclear eh ramoso me Regulation of cell motion .1;;=:::1== Collagen Col!agen metabolic process -J::=::j:=:::¡ Endoplasmic reticulum Regulation of cell migration -l;;;;==t== Actin cytoskeleton Blood circulation .!;;;;;:::::=!=== Golgi apparatus Programmed cell death .J==::j::== Basement membrane Cell adhesion .l;;=:::f:===:::J Cytoplasmic.vesicle Response to wounding -t:=:::f==== Extracellular matrix-t:=!=====:::::J Extracellular structure organization -!==+==== Proteinaceous extracellular matrix -t:::::i======Response to stress .l;;;;;::::t======Extracellular region.l;;;;:::l======

Regulation of cell communication 1!E~~~~=:;:=~---, lntracellular F""T'--r-r-.-r--r-r-r-. o 2 3 4 5 6 o 2 3 4 5 6 7 8 9 -lo910 P value -lo910 P value

Molecular Function --, : Enzyme binding -SSH Nucleoside kinase activity WDirecta Calcium ion binding Extracellular matrix binding 1 Actín binding ! Glycosaminoglycan bindíng - Polysaccharide binding - : Protein binding - Receptor binding -· o 2 3 4 -lo910 P value

Figura 17. Análisis de términos de ontología de genes enriquecidos mediante las estrategias Directa y SSH. El enriquecimiento de los términos GO es representado con el Log1o de la significancia estadística (P value).

72 1 WNT SIGNALING PATHWAY 1

' -- ---~~------'------' G'netrn=ip<>ou (MAPK~) l_ pe.thw:ay .

043103.12fJ2 (e) K~Iti:;a Lab:mOOries Figura 18. Genes de la vía de señalización de Wnt identificados mediante la estrategia SSH. Genes destacados en color rojo. En la figura: WNT=WNT9A; TCF/LEF= TCF7L2; cycD= CCND1; Uterine=MMP7; PP2A=PPP2R1 B; TBL 1=TBL 1Y. Imagen obtenida desde la base de datos KEGG (http://www.genome.jp/kegg/)

En la figura 19 se describen genes up-regu/ated en tejidos de ADP, presentados según su asociación a los términos de ontología de extracellular matrix (G0:0031 012), inf/ammatroy response (G0:0006954) y ce// motion (G0:0006928). Entre los genes asociados a matriz extracelular se encuentran varias fibras de colágeno (e.j. COL 1A1, COL 17 A 1 y COL4A 1 ), metaloproteinasas de matriz (MMP1 y MMP11 ), fibronectin 1 (FN1 ), periostin (POSTN) y versican (VCAN). La respuesta inflamatoria se ve reflejada con la alta abundancia de interleukin 8 (IL8), interleukin 1 receptor accessory protein (IL 1 RAP), mo/ecu/e, decay accelerating factor for complement (CD55), coagulation

73 factor 11 (thrombin) receptor (F2R), coagulation factor 111 (thromboplastin, tissue factor) (F3) y coagulation factor VIII, procoagulant component (F8).

Q) 1/1 1:o c. 1/1 Q)'

1: ·-"'0"' -"'0"'Eg =o oGl(!)

Figura 19. Ontologías de genes up-regulated identificados en la estrategia Directa. Componentes de la matriz extracelular, de respuesta inflamatoria y movimiento celular se encuentran enriquecidos en tejido tumoral debido principalmente a la presencia de una intensa desmoplasia. Los genes se muestran según ontología. 74 En la figura 20 se observan los niveles de expresión de los genes de la estrategia SSH y sus respectivos valores en la matriz de la estrategia Directa. Los transcritos identificados como diferenciales en la estrategia Directa presentan en promedio un valor de fold change cercano a O lo que demuestra que mediante la estrategia Directa estos genes no pudieron ser identificados como diferenciales.

2.0

o Ql oo0 C) e o ra c3 0.5 •

o ooo

SSH

Figura 20. Comportamiento de los genes diferenciales de la estrategia SSH en la matriz de la estrategia Directa. Los genes identificados en la estrategia SSH fueron extraídos de la matrix de la estrategia Directa.

4.4.3 Ambas estrategias identifican genes que codifican para proteínas de secreción.

Las proteínas secretadas por varios tipos celulares, incluidas las células malignas, representan una potencial fuente de biomarcadores debido a que reflejan varios estadios celulares en tiempo real y bajo determinadas condiciones [11 0]. Para identificar los genes que codifican para proteínas de secreción (predichas computacionalmente y validadas) realizamos un análisis en la base de datos DAVID. Las proteínas de secreción de la estrategia Directa y SSH se muestran en las tablas 3 y 4 respectivamente. Entre los genes de secreción de la estrategia Directa

75 encontramos moléculas tales como ANXA2, FAS, CD59, VCAN, IL8, SPARC, WNT5A. Entre los genes identificados mediante la estrategia SSH encontramos a MMP?, WNT9A, NMU, EGFR y OLFML 1, de los cuales WNT9A fue escogido para estudios posteriores.

Estrategia Directa Símbolo Nombre oficial ANGPT2 angiopoietin 2 ANXA2 annexin A2 pseudogene 3; annexin A2; annexin A2 pseudogene 1 FAS Fas (TNF receptor superfamily, member 6) C4BPB complement component 4 binding protein, beta CALU calumenin CD59 CD59 molecule, complement regulatory protein COL1A1 collagen, type 1, alpha 1 COL4A1 collagen, type IV, alpha 1 COL6A3 collagen, type VI, alpha 3 COL8A2 collagen, type VIII, alpha 2 COL17A1 collagen, type XVII, alpha 1 VCAN versican DAG1 dystroglycan 1 (dystrophin-associated glycoprotein 1) F8 coagulation factor VIII, procoagulant component FGFR2 fibroblast growth factor receptor 2 FN1 fibronectin 1 IGFBP3 insulin-like growth factor binding protein 3 IGFBP5 insulin-like growth factor binding protein 5 IGFBP7 insulin-like growth factor binding protein 7 IL1RAP interleukin 1 receptor accessory protein IL8 interleukin 8 KNG1 kininogen 1 LCN2 lipocalin 2 LGALS1 lectin, galactoside-binding, soluble, 1 UF leukemia inhibitory factor (cholinergic differentiation factor) LTBP1 latent transforming growth factor beta binding protein 1 MATN3 matrilin 3 MFGE8 milk fat globule-EGF factor 8 protein MMP1 matrix metallopeptidase 1 (interstitial collagenase) MMP11 matrix metallopeptidase 11 (stromelysin 3) NID1 nidogen 1 TNFRSF11B tumor necrosis factor receptor superfamily, member 11 b PVRL11 poliovirus receptor-related 1 (herpesvirus entry mediator C) CXCL5 chemokine (C-X-C motif) ligand 5 76 SFRP2 secreted frizzled-related protein 2 SLPI secretory leukocyte peptidase inhibitor SPARC secreted protein, acidic, cysteine-rich (osteonectin) STC1 stanniocalcin 1 TCN1 transcobalamin 1 (vitamin 812 binding protein, R binder family) TGFBI transforming growth factor, beta-induced, 68kDa TNXB tenascin X8; tenascin XA pseudogene VEGFC vascular endothelial growth factor e ~NTSA wingless-type MMTV integration site family, member 5A M lA melanoma inhibitory activity GGH gamma-glutamyl hydrolase ~IMP1 aminoacyl tRNA synthetase complex-interacting multifunctional protein 1 CRISP3 cysteine-rich secretory protein 3 SEMA3C sema domain, immunoglobulin domain (semaphorin) 3e POST N periostin, osteoblast specific factor ADAM28 ADAM metallopeptidase domain 28 LILRA3 leukocyte immunoglobulin-like receptor, subfamily A, member 3 ESM1 endothelial cell-specific molecule 1 PRSS23 protease, serine, 23 FSTL1 follistatin-like 1 DKK1 dickkopf homolog 1 (Xenopus laevis) LY96 lymphocyte antigen 96 OLFML2B olfactomedin-like 28 GREM1 gremlin 1, cysteine knot superfamily, homolog (Xenopus laevis) PILRA paired immunoglobin-like type 2 receptor alpha CKLF chemokine-like factor TREM2 triggering receptor expressed on myeloid cells 2 PDGFC platelet derived growth factor e TWSG11 twisted gastrulation homolog 1 (Drosophila} PRSS22 protease, serine, 22 C12orf49 12 open reading frame 49 LOXL4 lysyl oxidase-like 4 CTHRC1 collagen triple helix repeat containing 1 DEFB118 defensin, beta 118 IZG16B zymogen granule protein 16 homolog 8 (rat) OVOS2 ovostatin; ovostatin 2 OTOA otoancorin IGFL2 IGF-Iike family member 2 CDCP2 eU8 domain containing protein 2 LIPH lipase, member H C3orf58 chromosome 3 open reading frame 58 AGRN agrin FIBIN fin bud initiation factor homolog (zebrafish} Tabla 3. Genes que codifican para proteínas secretadas identificados mediante la estrategia Directa. Genes up-regulated de la estrategia Directa fueron analizados

77 mediante la base de datos DAVID empleando la categoría SP _PIR_KEYWORDS. Un total de 77 proteínas fueron identificadas como secretadas.

Estrategia SSH Símbolo Nombre oficial C8A complement component 8, alpha polypeptide CP ceruloplasmin (ferroxidase) CPB2 carboxypeptidase B2 (plasma) HAPLN1 hyaluronan and proteoglycan link protein 1 CST1 cystatin SN EGFR epidermal growth factor receptor FGG fibrinogen gamma chain IL16 interleukin 16 (lymphocyte chemoattractant factor) KNG1 kininogen 1 LAMA3 laminin, alpha 3 LIPC lipase, hepatic MMP7 matrix metallopeptidase 7 (matrilysin, uterine) NRTN neurturin WNT9A wingless-type MMTV integration site family, member 9A WISP2 WNT1 inducible signaling pathway protein 2 IL18BP interleukin 18 binding protein NMU neuromedin U C1orf56 chromosome 1 open reading frame 56 MMELI1 membrane metallo-endopeptidase-like 1 ADAMTS20 ADAM metallopeptidase with thrombospondin type 1 motif, 20 PVRL4 poliovirus receptor-related 4 RSP03 R-spondin 3 homolog (Xenopus laevis) C1QTNF3 C1 q and tumor necrosis factor related protein 3 C4orf26 chromosome 4 open reading frame 26 TAC4 tachykinin 4 (hemokinin) OLFML1 olfactomedin-like 1

Tabla 4. Genes que codifican para proteínas secretadas identificados mediante la estrategia SSH. Genes up-regulated de la estrategia SSH fueron analizados mediante la base de datos DAVID empleando la categoría SP _PIR_KEYWORDS. Un total de 26 proteínas fueron identificadas como secretadas.

78 Un total de 35 transcritos son compartidos entre ambas estrategias, representando solo un 5 y 6% de la lista Directa y la lista SSH, respectivamente (Figura 21A). Los genes compartidos entre ambas estrategias son 22 up-regulated y 13 down­ regulated (Figura 21 B). Entre los genes up-regulated se encuentran CD55 y EGFR, los cuales han sido descritos previamente como genes up-regulated en cáncer pancreático. El transcrito de CD55 es detectable en jugo pancreático de pacientes con cáncer pancreático, donde se encuentra aumentado con respecto a casos controles [111]. EGFR se encuentra up-regulated en más del 90% de los casos de ADP y es actualmente un objetivo terapéutico de cáncer pancreático mediante el uso de la droga erlotinib, aprobada por la FDA para el tratamiento de esta enfermedad [112]. Otro aspecto interesante de destacar es que la estrategia Directa contiene varios genes previamente descritos como potenciales marcadores de ADP (Figura 22). Es el caso de ANXA8 [113], CEACAM [114, 115], CLDN 18 [113], CTSB [116], CTSL [116] Y TOP2A [117].

79 Estrategia Estrategia Directa SSH B ID Símbolo Nombre oficial 1604 CD55 CD55 molecule, decav acceleratina factor for comolement 1770 DNAH9 dvnein, axonemal, heavv chain 9 1956 EGFR epidermal growth factor receptor 2152 F3 coaaulation factor 111 3827 KNG1 kininogen 1 3909 LAMA3 laminin, alpha 3 6772 STAT1 signa! transducer and activator of transcriotion 1, 91 kDa 7050 TGIF1 TGFB-induced factor homeobox 1 7273 TTN titin 7932 OR2H2 olfactorv receptor, familv 2, subfamilv H, member 2 8853 ASAP2 ArfGAP with SH3 domain, ankvrin reoeat and PH domain 2 9344 TAOK2 TAO kinase2 10753 CAPN9 calpain 9 23224 SYNE2 spectrin repeat containing, nuclear envelooe 2 55821 ALLC allantoicase 57718 PPP4R4 lorotein ohosohatase 4, reaulatorv subunit 4 64072 CDH23 cadherin-like 23 79925 SPEF2 sperm flagellar 2 80144 FRAS1 Fraser svndrome 1 114907 FBX032 F-box orotein 32 116966 WDR17 WD repeat domain 17 170482 CLEC4C C-type lectin domain familv 4, member C 472 IE,Tf'F: ataxia telanaiectasia mutated 1111 ICHEl~'l CHK1 checkooint homoloa (5. oombe) 2568 IG.'\BHF gamma-aminobutvric acid (GABA) A receptor, pi 7052 lrm,;r transglutaminase 2 7083 in~·¡ thvmidine kinase 1, soluble 9833 ¡¡,;¡;:u: maternal embrvonic leucine zioper kinase 10863 J:,Oj'\[·,129 ADAM metallopeptidase domain 28 10933 lic1'JHFPU mortalitv factor 4; mortality factor 4 like 1 26031 IOSEPL3 oxvsterol bindina orotein-like 3 54798 IDCHS2 dachsous 2 (Drosoohila) 54991 l:;,d133 chromosome 1 open readino frame 159 55520 IEUi.Gj elaC homoloo 1 (E. coli) 80178 IG'i6orf39 chromosome 16 ooen readina frame 59 Figura 21. Genes diferenciales identificados mediante ambas estrategias. A) Diagrama de Venn de los genes comunes identificados mediante la estrategia Directa y SSH. 8) Lista de 22 genes up-regulated (color rojo) y 13 genes down-regulated (color verde).

80 Figura 22. Genes identificados en la estrategia Directa que han sido previamente validados como biomarcadores de ADP.

4.5 Validación de los genes diferenciales obtenidos por la estrategia directa y SSH por PCR en tiempo real.

La tecnología de microarray es una herramienta poderosa para la caracterización de la expresión génica a alta escala. Sin embargo, esta tecnología puede generar falsos positivos que deben ser corroborados para análisis posteriores. El PCR cuantitativo en tiempo real (qPCR) es una herramienta sensible y robusta para medir niveles de expresión génica y ha sido utilizada para confirmar los resultados obtenidos de plataformas de microarray. Para confirmar la veracidad de los resultados de ambas estrategias aplicamos la metodología de qPCR para validar grupos de genes escogidos. Los genes fueron seleccionados a partir de los genes up-regulated de cada lista. Como el objetivo principal del proyecto es identificar moléculas cuya expresión sea de utilidad para la caracterización y potencial diagnóstico de la enfermedad, decidimos optimizar la validación y caracterización de genes candidatos con tendencia a la sobre expresión (up-regulated) en tejidos tumorales. Los valores de expresión de los genes de la estregia Directa y SSH identificados mediante microarray y validados mediante qRT-PCR son mostrados en la figura 23. De los diez genes analizados para la estrategia Directa, ocho son significativamente estadísticos. De los diez genes analizados para la estrategia SSH, cuatro genes son significativamente estadísticos, sin embargo, el resto de los genes presenta una tendencia hacia la sobre expresión (Figura 22). 81 Estrategia Directa Estrategia SSH

BHLHE40 CD55 CALU DPYSL4 COL4A1 KNG1 CTSK LAMA3 FN1 MMP7 IGFBP5 OASL PMEPA1 PLEKHM1 SBN02 STAT1 SPARC SYNE2 TOP2A SYT12

BHLHE40 CD55* CALU * DPYSL4 COL4A1* KNG1 CTSK LAMA3 FN1* MMP7* IGFBP5 OASL* PMEPA1* PLEKHM1 SBN02* STAT1 SPARC * SYNE2 TOP2A * SYT12 *

-3.0 0.0 -3.0 Log2 fold change Figura 23. Validación por qRT-PCR de genes diferenciales de las estrategias Directa y SSH. Diez genes de cada estrategia fueron escogidos para validación. La cuantificación relativa por qRT-PCR fue realizada empleando como calibrador un cONA de páncreas normal. Los genes en estudio fueron analizados en duplicado y normalizados a la expresión de los genes QARS y TBP. *= P value

82 4.6 Validación de genes diferenciales mediante tinción inmunohistoquímica

4.6.1. Genes candidatos de la estrategia Directa.

Es de relevancia validar si los niveles de los transcritos identificados se correlacionan con los niveles de las respectivas proteínas. Con el objetivo de confirmar los resultados obtenidos mediante estrategia Directa, realizamos un estudio inmunohistoquímico en tejidos de adenocarcinoma ductal y tejidos no neoplásicos adyacentes al tumor. Seleccionamos tres genes candidatos de la lista de genes up­ regu/ated identificados en la estrategia Directa y los estudiamos mediante tissue arrays. Los genes PMEPA1 (prostate transmembrane protein, androgen induced 1), BHLHE40 (basic helix-loop-helix fami/y, member e40), y CALU (calumenin) fueron escogidos en función de la escasa literatura científica y propiedad intelectual asociada a ellos, además de la disponiblidad anticuerpos.

4.6.1.1. PMEPA1 prostate transmembrane protein, androgen induced 1

No se observa expresión de PMEPA1 en duetos de morfología normal en tejidos no neoplásicos, con expresión leve en células acinares (Figura 24A y 248). En tejidos tumorales se observan células positivas para PMEPA 1, con un marcado patrón nuclear exclusivo (Figura 24C) o en combinación con expresión citoplasmática (Figura 24C- 24E). En algunos casos se observa intensa expresión nuclear (Figura 24F).

83 Figura 24. Tinción inmunohistoquímica de PMEPA1. A y B) Duetos pancreáticos de morfología normal con nula expresión de PMEPA1 (flecha roja). Células acinares presentan una tinción leve; C) Células tumorales con tinción nuclear intensa de PMEPA 1 (flecha roja); D y E) Células tumorales con tinción nuclear intensa y citoplasmática leve de PMEPA 1 (flecha roja); F) Células tumorales con expresión nuclear intensa y citoplasmática moderada de PMEPA1 (flechas rojas).

84 Non-neoplastic Adenocarcinoma

Figura 25. Análisis inmunohistoquímico de la expresión de PMEPA1 en tissue arrays de tejidos no neoplásicos y adenocarcinoma pancreático. Los staining scores obtenidos como múltiplo de la intensidad de la tinción y la celularidad de cada grupo fueron comparados mediante t test pareado de una cola. Media y SD de un n=30. ***=p<0.001.

El análisis de tissue microarray muestra que el tejido de ADP presenta niveles mayores de la proteína PMEPA1, en comparación con los tejidos no neoplásicos. PMEPA1 se encuentra ubicado en el brazo largo del cromosoma 20 (20q13.3). Esta región del cromosoma se presenta amplificada en un gran porcentaje de casos de cáncer pancreático [118-120], estómago [121] y colon [122]. Un aumento en el número de copias del gen explicaría por que PMEPA1 se encuentra up-regulated en adenocarcinoma pancreático, ya que se ha observado que otros genes de esta región se encuentran up-regulated en tejidos tumorales de cáncer de páncreas, estómago y colon.

20p13

20p12

20p11.2 20p11.1 20q11.1 20<:t11.2 +J·iJt-;!J.tWMg.¡r.!~MMf,,·l$P·id·l

20.,1~.1 SERINC3 UBE2C DDX27 ZNF217 MMP9 TMEM189 2••1~.2

20.,1~.~ BCASl AURKA CTSZ PMEPAl

Figura 26. Representación gráfica del cromosoma 20. La región coloreada de gris muestra la región del brazo largo a la que con mayor frecuencia se le ha asociado incremento en el número de copia de diferentes genes. PMEPA 1 se ubica en esta región. La tabla de la derecha muestra ejemplos de genes que pertenecen a esta región y que presentan incremento en el número de copias [118, 119, 121, 122]. 85 4.6.1.2 BHLHE40 basic helix-loop-helix family, member e40

En tejidos no neoplásicos se observan duetos positivos (Figura 27 A) y negativos (Figura 278) para BHLHE40. Las células acinares presentan una tinción leve. Las células tumorales presentan un patrón intenso de expresión citoplasmática supranuclear (Figura 27C) o difusa (Figura 27D-F). ' . ..•. ., . . ·"\ iJ

• .. .\ti. ' i .,... ' '

Figura 27. Tinción inmunohistoquímica de BHLHE40. A) Dueto pancreático de morfología normal con expresión leve de BHLHE40 (flecha roja). Células acinares presentan una tinción leve; B) Dueto pancreático de morfología normal con nula expresión BHLHE40 (flecha roja); C-F) Células tumorales con expresión citoplasmática intensa de BHLHE40 (flechas rojas) Células estromales no expresan la proteína. 86 BHLHE40 **

Non-neoplastic Adenocarcinoma Figura 28. Análisis inmunohistoquímico de la expresión de BHLHE40 en tissue arrays de tejidos no neoplásicos y adenocarcinoma pancreático. Los staining scores obtenidos como múltiplo de la intensidad de la tinción y la celularidad de cada grupo fueron comparados mediante t test pareado de una cola. Media y SO de un n=30. **=p<0.01.

El análisis de tissue array muestra que el tejido de ADP presenta niveles mayores de la proteína BHLHE40, en comparación con los tejidos no neoplásicos (Figura 28).

87 4.6.1.3. CALU calumenin

En tejidos no neoplásicos de páncreas se observan duetos que no expresan CALU (Figura 29A) o lo expresan de forma leve (Figura 298). A diferencia de PMEPA1 y BHLHE40, CALU se expresa predominantemente en el citoplasma de células de fenotipo ahusado presentes en el tejido estroma! reactivo alrededor de las células tumorales, presentando una intensa tinción (Figura 29 C y D). Este fenotipo es compatible con fibroblastos activados o células estelares pancreáticas (CEP) [123]. Adicionalmente, se observa expresión leve y ubicua de CALU en células tumorales. Las CEPs expresan conjuntamente marcadores de fibroblastos y células de músculo liso. La actina de músculo liso (ACTA2; actin, alpha 2, smooth muse/e, aorta; a-SMA) ha sido utilizada como marcador para identificar CEPs en tejidos neoplásicos de páncreas. Realizamos una doble tinción de inmunofluorescencia con anticuerpos anti-CALU y anti-a-SMA en tejidos de adenocarcinoma ductal pancreático. Ambas molecúlas se encuentran presentes en el compartimento estroma! reactivo y se observan células que expresan ambas moléculas (Figura 29F). Es de interés destacar que CALU es una molécula que es secretada al espacio extracelular, lo que la convierte en un potencial candidato a biomarcador (Tabla 3).

88 Figura 29. Tinción inmunohistoquímica de CALU. A y B) Duetos pancreáticos de morfología normal que no expresan CALU (flechas rojas); C) y D) Células de morfología ahusada (probablemente fibroblastos o células estelares) con expresión intensa de CALU (flechas rojas); E) Células tumorales que expresan CALU (flecha roja); F) Tinción de inmunofluorescencia de tejido de adenocarcinoma ductal pancreático. Células de fenotipo ahusado expresan simultáneamente CALU (verde) y actina de músculo liso (a-SMA, rojo), marcador de células estelares activadas. DAPI=tinción nuclear.

89 A B CALU Adenocarcinoma CALUStroma e e *** ·¡¡¡ ]l o o -..c.. .. c.. ::J ::J ..J ;¡¡! (.)< (.) ....o ....o Cl) Cl) ..o ..o u u U) U) C) C) e e '2 '2 ·¡¡¡ ·¡¡¡ -tJ) -tJ) Non-neoplastic Adenocarcinoma Non-neoplastic Adenocarcinoma

Figura 30. Análisis inmunohistoquímico de la expresión de CALU en tissue arrays de tejidos no neoplásicos y adenocarcinoma pancreático. La expresión de CALU fue analizada de forma independiente según compartimento celular. A) Los tejidos ductales y acinares fueron comparados con el tejido de adenocarcinoma. B) El tejido estroma! normal fue comparado con la reacción desmoplástica del ADP). Los staining scores obtenidos como múltiplo de la intensidad de la tinción y la celularidad de cada grupo fueron comparados mediante t test pareado de una cola. Media y SD de un n=30. ***=p<0.001.

El análisis de tissue array muestra que el tejido estroma! reactivo de tumores de ADP presenta niveles mayores de la proteína CALU, en comparación con las células estromales del tejido no neoplásico (Figura 308). Por el contrario, las células tumorales expresan niveles similares en comparación con los duetos y acinos no neoplásicos (Figura 30A).

90 4.6.2. Genes identificados en la estrategia Directa contrastados con la base de datos del Human Atlas Protein.

El Human Atlas Protein (HAP) es una base de datos disponible públicamente con millones de imágenes de alta resolución que muestran la distribución espacial de más de 12,000 proteínas humanas en 46 tejidos humanos diferentes y 20 tipos de cáncer, así como en 47 líneas celulares humanas (http://www.proteinatlas.org/). Genes up-regulated escogidos de la estrategia Directa fueron contrastados con la base de datos del Human Atlas Protein. La figura 31 muestra ejemplos de expresión en dos tejidos no neoplásicos y dos tejidos tumorales diferentes para cada proteína. Las proteínas codificadas por los genes SPARC (secreted protein, acidic, cysteine-rich (osteonectin)), FN1 (fibronectin 1) y CDH11 (cadherin 11, type 2, OB-cadherin (osteoblast)) presentan una intensa expresión en el tejido estroma! reactivo que rodea al tejido tumoral. Por el contrario, las proteínas codificadas por los genes CEACAM6 (carcinoembryonic antigen-related cell adhesion molecule 6), CLDN18 (claudin 18), TOP2A (topoisomerase (DNA) 11 alpha 170kDa), DKK1 (dickkopf 1 homolog (Xenopus laevis)) y RAB23 (RAB23, member RAS oncogene family) presentan tinción tumoral que va de leve a intensa. Las proteínas codificadas por los genes SPARC, FN1 y CDH11 presentan una intensa expresión en el estroma reactivo que rodea al tejido tumoral, sin embargo, su expresión en tejido no neoplásico es débil y aislada. Por el contrario, es posible encontrar tumores que expresan intensamente CEACAM6, CLDN18, TOP2A y DDK1 en sus células tumorales, observándose expresión nula o leve de estas proteínas en los tejidos pancreáticos no neoplásicos. El hallazgo de tumores que sobre expresan estas proteínas en comparación con tejidos no neoplásicos se encuentra en concordancia con los hallazgos transcriptómicos identificados mediante la estrategia Directa.

91 No neoplásico ADP

..- ..- I o ü

' . . . ..- z l/ LL

ü 0::: ¡ 0: .. !t en :J ·..... ~-> ...... CD ~ <( ü <( w ü ; ' (() ..- z o _J ü ' ..., . :·.·:'-

<( N a.. o 1--

Figura 31. Imágenes de tinciones inmunohistoquímicas de genes de la estrategia Directa obtenidas del Human Atlas Protein. Tinciones IHQs de 8 genes up-regulated identificados mediante la estrategia Directa. ADP=adenocarcinoma ductal pancreático.

92 4.6.3. Genes candidatos de la estrategia SSH.

Con el objetivo de confirmar los resultados obtenidos mediante estrategia · Directa, realizamos un estudio inmunohistoquímico en tejidos de adenocarcinoma ductal y tejidos no neoplásicos adyacentes al tumor. Seleccionamos cinco genes candidatos de la lista de genes up-regulated identificados mediante la estrategia SSH en function de la disponibilidad de anticuerpo y si ha sido reportado antes en cáncer pancreático: PLEKHM1 (pleckstrin homology domain containing, family M (with RUN domain) member 1 ), WNT9A (wingless-type MMTV integration site family, member 9A), DPYSL4 (dihydropyrimidinase-like 4), MMP7 (matrix metallopeptidase 7 (matrilysin, uterine)) y STAT1 (signal transducer and activator of transcription 1, 91 kDa).

4.6.3.1. PLEKHM1 p/eckstrin homology domain containing, family M (with RUN domain) member 1.

El producto génico de PLEKHM1 fue identificado inicialmente como una proteína unida fuertemente a una mucina positiva para el grupo Sialii-Lewis(x) en un screening de cONA humano de colon contra un antisuero de conejo desarrollado contra mucinas Lewis positivas [124]. Posteriormente, PLEKHM1 fue vinculado con el transporte vesicular en osteoclastos [125] y la mutación de este gen produce un tipo de osteopetrosis en humanos, asociada con osteopenia general y osteoclerosis focal [126]. Al momento de escribir este documento no existe información de PLEKHM1 asociada a neoplasias. Los tejidos pancreáticos normales no expresan PLEKHM1 en duetos, células acinares o islotes de Langerhans (Figura 32A-B). Por el contratrio, células tumorales positivas para queratina 19 (marcador epitelial), expresan PLEKHM1 (Figura 32C-F). Posteriormente se realizó una tinción doble de inmunofluorescencia en dos tejidos tumorales, donde fue confirmado que solo las células positivas para queratina 19 expresan PLEKHM1, no observándose expresión de esta proteína en otros

93 compartimentos del tejido pancreático (Figura 33). Es interesante destacar que en algunos tejidos analizados fue posible observar células solitarias positivas para PLEKHM1 en aparente proceso de invasión. Estas células a su vez presentan expresión de MMP?, una metaloproteinasa de matriz asociada a procesos invasivos (Figura 34 ). Posteriormente analizamos los niveles de PLEKHM1 mediante un tinción IHQ de un tissue array compuesto por 90 cores provenientes de 30 pacientes con adenocarcinoma ductal. Por cada core de tejido no neoplásico adyacente a la lesión tumoral, están representados dos cores de tejido tumoral del mismo paciente. El tissue array fue escaneado y cuantificado empleando el sistema ACIS 111 (DAKO), para determinar un staining score para cada core (ver Material y Métodos 3.12.1 ). Para el análisis estadístico el score tumoral fue calculado del promedio de ambos replicados y comparado con el score no neoplásico mediante t test pareado de dos colas. En la figura 35 se observa que los niveles de expresión de PLEKHM1 son significativamente mayores en tejidos neoplásicos que en tejidos no neoplásicos de páncreas. Tres tejidos no neoplásicos presentan un alto staining score para PLEKHM1, sin embargo, la morfología de estos cores se encuentra alterada (datos no mostrados). A excepción de estos tres casos, no hay expresión detectable de PLEKHM1 en tejidos no neoplásicos pancreáticos. Por el contrario, alrededor del -30% de los casos de adenocarcinoma ductal del tissue array presentan algún grado de expresión de PLEKHM1. Para determinar la expresión de PLEKHM1 en otros órganos empleamos un tissue array de tejidos normales. PLEKHM1 no se expresa en tejidos pancreáticos normales (Figura 36), sin embargo, es expresada por tipos celulares de la próstata (prostate), timo (thymus), amígdala (tonsil), cuello del útero (uterus cervix) y piel (skin), siendo este último el órgano donde se observa una mayor expresión de PLEKHM1 (Figura 36). No detectamos expresión de PLEKHM1 en tejidos de glándula adrenal, vejiga, médula ósea, ojo, mama, cerebelo, corteza cerebral, trompa de Falopio, estómago, esófago, intestino delgado, colon, recto, corazón, riñón, hígado, pulmón, ovario, glándula paratiroides, glándula pituitaria, placenta, bazo, músculo estriado, testículo y glándula tiroides

94 Pancreas no neoplásico

~L '"~- , .... ;~ •• ·~~-··...,_.'·:·:, f B ~ ,.' ··. ·.t.,····A.. ·;··•J• ~ . • ...... ·, . ~ ~ ..... ~ . ~"' '~ · ~ .~-, ;...... · .,.... . ,·. : ...... ·~·-' •. J;".,.·.¡·· •. • .. ·\·~--~ ' . ' ' \ 1, 4'.. .. . ,, . 1 • • ., 1 ~ . .: • 'r • r ~ '· ·: '. J L· · f' ••. ~- ·• • , ··, '1 . '. .• , •• '.. .. f' ....: . • . ~ . '-"""' ... ·' . ·: : -== ...... _;, ...... ' ,1:' .. ·'-··"-······- ...... ~-~ ..::::; •. •• ~ '; ' • •• " " ... •, .'·1 • :::E: ·;;·.:~.:' ··r :. ::' / .• .: .• : .. ·.• -~· ~ • • "' .• 1.. • .. ... • ., •. ~ ~ •• '4 1 ¡ ...... • .• '... ~- .,, .·, J • •• • • • ... -. w '...... ,. ·.·.-::-..·· • ~...' "'~ .~ 1 ',~ ;, • .. J,"-. 11 •• •-.\.1. 1 ..... • _. •. ;. •... _~....._ ~: ·.} .• ..J . ..., ,. ' ~ ' . ·...... ¿ .. •• . •¡ .: • \. a. ~-- ..... _., .••. : ... J.• .· .. , . • ' ,. • 'r .,.... • • .. . , • . . . ':_~·-· ~,.· ·•• •·:•.. ;<·.. ·:' -:~~·:,·.~·.;-·\· ...... ·/, . • "' ~· 1 • 1 ·". • ~~ . , ..... - ' - . ' . '. . ··' . ••• •...... 1 t ""· •· ·~·. • ...... -. ·..... , .... ·~. -"-~. . .. -~ ,. - .,. . ··~-'' / ...... ·' ··~ ~ ·'··'{\· t••. /~ J . ••• ;' - ~ ..t • • • •• .. ~ ....•• _,. .

# .. t·· ,·:.. ~ ••. ~~~-..-~·····.' ·: ... "· .. ,. .

Queratina 19 PLEKHM1

a. e <(

·'. ... ·...

:-:··

.-·' J 1. ~ ~ ·~ :\. .;· : : .. ~ . \ Figura 32. Expresión de PLEKHM1 en tejidos pancreáticos tumorales y no neoplásicos. A y B) Nula expresión de PLEKHM1 en duetos (flecha roja), islotes de Langerhans (flecha azul) y células acinares; D y F) El producto génico de PLEKHM1 se expresa en células tumorales positivas para queratina 19 (C y D respectivamente). ADP= adenocarcinoma ductal pancreático. 95 DAPI PLEKHMl

Keratin 19 Merged Figura 33. Doble tinción inmunohistoquímica de PLEKHM1 y queratina 19. Tejido de adenocarcinoma ductal pancreático marcado con anti-PLEKHM1 (AiexaFiuor 488, verde); Queratina 19 (AiexaFiuor 568, rojo); tinción nuclear DAPI (azul).

PLEKHM1 MMP7

.. ,_

·-":'~~------"'-_ .. ···-- ~ ··.·• ...._ ... , . ? ~- •ty~,; .... ·.... -- ' ·:~<- . ~-,. a. --~ '<. ..: ..... e :·-<·· <( ~-.-,. ·:

Figura 34. Expresión de PLEKHM1 en células solitarias en aparente proceso invasivo. Células positivas para PLEKHM1 se encuentran en regiones desprendidoas de la masa tumoral principal y expresan MMP7. ADP= adenocarcinoma ductal pancreático. 96 2.0 PLEKHMl •J ; o* / cu o o ...o ··~ / ."' 00 "¡jl o o S:: .. .J ....cu o ~·.. S:: o ·e-o 1.0 o S:: ·a o ,;-.• :E .\. r;f)"' ., ~: .. ,; u 0.5 : -< ._:;. n=30 0.0354 \../ ·.. ·. .. 0.0 * Non-neoplastic Adenocarcinoma

Figura 35. Análisis inmunohistoquímico de PLEKHM1 en muestras tumorales y no neoplásicas de páncreas. Análisis t test pareado de dos colas de los staining scores de PLEKHM1 para 30 casos de adenocarcinoma ductal y sus respectivo tejidos no neoplásicos adyacentes. Línea punteada representa la media del score para cada grupo.

97 Pan creas

Timo Próstata

/

' ... _,:.~,'"~ . ·. -.. \- Piel Amígdala Cuello del útero

..: \ .~ .· :.., \ •::! +Í'I;.\··: . "\, ·"1 . . '. ·, :.¡, .,. 1 ·,. Figura 36. Expresión de PLEKHM1 en tejidos humanos normales de distintos órganos evaluada mediante tinción inmunohistoquímica. Las flechas rojas indican células positivas para la expresión de PLEKHM1 en timo, próstata, piel, amígdala y cuello uterino.

98 4.6.3.2. WNT9A wingless-type MMTV integration site family, member 9A

WNT9A pertenece a la familia de proteínas WNT y ha sido relacionado al proceso de condrogénesis e integridad articular en modelos murinos [127], así como a la morfogénesis y proliferación celular en epitelio hepático de modelos aviares [128]. En zebrafish ha sido vinculado al desarrollo del paladar y mandíbula inferior [129]. La expresión de WNT9A fue analizada en tejidos de adenocarcinoma ductal pancreático mediante análisis inmunohistoquímico. WNT9A se expresa en células ductales del páncreas no neoplásico con un patrón de tinción granular de localización supranuclear, sugerente de ubicación en retículo endoplasmático, sin embargo, esta expresión es baja o nula en la mayoría de los duetos pequeños (Figura 37 A), siendo los duetos de tamaño mediano o mayor los que expresan WNT9A de forma más intensa (Figura 378). En tejidos de adenocarcinoma ductal, WNT9A se expresa en la mayoría de las células tumorales, mostrando frecuentemente patrones de tinción intenso (Figura 37C y 370). La expresión de WNT9A fue evaluada mediante TMA, siguiendo el mismo protocolo empleado para PLEKHM1. En la Figura 38 se observa que la expresión de WNT9A es significativamente mayor en los tejidos de adenocarcinoma ductal en comparación con los tejidos no neoplásicos. Es de interés destacar que al comparar los staining scores de los tejidos no neoplásicos de PLEKHM1 (Figura 35) y WNT9A (Figura 38), el primero presenta un score promedio inferior al segundo, esto debido a que WNT9A se expresa en tejidos normales a niveles bajos, a diferencia de PLEKHM1, que se encuentra ausente en tejidos normales y no neoplásicos de páncreas. Para el análisis estadístico de WNT9A fueron removidos los datos de dos pacientes por presentar plegamiento del core sobre sí mismo, invalidando el staining score.

99 Pancreas no neoplásico

Queratina 19 WNT9A

o. e

Figura 37. Expresión de WNT9A en tejidos pancreáticos tumorales y no neoplásicos. A) WNT9A se expresa en duetos de morfología normal con un patrón supranuclear (flecha roja), B) Algunos duetos no expresan WNT9A (flecha roja). Islotes de Langerhans (flecha azul) y células acinares no expresan WNT9A; El producto génico de WNT9A se expresa en células tumorales (D y F) positivas para queratina 19 (C y D respectivamente). ADP= adenocarcinoma ductal pancreático.

100 WNT9A 2.0

o ** QJ ... •> 1 o oog o - ~ 1.5 ..... >. ooo ·o;.... 00 ~ ....QJ o0o _'2.~'2. ~ o Ooooo 'é'o 1.0 ·¡::~ -~q_m ooo o o ] 000 o o 0ooo 8 rJl"' o ü 0.5 00 <

n=28 **p=O.OOJ9 0.0 Non-neoplastic Adenocarcinoma

Figura 38. Análisis inmunohistoquímico de WNT9A en muestras tumorales y no neoplásicas de páncreas. Análisis t test pareado de dos colas de los staining scores de WNT9A para 28 casos de adenocarcinoma ductal y sus respectivo tejidos no neoplásicos adyacentes. Línea punteada representa la media del score para cada grupo.

101 4.6.3.3. MMP7 matrix metallopeptidase 7 (matrilysin, uterine) y STAT1 signa/ transducer and activator of transcription 1, 91kDa

MMP7 y STAT1 fueron identificados como transcritos de expresión tumoral mediante la estrategia SSH. Adicionalmente, STAT1 fue identificada en la estrategia Directa. Ambas moléculas fueron estudiadas mediente inmunohistoquímica en tejidos no neoplásicos y de adenocarcinoma pancreático. Los duetos tumorales expresan MMP7 con un patrón apical de intensidad moderada (Figura 398), sin embargo, es posible observar tejidos no tumorales que presentan expresión de MMP7. Esto se ve reflejado en el análisis de tissue array (Figura 39E), donde se observa que los tejidos no neoplásicos no se diferencian significativamente de los tumores en la expresión de MMP7, aunque hay mayor tendencia a la expresión en ADP. Una situación similar ocurre con STAT1, donde se observa expresión de la proteína en tejidos tanto tumorales como no tumorales, sin diferencias significativas de expresión en el análisis de tissue arra y (Figura 39F), aunque hay una leve tendencia de mayor expresión en los tejidos de ADP.

102 No neoplásico ADP A

., ' ·.··U!

' ; . ... ~ . ·.. : . ·.. :. ;- \

-~ ... ,.·,, . .·• j>',.,, ... .) ·;·.· .. ¡-:.' . :-\ . ' .

,. . -:··. '.. ·

E 4.o MMP7 F 4.o STATl ns 3.5 00 3.5 o .,.... ns ., o l5 ~ 3.0 ~ 3.0 o o ~ o 0 oooO ~ ·~ 2.5 0 oo -~., 2.5 o o oo oo0 o ;:; oooo o¿:g 0 o -~2.0 ·e;, 2.0 ~Yao80 o -oc1'56g -~ oo o .5 ~ooo o0 O e oooo ] 1.5 o o ] 1.5 ~ o o "' "' ooo en o o en o ü 1.0 ü 1.0 o < < o 0.5 0.5 n=30 n=28 0.0 0.0 Non-neoplastic Adenocarcinoma Non-neoplastic Adenocarcinoma

Figura 39. Expresión de MMP7 y STAT1 en tejidos pancreáticos. A) Nula expresión de MMP? en tejido no neoplásico; B) Expresión moderada de MMP7 en tejido tumoral. C) Baja expresión de STAT1 en tejido no neoplásico; D) Expresión moderada de STAT1 en tejido tumoral y estroma reactivo. E) Análisis t test pareado de una cola de la expresión de MMP? en tejidos no neoplásicos y sus respectivos tejidos tumorales. F) Análisis t test pareado de una cola de la expresión de STAT1 en tejidos no neoplásicos y sus respectivos tejidos tumorales. Dos casos fueron removidos debido a plegamiento del core.

103 4.6.3.4. DPYSL4 dihydropyrimidinase-like 4

DPYSL4 fue identificada mediante la estrategia SSH como un transcrito up­ regulated en tejidos de ADP. El estudio inmunohistoquímco indica que una baja proporción de los pacientes expresan la proteína solo un baja proporción de las células tumorales. Las escasas células positivas presentan un patrón de tinción supranuclear de tipo granular (Figura 40A y 40C). Este gen no fue estudiado mediante tissue microarray.

DPYSL4 Queratina 19

. '·.\ .&

Figura 40. Expresión de DPYSL4 en tejidos de adenocarcinoma pancreático. A y C) DPYSL4 presenta un patrón de tinción granular de ubicación supranuclear en células tumorales. By D) Tinción de queratina 19 correspondiente a las imágenes en A y e respectivamente.

104 4.6.4. Genes identificados en la estrategia SSH contrastados con la base de datos del Human Atlas Protein.

Genes up-regulated escogidos de la estrategia SSH fueron contrastados con la base de datos del Human Atlas Protein y procedados de manera similar. La figura 41 muestra ejemplos de expresión en dos tejidos no neoplásicos y dos tejidos tumorales diferentes para cada proteína. Las proteínas codificadas por los genes EGFR (epidermal growth factor receptor), C1orf159 (chromosome 1 open reading frame 159), DNAH9 (dynein, axonemal, heavy chain 9), F3 (coagulation factor 111 (thromboplastin, tissue factor)), CD55 (molecule, decay accelerating factor for complement (Cromer blood group)), PPP4R4 (protein phosphatase 4, regulatory subunit 4), TAOK2 (TAO kinase 2), ZNF266 (zinc finger protein 266) presentan niveles bajos o nulos de expresión en tejidos no neoplásicos de páncreas. Por el contrario, es posible encontrar tejidos tumorales que expresan niveles moderados a intensos de la respectiva proteína (Figura 41 ). El hallazgo de tumores que sobre expresan estas proteínas en comparación con tejidos no neoplásicos se encuentra en concordancia con los hallazgos transcriptómicos identificados mediante la estrategia SSH.

105 ADP

0::: LL (.9 w

(j) 1.() T""" 't: o T""" (_)

(j) :r: <( z o

('() LL

1.() o1.() (_) :t~~~iJ?~~: r===:::;:==::::;

,"".'

. ~ ' co ,· co : . . C\1 -,'. LL ... /. z , . N ...... ··."

Figura 41. Imágenes de tinciones inmunohistoquímicas de genes de la estrategia SSH obtenidas del Human Atlas Protein. Tinciones IHQs de 8 genes up-regu/ated identificados mediante la estrategia Directa. ADP=adenocarcinoma ductal pancreático. Cada tejido no neoplásico y de adenocarcinoma provienen de muestras de pacientes independientes. 106 5. DISCUSION

El ADP es una enfermedad devastadora con un pésimo pronóstico. La única cura es la resección quirúrgica del órgano, pero esta opción solo es posible en -20% de los pacientes debido principalmente a la infiltración arterial del tumor y el diagnóstico tardío, el cual muchas veces requiere de tres o más visitas al médico para poder ser concretado [13]. Incluso en el grupo de pacientes que accede a la resección quirúrgica, la sobrevivencia media es menor a los 2 años y se ha observado, en aquellos pacientes que sobreviven por más tiempo, que un porcentaje de ellos nunca tuvieron adenocarcinoma ductal en el diagnóstico inicial [130]. Se ha propuesto que la única forma de sobrevivir al ADP es el diagnóstico temprano de la enfermedad seguido de la resección quirúrgica del páncreas [11 ]. Es por ello que es de urgente necesidad la identificación y desarrollo de nuevas estrategias de diagnóstico para mejorar así el desastroso pronóstico de esta enfermedad. En este proyecto de tesis hemos investigado el transcriptoma de tumores completos de ADP con el objetivo de identificar moléculas con el potencial de ser utilizadas como elementos diagnósticos de la enfermedad. Para este objetivo hemos empleado una plataforma de microarray de expresión génica mediante dos estrategias simultáneas, con el propósito de caracterizar los transcriptomas de alta y baja abundancia del ADP. Para estudiar el transcriptoma de baja abundancia hemos acoplado las tecnologías de microarray y de hibridación sustractiva por supresión (SSH). La SSH es una técnica de sustracción de cONA basada en la reacción en cadena de la polimerasa [96]. Este método permite normalizar y sustraer genes diferenciales entre dos poblaciones de cDNAs. La normalización iguala la abundancia de cONA y la sustracción excluye las secuencias que son comunes entre ambas poblaciones comparadas. Luego, mediante amplificación se obtienen secuencias específicas que provienen de las diferencias moleculares entre dos sistemas. Sin embargo, la caracterización y cuantificación de las clonas de cONA es laboriosa y ha limitado su utilización. Debido a esto, la técnica de SSH se ha acoplado a la tecnología de microarray para incrementar el rendimiento en la identificación de las secuencias.

107 En esta investigación, hemos mostrado que los transcritos de alta abundancia en el tumor, identificados mediante la estrategia Directa, se caracterizan por estar relacionados con la elevada presencia de tejido estroma! reactivo, el cual se compone principalmente de células pancreáticas estelares [123] que producen cantidades anormalmente elevadas de componentes de la matriz extracelular. Por el contrario, la aplicación de la estrategia SSH en las mismas muestras, empleando un tejido normal como muestra sustractora (driver), permite una reducción importante de trancritos asociados a reacción desmoplástica y revela genes de baja abundancia pero que presentan una relativa especificidad por el tejido tumoral epitelial. Un ejemplo de esto es la proteína codificada por el gen PLEKHM1, que presenta niveles bajos u nulos en tejido no tumoral pero se expresa en el tejido canceroso epitelial de aproximadamente el 30% de los casos estudiados, sin embargo, el número de células tumorales que expresan la proteína es variable.

5.1 La estrategia Directa permite la identificación de transcritos relacionados con la reacción desmoplástica del ADP

El estudio de la expresión génica empleando una estrategia Directa implica que la población de transcritos en estudio es hibridada directamente sobre el microarray o es amplificada previamente para alcanzar las cantidades necesarias. La metodología de amplificación mediante transcripción in vitro es la más frecuentemente usada, pues conserva la abundancia relativa de los transcritos [131]. Esta metodología permite el estudio de los transcriptoma nativos de los tejidos y ha sido usada ampliamente para la caracterización de perfiles de expresión génica en una variedad de tejidos tumorales [51, 52, 132, 133], incluyendo el ADP [134, 135]. Las masas tumorales de ADP se caracterizan por tener una alta heterogeneidad celular, donde es posible encontrar diversos tipos celulares distintos de las células epiteliales cancerosas [32]. El ADP presenta una extensa y compleja reacción desmoplástica compuesta por una matriz extracelular, fibroblastos y miofibroblastos activados, células inflamatorias, células sanqguíneas y vasos linfáticos que

108 distorsionan la arquitectura normal del tejido pancreático. Muchos tumores de origen epitelial como el cáncer de mama, próstata y ovario presentan una importante reacción desmoplástica con acumulción de células estromales. Sin embargo, el páncreas presenta la desmoplasia más extensa entre los cánceres epiteliales, observándose tumores donde el estroma reactivo representa hasta el 90% del volumen tumoral [21]. En nuestro estudio transcriptómico fue posible identificar un elevado número de genes relacionados con la biología estroma! del tumor paralelamente con genes relacionados con el tejido tumoral epitelial. Estos hallazgos se encuentran en concordancia con estudios previos que han utilizado la misma estrategia para caracterizar tejidos completos de ADP. Genes tales como COL 1A 1, COL4A 1 y COL6A3 (codifican para fibras de colágeno), CTHR1, FN1, POSTN, VCAN y FBN1 han sido identificados como proteínas de expresión predominantemente estromales. Las fibras de colágenos son componentes mayoritarios de la matriz extracelular de los tumores pancreáticos [136] y sus transcritos se expresan abundantemente en tejidos de ADP analizados mediante microarray [68, 137, 138]. El transcrito del gen COL 17A1 ha sido indentificado en nuestro estudio como up-regu/ated en ADP y asociado a matriz extracelular, sin embargo, se ha demostrado que su origen es el tejido tumoral epitelial y no el estroma reactivo [137]. A su vez, COL6A3 se expresa intensamente en la desmoplasia tumoral pancreática [139]. Versican (VCAN) y periostin (POSTN) son proteoglicanos que se expresan abundantemente en los tejidos tumorales pancreáticos, donde las células estelares representan la mayor fuente de estas moléculas, observándose escasa o nula expresión en células tumorales [140, 141]. En ADP y otros tumores se ha demostrado que periostin sustenta un microambiente favorable para el crecimiento tumoral [142], además de ser inducido por células metastásicas para iniciar la colonización de otros tejidos [143]. Fibronectin 1 (FN 1) es una glicoproteína de la matriz extracelular relacionada con procesos de adhesión y migración celular, cuyo transcrito se encuentra en alta abundancia en tejidos tumorales de ADP [68, 70, 137, 144]. En estudios in vitre se ha obervado que fibronectin 1 incrementa significativamente el comportamiento invasivo de líneas celulares pancreáticas [145].

109 Expresión de SPARC en fibroblastos peritumorales del cáncer de páncreas se encuentra asociado a un mal pronóstico [146, 147]. Estudios experimentales han sugerido que la expresión de SPARC es dependiente de procesos inflamatorios y acompaña la ca-expresión de otros componentes inflamatorios y de matriz extracelular tales como colágenos y fibronectinas [148]. La producción exacerbada de matriz extracelular no es exclusiva del ADP. La pancreatis crónica (PA) es un desorden inflamatorio progresivo donde el parénquima secretorio del páncreas es destruido y reemplazado por tejido fibrótico [149]. Varios genes identificados en nuestro estudio ya han sido asociados a los procesos fibróticos del ADP y de la PC. A nivel transcriptómico y proteómico el tejido desmoplástico del ADP y PC poseen muchos genes en común, tales como fibras de colágeno, FN1, CDH11, LTBP1, SPARC, HIF1A, MMP11, VCAN, POSTN y LYZ [75, 150-152]. El factor común entre la PC y el ADP es la activación y proliferación de las células estelares, mediada por injuria tisular, necrosis y procesos inflamatorios que estimulan la fibrosis [153]. La desmoplasia tumoral pancreática es de relevancia para el tratamiento de la enfermedad. En modelos murinos de cáncer pancreático donde el estroma reactivo es depletado se ha observado un incremento en la eficiencia de los tratamientos antineoplásicos, aumentando la sobrevivencia de los ratones y la biodistribución de las drogas antineoplásicas [23, 24]. Actualmente en el mercado no existen biomarcadores de origen estroma! para tumores epiteliales, sin embargo, se ha propuesto que el estroma es una fuente de marcadores para la evaluación de la enfermedad, que pueden ser combinados con marcadores específicos de tumor [154]. Un reciente estudio proteómico realizado en modelos HER2 de cáncer de mama, demostró que tanto en el inicio como en la progresión del tumor el proteoma plasmático es inundado con proteínas relacionados con matriz extracelular, reparación de heridas, sistema inmune, coagulación, cascada del complemento y metabolismo. Este estudio demuestra que una visión integrada de la masa tumoral y todos sus componentes celulares tienen el potencial de ser empleados en detección y diagnóstico del cáncer [155]. El rol del microambiente

110 tumoral y la fibrosis mediada por fibroblastos y células estelares son reconocidos actualmente como fenómenos de relevancia para la progresión del cáncer [156].

5.2 la estrategia SSH permite enriquecer grupos de genes distintos de los identificados mediante la estrategia Directa

La hibridación sustractiva por supresión permite la detección de transcritos de expresión diferencial entre dos muestras complejas independientemente de la abundancia relativa de los transcritos. Los transcritos de baja abundancia son enriquecidos mientras que los transcritos comunes entre las muestras y de una mayor abundancia son depletados. En nuestro estudio aplicamos la estrategia SSH a las mismas muestras estudiadas mediante la estrategia Directa. Para la sustracción empleamos como muestra driver un tejido normal de páncreas, el cual se compone aproximadamente en un 80-90% de células acinares y de un 10-20% de otros tipos celulares, tales como tejidos ductales de origen epitelial, células estelares y fibroblastos. La estrategia SSH permitió una fuerte reducción en la detección de transcritos relacionados con la biología estroma!. Una menor cantidad de transcritos relacionados con respuesta a heridas, adhesión celular, muerte celular programada, matriz extracelular, retículo endoplasmático y membrana basal, entro otros. Por el contrario, la estrategia SSH permitió un enriquecimiento de genes cuya función es predominantemente intracelular y que no fueron identificados en la estrategia Directa, tales como morfogénesis de órganos, respuesta al calcio, diferenciación de stem cells y mediadores de la vía Wnt, entre otros. Wnt-¡3-catenina es una vía de señalización de importancia para el desarrollo embrionario y requerido para la proliferación, diferenciación y morfogénesis de varios órganos [157]. Los ligandos de Wnt se unen a receptores de la familia Frizzled. La unión de los ligandos resulta en la inactivación de un complejo de proteínas citoplasmáticas (incluyendo adenomatous po/yposis coli (APC) y axin) que promueven la degradación proteosomal de {3-catenin, lo que resulta en la acumulación

111 citoplasmática y localización nuclear de esta proteína. A nivel nuclear, {3-catenín se une a los factores de transcripción TCF/LEF para activar la expresión de una serie de genes objetivo. Diecinueve ligandos Wnt han sido identificados en mamíferos y que son capaces de activar la vía canónica (dependiente de {3-catenín) y la vía no canónica (independiente de {3-catenín) a través de la unión de los ligandos con los receptores Frizzled y sus ce-receptores. Estudios recientes han comenzado a establecer el rol de la vía Wnt- {3-catenín en ADP. Aunque la acumulación de {3-catenín no es una característica universal de esta enfermedad, la acumulación tanto nuclear (10-60%) como citoplasmática (25-65%) es observada en lesiones PaniN y ADP [158-160]. Los ligandos canónicos (como WNT3 y WNT8B) se expresan intensamente en tejidos de ADP[160]. Un aumento en la expresión de DKK1, inhibidor de los ligandos Wnt secretados, está asociado con PaniNs avanzadas y ADP, además de incrementar el crecimiento y motilidad de células de ADP [161]. Sin embargo, la vía no canónica también se ha relacionado al desarrollo de la intensa reacción desmoplástica observada en el cáncer pancreático, a través del aumento de la expresión de WNT5A [162]. Mediante la estrategia SSH identificamos transcritos de la vía Wnt que no fueron identificados por la estrategia Directa. Entre los genes identificados encontramos a TCF7L2, NFAT5 y WNT9A. TCF7L2 (transcríptíon factor 7-/íke 2 (T-ce/1 specífíc, HMG-box)) ha sido relacionado principalmente a la diabetes de tipo 2, donde cambios en este gen definen el riesgo genético más determinante de esta enfermedad [163]. Sin embargo, se ha demostrado que TCF7L2 posee un rol represor de la transcripción mediante el cual restringe el crecimiento de tumores de cáncer colorectal [164]. Un reciente estudio publicado por The Cancer Genome Atlas Network reveló que miembros de la vía Wnt se encuentra mutados en más del 94% de los tumores de colon y recto, predominantemente en el gen APC. lnteresantemente, el gen TCF7L2 es uno de los 8 genes cuya mutación se encuentra con mayor frecuencia. Además, se encontró un nuevo gen fusión entre los genes TCF7L 1 (homólogo de TCF7L2) y NAV2. Esta fusión carece del dominio TCF3 que se une a {3-catenín. Estos hallazgos destacan el rol de la vía Wnt y de los genes TCF en el desarrollo del cáncer de colon [165]. 112 NFAT5 (nuclear factor of activated T-cel/s 5, tonicity-responsive) está relacionado a la vía de señalización de la integrina a5l34, mediante la cual incrementa la capacidad migratoria de células tumorales de mama [166]. Acerca de WNT9A y su relación con cáncer hay escasos reportes en la literatura. Mutaciones frecuentes de este gen han sido reportadas en ADP [31], sin embargo, no ha sido estudiado en detalle en este tipo de cáncer. Debido a su relativa novedad biológica, WNT9A fue escogido como gen candidato de estudio y será discutido más adelante. La estrategia SSH identificó genes relacionados con diferenciación celular. MSI2 (musashi homolog 2) ha sido relacionado a la malignidad de la leucemia mieloide crónica [167]. La expresión de MSI2 aumenta a medida que progresa la enfermedad, pero además es un indicador de un mal pronóstico. MSI2 participa de la vía de señalización que controla la diferenciación de células de leucemia mieloide crónica, por lo cual ha sido propuesta como una nueva alternativa terapéutica para esta enfermedad [168] RIF1 (RAP1 interacting factor homolog) fue identificada originalmente en levaduras como una proteína capaz de interactuar con Rap1, una molécula localizada en los telomeros [169]. RIF1 participa en el checkpoint de la fase S, contribuyendo a la inhibición de la replicación de DNA asociada con la activación de ATM [170]. La expresión de RIF1 cambia progresivamente durante la diferenciación de linajes celulares mamarios de primates no humanos, agrupándose con otros genes relacionados con diferenciación tales como ESR1, TFF1 AR y IGF1 [171]. La relación de esta proteína y el cáncer no ha sido estudiada en profundidad. ERCC2 ( excision repair cross-complementing rodent repair deficiency, complementation group 2) es un gen relacionado con la vía de reparación por escisión de nucleótidos que participa de la reparación del daño al DNA. Se ha demostrado que polimorfismos de ERCC2 son un factor pronóstico para el tratamiento con oxaliplatino en cáncer gástrico y de colon [172]. Además, en cáncer de pulmón se ha observado que polimorfismos de ERCC2 sirven como factores de riesgo de baja penetrancia al contribuir con la presentación de la enfermedad en fumadores [173].

113 Un estudio previo realizado por nuestro grupo de investigación aplicó la metodología de hibridación sustractiva por supresión al estudio de tumores avanzados de cáncer gástrico y cáncer de colon. De forma similar a lo observado en cáncer pancreático, esta metodología permitió una reducción en el enriquecimiento de genes relacionados con matrix extracelular y componentes inflamatorios. A su vez, esta metodología permitió el enriquecimiento de genes relacionados con vías de señalización intracelulares tales como Wnt, PI3K, angiogénesis y activación de células B y T en tumores de cáncer gástrico. Un análisis más profundo de la vía de señalización Wnt mostró que transcrito de CTBP1 se encuentra aumentado en más de 70 veces en tejidos tumorales en comparación a tejidos no neoplásicos adyacentes a la lesión tumoral. El silenciamiento mediante knockdown por RNAi del transcrito de CTBP1 sensibiliza a líneas celulares tumorales frente a drogas antineoplásicas de uso oncológico [174]. Esto demuestra la utilidad de esta plataforma en la identificación de objetivos moleculares con relevancia terapeútica. El desarrollo y perfeccionamiento de las plataformas de alto rendimiento para el análisis de expresión génica ha permitido identificar patrones que se asocian a una variedad de fenotipos tumorales. El panorama ha conducido esta área a ir más allá del clustering y de las correlaciones de expresión génica y en su lugar, utilizar un enfoque más dirigido para derivar firmas de expresión que representen "sub-fenotipos" o "meta­ genes" que reflejen fenómenos como el estatus de receptores de hormonas, sobrevivencia a la enfermedad, respuesta a terapias, la respuesta a hipoxia o la activación de vías de señalización oncogénicas que contribuyan al desarrollo tumoral [55, 175].

114 5.3 Ambas estrategias permiten identificar genes que codifican para proteínas de secreción.

Las proteínas de secreción representan una fuente de potenciales biomarcadores para biomedicina [110], especialmente para la detección de neoplasias [176, 177]. Mediante análisis bioinformático identificamos proteínas de secreción enriquecidas en ambas estrategias.

El año 2009, Harsha et al publicaron un compendio de potenciales biomarcadores de cáncer pancreático construido a partir de un análisis sistemático de la literatura científica y de repositorios de bases de datos de DNA microarray, tales como GEO (NCBI), Oncomine y ArrayExpress (EBI) [178]. Comparando nuestros datos con este estudio, la lista de genes obtenida mediante la estrategia Directa contiene 77 proteínas de secreción, de las cuales 47 (61%) ya han sido descritas por Harsha et al como potenciales biomarcadores. Por otro lado, la lista SSH contiene 26 proteínas de secreción, de las cuales solo 9 (35%) fueron descritas en el compendio.

Dentro de los genes descritos en la lista Directa encontramos a ANXA2, FAS, C4BP4, CD59, CALU, VCAN, IL8, WNT9A, CXCL5, TGFBI. Varios de estos genes ha sido estudiados con anterioridad en cáncer pancreático. ANXA2 ha sido propuesto como marcador de cáncer pancreático [179], siendo posible encontrar la proteína en orina de pacientes [180]. El transcrito de CD59 ha sido identificado previamente como sobre expresado en ADP mediante análisis de microarray [69]. Además, la proteína de CD59 ha sido identificada en muestras de orina de pacientes con cáncer pancreático [180]. El transcrito de FAS se ha encontrado previamente aumentado en tejidos de adenocarcinoma ductal [181] y su abundancia en suero permite discriminar pacientes con cáncer de controles con lesiones benignas y controles saludables [182].

El transcrito de la proteína y transcrito de Calumenin (CALU) se expresan a mayores niveles en tumores de cáncer de colon en comparación con adenomas y tejidos normales [183]. CALU fue escogido como candidato de estudio y es discutido más adelante.

115 En un reciente estudio proteómico, Turtoi et al identificaron proteínas de alta expresión en tejidos de adenocarcinoma pancreático. Entre las proteínas identificadas se encuentran FN1, TGFBI y AGRN las cuales fueron identificadas en la estrategia Directa de nuestro estudio, como transcritos que codifican para proteínas de secreción [184].

Mediante la estrategia SSH identificamos, entre otros, a MMP7, WNT9A, NMU, EGFR, CP, NMU, y OLFML 1. NMU (Neuromedin) es un neuropéptido que ha sido descrito como molécula de alta expresión en tejidos tumorales de cáncer pancreático [185]. EGFR (epidermal growth factor receptor) es el objetivo terapeútico de la droga Erlotinib, actualmente aprobada para el tratamiento de cáncer pancreático. En tejidos de ADP se ha observado que la expresión de EGFR se correlaciona con el pronóstico de la enfermedad [186] y la presencia de metástasis en nodos linfáticos y órganos distantes [187]. CP (ceruplasmin) codifica para una proteína sérica que participa en la homeostasis del hierro y que cuyos niveles se encuentran aumentados en los sueros de pacientes con cáncer pancreático [188], aunque su especificidad por la enfermedad es baja [189]. La proteína de OLFML 1 (o/factomedin-like 1), cuyo transcrito fue identificado mediante nuestra estrategia SSH, fue descrita recientemente como una molécula sobre expresada en un grupo de tumores de adenocarcinoma ductal pancreático [184]. En términos generales, ambas estrategias permitieron la identificación de genes que codifican para proteínas de secreción. Muchos de los genes han sido descritos previamente en la literatura y apoyan los resultados obtenidos mediante nuestro estudio. Sin embargo, otros genes han sido escasamente estudiados y fueron evaluados como potenciales candidatos.

116 5.4 Validación de genes candidatos de la estrategia Directa

PMEPA1 prostate transmembrane protein, androgen induced 1

PMEPA1 (llamado alternativamente como TMEPAI, STAG1, ERG1.2 o N4wbp4) es una proteína capaz de ser inducido por testosterona y relacionado con procesos tumorales [190, 191]. La expresión de PMEPA1 puede ser inducido por el factor de crecimiento transformante TGF-13 [192-194], además de ser identificado como up­ regu/ated en cáncer de estómago, rectal [122, 195] y de próstata [196]. Recientemente se ha demostrado que una posible función fisiológica de PMEPA 1 es participar en un circuito de retroalimentación negativa (negative feedback loop) que controla la duración e intensidad de la vía de señalización de TGF-13/Smad [197]. En estudios in vitro de cáncer de mama se ha observado que PMEPA 1 promueve las funciones promotoras de tumores de TGF-13, en desmedro de sus funciones supresoras de tumores [198]. Ambas investigaciones fueron publicadas posteriormente a nuestra decisión de caracterizar PMEPA1 en tejidos de ADP. En nuestro estudio identificamos una alta expresión de transcritos de PMEPA 1 en tejidos de ADP. Un estudio inmunohistoquímico de la proteína de PMEPA1 confirma que su expresión es predominantemente epitelial. .. Un estudio transcriptómico realizado en tumores de ADP identificó al transcrito de PMEPA 1 como up-regulated en tumores y asociado a una firma de expresión dependiente de TGF-13 [199]. Sin embargo, nuestro estudio es el primero en realizar estudios morfológicos de expresión de PMEPA1 en ADP.

BHLHE40 basic helix-/oop-helix family, member e40

El gen BHLHE40 (también conocido como BHLHB2, SHARP2, Stra13 o DEC1) es un factor de transcripción basic helix-loop-helix que ha sido identificado como sobre­ expresado en cáncer de mama [200] y gástrico [201]. BHLHE40 se caracteriza por ser un gen que responde al descenso de los niveles de oxígeno en células y tejidos. Esta asociación con hipoxia ha sido encontrada en diferentes tumores [200, 202]. Además 117 de hipoxia, este gen puede ser estimulado por TGF-~ [203, 204]. En una investigación reciente, realizada en cáncer de mama, se observó que la sobre expresión de

BHLHE40 inducida por TGF-~ previene la apoptosis y el knockdown de este gen mediante RNA de interferencia reduce la sobrevivencia inducida por TGF-~ [205]. El reporte más interesante vinculado a BHLHE40 fue publicado el año 2010 y describe el uso de la metodología ARACNe para determinar la interacción entre factores de transcripción y moléculas objetivos. Mediante este algoritmo, BHLHE40 fue identificado como uno de cinco factores de transcripción relacionados con la transformación mesenquimal de gliomas de alto grado. La transformación mesenquimal posee similitudes con la transición epitelio-mesenquimal [206]. Sin embargo, BHLHE40 no fue estudiado en profundidad en este estudio. Wang et al realizaron un estudio descriptivo y funcional de BHLHE40 en ADP (Investigación publicada posteriormente a nuestra elección de BHLHE40 como gen candidato de estudio). Esta proteína se expresa en alrededor del 50% de las células acinares y en una bajo porcentaje de células ductales, con un patrón predominantemente nuclear. En tejidos tumorales se observó una mayor expresión de esta proteína, con un patrón nuclear predominante, aunque alrededor del 8% de los tejidos presentaron patrón citoplasmático granular. Es de destacar que nuestro estudio inmunohistoquímico identificó un patrón predominantemente citoplasmático mientras que este Wang et al identificaron un patrón nuclear [207]. Los anticuerpos empleados en ambos estudios son diferentes. Nuestro anticuerpo fue validado por el manufacturador y detecta la proteína predicha más otras bandas inespecíficas, además de detectar tinción nuclear de BHLHE40 en líneas celulares mediante inmunofluorescencia pero no mediante inmunohistoquímica en tejidos FFPE (http://www.proteinatlas.org/ENSG000001341 07/antibody).

CALU calumenin El transcrito de CALU fue identificado como up-regulated en cáncer pancreático y gástrico en la estrategia directa de microarray empleada en nuestro laboratorio, además de encontrarse en la lista sustractiva de colon. Calumenin ha sido identificada como una proteína de retículo endoplásmico que actúa como inhibidor de la y- 118 carboxilación dependiente de vitamina K durante la síntesis de factores de coagulación [208, 209], además de participar en los procesos de migración y proliferación de neuronas [21 0]. Ca/umenin extracelular ha sido asociada a modificaciones de citoesqueleto y ciclo celular en fibroblastos, mediante un hipotético mecanismo paracrino o autocrino [211]. Esta molécula se encuentra localizada en la vía secretora, tanto en retículo endoplasmático como en aparato de golgi y es secretada al espaci o extracelular de células en cultivo [212]. La evidencia de que ca/umenin es secretada al espacio extracelular y que además posee un péptido señal para ser secretada, permite suponer que esta proteína posee el potencial de ser evaluada como biomarcador. Tres estudios independientes han corroborado altos niveles de expresión del transcrito y proteína de calumenin en tejidos de cáncer colon [183, 213] y cáncer endometrial [214]. Un aumento en la expresión de la proteína calumenina ha sido encontrado en células de cáncer de cuello uterino que han adquirido resistencia a cisplatino [215]. La relación entre CALU y procesos tumorales no ha sido estudiada en profundidad

119 5.5 Validación de genes candidatos de la estrategia SSH

PLEKHM1 pleckstrin homology domain containing, family M (with RUN domain) member 1

Esta molécula fue denominada originalmente como AP162 y fue identificada en cáncer de colon como una proteína de peso aparente de 162 kDa que forma un complejo con una molécula similar a mucina, sin embargo, su función y relación con el cáncer no ha sido estudiada en detalle [124]. La mutación de este gen es responsable de la osteopetrosis autosomal recesiva tipo VI (OPTB6) [125], enfermedad caracterizada por el incremento de la densidad ósea y que es producida como consecuencia de la alteración de la maduración/acidificación del compartimiento endosomal en osteoclastos [126]. En estudio reciente se demostró que la sobre expresión de PLEKHM1 regula negativamente el transporte desde los endosomas tempranos a los lisosomas. Por el contrario, el knockdown de PLEKHM1 acelera el transporte de moléculas a los lisosomas, incrementando la degradación de proteínas dentro de los lisosomas [216]. Esto se encuentra en concordancia con el rol propuesto para PLEKHM1 en la reabsorción de hueso por parte de los osteoclastos. PLEKHM1 fue identificado en nuestro estudio mediante la estrategia SSH como up-regulated en tejidos de ADP y validado posteriormente mediante qRT-PCR. Mediante un estudio inmunohistoquímico de tejidos de ADP confirmamos que la proteína se expresa en -30% de los casos de forma exclusiva en células tumorales de origen epitelial. La mayoría de los tejidos no tumorales adyacentes a la lesión tumoral no expresa la proteína y no hay evidencia de expresión de la proteína en tejidos normales de páncreas. La estrategia SSH tiene por objetivo identificar genes de relativa especificidad en tejido tumoral, indepedientemente de su abundancia. PLEKHM1 es un ejemplo de un gen de específico de tejido tumoral pero cuya abundancia es baja dentro de la masa tumoral, donde solo un porcentaje variable de las células tumorales expresa la proteína. Es de interés para futuras investigaciones establecer la relación entre este gen y el cáncer pancreático.

120 WNT9A wingless-type MMTV integration site family, member 9A

WNT9A pertenece a la familia de proteínas WNT y ha sido relacionado al proceso de condrogénesis e integridad articular en modelos murinos [127], así como a la morfogénesis y proliferación celular en epitelio hepático de modelos aviares [128]. En zebrafish ha sido vinculado al desarrollo del paladar y mandíbula inferior [129]. Es de relevancia destacar que WNT9A es un gen que se encuentra mutado con frecuencia en el adenocarcinoma ductal pancreático y pertenece a un grupo de 29 genes de la vía Wnt/Notch, la cual se encuentra alterada en el 100% de los canee res pancreaticos [31]. El rol de WNT9A en cáncer no ha sido estudiado en profundidad. Mediante un estudio inmunohistoquímico de tejidos de ADP confirmamos que la proteína se expresa de forma leve o nula en duetos pancreáticos de morfología normal, encontrándose ausente en otros tipos celulares de páncreas. Además, mediante tissue microarray corroboramos que los niveles de expresión de WNT9A se encuentran aumentados significativamente en tejidos tumorales en comparación a tejidos no neoplásicos de páncreas. La familia de ligandos Wnt está conformada por proteínas de secreción que cumplen su rol de señalización en distancias cortas [217]. Hay evidencia de que WNT9A se secreta al espacio extracelular [128] y mayores niveles del transcrito se han observado en cáncer de mama, colorectal, riñon y pulmón, analizado mediante búsqueda en bases de datos [218]. Por lo tanto, esta proteína representa un interesante candidato para ser estudiado como biomarcador de cáncer páncreatico. Recientemente se ha establecido un rol de la vía no canónica de Wnt en el proceso metastásico del cáncer de páncreas. Un grupo de células tumorales circulantes poseen niveles aumentados de Wnt2, el cual favorece la sobrevivencia independiente de anclaje y aumenta en el número de metástasis en modelos murinos [219]. Si bien la vía de señalización de Wnt parece no participar en la iniciación del ADP, recientes evidiencia indican un rol de esta vía en la mantención de la enfermedad [220]. Futuras investigaciones son requeridas para establecer el rol concreto de esta vía en el ADP y concretamente, analizar la utilidad de WNT9A en diagnóstico y posiblemente terapia. 121 MMP7 se encuentra up-regu/ated en PaniNs y ADP [221, 222] y su expresión se correlaciona con menor sobrevida y tamaño tumoral, estado de nodo linfático y metástasis en órganos distantes [222-224]. La deleción de MMP7 en modelos murinos de ADP, ·disminuye significativamente el tamaño del tumor y las metástasis distantes. De manera importante, la medición de MMP7 en sueros de pacientes se correlaciona con la enfermedad metastásica y la sobrevivencia de los pacientes [225]. lnteresantemente, MMP7 ha sido asociada previamente a la vía de señalización de Wnt en modelos murinos de tolerancia a transplantes de riñón [226], así como también en modelos de cáncer de ovario, donde la expresión específica de MMP7 es regulada activamente por WNT7 A y mediada por {3-catenin/TCF [227]. A pesar de que MMP7 ya ha sido estudiada previamente en ADP, incluimos esta molécula en nuestro estudio con el objetivo de evaluar su especificidad por tejidos tumorales, debido a que MMP7 fue identificada en la estrategia SSH de nuestro estudio. Corroboramos que MMP7 se expresa en células tumorales de forma intensa, sin embargo, es posible encontrar regiones positivas en el tejido no neoplásico adyacente a las lesiones tumorales. Esto se ve reflejado en el análisis de tissue microarrays, donde a pesar de que el tejido tumoral presenta una mayor tendencia a la sobre expresión, las diferencias no fueron significativamente estadísticas debido a la expresión de MMP7 en los tejidos no neoplásicos.

5.6 Realidad actual del diagnóstico de adenocarcinoma ductal pancreático

La ventana de tiempo terapéutica para un tumor puede ser descrita como el período de tiempo desde el diagnóstico de la lesión pre-neoplásica más temprana removible quirúrgicamente hasta el estado avanzado o metastásico donde la cura de la enfermedad ya no es posible. Debido a la localización retroperitoneal del páncreas, la falta de síntomas específicos y la limitada sensibilidad de los métodos de diagnóstico, casi todos los pacientes se diagnostican en una fase incurable, es decir, la ventana de tiempo está cerrada para estos pacientes.

122 A pesar de una teórica resección quirúrgica (sin tumor en los márgenes y sin metástasis a ganglios linfáticos) y bajo quimioterapia adyuvante, la recurrencia de la enfermedad es solo cuestión de tiempo. Esta recurrencia se debe a la diseminación subclínica de la enfermedad y la resistencia innata a la terapia. Por lo tanto, teniendo en consideración que la extracción quirúrgica del órgano es la única posibilidad de curación para los pacientes con ADP, es de suma importancia detectar los cánceres en una etapa pre-invasiva [228]. El problema aumenta cuando se trata de la detección temprana de las lesiones. La toma de biopsias pancreáticas no es un procedimiento realizado de forma rutinaria para la detección de lesionaes pre-neoplásicas. Además, no se sabe con exactitud que tipo de lesiones deben ser buscadas en los tejidos, debido a que el origen celular del páncreas es un tema de debate permanente. Por último, aunque los especialistas traten de buscar todas las variedades de lesiones conocidas, los precursores que no son de origen quístico están por debajo del umbral de detección actual de 5-8 mm, empleando herramientas radiológicas convencionales [229, 230]. Actualmente, la molécula CA 19-9 es el único biomarcador que posee utilidad clínica demostrada, permitiendo monitorear la terapia y detectar tempranamente la recurrencia de la enfermedad después del tratamiento en pacientes con cáncer pancreático ya diagnosticado. Sin embargo, CA 19-9 posee importantes limitaciones. No es un marcador específico de cáncer pancreático; niveles aumentados de esta molécula pueden estar elevados en otras condiciones como la colestasis. Además, los pacientes que son negativos para el antígeno Lewis A o B (aproximadamente el 10% de los pacientes con cáncer de páncreas) son incapaces de sintetizar CA19-9 y tienen niveles no detectables, incluso en etapas avanzadas de la enfermedad. Aunque la medición de este biomarcador es de utilidad en pacientes con cáncer pancreático conocido, el uso de este biomarcador como herramienta de diagnóstico ha tenido resultados decepcionates [12]. Otros biomarcadores han sido propuestos y se encuentran en distintos niveles de validación clínica [179]. La mayoría de los marcadores tumorales carecen de la sensibilidad y especificidad necesarias para detectar la enfermedad en estados tempranos. Recientes evidencias abordan esta problemática. Empleando como modelo el cáncer 123 de ovario y la detección del biomarcador CA-125, Hori et al propusieron rigurosamente que la tasa de liberación de biomarcadores a la sangre se encuentra miles de veces por debajo de los límites necesarios para detectar un tumor durante su primera década de desarrollo [231]. Esta evidencia destaca una problemática que ha afectado el campo del diagnóstico oncológico durante décadas: las herramientas moleculares desarrolladas para el screening de enfermedades neoplásicas son incapaces de detectar la enfermedad en etapas tempranas, a pesar de la abundante evidencia que indica que los pacientes poseen mejor expectativa de vida si el órgano es son removido en etapas tempranas [11]. En este contexto, nuevas moléculas y tecnologías emergentes ofrecen la oportunidad de aumentar la sofisticación del diagnóstico oncológico. La metodología de hibridación sustractiva acoplada a microarray ha sido empleada previamente para identificar nuevos genes con potencial de ser usados como elementos de diagnóstico o terapia en cáncer [88-90, 92, 93, 232]. Sin embargo, una característica común de estos estudios es que ninguno realizó validaciones para verificar si las variaciones observadas a nivel de transcritos se correlacionaban con los correspondientes niveles de proteínas. En nuestro estudio identificamos a WNT9A y PLEKHM1 como transcritos diferenciales y validamos que esa expresión diferencial se extiende hasta los niveles de la proteína. Ambas proteínas se expresan nula o escasamente en tejidos no neoplásicos y normales. La aplicación de moléculas específicas de tumor posee un gran potencial en cáncer pancreático. Plectin-1 (PLEC) es un ejemplo de este tipo de proteínas [233]. Plectin-1 fue identificada como una proteína expresada con alta especificidad en tejidos de ADP. Aunque se expresa en otros tejidos normales, es posible detectar Plectin-1 en lesiones tumorales primarias y metastásicas mediante bio-imagenología In vivo [234]. Otro ejemplo de la aplicación de esta tecnología es la detección de las enzimas catepsinas en tejidos de ADP. Las cate psi nas son enzimas que se expresan en altos niveles en lesiones preneoplásicas y tumores de ADP. Mediante microscopía confocal de fluorescencia es posible detectar lesiones tumorales preneoplásicas de forma sensible y específica en modelos murinos In vivo y en tiempo real [116].

124 Las tecnologías de bio-imagenología han emergido como alternativas para mejorar la situación actual del diagnóstico oncológico [235]. A futuro, es de interés evaluar si transcritos identificados mediante la estrategia indirecta pueden ser útiles para la detección de la enfermedad mediante tecnologías como la bio-imagenología y ser aplicadas posteriormente a escenarios clínicos reales. En resumen, la alta heterogeneidad celular y la biología de matriz extracelular del cáncer pancreático contribuyen enormemente a su firma transcriptómica, afectando así la identificación de transcritos de baja abundancia mediante tecnologías como el microarray. La aplicación de la tecnología de hibridación sustractiva por supresión acoplada a microarray permite diferenciar los transcriptomas de alta y baja abundancia (Figura 43A). lnteresantemente, es posible encontrar transcritos de baja abundancia de expresión diferencial entre tejidos no neoplásicos y tejidos tumorales pancreáticos que poseen el potencial de ser usados como marcadores de la enfermedad (Figura 438).

125 A Transcriptoma de Transcriptoma de alta abudancia Adenocarcinoma baja abundancia Wnt Transporte de Matriz Epitelio Inflamación pathway iones metal

Colágenos CEACAMs WNT9A CP Fibronectina 1 ANXA8 NFAT5 KCNMA1 Periostin t CLDN18 t TCF7L2 t KCNN2 t Versican Catepsinas PPP2R1B ARMC1 MMPs TOP2A TBL1Y SLC39A10 SPARC APC MFI2 CCND1 KCNG2 Retículo Citoesqueleto MMP7 endoplasmático de actina Respuesta a Diferenciación daño del DNA de Stem cel/s PLOD2 MYBPC1 EDEM3t ANLN t CHEK1 APC CALU ACTA2 ERCC2 MSI2 FKBP1B CAPG2UW t t Células Células LIG3 RIF1 DSE DAG1 endoteliales tumorales FBX018 ERCC2 RAD51 ACE

B Expresión diferencial de transcritos de baja abundancia No neoplásico Adenocarcinoma Adenocarcinoma

<( m 1-z 5

WNT9A PLEKHM1 Función: Participa en condrogénesis Función: Participa en endocitosis Vía de señalización de WnUB-catenina Regulador negativo del transporte (posiblemente vía canónica) endocítico, pero no la maduración de autofagosomas. Vía de señalización de Rab7 Figura 43. Transcriptomas de alta y baja abundancia del cáncer pancreático. A) El cáncer pancreático se caracteriza por su heterogeneidad celular y la presencia de una intensa reacción desmoplástica, la cual es predominante en los estudios de transcriptomas tumorales. La estrategia indirecta (SSH) reduce los transcritos de alta abundancia y enriquece el transcriptoma de baja abundancia de los tumores pancreáticos, donde destacan genes relacionados con la vía Wnt y diferenciación de stem cells. 8) El transcriptoma de baja abundancia contiene transcritos que codifican para proteínas que se expresan diferencialmente en el tumor, como es el caso de WNT9A y PLEKHM1. Futuros estudios son necesarios para definir el potencial diagnóstico de estas proteínas.

126 6. CONCLUSIONES GENERALES

La estrategia Directa permitió la identificación de marcadores tumorales conocidos y nuevos candidatos. Sin embargo, debido al diseño experimental empleado enriquece además genes asociados a la extensa reacción desmoplástica presente en las masas tumorales del adenocarcinoma ductal pancreático. Aún queda por definir si los genes de origen estroma! poseen relevancia para el diagnóstico de esta enfermedad.

Por otra parte, la aplicación de la estrategia SSH a las mismas muestras de la estrategia Directa, empleando como muestra sustractora un tejido normal de páncreas, permitió sustraer genes de alta abundancia presentes en la masa tumoral, principalmente componentes de matriz extracelular y genes de sistema inmune. Paralelo a esto, la estrategia SSH permitió el enriquecimiento de genes de mediana y baja abundancia cuya ubicación es predominantemente intracelular y relacionados a vías de señalización como es el caso de la vía Wnt, ofreciendo así nuevos genes candidatos.

De entre los genes candidatos de la estrategia Directa observamos que los productos génicos de PMEPA 1 y BHLHE40 se expresan con mayor abundancia en tejidos tumorales en comparación a tejidos adyacentes a la lesión tumoral. Por otro lado, el producto génico de CALU se expresa abundantemente es las células pancreáticas estelares activadas, presentes en el estroma reactivo de los tumores.

De entre los genes candidatos de la estrategia SSH observamos que los productos génicos de PLEKHM 1 y WNT9A se expresan con mayor abundancia en tejidos tumorales en comparación a tejidos adyacentes a la lesión tumoral. Ambos genes se expresan exclusivamente en células de origen epitelial. PLEKHM1 cumple con la teoría de la estrategia SSH: un gen de nula o muy baja presencia en tejidos no neoplásicos pero presente específicamente en los tejidos tumorales del cáncer pancreático. 127 128 7. BIBLIOGRAFÍA

1. Mukherjee S: The emperor of al/ maladies: a biography of cancer. 1st Scribner hardcover edn. New York: Scribner; 2010.

2. Nelson CM, Bissell MJ: Of extracellular matrix, scaffolds, and signaling: tissue architecture regulates development, homeostasis, and cancer. Annual review of ce// and developmenta/ biology 2006, 22:287-309.

3. Apte MV, Haber PS, Applegate TL, Norton ID, McCaughan GW, Korsten MA, Pirola RC, Wilson JS: Periacinar stellate shaped cells in rat pancreas: identification, isolation, and culture. Gut 1998,43:128-133.

4. Haber S, Uhlen M: Human protein atlas and the use of microarray technologies. Curr Opin Biotechno/2008, 19:30-35.

5. WHO (Ed.). World Cancer Report 2008. Lyon2008.

6. Medina E, Kaempffer AM: Cancer mortality in Chile: epidemiological considerations. Rev Méd Chile 2001, 129:1195-1202.

7. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, Thun MJ: Cancer statistics, 2008. CA Cancer J Clin 2008, 58:71-96.

8. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D: Global cancer statistics. CA Cancer J Clin 2011, 61:69-90.

9. Maitra A, Hruban RH: Pancreatic cancer. Annu Rev Patho/2008, 3:157-188.

10. Pavlidis N, Briasoulis E, Hainsworth J, Greco FA: Diagnostic and therapeutic management of cancer of an unknown primary. Eur J Cancer 2003, 39:1990-2005.

11. Lohr M: ls it possible to survive pancreatic cancer? Nat Clin Pract Gastroenterol Hepatol 2006, 3:236-237.

12. Hidalgo M: Pancreatic cancer. N Engl J Med 2010, 362:1605-1617.

13. Lyratzopoulos G, Neal RO, Barbiere JM, Rubin GP, Abel GA: Variation in number of general practitioner consultations befo re hospital referral for cancer: findings from the 201 O National Cancer Patient Experience Survey in England. Lancet Onco/2012, 13:353-365.

14. Vincent A, Herman J, Schulick R, Hruban RH, Goggins M: Pancreatic cancer. Lancet 2011, 378:607-620.

15. Poste G: Bring on the biomarkers. Nature 2011, 469:156-157.

129 16. Carmichael J, Fink U, Russell RC, Spittle MF, Harris AL, Spiessi G, Blatter J: Phase 11 study of gemcitabine in patients with advanced pancreatic cancer. Br J Cancer 1996, 73:101-105.

17. Burris HA, 3rd, Moore MJ, Andersen J, Green MR, Rothenberg ML, Mediano MR, Cripps MC, Portenoy RK, Storniolo AM, Tarassoff P, et al: lmprovements in survival and clinical benefit with gemcitabine as first-line therapy for patients with advanced pancreas cancer: a randomized trial. J Clin Onco/1997, 15:2403-2413.

18. Koorstra JB, Hustinx SR, Offerhaus GJ, Maitra A: Pancreatic carcinogenesis. Pancreatology 2008, 8:110-125.

19. Adsay NV, Thirabanjasak O, Altinel O: Spectrum of Human Pancreatic Neoplasia. In Pancreatic Cancer. Edited by Lowy AM, Leach SO, Philip PA: Springer; 2008: 3-26.[Pollock RE (Series Editor): M. D. Anderson So/id Ttumr Oncology Series].

20. Bapat AA, Hostetter G, Von Hoff DO, Han H: Perineural invasion and associated pain in pancreatic cancer. Nat Rev Cancer 2011, 11:695-707.

21. Neesse A, Michl P, Frese KK, Feig C, Cook N, Jacobetz MA, Lolkema MP, Buchholz M, Olive KP, Gress TM, Tuveson DA: Stromal biology and therapy in pancreatic cancer. Gut 201 O.

22. Yen TW, Aardal NP, Bronner MP, Thorning DR, Savard CE, Lee SP, Bell RH, Jr.: Myofibroblasts are responsible for the desmoplastic reaction surrounding human pancreatic carcinomas. Surgery 2002, 131:129-134.

23. Olive KP, Jacobetz MA, Davidson CJ, Gopinathan A, Mclntyre O, Honess O, Madhu B, Goldgraben MA, Caldwell ME, Allard O, et al: lnhibition of Hedgehog signaling enhances delivery of chemotherapy in a mouse model of pancreatic cancer. Science 2009, 324:1457- 1461.

24. Diop-Frimpong B, Chauhan VP, Krane S, Boucher Y, Jain RK: Losartan inhibits collagen 1 synthesis and improves the distribution and efficacy of nanotherapeutics in tumors. Proc Natl Acad Sci U S A 2011, 108:2909-2914.

25. Caldas C, Kern SE: K-ras mutation and pancreatic adenocarcinoma. lnt J Pancreatol 1995, 18:1-6.

26. Smit VT, Boot AJ, Smits AM, Fleuren GJ, Cornelisse CJ, Bos JL: KRAS codon 12 mutations occur very frequently in pancreatic adenocarcinomas. Nuc/eic Acids Res 1988, 16:7773- 7782.

27. Almoguera C, Shibata O, Forrester K, Martin J, Arnheim N, Perucho M: Most human carcinomas of the exocrine pan creas contain mutant e-K-ras genes. Ce//1988, 53:549-554.

28. Roa JC, Anabalon L, Tapia O, Melo A, de Aretxabala X, Roa 1: [Frequency of K-ras mutation in biliary and pancreatic tumors]. Rev Med Chi/2005, 133:1434-1440.

130 29. Hingorani SR, Petricoin EF, Maitra A, Rajapakse V, King C, Jacobetz MA, Ross S, Conrads TP, Veenstra TD, Hitt 8A, et al: Preinvasive and invasive ductal pancreatic cancer and its early detection in the mouse. Cancer Ce/12003, 4:437-450.

30. Aguirre AJ, 8ardeesy N, Sinha M, Lopez L, Tuveson DA, Horner J, Redston MS, DePinho RA: Activated Kras and lnk4a/Arf deficiency cooperate to produce metastatic pancreatic ductal adenocarcinoma. Genes Dev 2003, 17:3112-3126.

31. Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, Angenendt P, Mankoo P, Carter H, Kamiyama H, Jimeno A, et al: Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science 2008,321:1801-1806.

32. Samuel N, Hudson TJ: The molecular and cellular heterogeneity of pancreatic ductal adenocarcinoma. Nature reviews Gastroenterology & hepatology 2011, 9:77-87.

33. Singh A, Greninger P, Rhodes D, Koopman L, Violette S, 8ardeesy N, Settleman J: A gene expression signature associated with "K-Ras addiction" reveals regulators of EMT and tumor cell survival. Cancer Ce/12009, 15:489-500.

34. Weinstein 18: Cancer. Addiction to oncogenes--the Achilles heal of cancer. Science 2002, 297:63-64.

35. Moskaluk CA, Hruban RH, Kern SE: p16 and K-ras gene mutations in the intraductal precursors of human pancreatic adenocarcinoma. Cancer Res 1997, 57:2140-2143.

36. Campbell PJ, Yachida S, Mudie LJ, Stephens PJ, Pleasance ED, Stebbings LA, Morsberger LA, Latimer C, Melaren S, Lin ML, et al: The patterns and dynamics of genomic instability in metastatic pancreatic cancer. Nature 201 O, 467:1109-1113.

37. Yachida S, Jones S, 8ozic 1, Antal T, Leary R, Fu 8, Kamiyama M, Hruban RH, Eshleman JR, Nowak MA, et al: Distant metastasis occurs late during the genetic evolution of pancreatic cancer. Nature2010, 467:1114-1117.

38. Maitra A, Adsay NV, Argani P, lacobuzio-Donahue C, De Marzo A, Cameron JL, Yeo CJ, Hruban RH: Multicomponent analysis of the pancreatic adenocarcinoma progression model using a pancreatic intraepithelial neoplasia tissue microarray. Mod Patho/2003, 16:902-912.

39. lacobuzio-Donahue CA: Genetic evolution of pancreatic cancer: lessons learnt from the pancreatic cancer genome sequencing project. Gut 2011.

40. Wang Z, Gerstein M, Snyder M: RNA-Seq: a revolutionary tool for transcriptomics. Nature reviews Genetics 2009, 10:57-63.

41. Vogel C, Marcotte EM: lnsights into the regulation of protein abundance from proteomic and transcriptomic analyses. Nature reviews Genetics 2012:227-232.

42. de Sousa Abreu R, Penalva LO, Marcotte EM, Vogel C: Global signatures of protein and mRNA expression levels. Mol Biosyst 2009, 5:1512-1526. 131 43. Maier T, Guell M, Serrano L: Correlation of mRNA and protein in complex biological samples. FEBS Lett 2009, 583:3966-3973.

44. Schwanhausser B, Busse O, Li N, Oittmar G, Schuchhardt J, Wolf J, Chen W, Selbach M: Global quantification of mammalian gene expression control. Nature 2011, 473:337-342.

45. Hebenstreit O, Fang M, Gu M, Charoensawan V, van Oudenaarden A, Teichmann SA: RNA sequencing reveals two major classes of gene expression levels in metazoan cells. Molecular systems bíology 2011, 7:497.

46. Islam S, Kjallquist U, MolinerA, Zajac P, Fan JB, Lonnerberg P, Linnarsson S: Characterization of the single-cell transcriptional landscape by highly multiplex RNA-seq. Genome Res 2011,21:1160-1167.

47. Hah N, Oanko CG, Core L, Waterfall JJ, Siepel A, Lis JT, Kraus WL: A rapid, extensive, and transient transcriptional response to estrogen signaling in breast cancer ce lis. Ce//2011, 145:622-634.

48. Mercer TR, Gerhardt OJ, Oinger ME, Crawford J, Trapnell C, Jeddeloh JA, Mattick JS, Rinn JL: Targeted RNA sequencing reveals the deep complexity of the human transcriptome. Nat Bíotechnol2012, 30:99-104.

49. Alizadeh AA, Eisen MB, Oavis RE, Ma C, Lossos IS, Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, et al: Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000, 403:503-511.

50. Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, Pollack JR, Ross OT, Johnsen H, Akslen LA, et al: Molecular portraits of human breast tumours. Nature 2000, 406:747-752.

51. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, Hastie T, Eisen MB, van de Rijn M, Jeffrey SS, et al: Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Scí U S A 2001, 98:10869-10874.

52. van 't Veer LJ, Oai H, van de Vijver MJ, He YO, Hart AA, Mao M, Peterse HL, van der Kooy K, Marton MJ, Witteveen AT, et al: Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002, 415:530-536.

53. Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher Rl, Gascoyne RO, Muller­ Hermelink HK, Smeland EB, Giltnane JM, et al: The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med 2002, 346:1937-1947.

54. Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M, Baehner FL, Walker MG, Watson O, Park T, et al: A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004, 351 :2817-2826.

132 55. Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, Joshi MB, Harpole D, Lancaster JM, Berchuck A, et al: Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature 2006, 439:353-357.

56. Lamb J, Ramaswamy S, Ford HL, Contreras B, Martinez RV, Kittrell FS, Zahnow CA, Patterson N, Golub TR, Ewen ME: A mechanism of cyclin 01 action encoded in the patterns of gene expression in human cancer. Ce//2003, 114:323-334.

57. Black EP, Huang E, Dressman H, Rempel R, Laakso N, Asa SL, lshida S, West M, Nevins JR: Oistinct gene expression phenotypes of cells lacking Rb and Rb family members. Cancer Res 2003, 63:3716-3723.

58. Reis-Filho JS, Pusztai L: Gene expression profiling in breast cancer: classification, prognostication, and prediction. Lancet 2011, 378:1812-1823.

59. Reis-Filho JS, Weigelt B, Fumagalli D, Sotiriou C: Molecular profiling: moving away from tumor philately. Sci Transl Med 201 O, 2:47ps43.

60. lacobuzio-Donahue CA, Ryu B, Hruban RH, Kern SE: Exploring the host desmoplastic response to pancreatic carcinoma: gene expression of stromal and neoplastic cells at the site of primary invasion. Am J Patho/2002, 160:91-99.

61. Han H, Bearss DJ, Browne LW, Calaluce R, Nagle RB, Von Hoff DD: ldentification of differentially expressed genes in pancreatic cancer cells using cONA microarray. Cancer Res 2002, 62:2890-2896.

62. lacobuzio-Donahue CA, Maitra A, Shen-Ong GL, van Heek T, Ashfaq R, Meyer R, Walter K, Berg K, Hollingsworth MA, Cameron JL, et al: Oiscovery of novel tumor markers of pancreatic cancer using global gene expression technology. Am J Patho/2002, 160:1239- 1249.

63. lacobuzio-Donahue CA, Maitra A, Olsen M, Lowe AW, van Heek NT, Rosty C, Walter K, Sato N, Parker A, Ashfaq R, et al: Exploration of global gene expression patterns in pancreatic adenocarcinoma using cONA microarrays. Am J Patho/2003, 162:1151-1162.

64. Hassan R, Sera T, Pastan 1: Mesothelin: a new target for immunotherapy. Clin Cancer Res 2004, 10:3937-3942.

65. Argani P, lacobuzio-Donahue C, Ryu B, Rosty C, Goggins M, Wilentz RE, Murugesan SR, Leach SO, Jaffee E, Yeo CJ, et al: Mesothelin is overexpressed in the vast majority of ductal adenocarcinomas of the pancreas: identification of a new pancreatic cancer marker by serial analysis of gene expression (SAGE). Clin Cancer Res 2001, 7:3862-3868.

66. McCarthy DM, Maitra A, Argani P, Rader AE, Faigel DO, Van Heek NT, Hruban RH, Wilentz RE: Novel markers of pancreatic adenocarcinoma in fine-needle aspiration: mesothelin and prostate stem cell antigen labeling increases accuracy in cytologically borderline cases. Appllmmunohistochem Mol Morpho/2003, 11 :238-243.

133 67. Corso CD, Stubbs DD, Lee SH, Goggins M, Hruban RH, Hunt WD: Real-time detection of mesothelin in pancreatic cancer cell line supernatant using an acoustic wave immunosensor. Cancer Detect Prev 2006, 30:180-187.

68. Jin G, Hu XG, Ying K, Tang Y, Liu R, Zhang YJ, Jing ZP, Xie Y, Mao YM: Oiscovery and analysis of pancreatic adenocarcinoma genes using cONA microarrays. World J Gastroenterol 2005, 11 :6543-6548.

69. Crnogorac-Jurcevic T, Efthimiou E, Nielsen T, Loader J, Terris B, Stamp G, Baron A, Scarpa A, Lemoine NR: Expression profiling of microdissected pancreatic adenocarcinomas. Oncogene 2002, 21:4587-4594.

70. Tan ZJ, Hu XG, Cao GS, Tang Y: Analysis of gene expression profile of pancreatic carcinoma using cONA microarray. World J Gastroentero/2003, 9:818-823.

71. lacobuzio-Donahue CA, Ashfaq R, Maitra A, Adsay NV, Shen-Ong GL, Berg K, Hollingsworth MA, Cameron JL, Yeo CJ, Kern SE, et al: Highly expressed genes in pancreatic ductal adenocarcinomas: a comprehensive characterization and comparison of the transcription profiles obtained from three major technologies. Cancer Res 2003, 63:8614-8622.

72. Friess H, Ding J, Kleeff J, Fenkell L, Rosinski JA, Guweidhi A, Reidhaar-Oison JF, Korc M, Hammer J, Buchler MW: Microarray-based identification of differentially expressed growth­ and metastasis-associated genes in pancreatic cancer. Ce// Mol Life Sci 2003, 60:1180-1199.

73. Nakamura T, Fidler IJ, Coombes KR: Gene expression profile of metastatic human pancreatic cancer cells depends on the organ microenvironment. Cancer Res 2007, 67:139-148.

74. Crnogorac-Jurcevic T, Efthimiou E, Capelli P, Blaveri E, Baron A, Terris 8, Jones M, Tyson K, Bassi C, Scarpa A, Lemoine NR: Gene expression profiles of pancreatic cancer and stromal desmoplasia. Oncogene 2001,20:7437-7446.

75. Logsdon CD, Simeone DM, Binkley C, Arumugam T, Greenson JK, Giordano T J, Misek DE, Kuick R, Hanash S: Molecular profiling of pancreatic adenocarcinoma and chronic pancreatitis identifies multiple genes differentially regulated in pancreatic cancer. Cancer Res 2003, 63:2649-2657.

76. Spratlin JL, Mulder KE: Looking to the future: biomarkers in the management of pancreatic adenocarcinoma. lnt J Mol Sci 2011, 12:5895-5907.

77. Keating P, Cambrosio A: Too many numbers: Microarrays in clinical cancer research. Stud Hist Phi/os Biol Biomed Sci 2012, 43:37-51.

78. Karakach TK, Flight RM, Douglas SE, Wentzell PD: An introduction to ONA microarrays for gene expression analysis. Chemometrics and lntelligent Laboratory Systems 2010:28-52.

79. Quackenbush J: Computational approaches to analysis of ONA microarray data. Yearb Med lnform 2006:91-103.

134 80. Altman NS, Hua J: Extending the loop design for two-channel microarray experiments. Genet Res 2006, 88:153-163.

81. Knapen D, Vergauwen L, Laukens K, Blust R: Best practices for hybridization design in two­ colour microarray analysis. Trends Bíotechno/2009, 27:406-414.

82. Churchill GA: Fundamentals of experimental design for cONA microarrays. Nat Genet 2002, 32 Suppl:490-495.

83. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, Lukyanov S, Lukyanov K, Gurskaya N, Sverdlov ED, Siebert PO: Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cONA probes and libraries. Proceedíngs of the Natíonal Academy of Scíences of the Uníted States of Ameríca 1996, 93:6025-6030.

84. Rebrikov DV, Desai SM, Siebert PO, Lukyanov SA: Suppression subtractive hybridization. Methods Mol Bío/2004, 258:107-134.

85. Lukyanov KA, Launer GA, Tarabykin VS, Zaraisky AG, Lukyanov SA: lnverted terminal repeats permit the average length of amplified ONA fragments to be regulated during preparation of cONA libraries by polymerase chain reaction. Ana/ytíca/ Bíochemístry 1995, 229:198-202.

86. Siebert PO, Chenchik A, Kellogg DE, Lukyanov KA, Lukyanov SA: An improved PCR method for walking in uncloned genomic ONA. Nucleíc acíds research 1995, 23:1087-1088.

87. Lukyanov SA, Rebrikov D, A. BA: Suppression Subtractive Hybridization. In Nucleíc Acíds Hybrídízatíon. Edited by Buzdin A, Lukyanov SA: Springer; 2007: 53-84

88. Liu BH, Goh CH, Ooi LL, Hui KM: ldentification of unique and common low abundance tumour-specific transcripts by suppression subtractive hybridization and oligonucleotide probe array analysis. Oncogene 2008, 27:4128-4136.

89. Liu Y, Zhu X, Zhu J, Liao S, Tang Q, Liu K, Guan X, Zhang J, Feng Z: ldentification of differential expression of genes in hepatocellular carcinoma by suppression subtractive hybridization combined cONA microarray. Oncol Rep 2007, 18:943-951.

90. Pan YS, Lee YS, Lee YL, Lee WC, Hsieh SY: Oifferentially profiling the low-expression transcriptomes of human hepatoma using a novel SSH/microarray approach. BMC Genomícs 2006, 7:131.

91. Wang T, Hopkins O, Schmidt C, Silva S, Houghton R, Takita H, Repasky E, Reed SG: ldentification of genes differentially over-expressed in lung squamous cell carcinoma using combination of cONA subtraction and microarray analysis. Oncogene 2000, 19:1519- 1528.

92. Zhang B, Nie X, Xiao B, Xiang J, Shen S, Gong J, Zhou M, Zhu S, Zhou J, Qian J, et al: ldentification of tissue-specific genes in nasopharyngeal epithelial tissue and differentially expressed genes in nasopharyngeal carcinoma by suppression subtractive hybridization and cONA microarray. Genes Cancer 2003, 38:80-90. 135 93. Barraclough DL, Sewart S, Rudland PS, Shoker BS, Sibson DR, Barraclough R, Davies MP: Microarray analysis of suppression subtracted hybridisation libraries identifies genes associated with breast cancer progression. Ce// Onco/201 O, 32:87-99.

94. Lundberg E, Fagerberg L, Klevebring D, Matic 1, Geiger T, Cox J, Algenas C, Lundeberg J, Mann M, Uhlen M: Defining the transcriptome and proteome in three functionally different human cell lines. Molecular systems bíology 201 O, 6:450.

95. Chenchik A, Zhu Y, Diatchenko L, Li R, Hill J, Siebert PO (Eds.): Generation and use of high­ quality cONA from small amounts of total RNA by SMART PCR: BioTechniques Books, MA; 1998.

96. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, Lukyanov S, Lukyanov K, Gurskaya N, Sverdlov ED, Siebert PO: Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cONA probes and libraries. Proc Natl Acad Scí U S A 1996, 93:6025-6030.

97. Ritchie ME, Silver J, Oshlack A, Holmes M, Diyagama D, Holloway A, Smyth GK: A comparison of background correction methods for two-colour microarrays. Bíoínformatícs 2007, 23:2700-2707.

98. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, Speed TP: Normalization for cONA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleíc Acíds Res 2002, 30:e15.

99. Saeed Al, Bhagabati NK, Braisted JC, Liang W, Sharov V, Howe EA, Li J, Thiagarajan M, White JA, Quackenbush J: TM4 microarray software suite. Methods in enzymology 2006, 411:134- 193.

100. Huang DW, Sherman DT, Lempicki R: Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature protocols 2008, 4:44-57.

101. Ramakers C, Ruijter JM, Deprez RH, Moorman AF: Assumption-free analysis of quantitative real-time polymerase chain reaction (PCR) data. Neuroscí Lett 2003, 339:62-66.

102. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F: Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple interna! control genes. Genome Bío/2002, 3:RESEARCH0034.

103. Kidd M, Nadler B, Mane S, Eick G, Malfertheiner M, Champaneria M, Pfragner R, Modlin 1: GeneChip, geNorm, and gastrointestinal tumors: novel reference genes for real-time PCR. Physíol Genomícs 2007, 30:363-370.

104. Rubie C, Kempf K, Hans J, Su T, Tilton B, Georg T, Brittner B, Ludwig B, Schilling M: Housekeeping gene variability in normal and cancerous colorectal, pancreatic, esophageal, gastric and hepatic tissues. Mol Gel/ Probes 2005, 19:101-109.

136 105. Zhao H, Hastie T, Whitfield ML, Borresen-Dale AL, Jeffrey SS: Optimization and evaluation of T7 based RNA linear amplification protocols for cONA microarray analysis. BMe Genomícs 2002, 3:31.

106. eox WG, Beaudet MP, Agnew JY, Ruth JL: Possible sources of dye-related signal correlation bias in two-color DNA microarray assays. Anal Bíochem 2004, 331 :243-254.

107. eao W, Epstein e, Liu H, DeLoughery e, GeN, Lin J, Diao R, eao H, Long F, Zhang X, et al: Comparing gene discovery from Affymetrix GeneChip microarrays and Clontech PCR­ select cONA subtraction: a case study. BMe Genomícs 2004, 5:26.

108. Huang da W, Sherman BT, Lempicki RA: Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009, 4:44-57.

109. Blake JA, Harris MA: The Gene Ontology (GO) project: structured vocabularies for molecular biology and their application to genome and expression analysis. eurr Protoc Bíoínformatícs 2008, Chapter 7:Unit 7 2.

110. Stastna M, Van Eyk JE: Secreted proteins as a fundamental source for biomarker discovery. Proteomícs 2012, 12:722-735.

111. Rogers eD, Fukushima N, Sato N, Shi C, Prasad N, Hustinx SR, Matsubayashi H, Canto M, Eshleman JR, Hruban RH, Goggins M: Differentiating pancreatic lesions by microarray and QPCR analysis of pancreatic juice RNAs. eancer Bíol Ther 2006, 5:1383-1389.

112. Troiani T, Martinelli E, Capasso A, Morgillo F, Orditura M, De Vita F, Ciardiello F: Targeting EGFR in Pancreatic Cancer Treatment. eurr Drug Targets 2012, 13:802-810.

113. Karanjawala ZE, lllei PB, Ashfaq R, Infante JR, Murphy K, Pandey A, Schulick R, Winter J, Sharma R, Maitra A, et al: New markers of pancreatic cancer identified through differential gene expression analyses: claudin 18 and annexin AS. Am J Surg Patho/2008, 32:188-196.

114. Strickland LA, Ross J, Williams S, Ross S, Romero M, Spencer S, Erickson R, Sutcliffe J, Verbeke C, Polakis P, et al: Preclinical evaluation of carcinoembryonic cell adhesion molecule (CEACAM) 6 as potential therapy target for pancreatic adenocarcinoma. J Pathol 2009, 218:380-390.

115. Simeone DM, Ji B, Banerjee M, Arumugam T, Li D, Anderson MA, Bamberger AM, Greenson J, Brand RE, Ramachandran V, Logsdon CD: CEACAM1, a novel serum biomarker for pancreatic cancer. Pancreas 2007, 34:436-443.

116. Eser S, Messer M, Eser P, von Werder A, Seidler B, Bajbouj M, Vogelmann R, Meining A, von Burstin J, Algul H, et al: In vivo diagnosis of murine pancreatic intraepithelial neoplasia and early-stage pancreatic cancer by molecular imaging. Proc Natl Acad Scí U S A 2011, 108:9945-9950.

117. Hansel DE, Ashfaq R, Rahman A, Wanzer D, Yeo CJ, Wilentz RE, Maitra A: A subset of pancreatic adenocarcinomas demonstrates coamplification of topoisomerase llalpha and

137 HER21neu: use of immunolabeling and multicolor FISH for potential patient screening andtreatment. Am J elin Pathol2005, 123:28-35.

118. Holzmann K, Kohlhammer H, Schwaenen e, Wessendorf S, Kestler HA, Schwoerer A, Rau 8, Radlwimmer B, Dohner H, Lichter P, et al: Genomic DNA-chip hybridization reveals a higher incidence of genomic amplifications in pancreatic cancer than conventional comparative genomic hybridization and leads to the identification of novel candidate genes. eancer Res 2004, 64:4428-4433.

119. Mahlamaki EH, Barlund M, Tanner M, Gorunova L, Hoglund M, Karhu R, Kallioniemi A: Frequent amplification of 8q24, 11 q, 17q, and 20q-specific genes in pancreatic cancer. Genes ehromosomes eancer 2002, 35:353-358.

120. Mahlamaki EH, Kauraniemi P, Monni O, Wolf M, Hautaniemi S, Kallioniemi A: High-resolution genomic and expression profiling reveals 105 putative amplification target genes in pancreatic cancer. Neoplasia 2004, 6:432-439.

121. Tsukamoto Y, Uchida T, Karnan S, Noguchi T, Nguyen LT, Tanigawa M, Takeuchi 1, Matsuura K, Hijiya N, Nakada e, et al: Genome-wide analysis of DNA copy number alterations and gene expression in gastric cancer. J Pathol2008, 216:471-482.

122. Sheffer M, Bacolod MD, Zuk O, Giardina SF, Pincas H, Barany F, Paty PB, Gerald WL, Notterman DA, Domany E: Association of survival and disease progression with chromosomal instability: a genomic exploration of colorectal cancer. Proc Natl Acad Sci U S A 2009, 106:7131-7136.

123. Apte MV, Park S, Phillips PA, Santucci N, Goldstein O, Kumar RK, Ramm GA, Buchler M, Friess H, McCarroll JA, et al: Desmoplastic reaction in pancreatic cancer: role of pancreatic stellate ce lis. Pancreas 2004, 29:179-187.

124. Hartei-Schenk S, Gratchev A, Hanski ML, Ogorek O, Trendelenburg G, Hummel M, Hopfner M, Scherubl H, Zeitz M, Hanski C: Novel adapter protein AP162 connects a sialyl-Le(x)-positive mucin with an apoptotic signa! transduction pathway. Glycoconjugate journa/2001, 18:915- 923.

125. Van Wesenbeeck L, Odgren PR, Coxon FP, Frattini A, Moens P, Perdu B, MacKay CA, Van Hui E, Timmermans JP, Vanhoenacker F, et al: lnvolvement of PLEKHM1 in osteoclastic vesicular transport and osteopetrosis in incisors absent rats and humans. The Journal of clínícal ínvestígation 2007, 117:919-930.

126. Del Fattore A, Fornari R, Van Wesenbeeck L, de Freitas F, Timmermans JP, Peruzzi B, Cappariello A, Rucci N, Spera G, Helfrich MH, et al: A new heterozygous mutation (R714C) of the osteopetrosis gene, pleckstrin homolog domain containing family M (with run domain) member 1 (PLEKHM1), impairs vesicular acidification and increases TRACP secretion in osteoclasts. Journal of bone and mineral research : the official jo u mal of the American Society for Bone and Mineral Research 2008, 23:380-391.

127. Spater O, Hill TP, O'Sullivan R J, Gruber M, Conner DA, Hartmann C: Wnt9a signaling is required for joint integrity and regulation of lhh during chondrogenesis. Development 2006, 133:3039-3049. 138 128. Matsumoto K, Miki R, Nakayama M, Tatsumi N, Yokouchi Y: Wnt9a secreted from the walls of hepatic sinusoids is essential for morphogenesis, proliferation, and glycogen accumulation of chick hepatic epithelium. Dev Bio/2008, 319:234-247.

129. Curtin E, Hickey G, Kamel G, Davidson AJ, Liao EC: Zebrafish wnt9a is expressed in pharyngeal ectoderm and is required for palate and lower jaw development. Mech Dev 2011,128:104-115.

130. Carpelan-Holmstrom M, Nordling S, Pukkala E, Sankila R, Luttges J, Kloppel G, Haglund C: Ooes anyone survive pancreatic ductal adenocarcinoma? A nationwide study re­ evaluating the data of the Finnish Cancer Registry. Gut 2005, 54:385-387.

131. Stoyanova R, Upson JJ, Patriotis C, Ross EA, Henske EP, Datta K, 8oman 8, Clapper ML, Knudson AG, 8ellacosa A: Use of RNA amplification in the optimal characterization of global gene expression using cONA microarrays. J Ce// Physio/2004, 201:359-365.

132. 8arrier A, Lemoine A, 8oelle PY, Tse C, 8rault D, Chiappini F, 8reittschneider J, Lacaine F, Houry S, Huguier M, et al: Colon cancer prognosis prediction by gene expression profiling. Oncogene 2005, 24:6155-6164.

133. Trautmann K, Steudel C, Grossmann D, Aust D, Ehninger G, Miehlke S, Thiede C: Expression profiling of gastric cancer samples by oligonucleotide microarray analysis reveals low degree of intra-tumor variability. World J Gastroentero/2005, 11:5993-5996.

134. Grutzmann R, 8oriss H, Ammerpohl O, Luttges J, Kalthoff H, Schackert HK, Kloppel G, Saeger HD, Pilarsky C: Meta-analysis of microarray data on pancreatic cancer defines a set of commonly dysregulated genes. Oncogene 2005, 24:5079-5088.

135. Grutzmann R, Pilarsky C, Ammerpohl O, Luttges J, 8ohme A, Sipos 8, Foerder M, Alldinger 1, Jahnke 8, Schackert HK, et al: Gene expression profiling of microdissected pancreatic ductal carcinomas using high-density ONA microarrays. Neoplasia 2004, 6:611-622.

136. Shields MA, Dangi-Garimella S, Redig AJ, Munshi HG: Biochemical role of the collagen-rich tumour microenvironment in pancreatic cancer progression. Biochem J 2012, 441:541-552.

137. Lowe AW, Olsen M, Hao Y, Lee SP, Taek Lee K, Chen X, van de Rijn M, 8rown PO: Gene expression patterns in pancreatic tumors, cells and tissues. PLoS ONE 2007, 2:e323.

138. Nakamura T, Furukawa Y, Nakagawa H, Tsunoda T, Ohigashi H, Murata K, lshikawa O, Ohgaki K, Kashimura N, Miyamoto M, et al: Genome-wide cONA microarray analysis of gene expression profiles in pancreatic cancers using populations of tumor cells and normal ductal epithelial cells selected for purity by laser microdissection. Oncogene 2004, 23:2385-2400.

139. Arafat H, Lazar M, Salem K, Chipitsyna G, Gong Q, Pan TC, Zhang RZ, Yeo CJ, Chu ML: Tumor-specific expression and alternative splicing of the COL6A3 gene in pancreatic cancer. Surgery 2011, 150:306-315.

139 140. Fukushima N, Kikuchi Y, Nishiyama T, Kudo A, Fukayama M: Periostin deposition in the stroma of invasive and intraductal neoplasms of the pancreas. Mod Pathol 2008, 21:1 044- 1053.

141. Koninger J, Giese T, di Mola FF, Wente MN, Esposito 1, Bachem MG, Giese NA, Buchler MW, Friess H: Pancreatic tumor cells influence the composition of the extracellular matrix. Biochem Biophys Res Commun 2004, 322:943-949.

142. Erkan M, Kleeff J, Gorbachevski A, Reiser C, Mitkus T, Esposito 1, Giese T, Buchler MW, Giese NA, Friess H: Periostin creates a tumor-supportive microenvironment in the pancreas by sustaining fibrogenic stellate cell activity. Gastroentero/ogy 2007, 132:144 7-1464.

143. Malanchi 1, Santamaria-Martinez A, Susanto E, Peng H, Lehr HA, Delaloye JF, Huelsken J: lnteractions between cancer stem cells and their niche govern metastatic colonization. Nature 2012, 481:85-89.

144. Menke A, Adler G: TGFbeta-induced fibrogenesis of the pancreas. lnt J Gastrointest Cancer 2002, 31:41-46.

145. Sawai H, Okada Y, Funahashi H, Takahashi H, Matsuo Y, Yasuda A, Ochi N, Takeyama H, Manabe T: Basement membrane proteins play an important role in the invasive processes of human pancreatic cancer cells. J Surg Res 2008, 144:117-123.

146. Infante JR, Matsubayashi H, Sato N, Tonascia J, Klein AP, Riall TA, Yeo C, lacobuzio-Donahue C, Goggins M: Peritumoral fibroblast SPARC expression and patient outcome with resectable pancreatic adenocarcinoma. J Clin Onco/2007, 25:319-325.

147. Mantoni TS, Schendel RR, Rodel F, Niedobitek G, AI-Assar O, Masamune A, Brunner TB: Stromal SPARC expression and patient survival after chemoradiation for non-resectable pancreatic adenocarcinoma. CancerBiol Ther2008, 7:1806-1815.

148. Reding T, Wagner U, Silva AB, Sun LK, Bain M, Kim SY, Bimmler D, Graf R: lnflammation­ dependent expression of SPARC during development of chronic pancreatitis in WBN/Kob rats anda microarray gene expression analysis. Physiol Genomics 2009, 38:196-204.

149. Braganza JM, Lee SH, McCioy RF, McMahon MJ: Chronic pancreatitis. Lancet 2011, 377:1184-1197.

150. Binkley CE, Zhang L, Greenson JK, Giordano TJ, Kuick R, Misek D, Hanash S, Logsdon CD, Simeone DM: The molecular basis of pancreatic fibrosis: common stromal gene expression in chronic pancreatitis and pancreatic adenocarcinoma. Pancreas 2004, 29:254-263.

151. Kennedy RH, Bockman DE, Uscanga L, Choux R, Grimaud JA, Sarles H: Pancreatic extracellular matrix alterations in chronic pancreatitis. Pancreas 1987, 2:61-72.

152. Pan S, Chen R, Stevens T, Bronner MP, May D, Tamura Y, Mclntosh MW, Brentnall TA: Proteomics portrait of archiva! lesions of chronic pancreatitis. PLoS ONE 2011, 6:e2757 4.

140 153. Omary MB, Lugea A, Lowe AW, Pandol SJ: The pancreatic stellate cell: a star on the rise in pancreatic diseases. J Clín lnvest 2007, 117:50-59.

154. Sund M, Kalluri R: Tumor stroma derived biomarkers in cancer. Cancer Metastasis Rev 2009, 28:177-183.

155. Pitteri SJ, Kelly-Spratt KS, Gurley KE, Kennedy J, Buson TB, ChinA, Wang H, Zhang Q, Wong CH, Chodosh LA, et al: Tumor microenvironment-derived proteins dominate the plasma proteome response during breast cancer induction and progression. Cancer Res 2011, 71:5090-51 OO.

156. Hanahan D, Weinberg RA: Hallmarks of cancer: the next generation. Ce//2011, 144:646-674.

157. Clevers H: Wnt/beta-catenin signaling in development and disease. Ce//2006, 127:469-480.

158. AI-Aynati MM, Radulovich N, Riddell RH, Tsao MS: Epithelial-cadherin and beta-catenin expression changes in pancreatic intraepithelial neoplasia. Clin Cancer Res 2004, 10:1235- 1240.

159. Zeng G, Germinare M, Micsenyi A, Monga NK, BellA, Sood A, Malhotra V, Sood N, Midda V, Monga DK, et al: Aberrant Wnt/beta-catenin signaling in pancreatic adenocarcinoma. Neoplasia 2006, 8:279-289.

160. Pasea di Magliano M, Biankin AV, Heiser PW, Cano DA, Gutierrez PJ, Deramaudt T, Segara D, Dawson AC, Kench JG, Henshall SM, et al: Common activation of canonical Wnt signaling in pancreatic adenocarcinoma. PLoS ONE 2007, 2:e1155.

161. Takahashi N, Fukushima T, Yorita K, Tanaka H, Chijiiwa K, Kataoka H: Dickkopf-1 is overexpressed in human pancreatic ductal adenocarcinoma cells and is involved in invasive growth. lnt J Cancer2010, 126:1611-1620.

162. Pilarsky C, Ammerpohl O, Sipos B, Dahl E, Hartmann A, Wellmann A, Braunschweig T, Lohr M, Jesenofsky R, Friess H, et al: Activation of Wnt signalling in stroma from pancreatic cancer identified by gene expression profiling. J Ce// Mol Med 2008, 12:2823-2835.

163. Tong Y, Lin Y, Zhang Y, Yang J, Liu H, Zhang B: Association between TCF7L2 gene polymorphisms and susceptibility to type 2 diabetes mellitus: a large Epidemiology (HuGE) review and meta-analysis. BMC Med Genet 2009, 1 O: 15.

164. Tang W, Dodge M, Gundapaneni D, Michnoff C, Roth M, Lum L: A genome-wide RNAi screen for Wnt/beta-catenin pathway components identifies unexpected roles for TCF transcription factors in cancer. Proc Natl Acad Sci U S A 2008, 105:9697-9702.

165. TCGA-Network: Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012, 487:330-337.

166. Jauliac S, Lopez-Rodriguez C, Shaw LM, Brown LF, Rao A, Toker A: The role of NFAT transcription factors in integrin-mediated carcinoma invasion. Nat Ce// Biol2002, 4:540-544. 141 167. Nishimoto Y, Okano H: New insight into cancer therapeutics: induction of differentiation by regulating the Musashi/Numb/Notch pathway. Ce// Res 2010, 20:1083-1085.

168. lto T, Kwon HY, Zimdahl 8, Congdon KL, Blum J, Lento WE, Zhao C, Lagoo A, Gerrard G, Foroni L, et al: Regulation of myeloid leukaemia by the cell-fate determinant Musashi. Nature 201 O, 466:765-768.

169. Hardy CF, Sussel L, Shore O: A RAP1-interacting protein involved in transcriptional silencing and telomere length regulation. Genes Dev 1992, 6:801-814.

170. Silverman J, Takai H, Buonomo SB, Eisenhaber F, de Lange T: Human Rif1, ortholog of a yeast telomeric protein, is regulated by ATM and 53BP1 and functions in the 5-phase checkpoint. Genes Dev 2004, 18:2108-2119.

171. Stute P, Sielker S, Wood CE, Register TC, Lees CJ, Oewi FN, Williams JK, Wagner JO, Stefenelli U, Cline JM: Life stage differences in mammary gland gene expression profile in non-human primates. Breast Cancer Res Treat 2012, 133:617-634.

172. Yin M, Yan J, Martinez-Balibrea E, Graziano F, Lenz HJ, Kim HJ, Robert J, lm SA, Wang WS, Etienne-Grimaldi MC, Wei Q: ERCC1 and ERCC2 polymorphisms predict clinical outcomes of oxaliplatin-based chemotherapies in gastric and colorectal cancer: a systemic review and meta-analysis. C/in Cancer Res 2011, 17:1632-1640.

173. Feng Z, Ni Y, Oong W, Shen H, Ou J: Association of ERCC2/XPD polymorphisms and interaction with tobacco smoking in lung cancer susceptibility: a systemic review and meta-analysis. Mol Biol Rep 2012, 39:57-69.

174. Bizama C, Benavente F, Espinoza JA, Salvatierra E, Gutiérrez HA, Fernández E, Sagredo EA, Roa 1, Mazzolini G, Gidekel M, Podhajcer OL: Low abundance transcriptome analyses identifies novel markers, specific intracellular pathways and target genes in advanced human gastrointestinal cancer. Manuscript submitted 2012.

175. Bild AH, Potti A, Nevins JR: Linking oncogenic pathways with therapeutic opportunities. Nat Rev Cancer 2006, 6:735-741.

176. Gromov P, Gromova 1, Bunkenborg J, Cabezon T, Moreira JM, Timmermans-Wielenga V, Roepstorff P, Rank F, Celis JE: Up-regulated proteins in the fluid bathing the tumour cell microenvironment as potential serological markers for early detection of cancer of the breast. Mol Onco/201 O, 4:65-89.

177. Xue H, Lu 8, Lai M: The cancer secretome: a reservoir of biomarkers. J Transl Med 2008, 6:52.

178. Harsha HC, Kandasamy K, Ranganathan P, Rani S, Ramabadran S, Gollapudi S, Balakrishnan L, Owivedi SB, Telikicherla O, Selvan LO, et al: A compendium of potential biomarkers of pancreatic cancer. PLoS Med 2009, 6:e1 000046.

179. Chakraborty S, Saine MJ, Sasson AR, Batra SK: Current status of molecular markers for early detection of sporadic pancreatic cancer. Biochim Biophys Acta 2011, 1815:44-64. 142 180. Weeks ME, Hariharan O, Petronijevic L, Radon TP, Whiteman HJ, Kocher HM, Timms JF, Lemoine NR, ernogorac-Jurcevic T: Analysis of the urine proteome in patients with pancreatic ductal adenocarcinoma. Proteomics C/in App/2008, 2:1047-1057.

181. Kornmann M, lshiwata T, Kleeff J, Beger HG, Korc M: Fas and Fas-ligand expression in human pancreatic cancer. Ann Surg 2000, 231:368-379.

182. Brand RE, Nolen BM, Zeh HJ, Allen PJ, Eloubeidi MA, Goldberg M, Elton E, Arnoletti JP, ehristein JO, Vickers SM, et al: Serum biomarker panels for the detection of pancreatic cancer. Clínica/ cancer research : an official journal of the American Association for Cancer Research 2011, 17:805-816.

183. Mikula M, Rubel T, Karczmarski J, Goryca K, Dadlez M, Ostrowski J: lntegrating proteomic and transcriptomic high-throughput surveys for search of new biomarkers of colon tumors. Funct lntegr Genomics 2010.

184. Turtoi A, Musmeci O, Wang Y, Dumont B, Somja J, Bevilacqua G, De Pauw E, Delvenne P, eastronovo V: ldentification of novel accessible proteins bearing diagnostic and therapeutic potential in human pancreatic ductal adenocarcinoma. J Proteome Res 2011, 10:4302-4313.

185. Ketterer K, Kong B, Frank O, Giese NA, Bauer A, Hoheisel J, Korc M, Kleeff J, Michalski ew, Friess H: Neuromedin U is overexpressed in pancreatic cancer and increases invasiveness via the hepatocyte growth factor c-Met pathway. Cancer Lett 2009, 277:72-81.

186. Valsecchi ME, McDonald M, Brody JB, Hyslop T, Freydin B, Yeo eJ, Solomides e, Peiper Se, Witkiewicz AK: Epidermal growth factor receptor and insulinlike growth factor 1 receptor expression predict poor survival in pancreatic ductal adenocarcinoma. Cancer 2011.

187. Pryczynicz A, Guzinska-Ustymowicz K, Kemona A, ezyzewska J: Expression of EGF and EGFR strongly correlatas with metastasis of pancreatic ductal carcinoma. Anticancer Res 2008, 28:1399-1404.

188. Hanas JS, Hocker JR, eheung JY, Larabee JL, Lerner MR, Lightfoot SA, Margan DL, Denson KD, Prejeant KC, Gusev Y, et al: Biomarker identification in human pancreatic cancer sera. Pancreas 2008, 36:61-69.

189. lrigoyen Oyarzabal AM, Amiguet Garcia JA, Lopez Vivanco G, Genolla Subirats J, Munoz Villafranca Me, Ojembarrena Martinez E, Liso lrurzun P: [Tumoral markers and acute-phase reactants in the diagnosis of pancreatic cancer]. Gastroenterol Hepato/2003, 26:624-629.

190. Xu LL, Shanmugam N, Segawa T, Sesterhenn lA, Mcleod DG, Moul JW, Srivastava S: A novel androgen-regulated gene, PMEPA1, located on chromosome 20q13 exhibits high level expression in prostate. Genomics 2000, 66:257-263.

191. Giannini G, Ambrosini MI, Di Marcotullio L, eerignoli F, Zani M, MacKay AR, Screpanti 1, Frati L, Gulino A: EGF- and cell-cycle-regulated STAG1/PMEPA1/ERG1.2 belongs to a conservad gene family and is overexpressed and amplified in breast and ovarian cancer. Mol Carcinog 2003, 38:188-200. 143 192. Brunschwig EB, Wilson K, Mack D, Dawson D, Lawrence E, Willson JK, Lu S, Nosrati A, Rerko RM, Swinler S, et al: PMEPA1, a transforming growth factor-beta-induced marker of terminal colonocyte differentiation whose expression is maintained in primary and metastatic colon cancer. Cancer Res 2003, 63:1568-1575.

193. Levy L, Hill CS: Smad4 dependency defines two classes of transforming growth factor {beta} (TGF-{beta}) target genes and distinguishes TGF-{beta}-induced epithelial­ mesenchymal transition from its antiproliferative and migratory responses. Mol Gel/ Bíol 2005, 25:8108-8125.

194. ltoh S, Thorikay M, Kowanetz M, Moustakas A, ltoh F, Heldin CH, ten Dijke P: Elucidation of Smad requirement in transforming growth factor-beta type 1 receptor-induced responses. J Bíol Chem 2003, 278:3751-3761.

195. Rae FK, Hooper JO, Nicol DL, Clements JA: Characterization of a novel gene, STAG1/PMEPA1, upregulated in renal cell carcinoma and other solid tumors. Mol Carcínog 2001' 32:44-53.

196. Xu LL, Shi Y, Petrovics G, Sun C, Makarem M, Zhang W, Sesterhenn lA, Mcleod DG, Sun L, Moul JW, Srivastava S: PMEPA1, an androgen-regulated NEDD4-binding protein, exhibits cell growth inhibitory function and decreased expression during prostate cancer progression. Cancer Res 2003, 63:4299-4304.

197. Watanabe Y, ltoh S, Goto T, Ohnishi E, lnamitsu M, ltoh F, Satoh K, Wiercinska E, Yang W, Shi L, et al: TMEPAI, a transmembrane TGF-beta-inducible protein, sequesters Smad proteins from active participation in TGF-beta signaling. Mol Ce//201 O, 37:123-134.

198. Singha PK, Yeh IT, Venkatachalam MA, Saikumar P: Transforming growth factor-beta (TGF­ beta)-inducible gene TMEPAI converts TGF-beta from a tumor suppressor to a tumor promoter in breast cancer. Cancer Res 201 O, 70:6377-6383.

199. Badea L, Herlea V, Dima SO, Dumitrascu T, Popescu 1: Combined gene expression analysis of whole-tissue and microdissected pancreatic ductal adenocarcinoma identifies genes specifically overexpressed in tumor epithelia. Hepatogastroenterology 2008, 55:2016-2027.

200. Currie MJ, Hanrahan V, Gunningham SP, Morrin HR, Frampton C, Han C, Robinson BA, Fox SB: Expression of vascular endothelial growth factor D is associated with hypoxia inducible factor (HIF-1alpha) and the HIF-1alpha target gene DEC1, but not lymph node metastasis in primary human breast carcinomas. J Clín Pathol2004, 57:829-834.

201. Zheng Y, Jia Y, Wang Y, Wang M, Li B, Shi X, Ma X, Xiao D, Sun Y: The hypoxia-regulated transcription factor DEC1 (Stra13, SHARP-2) and its expression in gastric cancer. OMICS 2009, 13:301-306.

202. Chakrabarti J, Turley H, Campo L, Han C, Harris AL, Gatter KC, Fox SB: The transcription factor DEC1 (stra13, SHARP2) is associated with the hypoxic response and high tumour grade in human breast cancers. Br J Cancer 2004, 91:954-958.

144 203. Zawel L, Yu J, Torrance CJ, Markowitz S, Kinzler KW, Vogelstein B, Zhou S: DEC1 is a downstream target of TGF-beta with sequence-specific transcriptional repressor activities. Proc Natl Acad Sci U S A 2002, 99:2848-2853.

204. Kon N, Hirota T, Kawamoto T, Kato Y, Tsubota T, Fukada Y: Activation of TGF-beta/activin signalling resets the circadian clock through rapid induction of Dec1 transcripts. Nat Ce// Bio/2008, 10:1463-1469.

205. Ehata S, Hanyu A, Hayashi M, Aburatani H, Kato Y, Fujime M, Saitoh M, Miyazawa K, lmamura T, Miyazono K: Transforming growth factor-beta promotes survival of mammary carcinoma ce lis through induction of antiapoptotic transcription factor DEC1. Cancer Res 2007, 67:9694-9703.

206. Carro MS, Lim WK, Alvarez MJ, Bollo RJ, Zhao X, Snyder EY, Sulman EP, Anne SL, Doetsch F, Colman H, et al: The transcriptional network for mesenchymal transformation of brain tumours. Nature 2010,463:318-325.

207. Wang W, Reiser-Erkan C, Michalski CW, Raggi MC, Quan L, Yupei Z, Friess H, Erkan M, Kleeff J: Hypoxia inducible BHLHB2 is a novel and independent prognostic marker in pancreatic ductal adenocarcinoma. Biochem Biophys Res Commun 201 O.

208. Wajih N, Sane OC, Hutson SM, Wallin R: The inhibitory effect of calumenin on the vitamin K­ dependent gamma-carboxylation system. Characterization of the system in normal and warfarin-resistant rats. J Biol Chem 2004, 279:25276-25283.

209. Wajih N, Owen J, Wallin R: Enhanced functional recombinant factor VIl production by HEK 293 cells stably transfected with VKORC1 where the gamma-carboxylase inhibitor calumenin is stably suppressed by shRNA transfection. Thromb Res 2008, 122:405-410.

210. Vasiljevic M, Heisler FF, Hausrat TJ, Fehr S, Milenkovic 1, Kneussel M, Sieghart W: Spatio­ temporal expression analysis of the calcium-binding protein calumenin in the rodent brain. Neuroscience 2012, 202:29-41.

211. Ostergaard M, Hansen GA, Vorum H, Honore B: Proteomic profiling of fibroblasts reveals a modulating effect of extracellular calumenin on the organization of the actin cytoskeleton. Proteomics 2006, 6:3509-3519.

212. Vorum H, Hager H, Christensen BM, Nielsen S, Honore B: Human calumenin localizes to the secretory pathway and is secreted to the medium. Exp Ce// Res 1999, 248:473-481.

213. Galamb O, Gyorffy B, Sipos F, Spisak S, Nemeth AM, Miheller P, Tulassay Z, Dinya E, Molnar B: lnflammation, adenoma and cancer: objective classification of colon biopsy specimens with gene expression signature. Dis Markers 2008, 25:1-16.

214. Voisin SN, Krakovska O, Matta A, DeSouza LV, Romaschin AD, Colgan TJ, Siu KW: ldentification of novel molecular targets for endometrial cancer using a drill-down LC­ MS/MS approach with iTRAQ. PLoS ONE2011, 6:e16352.

145 215. eastagna A, Antonioli P, Astner H, Hamdan M, Righetti Se, Perego P, Zunino F, Righetti PG: A proteomic approach to cisplatin resistance in the cervix squamous cell carcinoma cell line A431. Proteomics 2004, 4:3246-3267.

216. Tabata K, Matsunaga K, Sakane A, Sasaki T, Noda T, Yoshimori T: Rubicon and PLEKHM1 negatively regulate the endocytic/autophagic pathway via a novel Rab7-binding domain. Molecular biology of the ce// 201 O, 21 :4162-4172.

217. elevers H, Nusse R: Wnt/beta-Catenin Signaling and Disease. ee//2012, 149:1192-1205.

218. Herr P, Hausmann G, Basler K: WNT secretion and signalling in human disease. Trends Mol Med 2012.

219. Yu M, Ting DT, Stott SL, Wittner BS, Ozsolak F, Paul S, eiciliano Je, Smas ME, Winokur D, Gilman AJ, et al: RNA sequencing of pancreatic circulating tumour cells implicates WNT signalling in metastasis. Nature 2012.

220. Morris JPt, Wang Se, Hebrok M: KRAS, Hedgehog, Wnt and the twisted developmental biology of pancreatic ductal adenocarcinoma. Nat Rev eancer 201 O, 10:683-695.

221. erawford He, Scoggins eR, Washington MK, Matrisian LM, Leach SD: Matrix metalloproteinase-7 is expressed by pancreatic cancer precursors and regulates acinar­ to-ductal metaplasia in exocrine pancreas. J elin lnvest 2002, 109:1437-1444.

222. Yamamoto H, ltoh F, lku S, Adachi Y, Fukushima H, Sasaki S, Mukaiya M, Hirata K, lmai K: Expression of matrix metalloproteinases and tissue inhibitors of metalloproteinases in human pancreatic adenocarcinomas: clinicopathologic and prognostic significance of matrilysin expression. J elin Oncol2001, 19:1118-1127.

223. Fukushima H, Yamamoto H, ltoh F, Nakamura H, Min Y, Horiuchi S, lku S, Sasaki S, lmai K: Association of matrilysin mRNA expression with K-ras mutations and progression in pancreatic ductal adenocarcinomas. earcinogenesis 2001,22:1049-1052.

224. Jones LE, Humphreys MJ, eampbell F, Neoptolemos JP, Boyd MT: Comprehensive analysis of matrix metalloproteinase and tissue inhibitor expression in pancreatic cancer: increased expression of matrix metalloproteinase-7 predicts poor survival. e/in eancer Res 2004, 1 0:2832-2845.

225. Fukuda A, Wang Se, Morris JPt, Folias AE, Liou A, Kim GE, Akira S, Boucher KM, Firpo MA, Mulvihill SJ, Hebrok M: Stat3 and MMP7 contribute to pancreatic ductal adenocarcinoma initiation and progression. Cancer Ce// 2011, 19:441-455.

226. Jovanovic V, Dugast AS, Heslan JM, Ashton-ehess J, Gira! M, Degauque N, Moreau A, Pallier A, ehiffoleau E, Lair D, et al: lmplication of matrix metalloproteinase 7 and the noncanonical wingless-type signaling pathway in a model of kidney allograft tolerance induced by the administration of anti-donor class 11 antibodies. J lmmunol2008, 180:1317-1325.

146 227. Yoshioka S, King ML, Ran S, Okuda H, Maclean JA, 2nd, McAsey ME, Sugino N, Brard L, Watabe K, Hayashi K: WNT7A regulates tumor growth and progression in ovarian cancer through the WNT/beta-catenin pathway. Mol Cancer Res 2012, 10:469-482.

228. Erkan M, Hausmann S, Michalski CW, Fingerle AA, Dobritz M, Kleeff J, Friess H: The role of stroma in pancreatic cancer: diagnostic and therapeutic implications. Nature revíews Gastroenterology & hepatology 2012.

229. Holzapfel K, Reiser-Erkan C, Fingerle AA, Erkan M, Eiber MJ, Rummeny EJ, Friess H, Kleeff J, Gaa J: Comparison of diffusion-weighted MR imaging and multidetector-row CT in the detection of liver metastases in patients operated for pancreatic cancer. Abdom lmagíng 2011,36:179-184.

230. Aichler M, Seiler C, Tost M, Siveke J, Mazur PK, Da Silva-Buttkus P, Bartsch DK, Langer P, Chiblak S, Durr A, et al: Origin of pancreatic ductal adenocarcinoma from atypical flat lesions: a comparative study in transgenic mice and human tissues. J Pathol 2012, 226:723-734.

231. Hori SS, Gambhir SS: Mathematical model identifies blood biomarker-based early cancer detection strategies and limitations. Scí Transl Med 2011, 3:109ra116.

232. Bangur CS, Switzer A, Fan L, Marton MJ, Meyer MR, Wang T: ldentification of genes over­ expressed in small cell lung carcinoma using suppression subtractive hybridization and cONA microarray expression analysis. Oncogene 2002, 21:3814-3825.

233. Kelly KA, Bardeesy N, Anbazhagan R, Gurumurthy S, Berger J, Alencar H, Depinho RA, Mahmood U, Weissleder R: Targeted nanoparticles for imaging incipient pancreatic ductal adenocarcinoma. PLoS Med 2008, 5:e85.

234. Bausch D, Thomas S, Mino-Kenudson M, Fernandez-del CC, Bauer TW, Williams M, Warshaw AL, Thayer SP, Kelly KA: Plectin-1 as a novel biomarker for pancreatic cancer. Clín Cancer Res 2011, 17:302-309.

235. Weissleder R, Pittet MJ: lmaging in the era of molecular oncology. Nature 2008, 452:580-589.

147 8. ANEXOS

ANEXO l. Patente no provisoria. GIDEKEL Manuel, PODHAJCER Osvaldo, ROA lván, BIZAMA Carolina, BENAVENTE Felipe, ESPINOZA Jaime, SALVATIERRA Edgardo, FERNANDEZ Elmer, GUTIERREZ Ana, ROA Juan Carlos, MAZZOLINI Guillermo: Novel genes and uses thereof, expression profile of colon, gastric and pancreatic cancer. Patente USPTO. Número presentación 61/404,141 fecha 09/28/2010. CTI-Salud.

ANEXO 11. Patente provisoria. GIDEKEL Manuel, ESPINOZA Jaime, MAZZOLINI Guillermo, CABEZON Teresa. Tumor biomarkers for pancreatic cancer. Patente Provisional. CTI-Salud.

ANEXO 111. Artículo sometido a Molecular Oncology Manuscrito en preparación. Jaime A. Espinoza, Carolina Bizama, Felipe Benavente, Elmer A. Fernández, Edgardo Salvatierra, Hilda A. Gutierrez, lván Roa, Guillermo Mazzolini, Osvaldo Podhajcer. ldentification of novel human pancreatic cancer biomarkers using low abundance transcriptome analyses.

ANEXO IV. Artículo sometido a Cancer Research Carolina Bizama, Felipe Benavente, Jaime A. Espinoza, Edgardo Salvatierra, Hilda A. Gutiérrez, Elmer A. Fernández, Eduardo A. Sagredo, lván Roa, Mazzolini Guillermo and Osvaldo L. Podhajcer. Low abundance transcriptome analyses identifies novel markers, specific intracellular pathways and target genes in advanced human gastrointestinal cancer. Enviado a Cancer Research.

148 ANEXO V. Artículo publicado (Co-autoría) M Malvicini, M lngolotti, F Piccioni, M García, J Bayo, C Atorrasagasti, L Alaniz, J Aquino, JA Espinoza, M Gidekel, OG Scharovsky, P Matar& G Mazzolini. Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12. Molecular Oncology 2011, 5(3):242-55.

149 ANEXO 1

PATENTE NO PROVISORIA USPTO 2011

Número presentación 61/404,141 fecha 09/28/201 O Novel genes and uses thereof, expression profile of colon, gastric and pancreatic cancer.

150 UNITED STATES PATENT A"lD TRPJJEMARK ÜFFIGE UXJTJ-:0 f.!T-\Tf:::S DRP.·\ RTMF.Yf OF CXJ.\11\JgRCE Uniwd St.nW:s f':t.tE-nt aml'lrad~marl. Offi~ A<.~·COMM!S!W.J.>lER .FOR PbTENi'S PO ll<,,:::t«:'ID AkH!~~ Yll$ir1i>'l :l2:\l.:H.t.."0 v..~.:t$_?1('<$,~'<'

ATTY.DOCKET,NO Ven!urdab-Colon CONFIRMATION NO. 3788 27772 FILING RECEIPT DODDS & ASSOCIATES 1707 N STREET NW IIIIIIIIIII~~~JPJil~~~~~lmUIIIrnll WASHINGTON, DC 20036 Date Mailed: 10/25/2010

Receipt is acl

151 Title Novel genes and uses thereof, expression profile of colon, gastric and pancreatic cancer PROTECTING YOUR INVENTION OUTSIDE THE UNITED STATES

Since the rights granted by a U.S. patent extend only throughout the territory of the United States and have no effect in a foreign country, an inventor who wishes patent protection in another country must apply for a patent in a specific country or in regional patent offices. Applicants may wish to consider the filing of an international application under the Patent Cooperation Treaty (PCT). An international (PCT) application generally has the same effect as a regular national patent application in each PCT-member country. The PCT process slmpllfies the filing of patent app!ications on !he same invention in member countries, but does not result in a grant of "an international patent" and does not eliminate the need of applicants to file additional documents and lees in countries where patent protection is desired.

Almost every country has its own patent law, anda person desiring a patent in a particular country must make an application for patent in that country in accordance wilh its particular laws. Since the laws of many countries differ in various respects from !he patent law of the United States, applicants are advised to seek guidance from specific foreign countries lo ensure that patent rights are not los! prematurely.

Applicants also are advised that in !he case of inventions made in !he United States, the Director ol the USPTO must issue a license before applicants can apply for a patent in a foreign country. The fi!ing ola U.S. patent application serves as a request for a foreign filing license. The application's filing receipt contains further information and guídance asto the status of applicant's license for foreign filing.

Applicants may wish to consult the USPTO booklet, "Generallnforrnatíon Concerníng Patents" {specífically, the sectíon entitled "Treatíes and Foreign Patents") for more ínlormation on timeframes and deadlines for fíling foreign patent applicatíons. The guide ís available eíther by contacting the USPTO Contact Center at 800-786-9 '! 99, or it can be viewed on the USPTO websíte al http://www.uspto.gov/web/offices/pac/docfgeneral/index.html.

For information on preventing theft of your intellectual property {patents, trademarks and copyrigllts), you may wisll to consult the U.S. Government website, http://www.stopfakes.gov. Part of a Department of Commerce initíative, this website includes self-help "toolkits" giving innovators guidance on how to protect intellectual property in specific countries such as China, Korea and Mexíco. For questions regarding patent enforcement issues, applicants may call the U.S. Government hotline at 1-866-999-HALT (1-866-999-4158).

LICENSE FOR FOREIGN FILING UNDER Title 35, United States Code, Section 184 Title 37, Code of Federal Regulations, 5.11 & 5.15 GRANTED The applicant has been granted a license under 35 U.S.C. 184, if the pllrase "IF REOUIRED, FOREIGN FILING LICENSE GRANTED" followed by a date appears on this form. Such licenses are issued in all applicatíons where the conditions for issuance of a license have been met, regardless of whether or nota license may be required as set forth in 37 CFR 5.15. The scope and limitations of this license are set forth in 37 CFR 5.15(a) unless an earlier license has been íssued under 37 CFR 5. 15(b). The license is subject to revocation upon written notification. The

page2of3

152 date indicated is the effective date of the license, unless an earlier license of similar scope has been granted under 37 CFR 5.13 or 5.14.

Thís lícense is to be retained by the lícensee and may be used at any time on or alter the effective date thereof unless ít ís revoked. This license is automatically transferred to any related applícations(s) fíled under 37 CFR 1.53(d). This license is not retroactiva.

The grant of a license does not in any way lessen the responsibility of a licensee for the securíty of the subject matter as imposed by any Government contract or the provisions of existing laws relating to espionage and the natlonal securíty or the export of technical data. Ucensees should apprise themselves of curren! regulatíons especially with respect to certaín countries, of other agencies, particularly the Oflíce of Defensa Trade Controls, Department of State (wíth respect to Arms, Munítions and lmplements of War (22 CFR 121-128)); the Bureau of lndustry and Security, Department of Commerce (15 CFR parts 730-774); the Offíce of Foreígn AssetsControl, Department of Treasury (31 CFR Parts 500+) and the Department of Energy.

NOTGRANTED No license under 35 U.S.C. 184 has been granted at this time, if the phrase "IF REOUIRED, FOREIGN FILING U CENSE GRANTED" DOES NOT appear on this form. Applicant may stlll petition for a lícense under 37 CFR 5. i 2, íl a lícense is desired befo re the expiration of 6 months from the filing date of the application. lf 6 months has lapsed from the filing date of this application and the licensee has not received any índication of a secrecy order under 35 U.S.C. 181, the licensee may toreign file the application pursuant to 37 CFR 5.15(b).

page 3 of3

153 ANEXO 11

PATENTE PROVISORIA 2012

Tumor biomarkers for pancreatic cancer.

154 IJS. PTO JIIIIIIJII 61634832 030712 Provisional Patent Applicatíon Inventor- 1\ftanuel Gídekel 1nventíon: Tumor Bíomarkers tor Pancreatíc Canear Specifícatíon & Figures Payment check $125 Data Sheet Oaths Submítted by Dodds and Associates

155 ANEXO 111

ARTICULO SOMETIDO A MOLECULAR ONCOLOGY Jdentification ofnovel human pancreatic cancer biomarkers using low-abundance transcriptome analyses

156 Elsevier Editorial System(tm) for Molecular Oncology Manuscript Draft

Manuscript Number:

Title: Identification ofnovel human pancreatic cancer biomarkers using low-abundance transcriptome analyses

Article Type: Research Paper

Keywords: Pancreatic cancer, microarray, suppression subtractive hybridisation, low-abundance transcriptome, PLEKHMl, WNT9A.

Corresponding Author: Dr Manuel Gidekel, Ph.D.

Corresponding Author's Institution: La Frontera University

First Author: Jaime A Espinoza, PhD

Order of Authors: Jaime A Espinoza, PhD; Carolina Bizama, PhD; Felipe Benavente, PhD; Guillermo Mazzolini, MD, PhD; Elmer A Fernández, PhD; Edgardo Salvatierra, PhD; Ana Gutiérrez-Moraga, PhD; Antonio Ferrández-Izquierdo, MD; Osvaldo Podhajcer, PhD; Iván Roa, MD; Manuel Gidekel, Ph.D.

Abstract: Pancreatic cancer is a deadly disease with no effective treatment for patients with advanced­ stage disease. Thus, analyses of the low-abundance transcriptome of cancer cells are important to identify mechanisms underlying tumour progression, especially for advanced pancreatic cancer. Using suppression subtractive hybridisation followed by transcriptome analysis ofhuman samples obtained from pan creatic cancer tissues and paired, adjacent non-cancerous tissues, we identified novel cancer biomarkers, such as WNT9A and PLEKHMl, that were not detectable by direct transcriptomic analyses. Then, by performing tissue microarrays, WNT9A and PLEKHM 1 were confirmed as proteins that are differentially expressed in pancreatic cancer tissues. The use ofthese combined platforms may have important implications for the understanding of pancreatic cancer biology, although further investigation is required to evaluate the diagnostic potential ofthese markers. •er Letter

November, 2012

Julio Celis, Editor-in-Chief

Molecular Oncology

Please, find enclosed a manuscript by Jaime Espinoza and colleagues entitled "Identification of novel human pancreatic canccr biomarkcrs using low abundancc transcriptomc analyses" to evaluate its publication in your prestigious journal.

It is widely kno\vn that the prognosis of patients with advanced pan crea tic cancer is very poor and no efficacious treatment has yet been established. We believe, as others, that studies on the lo\v abundance transcriptome are of paramou11t importance to clarify the intimate mecha11isms of tumor progression and identify novel genes that can serve as biomarkers and/or targets.

With this in mind we applied a combination of subtractive hybridization followed by microarrays 011 samples obtained from patients with pancreatic cancer. The RNA obtained from the malignant and paired adjacent non-malignant tissue was subtracted with a common reference RNA and further hybridized with microarrays spanning most of the genome. We perforrned exhaustive controls to assured that the subtractive step indeed worked as expected.

By using this combined platform we identified:

a) Differe11tially expressed genes that have been validated by qRT-PCR

b) Novel biomarkers that have been validated by tissue microarrays 011 a new cohort of samples

This two steps platform combines an initial procedure that eliminated most of the highly expressed genes mainly associated with extracellular matrix, which are the commo11 genes found associated with cancer when direct microarrays hybridization is perforn1ed. We believe that the present approach that can be afforded by a11y molecular biology laboratory in the world could be very useful in the identification of novel genes with ímportant roles in cancer as other diseases as well.

All the authors declare no financia! co11flict of interest that might be construed to influence the rcsults or interpretation of the present rnanuscript. In addition, there are no restrictions to the availability of any materials, data, or infonnation related to the manuscript. ~~ Manuel iidekel ~ Universidad de La Frontera

-..,..-,.....,._...,-·······------····· ¡hlights (for review)

Highlights

We studied the low abundance transcriptome of human pancreatic cancer samples.

Novel markers that might help in tumor diagnosis were identified.

PLEKHMI and WNT9A are differentially expressed in pancreatic cancer at the transcription and protein level. 'Manuscript :lick here to view linked References

1 Identification of novel human pancreatic cancer biomarkers usmg low-

2 abundance transcriptome analyses

1 1 1 2 3 Jaime A. Espinoza , Carolina Bizama , Felipe Benavente , Guillermo Mazzolini , Elmer 3 4 5 4 A. Fernández , Edgardo Salvatierra , Ana Gutiérrez-Moraga , Antonio Ferrández- 6 7 8 5 Izquierdo , Osvaldo Podhajcer \ Iván Roa , Manuel Gidekel

6 1 Applied Cell and Molecular Biology PhD Program, La Frontera University, Temuco 7 4811230, Chile

8 2 Gene Therapy Laboratory, School of Medicine, Austral University, Pilar-Buenos

9 Aires B1664INZ, Argentina

10 3 School of Engineering, Biosciences Data Mining Group, Catholic University of

11 Córdoba, Córdoba X5016DHK, Argentina

12 4 Laboratory of Molecular and Cellular Therapy, Fundación Instituto Leloir, Buenos 13 Aires C1405BWE, Argentina

14 5 Laboratory of Applied Molecular Biology, Faculty of Agriculture and Forestry, La 15 Frontera University, Temuco 4811230, Chile

16 6 Department of Pathology, Clinical Hospital, University of Valencia, Valencia 46007,

17 Spain

18 7 Pathology Service, Clínica Alemana de Santiago, Faculty of Medicine, Universidad 19 del Desarrollo, Santiago 7650568, Chile

20 8 Vicerrectoría de Investigación y Postgrado. La Frontera University, Temuco 4811230, 21 Chile

22

23

24

25 Correspondence to: Manuel Gidekel, Av. Alvaro Casanova 4090, Peñalolén, Santiago, 26 7941068. Phone: +56-999970216; E-mail: [email protected]

27

1 28 Abbreviations:

29 CK 19, Cytokeratin 19;

30 GAPDH, glyceraldehyde-3-phosphate dehydrogenase;

31 GO, Gene ontology;

32 lliQ, Immunohistochemistry;

33 MMP7, Matrix metalloproteinase 7;

34 PLEKHM1, pleckstrin homology domain containing, family M (with RUN domain)

35 member 1;

36 PC, Pancreatic cancer;

37 SSH, Suppression subtractive hybridisation;

38 TMA, Tissue microarray;

39 qRT-PCR, quantitative reverse transcription polymerase chain reaction.

40 WNT9A, wingless-type MMTV integration site family, member 9A.

41

42

43

44

45

46

47

48

49

50

2 51 Abstract

52 Pancreatic cancer is a deadly disease with no effective treatment for patients 53 with advanced-stage disease. Thus, analyses of the low-abundance transcriptome of 54 cancer cells are important to identi:f)r mechanisms underlying tumour progression, 55 especially for advanced pancreatic cancer. Using suppression subtractive hybridisation 56 followed by transcriptome analysis of human samples obtained from pancreatic cancer

57 tissues and paired, adjacent non-cancerous tissues, we identified novel cancer

58 biomarkers, such as WNT9A and PLEKHMl, that were not detectable by direct

59 transcriptomic analyses. Then, by performing tissue microarrays, WNT9A and 60 PLEKHMl were confirmed as proteins that are differentially expressed in pancreatic 61 cancer tissues. The use of these combined platforms may have important implications

62 for the understanding of pancreatic cancer biology, although further investigation is

63 required to evaluate the diagnostic potential of these markers.

64

65

66

67

68

69

70

71

72

73

74

75

76

3 77 Keywords: Pancreatic cancer, microarray, suppression subtractive hybridisation, low- 78 abundance transcriptome, PLEKHMI, WNT9A.

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

4 100 l. Introduction

101 Pancreatic cancer (PC) is a devastating disease with an overall 5-year survival 102 rate of less than 5% (Hidalgo, 201 0). The factors responsible for this extremely low 103 overall survival rate include the lack of early diagnostic tests and the limited efficacy of 104 therapies for patients with advanced-stage disease (Vincent et al., 2011 ). Recent 105 evidence indicates that PC grows very slowly, which provides a potential window for 106 early diagnosis and curative therapy (Yachida et al., 201 0). Unfortunately, PC is 107 extremely difficult to detect in its early stages.

108 Gene expression profiling by DNA microarray technology has been used for the 109 molecular characterisation of cancers, including PC, to better understand cancer biology 110 and aid in the development of new diagnostic and therapeutic targets (Badea et al., 111 2008; Cmogorac-Jurcevic et al., 2002; Iacobuzio-Donahue et al., 2003b; Iacobuzio- 112 Donahue et al., 2002a; Iacobuzio-Donahue et al., 2002b ). However, a frequent 113 limitation of these expression profiles is the transcriptomic predominance of the 114 desmoplastic reaction, a common aspect ofthese types oftumours (Neesse et al., 2010). 115 This strong desmoplastic signature may affect the identification of markers specific for 116 PC.

117 Suppression subtractive hybridisation (SSH) is a PCR-based technique that 118 permits cDNA normalisation and subtraction in a single procedure (Diatchenko et al., 119 1996), allowing for the enrichment of specific transcripts present in biological samples 120 such as normal and tumoural tissues. Combining SSH with oligonucleotide microarray 121 increases the efficiency of the characterisation of low-abundance transcripts that are 122 expressed specifically in target samples. This technique has been successfully used for 123 the analysis ofthe low-expression transcriptomes ofhepatocellular (Liu et al., 2008; Liu 124 et al., 2007; Pan et al., 2006), breast, nasopharyngeal (Liu et al., 2008), lung and renal 125 cell carcinoma (Amatschek et al., 2004).

126 In the present study, we applied SSH-microarray (indirect strategy) technology 127 to the transcriptomic study of PC to identify low-abundance transcripts. The indirect 128 strategy enriched distinct ontologies and pathways than those identified by conventional 129 microarray (direct strategy) and highlighted the presence of genes related to the Wnt 130 signalling pathway. Furthermore, for two genes identified using the indirect strategy,

S 131 WNT9A and PLEKHMl, we used tissue microarray to confirm that the differential 132 expression was reflected at the protein level.

133

6 134

135 2. Materials and Methods

136

137 2.1. Clinical specimens and tissue microarrays (TMAs)

138 Six fresh frozen PC samples and non-neoplastic matched tissues were obtained from 139 the National Tumour Bank of Chile. Approval from the Ethics Committee Bank was 140 obtained before the samples were analysed. Each patient signed the informed consent 141 form, and an anonymization algorithm was applied for deceased patients. In all cases, 142 the tissues were collected from patients who did not receive any adjuvant therapy prior 143 to surgery. The collected tissues were preserved in RNALater (Ambion Inc, Austin TX, 144 USA) at -80 oc until the transcriptomic analysis was performed. Commercially 145 available pancreatic adenocarcinoma TMA (A307; AccuMax Array), obtained from ISU 146 ABXIS (Seoul, South Korea), contained 30 PC cases with matched non-neoplastic 147 tissues. Each tumoural tissue was spotted in duplicate, while each non-neoplastic tissue 148 was spotted in a single core. Additional formalin-fixed paraffin-embedded tissue 149 sections of pancreatic cancer tissues were obtained from tumor bank of University 150 Hospital ofValencia, Spain. Pantomic MN0661 multi-normal human TMA containing 151 33 normal tissues spotted by duplicate was kindly donated by the Departament of 152 Proteomics in Cancer (Danish Cancer Society Research Center, Copenhagen, 153 Denmark).

154

155 2.2. Suppressive subtractive hybridisation (SSH) and microarray 156 procedures

157 Total RNA was extracted from PC tissues using Trizol reagent (Invitrogen Corp., 158 Carlsbad, CA, USA) and purified using RNeasy mini kit columns (Qiagen, Hilden, 159 Germany) according to the manufacturer' s instructions. Human Universal Reference 160 RNA and Human Pancreas Total RNA were purchased from Clontech (Palo Alto, CA, 161 USA). The integrity of the RNA was assessed by denaturing agar and ultraviolet 162 spectrophotometry. Total RNA from PC and non-neoplastic tissues was used as the 163 tester, and commercial Human Pancreas Total RNA served as the driver. First strand 164 cDNA was synthesised from 1 Jlg of total RNA using a Super SMART PCR cDNA

7 165 Synthesis kit (Clontech, Palo Alto, CA, USA) and SuperScript III Reverse Transcriptase 166 (Invitrogen, Carlsbad, CA, USA) according to the supplier's protocol. SSH was 167 performed using the Clontech PCR-Select cDNA Subtraction kit (Clontech, Palo Alto, 168 CA, USA). SSH was performed according to the manufacturer' s instructions except for 169 the use of a modified nested PCR Primer IR 5 '- 170 CTAATACGACTCACTATAGGGCTCGAGCGGCC-3' in the secondary PCR, which 171 included a T7 promoter site to carry out in vitro transcription of the subtractive

172 amplicon. After the secondary PCR, the subtractive cDNA was purified using an 173 E.Z.N.A Cycle Pure kit (Omega Bio-Tek, Norcross, GA, USA). Subtraction was 174 confirmed by measuring the GAPDH transcript by qPCR and was considered successful 175 if the subtracted samples amplified more than five cycles later than the non-subtracted 176 samples.

177 For in vitro transcription, 300 ng ofthe purified subtractive cDNA was used as the 178 template. The aRNA was synthesised and labelled with aminoallyl-UTP and AlexaFluor 179 647 (Invitrogen, Carlsbad, CA, USA) using the SuperScript Indirect RNA 180 Amplification System (Invitrogen, Carlsbad, CA, USA). The aRNA for the direct 181 strategy was amplified from 1 ¡.tg of total RNA and labelled using the same system 182 described above. Labelled aRNA from the direct and the SSH-strategy were employed 183 for hybridisation on 48.5K Exonic Evidence-Based Oligonucleotide (HEEBO) arrays 184 purchased from Microarray Inc. (Nashville, TN, USA). Prior to hybridisation, slides 185 were pre-blocked with 5X SSC, 0.1% BSA and 0.1% SDS. The fluorescent-labelled 186 probe was mixed with IX hybridisation solution (SX SSC, 50% formamide, 0.1% SDS, 187 and 0.01% salmon sperm DNA) and heated at 95 oc for 2 min. The samples were 188 hybridised on microarray slides for 16 hours at 42 oc and then scanned using a 189 ScanArray Gx (Perkin Elmer, Waltham, MA, USA).

190

191 2.3. Data analysis and gene set enrichment

192 The microarray signal intensity was evaluated using SpotReader software (Niles 193 Scientific, Portola Valley, CA, USA). The raw data were deposited in the Gene 194 Expression Omnibus (GEO, http://vvww.ncbi.nlm.nih.gov/projects/geo/) under the 195 accession number GSE39751. The normalisation was performed in an R statistical 196 environment using the Limma package (http://w\vw.r-proyect.org). The raw data from

8 197 the individual arrays were processed using standard and normexp background correction 198 (Ritchie et al., 2007) and printtiploess normalisation (Smyth and Speed, 2003 ). The 199 global scale normalisation function using the median absolute deviation was used for 200 normalisation between arrays (Yang et al., 2002). Heatmaps were constructed using 201 MeV software (Saeed et al., 2006). The gene ontology (GO) analysis was performed 202 using DAVID Bioinformatics Resources 6.7 (http://david.abcc.ncifcrf.gov/) (Huang da 203 et al., 2009), and a pathway analysis was performed with the KEGG database 204 (http://WW\v.genome.jplkegg/) (Kanehisa et al., 2012).

205 2.4. Quantitative reverse-transcription PCR (qRT-PCR)

206 To confirm the data obtained using the direct and indirect strategies, we chose ten 207 genes from each strategy to analyse with qRT-PCR for the same six-pair samples used 208 in the microarray experiments. Por the direct strategy, BHLHE40, CALU, COL4A1, 209 CTSK, PN1, IGPBP5, PMEPA1, SBN02, SPARC and TOP2A were analysed. Por the 210 indirect strategy, CD55, DPYSL4, KNG1, LAMA3, MMP7, OASL, PLEKHM1,

211 STATl, SYNE2 and SYT12 were analysed. A list ofprimer sequences can be found in 212 Supplementary table Sl. QARS and TBP were used as interna! controls for

213 normalisation using GeNorm software (Vandesompele et al., 2002) (data not shown). 214 Por cDNA synthesis, 1 flg of RNA was reverse-transcribed using the AffinityScript 215 qRT-PCR cDNA Synthesis kit (Stratagene, USA). Brilliant II SYBR Green qRT-PCR 216 Master Mix (Stratagene, USA) was used according to the manufacturer's instructions. 217 Reactions were quantified with a real-time thermocycle Mx3000p (Stratagene, USA). 218 Relative quantification was performed using MxPRO software (Stratagene, La Jolla, 219 CA, USA). The statistical significance of the difference between the two groups was 220 determined by one-tailed paired Student's t-tests using GraphPad Prism version 5.0 for

221 windows (GraphPad Software Inc., San Diego, USA); p values < 0.05 were considered 222 statistically significant.

223 2.5. Immunohistochemistry (IHQ) and immunofluorescence staining and

224 analysis

225 Rabbit polyclonal antibodies against PLEKHM1 (HPA021558; dilution 1:75) 226 and WNT9A (HPA011223; dilution 1:25) were purchased from Atlas Antibodies AB 227 (Stockholm, Sweden). The MMP7 antibody (clone 111433; dilution 1:500) was

228 purchased from R&D Systems (Minneapolis, MN, USA), and the CK19 antibody

9 229 (m o use mono-clona!; dilution 1: 1000) was purchased from Labvision-Thermo Scientific 230 (Kalamazoo, MI, USA). The tissue microarray slides were baked at 70 oc for 60 min,

231 deparaffmised, blocked with 3% H20 2 in 99% ethanol and rehydrated through graded 232 alcohol rinses. Heat-induced antigen retrieval was performed by immersing the TMA 233 slides in Tris/EDTA (pH 9.0) buffer (10 mM Tris, 1 mM EDTA) and then microwaving 234 in a 750 W microwave oven for 10 min. Nonspecific staining was blocked by 235 incubating with 1% foetal bovine serum in TBS for 1O min. The slides were incubated 236 with the primary antibody for 45 min, an anti-rabbit antibody conjugated to a 237 peroxidase complex for 45 min (Envision+ detection kit DAKO, Denmark) and finally 238 with the chromogen, and then the slides were counterstained with hematoxylin. Indirect 239 immunofluorescence analysis was performed as previously described with minor 240 modifications (Moreira et al., 2010). Immune complexes were detected with anti- 241 ideotypic secondary antibodies conjugated to Alexa Fluor 488 (Green, PLEKHM1) and 242 Alexa Fluor 568 (Red, cytokeratin 19) (Molecular Probes, USA).

243 The automated imaging system ACIS III was used for the digitalisation of the 244 TMAs and the quantitative assessment of the IHQ staining. For the TMA analysis, a 245 staining score was generated for each core based on the staining intensity and the areas 246 with positive immunostaining, as described elsewhere (Gromov et al., 2010). The 247 staining scores were compared using a two-tailed Student's t test (p < 0.05 was 248 considered statistically significant). An average of two tumour cores from each patient 249 was compared with the respective non-neoplastic tissues. The statistical analysis of the 250 data was GraphPad Prism 5 software (GraphPad Software, Inc, San Diego, CA, USA).

251

252

253

254

255

256

257

10 258 3. Results

259 260 3.1. Validation of the indirect strategy

261 To identify genes differentially expressed in PC tissues, we performed a gene 262 expression study using a conventional microarray approach for gene expression analysis 263 (direct strategy) and a modified approach coupling suppression subtractive 264 hybridisation with microarray (indirect strategy). Total RNA extracted from six PC 265 tissues with their corresponding matched non-neoplastic tissues was used as the tester 266 for subtractions, and human normal pancreas RNA was used as the driver. fu parallel, 267 the same 12 tissues were analysed using the direct strategy to compare data obtained 268 from both strategies (Figure 1). The subtraction efficiency was confirmed using qPCR 269 prior to hybridisation on microarray slides. The GAPDH transcript was measured in 270 subtracted and non-subtracted samples, and the subtraction was considered efficient if 271 there was an amplification delay of greater than five cycles for the subtracted sample

272 compared to the non-subtracted sample (Supplementary figure S 1).

273 To confmn the subtraction efficiency of the indirect strategy, we analysed the 27 4 relative expression of 100 housekeeping genes related to ribosome structure, 275 mitochondria, carbohydrate and nucleotide metabolism, and other biological processes. 276 Then, we compared both the direct and indirect strategies. A reduction of relative 277 expression >50% was observed for a mean of 37 genes in all samples (Supplementary 278 table S2). The following genes showed reduced expression in >80% of the samples: 279 RPSll, ZFP36Ll, FAU, COX4Il, TKT, RPL19, LDHA9, TUB6A, RPL29, RPL13A

280 and NACA.

281

282 3.2. Differential gene expression of direct and indirect strategies

283 For both the direct and indirect strategies, a statistical analysis was performed 284 according to the comparison between PC and non-neoplastic tissues. For the direct 285 strategy, 736 sequences were identified with a p value < 0.01 and fold changes >0.4 and 286 <-0.4 (505 up-regulated and 231 down-regulated). For the indirect strategy, 616 287 sequences were identified with a p value < 0.05 and fold changes >0.35 and <-0.35 (312

288 up-regulated and 304 down-regulated) (Figure 2A and Supplementary table S3). The p

11 289 values and fold change thresholds were selected to achieve a level of differentially 290 expressed genes that allowed for a proper ontology analysis (Fresno et al., 20I2). For 291 both strategies, the differentially expressed genes enabled discrimination between PC 292 and non-neoplastic samples with an overlap of 35 genes (Figure 2B). Among those 293 genes, CD55, EGFR, F3 STATl and TAOK2 were identified. The up- and down- 294 regulated genes identified by the indirect strategy were tracked in the data matrix of the 295 direct strategy to identifY the levels of change. The mean expression of these genes was 296 approximately a zero-fold change, indicating that the direct strategy was not suitable for 297 measuring fold changes of genes expressed at low levels (data not shown).

298 To confirm the microarray data obtained from the direct and indirect strategies, 299 a qRT-PCR assay was performed to measure the transcription levels of IO up-regulated 300 genes for each strategy. Each measured gene tended to be overexpressed in PC tissue 301 compared with non-neoplastic tissues. This was consistent with the results obtained 302 from the microarray analysis. Eighty percent (8/10) ofthe genes from the direct strategy

303 and 40% (4/I O) of the genes from the indirect strategy were statistically significant and 304 tending to overexpression in tumoural tissues (Figure IC and ID, respectively).

305

306 3.3. Gene ontology

307 To examine whether particular gene ontologies (GO) were enriched with each 308 strategy, indirect and direct gene lists were submitted to the DA VID bioinformatics 309 resource, which employs a Fisher's exact test to assess GO terms over-representation, 310 using a p value cut-off ofp < 0.05.

311 As PC is a highly fibrotic tumour, the presence of stromal components was 312 expected because whole tissue was used as the starting material. The results of the 313 direct strategy were significantly enriched for genes that encode proteins of 314 predominantly extracellular location, as well as genes related to collagen (COLlAI,

315 COL17AI, CTHRCI), the extracellular matrix (VCAN, POSTN, FNI), the 316 endoplasmic reticulum (EDEM3, STIMI, FNDC3B), the basement membrane (NIDI,

317 DAGI) and the Golgi apparatus (RAB7A, RABIO) (Figure 3A). We identified the 318 regulation of cell communication, response to wound healing, cell adhesion and

12 319 programmed cell death as biological processes enriched by the direct strategy (Figure 320 3B).

321 The indirect strategy dramatically reduced the enrichment of genes that encode 322 extracellular matrix proteins and enabled further enrichment of genes that encode 323 intracellular proteins, although this enrichment was not statistically significant (Figure 324 3A). Interestingly, the indirect strategy enriched genes encoding proteins involved in

325 stem cell differentiation (APC, MSI2, RIFI, ERCC2 and ACE) and the Wnt/~-catenin 326 signalling pathway. Further analysis of the indirect strategy using the KEGG database 327 revealed the following six up-regulated genes involved in the Wnt signalling pathway 328 that were not found with the direct strategy: WNT9A, PPP2RIB, TCF7L2, CCNDl, 329 MMP7 and NFATS.

330 Overall, the indirect strategy reduced the identification of transcripts highly 331 represented in PC transcriptomes and allowed for the enrichment of genes encoding 332 proteins involved in biological processes that were not detected using the direct 333 strategy.

334

335 3.4. Immunohistochemical validation of the expression microarray 336 results

337 To confirm the data obtained by the indirect strategy, an immunohistochemical 338 analysis was performed using PDAC and non-neoplastic tissues sections. Two candidate 339 genes, WNT9A and PLEKHMl, were chosen based on their biological novelty.

340 WNT9A-positive staining was observed in non-neoplastic ducts with a moderate 341 to intense supranuclear granular pattem. Acinar cells and the islets of Langerhans were 342 negative for WNT9A (Figure 4A). A similar pattem of expression was observed in 343 tumoural cells but with greater signa! intensity (Figure 4B and C). WNT9A expression 344 in non-neoplastic pancreas and tissues from primary PC was compared using a tissue 345 microarray (n=28). The TMA analysis established that WNT9A expression was 346 increased significantly in tumour samples compared to non-neoplastic tissues (Figure 347 4D); nearly 50% ofthe tumour samples had above-average WNT9A expression levels.

13 348 PLEKHM1 was not detected in acinar cells, the is1ets of Langerhans or non- 349 neoplastic ducts (Figure 4E). A group of pancreatic tumours showed a diffuse 350 cytoplasmic pattern of staining for PLEKHM1 as well as sorne cell membrane staining 351 (Figure 4F and G). The TMA analysis established that PLEKHM1 expression increased 352 significantly in tumour samples compared to non-neoplastic tissues (n=30) (Figure 4H);

353 above-average expression levels were observed in ~30% of the samples. To confirm 354 that the reactive cells were of epithelial origin, a three-colour indirect 355 immunofluorescence analysis was performed using an anti-CK 19 (keratin 19) antibody 356 as an epithelial marker. PLEKHM1 expression was restricted to CK 19-positive cells in 357 PC tissues (Figure 5A). Interestingly, PLEKHM1 expression was observed in cancer 358 cells that were involved in an apparent invasive process and displayed positive staining 359 for MMP7 (Figure 5B), a matrix metalloproteinase implicated in the initiation and 360 progression of pancreatic cancer (Fukuda et al., 2011). In addition, MMP7 was 361 identified by the indirect strategy as a transcript that was differentially expressed in PC 362 tissues. The IHQ analysis of the human normal TMA revealed that PLEKHM1 was not 363 detected in most normal tissues, including the pan creas, with the exception of the skin, 364 uterus, cervix, thymus, tonsil and prostate (Supplementary figure S2).

365

14 366

367 4. Discussion

368 PC remains a challenging disease with an overall cumulative 5-year survival rate 369 below 1% (Hidalgo, 201 0). One of the characteristics of PC is its early loco-regional 370 dissemination that precludes the application of curative treatments in the majority of 371 patients. Despite important progress in understanding the molecular alterations involved

372 in PC growth, the processes of cancer initiation, progression and metastasis are still not 373 fully understood (Lyratzopoulos et al., 2012). However, it is clear that the progression 374 of PC, from the generation of the initiating mutation to the appearance of the first 375 metastatic clone, takes approximately 15 years, which provides a hypothetical time 376 window for early detection and curative therapy (Yachida et al., 201 0). Therefore, there

377 is an urgent need to identify new potential markers for this disease.

378 Previous studies have examined the native transcriptome of PC tissues to 379 characterise the biological behaviour of this disease and identify new potential 380 molecular markers. A common feature of many of these studies has been the 381 transcriptomic predominance of desmoplastic components such as fibroblasts, 382 pancreatic stellate cells, and other components of the tumour microenvironment, 383 including endothelial and inflammatory cells (Badea et al., 2008; Iacobuzio-Donahue et 384 al., 2003a; Iacobuzio-Donahue et al., 2002a; Iacobuzio-Donahue et al., 2002b). 385 Consistent with this, we found a strong enrichment of genes that encode for 386 extracellular proteins, especially structural proteins of the extracellular matrix. In 387 addition, genes related to biological processes such as the inflammatory response, cell 388 adhesion and programmed cell death were also represented at high levels in our

389 . samples.

390 The indirect strategy was performed for the same samples evaluated by the 391 direct strategy. PC and non-neoplastic tissues were used as testers, and normal pancreas 392 tissue was used as the driver for the subtraction assay. After subtraction and microarray 393 hybridisation, the PC and non-neoplastic tissues were compared to identify 394 differentially expressed transcripts in tumour samples. Significantly, indirect strategy- 395 enriched genes related to biological processes and cellular components were 396 dramatically different from those identified using the direct strategy. In addition, for 397 indirect strategy the amount of tumour microenvironment-related genes showed a

15 398 decrease, while the intracellular-related genes were increased compared with direct 399 strategy.

400 The Wnt/j3-catenin pathway is an important embryonic signalling pathway 401 required for morphogenesis and organogenesis (Clevers, 2006). However, its aberrant 402 activation is associated with several cancers, including PC (Jones et al., 2008; Lowe et 403 al., 2007; McCleary-Wheeler et al., 2012 ). More recenti y, the Wnt/j3-catenin pathway 404 has been linked to the metastatic process ofPC and is considered a potential therapeutic 405 target (Yu et al., 2012). In our study, severa} up-regulated transcripts in the Wntlj3- 406 catenin signalling pathway were identified by the indirect strategy but not by the direct 407 strategy. Among the up-regulated genes, WNT9A, TCF7L2 and NFATS were 408 identified. Recently, TCF7L2 was found to be up-regulated in PC tissues irrespective of 409 patient outcome (Van den Broeck et al., 2012) and was also reported as one ofthe most 410 frequently mutated genes in colorectal cancers (TCGA-Network, 2012).

411 WNT9A is a member of the Wnt signalling pathway, and it has a role in joint 412 integrity during chondrogenesis (Spater et al., 2006). WNT9A is also required for 413 proper morphogenesis of hepatic architecture during chicken embryogenesis 414 (Matsumoto et al., 2008). In our study, WNT9A was identified by the indirect strategy 415 as an up-regulated transcript in pancreatic cancer and was further validated as a protein 416 over-expressed in PC cells but not in non-tumoural pancreatic cells. The WNT9A gene 417 is commonly mutated in PC (Jones et al., 2008), although the role ofthis gene in PA has 418 not been studied in depth. The WNT9A protein has been reported to be a molecule 419 secreted into the extracellular space from the walls of hepatic sinusoids (Matsumoto et 420 al., 2008). Because many biomarkers used in the clinic are detected in patient fluids, 421 WTN9A may be an interesting candidate marker for further studies. The PLEKHM1 422 gene encodes a non-secretory adaptor protein that negatively regulates endosomal 423 trafficking (Tabata et al., 201 0), and a mutation in this gene causes osteopetrosis, a 424 disease generated by an altered endosomal acidification/maturation process in 425 osteoclasts (Del Fattore et al., 2008). However, there is a lack of information regarding 426 PLEKHM1 and its role in cancer. In our study, the PLEKHM1 protein was absent in the

427 majority of non-neoplastic pancreatic tissues, but positive lesions were found in ~30% 428 of PC samples. In our samples, PLEKHM1 expression was restricted to cykeratin (19 429 positive cells). Interestingly, MMP7, a matrix metalloproteinase intimately involved in 430 the initiation and progression of PC (Crawford et al., 2002; Fukuda et al., 2011 ), was

16 431 also detected in PLEKHMl-positive cells, suggesting a role for this factor in the 432 invasive process (Crawford et al., 2002; Fukuda et al., 2011). Moreover, the plasma 433 level of MMP7 has been reported to serve as a potential diagnostic biomarker when 434 used in combination with CA19-9 or other markers (Kuhlmann et al., 2007). However, 435 further research is needed to evaluate the role of PLEKHMl in PC progression and its 436 diagnostic potential.

437 WNT9A and PLEKMl were expressed differentially at the transcript and protein 438 levels in PC samples compared with matched non-neoplastic tissues. Interestingly, both 439 proteins were expressed at low levels in non-neoplastic tissues, with PLEKHMl 440 expression nearly absent in normal tissues. This is consistent with the fact that the goal 441 of the indirect strategy is to identify not only differentially expressed genes of low 442 abundance but also genes with high tumour-specific expression. Low abundance 443 transcripts are often misrepresented in conventional microarrays due to their low 444 intensities and are often excluded from further analysis (Qin et al., 2006). Coupling 445 microarray technology with subtractive suppressive hybridisation has been successful in 446 identifying low-abundance and tumour-specific transcripts in hepatocellular carcinoma 447 (Liu et al., 2008; Liu et al., 2007; Pan et al., 2006) as well as breast (Barraclough et al., 448 2010; Liu et al., 2008), nasopharyngeal (Liu et al., 2008; Zhang et al., 2003) and lung 449 cancer (Bangur et al., 2002; Wang et al., 2000). Interestingly, none of these studies 450 included a subsequent analysis of protein expression to ascertain whether the 451 differential expression oftranscripts affects protein levels.

452 This study provides evidence that the low-abundance transcriptome ofPC can be 453 a source of potential tumour markers. To the best of our knowledge, this was the first 454 study to apply suppressive subtractive hybridisation coupled with microarray analysis in 455 PC tissues, and the validation of differentially expressed transcripts at the protein level 456 increased the strength of our work. However, future studies are needed to evaluate the 457 clinical usefulness ofthese markers.

458

459

460

461

17 462 Conflict of interest

463 The authors declare that they have no competing interests.

464

465 Acknowledgements

466 We would like to thank Sofia Svensson, Anni Handesten, Lene Brogaard,

467 Soledad Lantadilla, Tamara Sánchez, Gabriela Bascuñán, Mónica Ramírez and

468 Francisco Gamín for their expert technical assistance. We thank Teresa Cabezón and 469 José Moreira for their contribution to the immunohistochemistry procedures, image

470 acquisition and analysis.

471

472 Funding

473 The present study was funded by PIA CTE-06 and Innova 12IDL2-13610. J.A.E

474 acknowledges the CONICYT fellowships N°21080240 and N°24100124 and the CHILE 4 7 S fellowship N°7 511 0023.

476

477 Appendix A. Supplementary data

478 Supplementary data related to this article can be found online at

479

480

18 481 References

482 Amatschek, S., Koenig, U., Auer, H., Steinlein, P., Pacher, M., Gruenfelder, A., Dekan, G., 483 Vogl, S., Kubista, E., Heider, K.H., Stratowa, C., Schreiber, M., Sommergruber, W., 2004. 484 Tissue-wide expression profiling using cDNA subtraction and microarrays to identify tumor- 48S specific genes. Cancer research 64, 844-856.

486 487 Badea, L., Herlea, V., Dima, S.O., Dumitrascu, T., Popescu, 1., 2008. Combined gene 488 expression analysis ofwhole-tissue and microdissected pancreatic ductal adenocarcinoma 489 identifies genes specifically overexpressed in tumor epithelia. Hepatogastroenterology 55, 2016- 490 2027.

491 492 Bangur, C.S., Switzer, A., Fan, L., Marton, M.J., Meyer, M.R., Wang, T., 2002. ldentification 493 of genes over-expressed in small celllung carcinoma using suppression subtractive 494 hybridization and cDNA microarray expression analysis. Oncogene 21, 3814-3825.

49S 496 Barraclough, D.L., Sewart, S., Rudland, P.S., Shoker, B.S., Sibson, D.R., Barraclough, R., 497 Davies, M.P., 201 O. Microarray analysis of suppression subtracted hybridisation libraries 498 identifies genes associated with breast cancer progression. Cell Oncol32, 87-99.

499 SOO Clevers, H., 2006. Wnt/beta-catenin signaling in development and disease. Cell 127, 469-480.

S01 S02 Crawford, H. C., Scoggins, C.R., Washington, M.K., Matrisian, L.M., Leach, S.D., 2002. Matrix S03 metalloproteinase-7 is expressed by pancreatic cancer precursors and regulates acinar-to-ductal S04 metaplasia in exocrine pancreas. J Clin lnvest 109, 1437-1444. sos S06 Cmogorac-Jurcevic, T., Efthimiou, E., Nielsen, T., Loader, J., Terris, B., Stamp, G., Baron, A., S07 Scarpa, A., Lemoine, N.R., 2002. Expression profiling ofmicrodissected pancreatic S08 adenocarcinomas. Oncogene 21,4587-4594.

S09 S10 Del Fattore, A., Fomari, R., Van Wesenbeeck, L., de Freitas, F., Timmermans, J.P., Peruzzi, B., Sll Cappariello, A., Rucci, N., Spera, G., Helfiich, M.H., Van Hui, W., Migliaccio, S., Teti, A., S12 2008. A new heterozygous mutation (R714C) ofthe osteopetrosis gene, pleckstrin homolog S13 domain containing family M (with run domain) member 1 (PLEKHM1), impairs vesicular S14 acidification and increases TRACP secretion in osteoclasts. Joumal ofbone and mineral 515 research: the officialjoumal ofthe American Society for Bone and Mineral Research 23, 380- 516 391.

517 518 Diatchenko, L., Lau, Y.F., Campbell, A.P., Chenchik, A., Moqadam, F., Huang, B., Lukyanov, 519 S., Lukyanov, K., Gurskaya, N., Sverdlov, E.D., Siebert, P.D., 1996. Suppression subtractive 520 hybridization: a method for generating differentially regulated or tissue-specific cDNA probes 521 and libraries. Proceedings ofthe National Academy ofSciences ofthe United States of America 522 93, 6025-6030.

523 524 Fresno, C., Llera, A.S., Girotti, M.R., Valacco, M.P., Lopez, J.A., Podhajcer, O.L., Balzarini, 525 M.G., Prada, F., Femandez, E.A., 2012. The multi-reference contrast method: facilitating set 526 enrichment analysis. Comput Biol Med 42, 188-194.

19 527 528 Fukuda, A., Wang, S.C., Morris, J.P.t., Folias, A.E., Liou, A., Kim, G.E., Akira, S., Boucher, 529 K.M., Firpo, M.A., Mulvihill, S.J., Hebrok, M., 2011. Stat3 and MMP7 contribute to pancreatic 530 ductal adenocarcinoma initiation and progression. Cancer Cell19, 441-455.

531 532 Gromov, P., Gromova, I., Friis, E., Timmermans-Wielenga, V., Rank, F., Simon, R., Sauter, G., 533 Moreira, J.M., 2010. Proteomic profiling ofmammary carcinomas identifies C7orf24, a gamma- 534 glutamyl cyclotransferase, as a potential cancer biomarker. J Proteome Res 9, 3941-3953.

535 536 Hidalgo, M., 2010. Pancreatic cancer. N Engl J Med 362, 1605-1617.

537 538 Huang da, W., Sherman, B. T., Lempicki, R.A., 2009. Systematic and integrative analysis of 539 large gene lists using DAVID bioinformatics resources. Nat Protoc 4, 44-57.

540 541 Iacobuzio-Donahue, C.A., Ashfaq, R., Maitra, A., Adsay, N.V., Shen-Ong, G.L., Berg, K., 542 Hollingsworth, M.A., Cameron, J.L., Yeo, C.J., Kem, S.E., Goggins, M., Hruban, R.H., 2003a. 543 Highly expressed genes in pancreatic ductal adenocarcinomas: a comprehensive 544 characterization and comparison ofthe transcription profiles obtained from three major 545 technologies. Cancer Res 63, 8614-8622.

546 547 Iacobuzio-Donahue, C.A., Maitra, A., Olsen, M., Lowe, A.W., van Heek, N.T., Rosty, C., 548 Walter, K., Sato, N., Parker, A., Ashfaq, R., Jaffee, E., Ryu, B., Jones, J., Eshleman, J.R., Yeo, 549 C.J., Cameron, J.L., Kem, S.E., Hruban, R.H., Brown, P.O., Goggins, M., 2003b. Exploration of 550 global gene expression pattems in pancreatic adenocarcinoma using cDNA microarrays. Am J 551 Pathol162, 1151-1162.

552 553 Iacobuzio-Donahue, C.A., Maitra, A., Shen-Ong, G.L., van Heek, T., Ashfaq, R., Meyer, R., 554 Walter, K., Berg, K., Hollingsworth, M.A., Cameron, J.L., Yeo, C.J., Kem, S.E., Goggins, M., 555 Hruban, R.H., 2002a. Discovery of novel tumor markers of pancreatic cancer using global gene 556 expression technology. Am J Pathol160, 1239-1249.

557 558 Iacobuzio-Donahue, C.A., Ryu, B., Hruban, R.H., Kem, S.E., 2002b. Exploring the host 559 desmoplastic response to pancreatic carcinoma: gene expression of stromal and neoplastic cells 560 at the site ofprimary invasion. Am J Pathol160, 91-99.

561 562 Jones, S., Zhang, X., Parsons, D.W., Lin, J.C., Leary, R.J., Angenendt, P., Mankoo, P., Carter, 563 H., Kamiyama, H., Jimeno, A., Hong, S.M., Fu, B., Lin, M. T., Calhoun, E.S., Kamiyama, M., 564 Walter, K., Nikolskaya, T., Nikolsky, Y., Hartigan, J., Smith, D.R., Hidalgo, M., Leach, S.D., 565 Klein, A.P., Jaffee, E.M., Goggins, M., Maitra, A., Iacobuzio-Donahue, C., Eshleman, J.R., 566 Kem, S.E., Hruban, R.H., Karchin, R., Papadopoulos, N., Parmigiani, G., Vogelstein, B., 567 Velculescu, V.E., Kinzler, K.W., 2008. Core signaling pathways in human pancreatic cancers 568 revealed by global genomic analyses. Science 321, 1801-1806.

569 570 Kanehisa, M., Goto, S., Sato, Y., Furumichi, M., Tanabe, M., 2012. KEGG for integration and 571 interpretation oflarge-scale molecular data sets. Nucleic acids research 40, Dl09-114.

572

20 573 Kuhlmann, K.F., van Till, J.W., Boermeester, M.A., de Reuver, P.R., Tzvetanova, I.D., 574 Offerhaus, G.J., Ten Kate, F.J., Busch, O.R., van Gulik, T.M., Gouma, D.J., Crawford, H.C., 575 2007. Evaluation ofmatrix metalloproteinase 7 in plasma and pancreatic juice as a biomarker 576 for pancreatic cancer. Cancer Epidemiol Biomarkers Prev 16, 886-891.

577 578 Liu, B.H., Goh, C.H., Ooi, L.L., Hui, K.M., 2008. Identification ofunique and common low 579 abundance tumour-specific transcripts by suppression subtractive hybridization and 580 oligonucleotide probe array analysis. Oncogene 27, 4128-4136.

581 582 Liu, Y., Zhu, X., Zhu, J., Liao, S., Tang, Q., Liu, K., Guan, X., Zhang, J., Feng, Z., 2007. 583 ldentification of differential expression of genes in hepatocellular carcinoma by suppression 584 subtractive hybridization combined cDNA microarray. Oncol Rep 18, 943-951.

585 586 Lowe, A. W., Olsen, M., Hao, Y., Lee, S.P., Taek Lee, K., Chen, X., van de Rijn, M., Brown, 587 P.O., 2007. Gene expression pattems in pancreatic tumors, cells and tissues. PLoS ONE 2, 588 e323.

589 590 Lyratzopoulos, G., Neal, R.D., Barbiere, J.M., Rubin, G.P., Abel, G.A., 2012. Variation in 591 number of general practitioner consultations before hospital referral for cancer: fmdings from 592 the 2010 National Cancer Patient Experience Survey in England. Lancet Oncoll3, 353-365.

593 594 Matsumoto, K., Miki, R., Nakayama, M., Tatsumi, N., Yokouchi, Y., 2008. Wnt9a secreted 595 from the walls ofhepatic sinusoids is essential for morphogenesis, proliferation, and glycogen 596 accumulation of chick hepatic epithelium. Dev Biol319, 234-247.

597 598 McCleary-Wheeler, A.L., McWilliams, R., Femandez-Zapico, M.E., 2012. Aberrant signaling 599 pathways in pancreatic cancer: a two compartment view. Mol Carcinog 51, 25-39.

600 601 Moreira, J.M., Cabezon, T., Gromova, 1., Gromov, P., Timmermans-Wielenga, V., Machado, 1., 602 Llombart-Bosch, A., Kroman, N., Rank, F., Celis, J.E., 2010. Tissue proteomics ofthe human 603 mammary gland: towards an abridged definition ofthe molecular phenotypes underlying 604 epithelial normalcy. Mol Oncol4, 539-561.

605 606 Neesse, A., Michl, P., Frese, K.K., Feig, C., Cook, N., Jacobetz, M.A., Lolkema, M.P., 607 Buchholz, M., Olive, K.P., Gress, T.M., Tuveson, D.A., 2010. Stromal biology and therapy in 608 pancreatic cancer. Gut.

609 610 Pan, Y.S., Lee, Y.S., Lee, Y.L., Lee, W.C., Hsieh, S. Y., 2006. Differentially profiling the low- 611 expression transcriptomes ofhuman hepatoma using a novel SSH/microarray approach. BMC 612 Genomics 7, 131.

613 614 Qin, L.X., Beyer, R.P., Hudson, F.N., Linford, N .J., Morris, D.E., Kerr, K.F., 2006. Evaluation 615 ofmethods for oligonucleotide array data via quantitative real-time PCR. BMC Bioinformatics 616 7, 23.

617

21 618 Ritchie, M.E., Silver, J., Oshlack, A., Holmes, M., Diyagama, D., Holloway, A., Smyth, G.K., 619 2007. A comparison ofbackground correction methods for two-colour microarrays. 620 Bioinformatics 23, 2700-2707.

621 622 Saeed, A.I., Bhagabati, N.K., Braisted, J.C., Liang, W., Sharov, V., Howe, E.A., Li, J., 623 Thiagarajan, M., White, J.A., Quackenbush, J., 2006. TM4 microarray software suite. Methods 624 in enzymology 411, 134-193.

625 626 Smyth, G.K., Speed, T., 2003. Normalization of cDNA microarray data. Methods 31, 265-273.

627 628 Spater, D., Hill, T.P., O'Sullivan R, J., Gruber, M., Conner, D.A., Hartmann, C., 2006. Wnt9a 629 signaling is required for joint integrity and regulation oflhh during chondrogenesis. 630 Development 133, 3039-3049.

631 632 Tabata, K., Matsunaga, K., Sakane, A., Sasaki, T., Noda, T., Yoshirnori, T., 2010. Rubicon and 633 PLEKHM1 negatively regulate the endocytic/autophagic pathway via a novel Rab7-binding 634 domain. Molecular biology of the cell21, 4162-4172.

635 636 TCGA-Network, 2012. Comprehensive molecular characterization ofhuman colon and rectal 637 cancer. Nature 487, 330-337.

638 639 Van den Broeck, A., Vankelecom, H., Van Eijsden, R., Govaere, 0., Topal, B., 2012. Molecular 640 markers associated with outcome and metastasis in human pancreatic cancer. Journal of 641 experimental & clinical cancer research : CR 31, 68.

642 643 Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A., Speleman, 644 F., 2002. Accurate normalization ofreal-tirne quantitative RT-PCR data by geometric averaging 645 ofmultiple interna] control genes. Genome Biol3, RESEARCH0034.

646 647 Vincent, A., Herman, J., Schulick, R., Hruban, R.H., Goggins, M., 2011. Pancreatic cancer. 648 Lancet 378, 607-620.

649 650 Wang, T., Hopkins, D., Schmidt, C., Silva, S., Houghton, R., Takita, H., Repasky, E., Reed, 651 S.G., 2000. Identification of genes differentially over-expressed in lung squamous cell 652 carcinoma using combination of cDNA subtraction and microarray analysis. Oncogene 19, 653 1519-1528.

654 655 Yachida, S., Jones, S., Bozic, 1., Anta], T., Leary, R., Fu, B., Kamiyama, M., Hruban, R.H., 656 Eshleman, J.R., Nowak, M.A., Velculescu, V.E., Kinzler, K.W., Vogelstein, B., Iacobuzio- 657 Donahue, C.A., 2010. Distant metastasis occurs late during the genetic evolution ofpancreatic 658 cancer. Nature467, 1114-1117.

659 660 Yang, Y.H., Dudoit, S., Luu, P., Lin, D.M., Peng, V., Ngai, J., Speed, T.P., 2002. Normalization 661 for cDNAmicroarray data: a robust composite method addressing single and multiple slide 662 systematic variation. Nucleic Acids Res 30, el5.

22 663 664 Yu, M., Ting, D. T., Stott, S.L., Wittner, B.S., Ozsolak, F., Paul, S., Ciciliano, J.C., Smas, M.E., 665 Winokur, D., Gilman, A.J., Ulman, M.J., Xega, K., Contino, G., Alagesan, B., Brannigan, B.W., 666 Milos, P.M., Ryan, D.P., Sequist, L.V., Bardeesy, N., Ramaswamy, S., Toner, M., Maheswaran, 667 S., Haber, D.A., 2012. RNA sequencing ofpancreatic circulating tumour cells implicates WNT 668 signalling in metastasis. Nature.

669 670 Zhang, B., Nie, X., Xiao, B., Xiang, J., Shen, S., Gong, J., Zhou, M., Zhu, S., Zhou, J., Qian, J., 671 Lu, H., He, X., Li, X., Hu, G., Li, G., 2003. Identification oftissue-specific genes in 672 nasopharyngeal epithelial tissue and differentially expressed genes in nasopharyngeal carcinoma 673 by suppression subtractive hybridization and cDNA microarray. Genes Chromosomes Cancer 674 38, 80-90.

675

676

677

678

679

680

681

682

683

684

685

686

687

688

689

690

691

692

23 693 Figure l. Experimental procedures. Total RNA extracted from pancreatic 694 adenocarcinoma tissues and their matched non-neoplastic samples were processed in 695 parallel for the direct and indirect strategies. The direct strategy included cDNA 696 synthesis, aRNA amplification and microarray hybridisation to measure the native 697 transcriptome of the whole mass tumour. The indirect strategy included a PCR-based

698 suppressive subtractive hybridisation prior to aRNA amplification and microarray 699 hybridisation to identify highly tumour-specific differentially expressed genes of low 700 abundance. The data were obtained separately for each strategy and analysed to identify 701 differentially expressed genes in adenocarcinoma and non-neoplastic tissues. 702 Subsequent validation using tissue microarray was performed for candidate genes 703 identified using the indirect strategy. PDAC=pancreatic ductal adenocarcinoma.

704

705

706

707

708

709

710

711

712

713

714

715

716

717

24 718 Figure 2. Direct and indirect strategies identify differentially expressed genes with 719 minimal overlap.

720 A) Differentially expressed genes from both strategies discriminate between PC and 721 non-neoplastic tissues. Red represents up-regulation, and green represent down-

722 regulation. PDAC= pancreatic adenocarcinoma. B) Only 35 genes overlapped between

723 strategies. C) qRT-PCR validation ofup-regulated genes from both strategies measured

724 in the same 12 samples processed by microarray. Log2-fold changes were obtained as 725 the ratio in PC/non-neoplastic tissue. The red colour represents up-regulation in PC

726 samples. *=p value < 0.05 for qRT-PCR assay.

727

728

729

730

731

732

733

734

735

736

737

738

739

740

741

742

25 743 Figure 3. Comparison of gene ontology enrichment between strategies.

744 A) Cellular components. B) Biological processes. Differentially expressed genes were 745 analysed using the DAVID database. Significantly enriched GO terms for each strategy

746 are represented in a histogramas the -log10 ofp values. Representative GO terms with p 747 values < 0.05 are shown. The stippled line indicates the significance threshold.

748

749

750

751

752

753

754

755

756

757

758

759

760

761

762

763

764

765

26 766 Figure 4. Immunohistochemistry analysis of WNT9A and PLEKHMl expression 767 in PC samples.

768 A) WNT9A positive staining in ductal cells of non-neoplastic pancreas, showing a 769 supranuclear granular pattem. Islets of Langerhans and acinar cells were negative (red 770 arrow). B-C) Tumoural cells showed stronger WNT9A immunoreactivity than non- 771 neoplastic pancreatic cells (red arrows) Scale bars, 50 Jlm. D) Tissue microarray 772 analysis revealed significantly higher intensity scores for WNT9A expression in 773 adenocarcinoma tissues compared with non-neoplastic pancreatic tissue (Student's two- 774 tailed paired t test). The mean intensity scores for each group are indicated by stippled 775 lines. E) PLEKHMl expression was negative in ductal cells, the islets of Langerhans 776 and acinar cells. F-G) Tumoural cells showed stronger PLEKHMl immunoreactivity 777 than non-neoplastic pancreatic cells (red arrows). Scale bars, 50 Jlm. H) Tissue 778 microarray analysis revealed significantly higher intensity scores for PLEKHMl 779 expression in adenocarcinoma tissues compared with non-neoplastic pancreas (student's 780 two-tailed paired t test). The mean intensity scores for each group are indicated by the 781 stippled Iines.

782

783

784

785

786

787

788

789

790

791

792

27 793 Figure 5. PLEKHMl positive expression in tumoural epithelial cells.

794 A) Indirect three-color immunofluorescence analysis of PC tissues revealed that only 795 CK19 positive cells (AlexaFluor 568; red) were positive for PLEKHMI expression 796 (AlexaFluor 488, green); cells were counterstained with the nuclear stain DAPI (blue ).

797 Scale bars, 25 )liD. B) Tumoural cells in an apparent invasive process were positive for

798 PLEKHMI and MMP7 (red arrows). Scale bars, 50 )liD.

799

800

801

802

803

804

805

806

807

808

28 1re 1 k here to download high resolution image

Six PDAC tissues with corresponding non-neoplastic adjacent tissues 1 RNA extraction and integrity analysis

Direct strategy lndirect strategy

• SMART cONA synthesis • Long distance PCR • RSAI digestion 1 TESTER DRIVER • PDAC • Normal • Non-neoplastic pancreas - Adaptar ~ 1 ligation ...... ,_...... _ ....

• Subtractive hybridizations • Primary and secondary PCR • Subtraction efficiency \.

r • In vitro transcription - • Microarray hybridization • Fluorescence extraction l r • Differential expression between ' PDAC and non-neoplastic tissues • qRT-PCR validation • Gene ontology analysis • Tissue microarray confirmation for indirect strategy ~ ~ 2 aere to download high resolution image

A Direct strategy lndirect strategy

e D

FN1 * IGFBPS * BHLHE40 TOP2A * SPARC * COL4A1 CALU SBN02 CTSK PMEPA1

Log2 fold dlange o 1 2 Log2 fold changa Jre 3 k here to download high resolution image A CeUular component -lndirect Cell pro~ctloo ---­ c:::J Direct Synapse Plasma membmne part Ce1! junotion Adnerens junctlon NuClear chromosome CoUagen Endoplasmic reticulum

AC1m cy1oske!etoo .r-~ Golg1 apparatus.t==!::: Basement membrane +==P Cytoplasmic vesic!e..l::::!=:: Extracelluiar ma1rlx ..t:=!t=====~ Protemaceous extrace!lular ma1ríx ,.t::=!~======:J Exttacellular region lntracel!.uJar · o 1 2 3 4 s 6 1 a 9 -log10 P value B Biological process -lndirect .• Organ morphogenes1s t-. c::JDirect ¡---, 1 Rcspooso lo calcium 10n t Stem cell differenhallon l \lnt receptor SJgnahng pathway through t:>eta-c.atenín lnlracellu!ar signa!ing c.asc.ade . ! Gly'CO¡>ro!etn metaoohc process : Rogulation of cata~íc nc~1víty . lnfl..'lmtn.3tory respMse -• j Regulaban of cell mohon . 1 Coflagen metabolic pmcess - ·~ Rogu!ation of celi m!gr«li!On ~ ' Slood circulatK>n ~ . Programmed ceU death 1 Celí adhes10n i Response lo wound.ing -1 1 Edr&ceflu!ar strudUf(l Ofganizat)(m • l Respoose to stress ,.... . Regul(ttíon of cell commünication ~ . O 12 3 4 S 6' -togu1 P value 4 1ere to download high resolution image

Non-neoplastlc Adenocarclnoma

_ . .,. .·· ~- ~ WNT9A A B e . ~ . . o 2.0 • f .. .,. ~ (.' ! • 1-! .. ~ .: • • .1 o .. . "'· . u Oooooo ',\., •. . , ;u 00 ~· 1 ... a., 00 < .. J,'.r , . ___o QQQ---00 (1) 1 :> .. ~'...r; 1 DoooO 1- 1 .. ~~~ .;, ~1.0 z : ( ., . • oD ', ~ o o .. ...\ .. 1,... t . :S 0 3: .. ,y ••. l.'''l .1:1 00 . .., :0.5 ·+ ji ... ,~ ...... ,. , . ~. :,. ' ..·~ ., ü ~ .... ~ ., ,. ,. ~~ ••M.0019 ·' ;. 1 0.0 :·.·· . . ·...... ·- - .... ~ ...... ·e:_·.· H 2.o PLEKHM1 ! o u .. ;u - co '"" ·. ! o o :e S 00 r ...·...... S 1 O lll: !!'' e IU :E ---~ge--- .J .1:1 a. .: 0.5 ooo .. o88~oo :1- ~ o " ; n•30 •P.O.Ol~ .. · O.OJ..:.:....;;.;:.-.,-----:....,.;.::.:.::..:...... -~ Hon.,..,plutic Ad.nocarclnom.a 5 ere to download high resolution image

B PLEKHM1 MMP7

~- ...... {. • ..... ""' ..... ~. (· .. · ~ ns , .;# .. r ·. -~."' -==-~ " E .. : .. ~ .... ~ o e: ·~ ns u ·. o /" . ~. e: J ,Cl) ·'· e:( '. ~· ... ~.J.~>·· i, . .... - ., . Supplementary data

SUPPLEMENTARY DATA

Supplementary figure Sl. Subtraction efficiency assayed by qPCR. The abundance of GAPDH transcripts was measured in non-subtracted and subtracted samples. A delay ofmore than five cycles in the amplification ofthe subtracted sample with respect to the non-subtracted was regarded as efficient. N=Non-neoplastic; T=Tumour.

< ' ¡ ovl 1 : 4T •• -.:-n.::::::::: 5T ···-=·- ¡6T .:::::::::::::::-.:::-.:: 1l·:Ac;yelo.-10.0f / 1l-IAc:ye._.l.~'· / 1JiAc:yeiM•I.I .··7 B< ,· ¡ B<~: ; ; ~- / ; :¡¡ ¡ ; ! al 1 ¡ 1 ~ ' 1 e;-· í / 5 ~1 i / ~- ! 1 ¡f : if. r------/- -}------r·----.+.'-: ~------

>, • • • ,' -·,-, ,-, •• 'a'• •, • •; • • o o" , , • o, o t~~~~__;~~~_;:;:.;:~~~, 0 0 0 0 ~:-:::·: 0 0 o.·~.:'. •• • • • • • • • • • • ~ • Cydes Cycles Cycles ..... NoSSH ..... SSH

1 Supplementary figure S2. PLEKHMl expression in normal tissues. Negative staining was observed in normal pancreatic tissues. Positive cells were found in the thymus, prostate, skin, uterus (cervix) and tonsil (red arrows).

Pancreas

~:: '·. '¡ ·~

Thymus Prostate

·. ~ ·¡

.~ -·-----_ .... ~:¿; S k in Tonsil Cervix uterus

2 Supplementary table Sl. qRT-PCR primer sequences.

Official Amplicon Sense Sequence 5'-->3' Efficiency symbol size OARS Forward ACCTGAACCTGGCATCACTACA 100 bp 101% Reverse CCAAGACGCTCAAACTGGAAC TBP Forward ACCAGGTGATGCCCTTCTGTAA 180 bp 99% Reverse CCTCAAACCAACTTGTCAACAG CALU Forward CAGCAACTGAACCTGCCATT 66 bp 93.50% Reverse TTGGGCCAAGCTTTCCTAGA BHLHE40 Forward CACGATCAGCAATCAGGCATA 150 bo 102% Reverse CAACGGCATATGGAGTGTCCTT TOP2A Forward TTCAGCTCTTGACCTGTCCC 108 bp 93.50% Reverse CAAATGTTGTCCCCGAGTCT COL4Al Forward CGTAAGCACATTCGGGCCATTT 108 bp 102% Reverse TCAGGCCTAGTGGTCCGAATCT CTSK Forward TTTCCCTGACAGCTGTGTACTC 109 bo 99% Reverse TGTGAAAATCTCCAGCCTGTAC SPARC Forward ACTTTTGGGAGCACGGACTGT 143 bo 99% Reverse TTTTGGCCTTCCTGGCTGAAAC FNl Forward AAGGCTTGAACCAACCTACGG 128 bp 91.50% Reverse AAAGCCTAAGCACTGGCACAA IGFBPS Forward TGTGTACCTGCCCAATTGTGAC 116 bp 97% Reverse GCAGCTTCATCCCGTACTTGT PMEPAl Forward TGCGTAGGTGAAAAGGCAGAA 149 bo 91.50% Reverse AGCTTGTGCATTCAGACCAGAC SBN02 Forward ACCCTTGAAATCCGTGAAACCG 170 bo 86.30% Reverse AGAGGTGAGCGGGCAATAACT CDSS Forward GCTTTGGAAGGCCGTACAAGTT 196 bp 99.50% Reverse AAGGCTGTTTGAGGGATGCAG OASL Forward AGCAGGTGCTCCTTAGCCAAAT 146 bp 88.50% Reverse AGGATGAAGCTGTTGGGGTTG KNGl Forward ATTGACTTCGTGGCCAGGGAAA 168 bo 97.3 Reverse TCCCAGTGGTTGACAGTTGACA SYT12 Forward TTCCCCATCGCAAATGCAGTTC 191 bp 77% Reverse TTGGACCTCTGGTTCTTGCCAT MMP7 Forward TGGGACATTCCTCTGATCCT 165 bp 88% Reverse TGAATGGATGTTCTGCCTGA LAMA3 Forward TGCAGATGCAAGCCCAGAATC 109 bo 99.40% Reverse GCTCAGTCCCATCTCTGTTGCA SYNE2 Forward CTGATGCTCGCTGGAAAGAGTT 171 bp 96.80% Reverse GCCGCAGTGCTGTCTCTAAA PLEKHM Forward AAGTGACCTGGTCCTAAGCTGT 105 bp 94.50% Reverse GGCAAGTTCAGTGATGCTTTGG DPYSL4 Forward AGCCCTCATCCAGGGAAGTTTT 175 bo 90% Reverse TTCACATCAGAGGCAGGATGA STATl Forward TCCGTGGCACTGCATACAATCT 184 bp 93.70% Reverse TGCCGAATTCCCAAAGGGCAA

3 Supplementary table S2. Signal subtraction of 100 housekeeping genes. N= Non- neoplastic tissue; T=Tumoural tissue.

Signal 1T 1N 2T 2N 3T 3N 4T 4N ST SN 6T 6N Mean subtraction >20% 42 39 37 53 54 60 64 59 58 42 49 42 50% >50% 28 22 25 44 49 39 49 44 42 32 39 32 37% >80% 10 5 9 14 16 21 26 19 13 18 16 6 14%

Housekeeping gene list

ACTB HMBS SDHA ZWlO COX7A2L RABIA YARS ADA HPRTl SUil ALDOA BTF3 UBE2M UBE2I AGPATl HSPCB TAF2 LDHA ATP5J2 GNAS PABPCl ALG9 LRPl TBP RPS27A SFRS9 PTBPl COX4Il ATP5B MCRSl TEGT RPL19 CSTB RPL36AL SPAG7 B2M NACA TFCP2 RPLll TETRAN C2lorf33 PSMD8 CAPN2 PGKl TPTl NONO HADHA GPI ZFP36Ll CETN2 PMMl TUBA6 RPS18 VEGFB COX7C ODCl CLTC POLR2L TXNRDl ILF2 STK24 FAU RPL18 CYCl PSMB6 UBB USPll EFNA3 GRIK5 RPSll FLOT2 QARS UBC UBEl ARHGAPl COX5B GAPDH RPL13A UBE2D2 CTNNBl TAPBP COX5A GPNMB RPL29 UBR2 EIF3S7 BATl VAMP3 GUSB RPS13 YWHAZ NDUFAl TKT GPAAl HDAClO RPS23 ZNF410 SAFB HLA-C HSBPl

4 Supplementary table 53

Direct strategy Gene ID Accession Symbol Average Fold Change p-value 1277 NM 000088 COL1A1 1.390 0.001 6374 NM 002994 CXCL5 1.159 0.004 3556 NM 002182 IL1RAP 1.153 0.000 1009 NM 001797.2 CDH11 1.115 0.000 10205 NM 005797.2 EVA1 1.074 0.000 115908 NM 138455.2 CTHRC1 1.066 0.000 7474 NM 003392.3 WNT5A 1.057 0.000 9859 - KIAA0470 1.018 0.000 283209 NM 173582.3 PGM2L1 1.014 0.000 2335 NM 002026.2 FN1 1.005 0.000 8942 - KYNU 1.004 0.000 3576 NM 000584.2 IL8 0.962 0.003 2274 NM 201557.1 FHL2 0.957 0.000 5918 NM 206963.1 RARRE$1 0.952 0.000 6590 NM 003064.2 SLPI 0.948 0.000 51715 NM 183227.1 RAB23 0.944 0.002 142913 - CFL1P1 0.932 0.000 3488 NM 000599 IGFBP5 0.929 0.001 26509 NM 133337.1 FER1L3 0.914 0.000 64063 NM 022119.3 PRS$22 0.908 0.000 8553 NM 003670 BHLHB2 0.903 0.000 4680 - CEACAM6 0.901 0.000 6513 NM 006516.1 SLC2A1 0.899 0.000 10631 NM 006475.1 POSTN 0.890 0.005 135228 NM 133493.1 CD109 0.887 0.003 2706 NM 004004.3 GJB2 0.878 0.000 2595 NM 198141 GANC 0.871 0.008 1293 NM 057167.2 COL6A3 0.867 0.003 2335 NM 002026.2 FN1 0.866 0.008 84419 NM 032413.2 NMES1 0.866 0.004 388272 NM 001001436.1 C16orf87 0.858 0.001 4070 NM 002353.1 TACSTD2 0.853 0.001 1728 NM 000903.1 NQ01 0.847 0.000 2568 NM 014211.1 GABRP 0.847 0.001 63898 NM 022071.2 SH2D4A 0.846 0.004 55454 NM 018590 GALNACT-2 0.844 0.001 8190 NM 006533.1 M lA 0.838 0.006 11082 NM 007036.2 ESM1 0.823 0.004 4085 NM 002358 MAD2L1 0.823 0.000 5743 NM 000963 PTGS2 0.813 0.000 55858 NM 018475.2 TPARL 0.809 0.005 3433 NM 001547.3 IFIT2 0.806 0.000 7321 NM 003338.3 UBE2D1 0.805 0.000 54209 NM 018965.1 TREM2 0.804 0.001 7153 NM 001067.2 TOP2A 0.798 0.000 6678 NM 003118.2 SPARC 0.793 0.000 221035 - REEP3 0.790 0.007 5918 NM 206963.1 RARRES1 0.790 0.002 139201 AK056535 MAP2K4P1 0.790 0.001 114884 NM 017784.3 OSBPL10 0.788 0.000 83540 NM 145697.1 CDCA1 0.781 0.001 8898 NM 201281.1 MTMR2 0.771 0.001 129642 NM 138799.2 OACT2 0.767 0.000 10103 NM 005727.2 TSPAN1 0.764 0.004 1282 NM 001845 COL4A1 0.764 0.001 4312 NM 002421.2 MMP1 0.762 0.005 51208 NM 016369.3 CLDN18 0.761 0.005 163223 NM 001001411 ZNF676 0.760 0.009 665 NM 004331.2 BNIP3L 0.754 0.002 122786 NM 152330.2 C14orf31 0.754 0.007 2123 NM 001003927.1 EVI2A 0.753 0.009 3486 NM 001013398.1 IGFBP3 0.750 0.002 64778 NM 022763.2 FNDC3B 0.749 0.000 10973 NM 006828.1 ASCC3 0.747 0.000 7514 NM 003400.3 XP01 0.744 0.000 7045 NM 000358.1 TGFBI 0.744 0.003 441258 XM 499330 LOC441258 0.740 0.006 7127 NM 006291.2 TNFAIP2 0.738 0.000 57162 NM 020651 PELI1 0.732 0.003 7058 NM 003247.2 THBS2 0.727 0.000 1084 NM 001815.1 CEACAM3 0.727 0.001 6423 NM 003013.2 SFRP2 0.726 0.005 155061 NM 152557.3 ZNF746 0.725 0.009 2149 NM 001992.2 F2R 0.724 0.000 81470 NM 001001915 OR2G2 0.723 0.000 9859 NM 014812 CEP170 0.722 0.003 3827 NM 000893.2 KNG1 0.722 0.010 128611 XM 066058.3 C20orf174 0.721 0.006 2591 NM 004482.2 GALNT3 0.720 0.004 3176 NM 006895.2 HNMT 0.720 0.002 611477 NM 024604.1 RPAP3 0.719 0.001 54443 NM 018685.2 ANLN 0.717 0.006 23643 NM 015364.2 LY96 0.716 0.004 149992 NM 153773.1 C21orf99 0.713 0.001 10159 NM 005765.2 ATP6AP2 0.712 0.001 11098 NM 007173.3 PRSS23 0.710 0.000 114907 NM 058229.2 FBX032 0.709 0.000 2182 NM 004458.1 ACSL4 0.709 0.004 64581 NM 197949.1 CLEC7A 0.708 0.004 7052 NM 004613 TGM2 0.708 0.001 6781 NM 003155 STCl 0.708 0.000 55709 - KBTBD4 0.706 0.004 56122 NM 018934.2 PCDHB14 0.706 0.009 360023 NM 194314.2 ZBTB41 0.703 0.006 2263 NM 022971.1 FGFR2 0.702 0.005 5159 NM 002609.3 PDGFRB 0.700 0.001 30001 NM 014584.1 EROlL 0.699 0.001 84196 NM 032236 USP48 0.698 0.007 51334 NM 016644.1 PRR16 0.696 0.000 4982 NM 002546.2 TNFRSFllB 0.695 0.001 122809 NM 080867.2 SOCS4 0.695 0.001 51339 NM 016651.4 DACTl 0.694 0.002 5738 NM 020440.2 PTGFRN 0.693 0.003 23636 NM 012346.3 NUP62 0.691 0.002 23552 NM 178432.1 CCRK 0.689 0.008 3775 - KCNK1 0.689 0.005 23414 NM 012082.2 ZFPM2 0.684 0.009 813 NM 001219 CALU 0.682 0.000 22904 - SBN02 0.680 0.002 10944 - SMAP 0.678 0.002 59 NM 001613.1 ACTA2 0.678 0.002 9683 NM 153029.3 N4BP1 0.678 0.000 3866 NM 002275.2 KRT15 0.677 0.002 285 NM 001147.1 ANGPT2 0.676 0.003 55612 NM 017671 C20orf42 0.675 0.007 1462 NM 004385.2 VCAN 0.675 0.000 80267 NM 025191.2 C1orf22 0.675 0.002 4320 NM 005940.3 MMPll 0.671 0.000 8935 NM 003930.3 SCAP2 0.668 0.002 147920 NM 001002915 IGFL2 0.668 0.007 85441 NM 033405 PRIC285 0.663 0.001 3399 NM 002167.2 103 0.663 0.006 170463 NM 032627.2 SSBP4 0.660 0.007 196 NM 001621 AHR 0.660 0.002 9255 NM 004757.2 SCYE1 0.656 0.002 5885 NM 006265 RAD21 0.653 0.000 1508 NM 147780.2 CTSB 0.652 0.002 375790 NM 198576.2 AGRN 0.651 0.000 255374 NM 203397.1 MBLAC1 0.650 0.006 146183 NM 144672.2 OTOA 0.650 0.010 84935 NM 032849.2 C13orf33 0.647 0.002 29940 NM 013352.1 SART2 0.645 0.006 3556 NM 002182.2 IL1RAP 0.644 0.003 1655 NM 004396.2 DDXS 0.644 0.002 90316 NM 138960 TGIF2LX 0.644 0.003 6947 NM 001062.2 TCN1 0.641 0.000 55959 NM 198596.1 SULF2 0.641 0.001 3091 NM 181054.1 HIF1A 0.640 0.010 56034 NM 016205.1 PDGFC 0.639 0.001 10321 NM 006061.1 CRISP3 0.639 0.003 2623 NM 002049.2 GATA1 0.638 0.002 26165 - C9orf36 0.638 0.000 6286 - S100P 0.638 0.002 26585 NM 013372 GREM1 0.637 0.003 1296 NM 005202.1 COL8A2 0.636 0.000 22943 NM 012242.2 DKK1 0.636 0.000 8491 NM 003618.2 MAP4K3 0.634 0.004 441707 XM 497432 LOC441707 0.633 0.005 9168 NM 021103 TMSB10 0.632 0.008 9749 NM 014721.1 PHACTR2 0.631 0.001 8836 - GGH 0.631 0.001 6045 NM 007212.3 RNF2 0.629 0.000 7045 NM 000358.1 TGFBI 0.626 0.005 79690 NM 024637.3 GAL3ST4 0.625 0.001 3934 NM 005564.2 LCN2 0.624 0.001 5277 NM 020473.1 PIGA 0.624 0.000 10487 NM 006367 CAP1 0.621 0.000 11177 NM 182648.1 BAZ1A 0.619 0.005 8767 NM 003821.4 RIPK2 0.618 0.000 79794 NM 024738.1 C12orf49 0.618 0.009 3976 NM 002309.2 LIF 0.618 0.005 84168 NM 032208.1 ANTXR1 0.617 0.001 6507 NM 004172 SLC1A3 0.617 0.003 150094 NM 173354.2 SNF1LK 0.616 0.004 2152 NM 001993.2 F3 0.614 0.001 84790 NM 032704.2 TUBA6 0.612 0.000 54799 NM 017643.1 MBTD1 0.611 0.000 23362 - PSD3 0.610 0.002 400221 XM 379075 FU22447 0.608 0.003 90390 NM 080651.1 THRAP6 0.607 0.001 306 NM 005139.1 ANXA3 0.605 0.000 402247 XM 377928 LOC402247 0.605 0.001 699 NM 004336.2 BUB1 0.604 0.003 55614 NM 024704.3 C20orf23 0.602 0.006 10561 NM 006417.2 IFI44 0.601 0.000 25903 NM 015441.1 OLFML2B 0.601 0.007 1513 NM 000396.2 CTSK 0.601 0.004 11167 NM 007085.3 FSTL1 0.599 0.009 5530 NM 000944.2 PPP3CA 0.599 0.005 153830 NM 144726.1 RNF145 0.598 0.001 2218 NM 006731.1 FCMD 0.598 0.004 23213 NM 015170.1 SULF1 0.597 0.003 822 NM 001747.2 CAPG 0.597 0.005 57234 AB007962 FAM91A2 0.594 0.002 1475 NM 005213.3 CSTA 0.594 0.004 6888 NM 006755.1 TALD01 0:594 0.001 3956 NM 002305.2 LGALS1 0.593 0.001 285927 BC044242 CCZ1 0.593 0.003 55858 NM 018475.2 TPARL 0.593 0.001 1514 NM 145918.1 CTSL 0.592 0.003 6443 NM 000232 SGCB 0.591 0.004 25932 NM 013943.1 CLIC4 0.590 0.001 23291 NM 033644.2 FBXW11 0.589 0.006 57045 - TWSG1 0.588 0.001 441273 XM 496912 SPDYE2 0.588 0.005 387805 XM 370649 LOC387805 0.587 0.000 136 NM 000676.2 ADORA2B 0.587 0.004 81606 NM 030915.1 LBH 0.587 0.002 57561 NM 020801.1 ARRDC3 0.586 0.009 50486 NM 015714.2 GOS2 0.586 0.007 28962 NM 014028.3 OSTM1 0.582 0.003 9344 NM 018348.4 TAOK2 0.582 0.001 1087 NM 006890.1 CEACAM7 0.580 0.008 2825 - GPR1 0.579 0.001 1263 NM 004073.2 PLK3 0.576 0.001 57480 XM 027307.3 PLEKHG1 0.576 0.001 51347 NM 016281.2 TAOK3 0.575 0.001 261729 NM 152999.2 STEAP2 0.573 0.001 284033 - SHISA6 0.573 0.004 4680 NM 002483 CEACAM6 0.572 0.010 692196 - SNORD76 0.570 0.002 27071 NM 014395.1 DAPP1 0.570 0.002 117285 NM 054112.1 DEFB118 0.569 0.001 9957 NM 005114.2 HS3ST1 0.569 0.001 11178 NM 021020 LZTS1 0.569 0.002 6659 NM 003107 SOX4 0.568 0.001 84641 NM 032558.1 HIATL1 0.568 0.002 387758 NM 203371.1 FIBIN 0.567 0.000 6336 NM 006514.1 SCN10A 0.567 0.001 23224 NM 015180.3 SYNE2 0.566 0.001 1462 - VCAN 0.565 0.001 100129460 XM 496860.1 DPY19L1P1 0.565 0.003 6382 NM 002997.4 SDC1 0.565 0.003 29127 NM 013277 RACGAP1 0.565 0.004 253981 AK025431 RELL1 0.564 0.009 6856 NM 006754.2 SYPL1 0.563 0.003 130940 NM 138803.2 CCDC148 0.563 0.004 10769 NM 006622.1 PLK2 0.562 0.001 60485 NM 021818.2 SAV1 0.561 0.006 1001 NM 001793.3 CDH3 0.559 0.001 283824 BX647541 XYLTl 0.559 0.004 1462 NM 004385.2 VCAN 0.558 0.008 3280 NM 005524.2 HES1 0.556 0.004 6591 NM 003068.3 SNAI2 0.555 0.001 55035 - NOL8 0.554 0.003 1718 NM 014762.2 DHCR24 0.551 0.003 6772 NM 007315 STATl 0.551 0.006 374 NM 001657.2 AREG 0.551 0.003 10146 NM 198395.1 G3BP 0.550 0.002 55030 NM 017943 FBX034 0.548 0.005 600246 - ELK4 0.547 0.004 29005 AF001542 MALATl 0.545 0.000 10487 NM 006367.2 CAP1 0.545 0.001 161742 NM 152594.1 SPRED1 0.544 0.001 10728 NM 006601 PTGE$3 0.544 0.007 57688 XM 035299.6 ZSWIM6 0.543 0.002 387103 NM 001012507 C6orf173 0.542 0.008 863 NM 175931.1 CBFA2T3 0.542 0.002 27242 NM 014452.3 TNFRSF21 0.542 0.001 390680 XM 372611 LOC390680 0.541 0.004 7148 - TNXB 0.541 0.001 4604 NM 206821.1 MYBPC1 0.539 0.010 57552 NM 020792.3 AADACL1 0.538 0.003 9052 NM 003979.3 GPRC5A 0.538 0.002 1894 NM 018098.4 ECT2 0.537 0.002 9689 NM 014670.2 BZW1 0.537 0.002 10092 NM 005717.2 ARPC5 0.536 0.003 308 - ANXA5 0.536 0.004 9140 NM 004707.2 APG12L 0.535 0.007 284397 XM 209180 LOC284397 0.535 0.009 7050 NM 173211.1 TGIF 0.534 0.001 7171 NM 003290.1 TPM4 0.534 0.001 2877 NM 002083.2 GPX2 0.532 0.003 11026 NM 006865.2 LILRA3 0.530 0.005 4148 NM 002381.3 MATN3 0.530 0.007 55281 NM 018295.1 TMEM140 0.530 0.001 1385 NM 004379.2 CREB1 0.530 0.001 84002 NM 032047 B3GNT5 0.530 0.007 57157 NM 020432.2 PHTF2 0.530 0.002 114884 NM 017784 OSBPL10 0.529 0.002 159 NM 001126.2 ADSS 0.527 0.005 197021 - KLPH 0.526 0.008 10926 NM 006716 ASK 0.526 0.002 79188 NM 024334.1 TMEM43 0.526 0.002 10617 NM 213622.1 STAMBP 0.526 0.003 205428 NM 173552.2 C3orf58 0.525 0.002 51495 NM 016395.1 HSPC121 0.524 0.006 118672 NM 153336.1 Cl0orf89 0.523 0.005 51056 NM 015907 LAP3 0.522 0.007 55714 XM 371717.2 ODZ3 0.521 0.002 2697 NM 000165.2 GJA1 0.521 0.003 55689 NM 018023.3 YEATS2 0.521 0.003 2737 NM 000168.2 GLI3 0.520 0.001 4811 - NID1 0.520 0.008 84920 NM 032834 ALG10 0.520 0.002 51174 NM 016261.2 TUBD1 0.519 0.002 151306 NM 170699.1 GPBAR1 0.519 0.010 8543 NM 006769.2 LM04 0.519 0.004 91404 NM 178123.3 SESTD1 0.518 0.002 10672 - GNA13 0.518 0.003 23508 XM 027236.5 TTC9 0.518 0.001 244 NM 001630 ANXA8 0.516 0.003 79919 - C2orf54 0.515 0.006 1604 NM 000574.2 DAF 0.515 0.002 157922 NM 015447 CAMSAP1 0.515 0.002 115207 - KCTD12 0.513 0.006 1999 NM 004433.3 ELF3 0.512 0.010 286410 NM 001010986.1 ATP11C 0.512 0.004 169436 NM 153710.2 C9orf96 0.512 0.009 9897 NM 014846.2 KIAA0196 0.512 0.004 51278 NM 016545.3 IER5 0.512 0.006 22856 NM 014918.3 CHSY1 0.511 0.002 665 NM 004331 BNIP3L 0.510 0.006 8460 NM 003596.2 TPSTl 0.510 0.004 9315 NM 004772.1 C5orf13 0.510 0.002 200008 NM 201546.2 CDCP2 0.509 0.006 9069 NM 012129.2 CLDN12 0.509 0.008 1075 NM 001814.2 CTSC 0.509 0.005 81611 - ANP32E 0.508 0.001 83468 NM 031302.2 GLT8D2 0.508 0.006 10049 NM 005494.2 DNAJB6 0.507 0.002 79444 NM 022161.2 BIRC7 0.507 0.005 286367 - LOC286367 FP944 0.507 0.005 54541 NM 019058.2 DDIT4 0.507 0.003 26504 NM 020184.2 CNNM4 0.506 0.009 254848 - AFAP1-AS1 0.506 0.003 7879 NM 004637.5 RAB7 0.506 0.004 8853 NM 003887.1 DDEF2 0.505 0.007 284441 XM 208204 LOC284441 0.505 0.001 23024 - PDZRN3 0.503 0.007 5996 NM 002922.3 RGS1 0.502 0.008 84674 NM 032587.2 CARD6 0.501 0.006 161742 - SPRED1 0.501 0.002 124220 - ZG16B 0.501 0.009 3832 NM 004523.2 KIF11 0.501 0.009 151011 NM 178584.1 10-Sep 0.501 0.008 355 NM 152873.1 FAS 0.501 0.002 1111 NM 001274.2 CHEK1 0.500 0.008 131566 NM 080927.3 DCBLD2 0.498 0.006 6566 - SLC16A1 0.497 0.008 50507 NM 016931.2 NOX4 0.497 0.004 23204 NM 015161.1 ARL61P 0.496 0.005 91607 NM 152270.2 SLFN11 0.496 0.008 7532 NM 012479.2 VWHAG 0.496 0.005 113146 XM 290629.5 AHNAK2 0.496 0.007 124583 - CANTl 0.496 0.004 25871 C3orf17 0.496 0.009 136319 NM 145808.1 MTPN 0.495 0.001 283417 NM 173812.3 DPV19L2 0.493 0.007 25923 - ATL3 0.493 0.009 5530 NM 000944.2 PPP3CA 0.492 0.006 6515 NM 006931 SLC2A3 0.492 0.004 303 NR 001562.1 ANXA2P1 0.492 0.003 25800 - SLC39A6 0.491 0.003 8915 NM 003921.2 BCL10 0.488 0.007 387 NM 001664.2 RHOA 0.487 0.003 9833 NM 014791.2 MELK 0.486 0.001 57038 NM 020320.2 RARSL 0.486 0.005 23612 - PHLDA3 0.486 0.001 10092 NM 005717 ARPC5 0.485 0.008 11031 NM 006868 RAB31 0.484 0.004 3460 NM 005534.2 IFNGR2 0.484 0.010 2017 NM 005231.2 CTTN 0.483 0.003 390790 XM 372668 ARL12 0.482 0.005 824 NM 001748 CAPN2 0.481 0.003 5230 NM 000291.2 PGK1 0.481 0.009 11236 NM 007218.3 RNF139 0.480 0.008 1075 NM 001814.2 CTSC 0.480 0.003 7932 - OR2H2 0.480 0.009 1495 NM 001903.2 CTNNA1 0.480 0.003 51192 NM 181641.1 CKLF 0.479 0.006 57523 XM 370756.2 NYNRIN 0.478 0.004 546 NM 000489.2 ATRX 0.477 0.006 84171 NM 032211.6 LOXL4 0.477 0.005 1605 NM 004393 DAG1 0.476 0.006 10097 NM 005722.2 ACTR2 0.476 0.001 11335 NM 016587.2 CBX3 0.476 0.004 10863 NM 014265.2 ADAM28 0.476 0.008 10933 NM 206839 MORF4L1 0.476 0.008 64770 NM 022757.3 CCDC14 0.475 0.008 389137 XM 371655 ARMC10P1 0.475 0.004 283126 AK094403 - 0.474 0.004 653464 XM 209227 SRGAP2P1 0.474 0.007 147339 NM 145055.2 C18orf25 0.474 0.005 91543 NM 080657.3 RSAD2 0.474 0.009 79139 NM 024295 DERL1 0.474 0.004 84945 NM 032859.2 C13orf6 0.473 0.003 6916 - TBXAS1 0.473 0.007 160492 NM 152590.1 IFLTD1 0.473 0.007 1308 NM 000494.2 COL17A1 0.473 0.004 54661 AF217751 PCDHB17 0.472 0.003 8323 NM 003506.2 FZD6 0.472 0.009 4069 NM 000239.1 LYZ 0.471 0.006 23075 NM 015055 SWAP70 0.471 0.003 25907 NM 015444.1 RIS1 0.471 0.003 4673 NM 004537.3 NAP1L1 0.470 0.005 10890 NM 016131.2 RAB10 0.470 0.008 283050 XM 378238.1 LOC283050 0.470 0.008 7171 NM 003290 TPM4 0.469 0.002 8099 NM 004642.2 CDK2AP1 0.469 0.008 80213 NM 078474.1 BLP2 0.469 0.005 10512 NM 006379.2 SEMA3C 0.469 0.005 22822 - PHLDA1 0.469 0.004 114908 NM 052932.1 PORIMIN 0.468 0.003 5782 NM 002835.2 PTPN12 0.467 0.003 10743 NM 030665.3 RAI1 0.467 0.007 2157 NM 000132.2 F8 0.467 0.008 4240 MFGE8 0.467 0.006 23512 NM 015355.1 SUZ12 0.467 0.006 51449 NM 016297.2 PCYOX1 0.466 0.001 444 NM 032468 ASPH 0.466 0.007 123970 NM 144602.2 C16orf78 0.465 0.008 7408 NM 003370.3 VASP 0.465 0.006 29982 NM 030759 NRBF2 0.463 0.007 29992 NM 013439.2 PILRA 0.462 0.004 3840 NM 002268.3 KPNA4 0.462 0.001 283911 AL833480 LOC283911 0.461 0.006 9315 NM 004772 C5orf13 0.461 0.004 5818 AK095375 PVRL1 0.459 0.006 340107 BC040315 LOC340107 0.459 0.004 1513 NM 000396.2 CTSK 0.458 0.002 26031 NM 015550.2 OSBPL3 0.458 0.005 2182 NM 004458.1 ACSL4 0.457 0.005 7424 NM 005429.2 VEGFC 0.457 0.006 754 NM 004339 PTIG11P 0.456 0.003 10529 NM 006393.1 NEBL 0.455 0.004 6240 - RRM1 0.454 0.008 151579 XM 045290 BZW1P2 0.454 0.004 597 NM 004049.2 BCL2A1 0.451 0.004 3099 NM 000189 HK2 0.451 0.006 10644 NM 006548.4 IMP-2 0.450 0.008 84168 NM 032208.1 ANTXR1 0.449 0.005 725 NM 001017367.1 C4BPB 0.449 0.002 3122 NM 019111.3 HLA-DRA 0.448 0.006 9780 NM 014745.1 FAM38A 0.448 0.002 10097 NM 005722.2 ACTR2 0.448 0.005 1258 - CNGB1 0.448 0.005 29107 NM 013248 NXTl 0.448 0.003 55233 NM 018221.1 MOBK1B 0.447 0.006 10054 NM 005499 UBA2 0.445 0.003 26005 - C2CD3 0.445 0.003 8604 NM 003705.2 SLC25A12 0.444 0.006 5493 - PPL 0.443 0.004 195828 - ZNF367 0.443 0.010 5901 - RAN 0.442 0.008 8731 NM 003799 RNMT 0.442 0.003 2983 - GUCY1B3 0.442 0.007 200879 NM 139248.2 LIPH 0.442 0.003 345778 NM 001010891.1 MTX3 0.441 0.008 3142 NM 021958.2 HLX1 0.441 0.004 128102 XM 371285.2 HSD3BP4 0.440 0.004 55610 NM 017667.2 FU20097 0.439 0.006 11217 - AKAP2 0.437 0.006 7321 NM 003338.3 UBE2D1 0.436 0.004 1087 NM 006890 CEACAM7 0.435 0.009 4683 NM 001024688.1 NBN 0.435 0.006 10753 - CAPN9 0.435 0.003 50809 NM 016287.2 HP1BP3 0.435 0.007 858 NM 198212.1 CAV2 0.432 0.009 441668 XM 497388 POTE M 0.432 0.003 6515 NM 006931.1 SLC2A3 0.432 0.010 144203 BN000357 OVOS2 0.432 0.008 80014 NM 024949 BOMB 0.432 0.006 728524 XM 499321 LOC442580 0.430 0.004 3992 NM 013402 FADS1 0.430 0.005 10426 NM 006322.3 TUBGCP3 0.430 0.007 26010 NM 015535.1 DNAPTP6 0.429 0.003 10123 NM 005737.3 ARL7 0.429 0.004 1612 NM 004938.1 DAPK1 0.428 0.008 8886 NM 006773.3 DDX18 0.428 0.002 55017 NM 017924.2 C14orf119 0.427 0.003 4052 NM 206943.1 LTBP1 0.427 0.004 51571 NM 016623.3 FAM49B 0.427 0.003 158427 NM 139246.3 C9orf97 0.426 0.005 6427 NM 003016.2 SFRS2 0.425 0.002 387745 XM 370603 LOC387745 0.424 0.007 80853 XM 376680.2 KIAA1718 0.424 0.008 84065 - TMEM222 0.424 0.003 768 NM 001216.1 CA9 0.424 0.009 991 NM 001255 CDC20 0.423 0.006 22803 NM 012255.3 XRN2 0.423 0.008 6596 NM 139048.1 SMARCA3 0.423 0.005 7083 - TK1 0.422 0.004 23317 NM 173823.1 DNAJC13 0.420 0.004 51729 NM 016312.2 WBPll 0.419 0.007 5879 NM 006908.3 RAC1 0.419 0.007 57522 NM 020762.1 SRGAP1 0.418 0.010 2591 NM 004482.2 GALNT3 0.417 0.007 983 NM 001786.2 CDC2 0.417 0.009 1371 NM 000097.4 CPOX 0.417 0.008 51088 NM 001007075.1 KLHL5 0.416 0.008 1660 NM 030588.1 DHX9 0.416 0.006 56937 NM 199171.1 PMEPAI 0.415 0.004 54534 NM 019051.1 MRPL50 0.413 0.004 10010 NM 004180.2 TANK 0.411 0.008 57047 NM 020359.1 PLSCR2 0.410 0.008 84168 NM 018153.2 ANTXR1 0.410 0.008 26499 NM 016445.1 PLEK2 0.410 0.004 440253 XM 496050 WHAMMP2 0.409 0.004 2036 NM 012156.2 EPB41L1 0.409 0.008 11260 NM 007235 XPOT 0.408 0.007 829 - CAPZA1 0.408 0.004 966 NM 000611.4 CD59 0.407 0.010 3839 NM 002267.2 KPNA3 0.407 0.008 8467 NM 003601.2 SMARCA5 0.406 0.008 90861 NM 144570.2 C16orf34 0.406 0.009 5573 NM 212471.1 PRKAR1A 0.404 0.005 64710 NM 022731 NUCKS 0.404 0.004 51095 NM 182916.1 TRNTl 0.402 0.009 23353 NM 025154.2 UNC84A 0.402 0.009 196074 NM 152636 METI5D1 0.402 0.009 25806 - VAX2 0.402 0.010 441731 XM 497465 LOC441731 0.402 0.008 3490 NM 001553.1 IGFBP7 0.401 0.003 113174 NM 138421.1 LOC113174 0.401 0.006 5577 NM 002736.2 PRKAR2B -0.401 0.010 1936 NM 032378.2 EEF1D -0.402 0.003 51166 NM 016228.3 AADAT -0.403 0.005 129080 NM 133455.2 EMID1 -0.403 0.007 402275 XM 377943 LOC402275 -0.405 0.007 81554 NM 148842.1 WBSCR16 -0.406 0.010 6649 NM 003102.1 SOD3 -0.408 0.006 292 NM 001152.1 SLC25A5 -0.409 0.008 64342 NM 022460.3 HS1BP3 -0.413 0.008 4118 NM 002371.2 MAL -0.413 0.009 115557 NM 182947.1 GEFT -0.413 0.003 442675 XM 499419 LOC442675 -0.414 0.008 54498 NM 019025.2 SMOX -0.416 0.009 2819 NM 005276 GPD1 -0.418 0.005 441464 XM 499160 LOC441464 -0.418 0.008 55361 NM 018425.2 PI4KII -0.419 0.007 6506 NM 004171 SLC1A2 -0.421 0.005 64219 NM 145119.1 PJA1 -0.421 0.008 10439 NM 014279.2 OLFM1 -0.422 0.010 5314 NM 138694.2 PKHD1 -0.422 0.007 27034 NM 014384.1 ACAD8 -0.424 0.004 5998 NM 144488.3 RGS3 -0.425 0.006 391481 XM 372970 LOC391481 -0.429 0.008 9576 NM 172242.1 SPAG6 -0.430 0.007 6584 - SLC22A5 -0.430 0.003 4622 NM 017533.1 MYH4 -0.431 0.009 7125 NM 003279.2 TNNC2 -0.431 0.006 441599 XM 497281 LOC441599 -0.432 0.009 7367 NM 001077 UGT2B17 -0.433 0.008 54715 NM 018723.2 A2BP1 -0.433 0.008 9686 - VGLL4 -0.433 0.003 8775 NM 003827.2 NAPA -0.434 0.004 389549 NM 001024613.1 LOC389549 -0.438 0.004 7273 NM 133378.2 TIN -0.438 0.009 5687 - PSMA6 -0.441 0.008 441449 XM 499150 LOC441449 -0.442 0.008 2494 NM 205860.1 NR5A2 -0.442 0.004 5777 NM 080548.2 PTPN6 -0.442 0.007 55592 NM 017600.1 DKFZp434M0331 -0.442 0.008 388444 XM 371097 LOC388444 -0.444 0.003 151354 - NSE1 -0.444 0.003 89886 NM 033438.1 SLAMF9 -0.446 0.008 58510 NM 021232.1 PRODH2 -0.447 0.003 5966 NM 002908.2 REL -0.447 0.003 308 - ANXA5 -0.447 0.006 54205 NM 018947 CYCS -0.450 0.006 5968 NM 006507 REG1B -0.450 0.006 1153 NM 001280 CIRBP -0.451 0.005 441809 XM 497573 LOC441809 -0.452 0.007 441997 XM 497818 LOC441997 -0.453 0.005 55520 NM 018696.2 ELAC1 -0.453 0.007 80144 NM 025074.3 FRAS1 -0.458 0.002 284612 NM 001006603.1 SYPL2 -0.463 0.004 5930 NM 006910.4 RBBP6 -0.465 0.009 80174 NM 145663.1 DRF1 -0.465 0.007 22807 NM 016260.1 ZNFN1A2 -0.466 0.003 23426 XM 290559.5 GRIP1 -0.466 0.005 11317 NM 014276.2 RBPSUHL -0.471 0.006 150000 NM 172024.1 ABCC13 -0.471 0.009 64577 NM 022568.2 ALDH8A1 -0.475 0.007 1523 NM 181552.1 CUTL1 -0.477 0.004 80196 NM 194271.1 RNF34 -0.477 0.007 54937 NM 017826.1 FU20449 -0.478 0.007 1196 NM 003993.2 CLK2 -0.479 0.010 403315 - - -0.480 0.002 2488 NM 001018080.1 FSHB -0.482 0.002 4135 NM 207577.1 MAP6 -0.484 0.008 64072 NM 022124.2 CDH23 -0.485 0.007 5354 NM 199478.1 PLP1 -0.485 0.007 1674 NM 001927.3 DES -0.490 0.004 5245 NM 002634 PHB -0.495 0.007 10230 - NBR2 -0.496 0.009 22947 NM 033178 DUX4 -0.497 0.009 64581 NM 197952.1 CLEC7A -0.498 0.005 57628 NM 001004360.1 DPP10 -0.498 0.005 284339 NM 173633.1 FU90805 -0.498 0.009 6786 - STIM1 -0.499 0.007 51092 NM 015996.1 SIDT2 -0.501 0.003 7881 NM 172159.2 KCNAB1 -0.501 0.005 25837 - RAB26 -0.502 0.007 3909 NM 198129.1 LAMA3 -0.502 0.005 55260 NM 018273.2 TMEM143 -0.504 0.003 22905 NM 014964.3 EPN2 -0.504 0.006 2053 NM 001979.4 EPHX2 -0.505 0.002 27020 NM 012428.1 SDFR1 -0.507 0.008 38 NM 000019.2 ACATl -0.507 0.004 196120 XM 114987 SSU72P5 -0.510 0.005 23336 NM 145728.1 DMN -0.511 0.009 23229 NM 015185.1 ARHGEF9 -0.511 0.005 85445 NM 033401.2 CNTNAP4 -0.513 0.003 63906 NM 022078.1 GPATC3 -0.513 0.006 51386 NM 016091.2 EIF3S61P -0.514 0.004 2742 NM 002063.2 GLRA2 -0.516 0.010 26063 NM 020664.3 DECR2 -0.518 0.005 170482 NM 203503.1 CLEC4C -0.519 0.002 253980 NM 178863.2 KCTD13 -0.520 0.006 80178 NM 025108.1 C16orf59 -0.521 0.008 1956 NM 005228.3 EGFR -0.521 0.002 81033 NM 030779.2 KCNH6 -0.521 0.010 317705 NM 173858 VN1R5 -0.525 0.004 57337 NM 020654.2 SENP7 -0.525 0.002 128344 NM 181643.2 C1orf88 -0.525 0.007 496 NM 000705.2 ATP4B -0.527 0.007 56673 NM 020643.1 Cllorf16 -0.528 0.007 56163 NM 031994.1 RNF17 -0.528 0.002 2903 NM 000833.2 GRIN2A -0.529 0.002 79656 NM 024603.1 BEND5 -0.529 0.005 441060 XM 498993 LOC441060 -0.529 0.001 151295 NM 144712.2 SLC23A3 -0.529 0.008 389599 XM 372002 STRADBP1 -0.535 0.006 2281 NM 004116.2 FKBP1B -0.535 0.009 112483 NM 133491.2 SAT2 -0.537 0.001 148638 - LOC148638 -0.538 0.003 55208 NM 018185.2 C13orf17 -0.539 0.003 57644 NM 020884.1 MYH7B -0.541 0.007 161380 BF109853 Cl4orf42 -0.543 0.008 285622 - NBPF22P -0.544 0.009 154872 XM 380140 LOC154872 -0.545 0.002 472 NM 000051.2 ATM -0.546 0.006 64130 NM 022165.2 LIN7B -0.547 0.002 7273 NM 133437.1 TIN -0.549 0.005 2346 NM 004476.1 FOLH1 -0.549 0.004 2984 NM 004963.1 GUCY2C -0.550 0.002 948 NM 000072.2 CD36 -0.553 0.003 10811 NM 006647.1 NOXA1 -0.554 0.007 2030 - SLC29A1 -0.558 0.004 56980 NM 199439.1 PRDM10 -0.559 0.006 390705 XM 372626 LOC390705 -0.560 0.001 9098 NM 004505 USP6 -0.561 0.008 2157 NM 000132.2 F8 -0.563 0.006 51011 NM 016044 FAHD2A -0.563 0.001 57678 NM 020918 GPAM -0.567 0.001 55821 - ALLC -0.571 0.001 5352 NM 182943.2 PLOD2 -0.571 0.002 3475 NM 001007245.1 IFRD1 -0.571 0.007 23495 NM 012452 TNFRSF13B -0.575 0.008 5037 NM 002567.2 PBP -0.577 0.001 4494 NM 005949.1 MTlF -0.578 0.001 284835 AK094492 LINC00323 -0.579 0.000 9148 - NEURL -0.579 0.001 64900 XM 372866 LPIN3 -0.581 0.010 56647 NM 078469.1 BCCIP -0.582 0.007 4968 NM 016819.2 OGG1 -0.583 0.008 388959 XM 373989 LOC388959 -0.587 0.006 24140 NM 177439.1 FTSJ1 -0.590 0.001 91948 XM 378549.2 LOC91948 -0.591 0.006 148362 NM 144695.1 Clorf58 -0.595 0.008 441339 XM 499103 LOC441339 -0.598 0.006 4329 NM 005589.2 ALDH6A1 -0.607 0.002 440464 XM 371081 LOC440464 -0.608 0.006 6425 NM 003015.2 SFRP5 -0.608 0.000 140801 NM 080746.2 RPL10L -0.608 0.009 8526 NM 003647.1 DGKE -0.609 0.004 284 NM 139290.1 ANGPTl -0.611 0.003 54885 NM 017752.2 TBC1D8B -0.611 0.009 119749 NM 001004703 OR4C46 -0.612 0.003 25924 NM 015460.2 MYRIP -0.612 0.010 340268 XM 374401 LOC340268 -0.612 0.005 144165 - PRICKLE1 -0.612 0.005 9154 NM 004213.3 SLC28A1 -0.613 0.006 54991 NM 017891.2 C1orf159 -0.614 0.004 9063 NM 173206.2 PIAS2 -0.617 0.005 65258 NM 023075.3 MPPE1 -0.617 0.001 1576 NM 017460.3 CYP3A4 -0.622 0.007 340178 BC044924 LOC340178 -0.623 0.007 147184 - TMEM99 -0.629 0.010 18 NM 020686.3 ABAT -0.631 0.001 5106 NM 004563.2 PCK2 -0.633 0.003 84315 NM 032355.2 MON1A -0.636 0.006 127540 NM 145205.3 HMGB4 -0.638 0.006 2625 NM 002051.2 GATA3 -0.639 0.005 4498 NM 175622 MTlJ -0.639 0.006 57556 NM 020796.2 SEMA6A -0.641 0.005 3083 NM 001528.2 HGFAC -0.646 0.006 100507203 XM 379432.1 LOC100507203 -0.648 0.010 399920 XM 378300 LOC399920 -0.652 0.001 10143 NM 005752.2 CLEC3A -0.655 0.009 1842 NM 001393.2 ECM2 -0.656 0.001 388591 NM 207396.1 RNF207 -0.657 0.001 1472 NM 001899 CST4 -0.659 0.006 202151 NM 145000.2 RANBP3L -0.661 0.001 9340 - GLP2R -0.663 0.004 2309 NM 001455 FOX03A -0.666 0.001 256691 NM 153267.3 MAMDC2 -0.669 0.007 83747 AK000652 C20orf57 -0.669 0.004 1770 NM 001372.2 DNAH9 -0.673 0.004 5210 NM 004567.2 PFKFB4 -0.677 0.003 11162 NM 198041.1 NUDT6 -0.677 0.001 57535 NM 020775.2 KIAA1324 -0.678 0.002 57057 NM 020417.1 TBX20 -0.679 0.002 5893 NM 134423.1 RAD52 -0.679 0.008 390667 NM 001013658 PTX4 -0.682 0.009 222052 XM 499284 TIC4P1 -0.683 0.003 10908 NM 006702.2 NTE -0.689 0.005 3651 NM 000209.1 IPFl -0.690 0.004 84107 NM 032153.3 ZIC4 -0.690 0.008 81472 NM 198074.3 OR2C3 -0.691 0.001 57718 NM 058237.1 PPP4R4 -0.695 0.002 79925 NM 024867.2 SPEF2 -0.706 0.005 84820 XM 499280.1 POLR2J4 -0.714 0.009 4496 NM 005951 MTlH -0.717 0.000 134548 NM 175873.3 ANKRD43 -0.718 0.004 51480 NM 016378.2 VCX2 -0.723 0.009 54798 NM 017639.2 DCHS2 -0.725 0.006 85282 NM 033059 KRTAP4-14 -0.734 0.000 5593 NM 006259.1 PRKG2 -0.736 0.006 2938 NM 145740 GSTA1 -0.742 0.004 478 NM 152296.3 ATP1A3 -0.747 0.004 56344 NM 019855.3 CABP5 -0.747 0.002 1558 NM 000770.2 CYP2C8 -0.750 0.003 53820 NM 018962.1 DSCR6 -0.752 0.008 93463 XM 376186.2 LOC93463 -0.759 0.002 245932 NM 153289.2 DEFB119 -0.762 0.003 10911 NM 006786.2 UTS2 -0.763 0.007 79366 NM 030763.1 NSBP1 -0.772 0.005 54897 NM 017766.2 CASZ1 -0.774 0.004 4653 NM 000261.1 MYOC -0.778 0.003 4607 NM 000256.2 MYBPC3 -0.779 0.000 7273 NM 133378.2 TIN -0.785 0.003 57282 - SLC4A10 -0.788 0.007 6563 NM 015865.1 SLC14A1 -0.796 0.010 54097 NM 058186.3 FAM3B -0.821 0.006 353288 NM 181539.3 KRT25B -0.856 0.008 50629 AF109299 PCANAP6 -0.862 0.004 116966 NM 181265.2 WDR17 -0.872 0.005 116444 NM 138690.1 GRIN3B -0.907 0.004 7681 NM 005664.1 MKRN3 -0.917 0.001 7364 NM 001074 UGT2B7 -1.086 0.003 186 NM 000686.3 AGTR2 -1.374 0.003 2939 NM 000846 GSTA2 -1.424 0.002 948 NM 000072.2 CD36 -1.451 0.004

lndirect strategy Gene ID Accession Symbol Average Fold Change p-value 8638 NM 003733.2 OASL 1.915 0.008 91683 NM 177963.2 SYT12 0.979 0.007 56114 NM 031993.1 PCDHGA1 0.895 0.029 2583 NM 001478.2 GALGT 0.857 0.013 81607 - PVRL4 0.851 0.002 7273 NM 133378.2 TIN 0.771 0.005 10874 NM 006681.1 NMU 0.742 0.009 4316 NM 002423.2 MMP7 0.725 0.007 10212 NM 138998.1 DDX39 0.725 0.010 201163 - FLCN 0.723 0.004 7483 NM 003395.1 WNT9A 0.721 0.010 3827 - KNG1 0.716 0.005 79925 NM 024867.2 SPEF2 0.714 0.020 84276 NM 032316.2 NICN1 0.686 0.041 9344 NM 004783.2 TAOK2 0.677 0.015 2266 NM 000509.3 FGG 0.673 0.010 2152 NM 001993.2 F3 0.669 0.013 1469 NM 001898.2 CSTl 0.666 0.013 5373 NM 000303.1 PMM2 0.663 0.011 54843 NM 032943.2 SYTL2 0.656 0.042 7706 NM 005082.4 TRIM25 0.656 0.014 2329 NM 002022.1 FM04 0.649 0.022 51227 NM 016430.2 DSCR5 0.648 0.002 23001 NM 014991.3 WDFY3 0.642 0.003 157680 - VPS13B 0.641 0.010 7140 NM 006757.1 TNNT3 0.633 0.008 7769 - ZNF226 0.614 0.008 1604 NM 000574.2 CD55 0.613 0.005 25893 NM 015431.2 TRI M 58 0.610 0.012 10066 - SCAMP2 0.607 0.003 7050 NM 170695.2 TGIF 0.604 0.011 80852 XM 042936.5 GRIP2 0.601 0.023 54726 NM 199324.1 HSHIN1 0.600 0.008 10781 NM 006631.2 ZNF266 0.597 0.012 8013 NM 173199.1 NR4A3 0.584 0.008 55764 NM 052985.1 WDR10 0.584 0.028 7273 NM 133378.2 TIN 0.581 0.004 1404 NM 001884.2 HAPLN1 0.580 0.023 535 NM 005177.2 ATP6VOA1 0.572 0.004 132 NM 006721.2 ADK 0.571 0.004 79864 NM 024806.2 Cllorf63 0.570 0.004 54880 NM 017745.4 BCOR 0.568 0.043 7634 NM 007136.1 ZNF80 0.566 0.009 2986 NM 001522.1 GUCY2F 0.566 0.005 51014 NM 181836.3 TMED7 0.561 0.050 136991 NM 130768.1 ASZ1 0.561 0.021 5790 - PTPRCAP 0.560 0.029 558 NM 001699.3 AXL 0.560 0.011 9842 NM 014798.1 PLEKHM1 0.558 0.002 64072 NM 022124.2 CDH23 0.553 0.036 22948 NM 012073.3 CCT5 0.553 0.005 55669 NM 033540.2 MFN1 0.552 0.016 94120 - SYTL3 0.550 0.037 29123 - ANKRDll 0.550 0.021 648987 - LOC648987 0.549 0.010 10068 NM 173044.1 IL18BP 0.546 0.025 4739 NM 006403.2 NEDD9 0.545 0.023 6915 NM 201636.1 TBXA2R 0.544 0.001 55074 - OXR1 0.544 0.047 2784 NM 002075.2 GNB3 0.541 0.019 57144 NM 020341.2 PAK7 0.539 0.036 80333 - KCNIP4 0.538 0.019 23224 NM 182914.1 SYNE2 0.537 0.042 116540 NM 053050.2 MRPL53 0.537 0.022 7273 NM 133378.2 TTN 0.536 0.008 255877 NM 181844.2 BCL6B 0.535 0.014 55063 NM 017984.2 ZCWPW1 0.532 0.012 79074 NM 024093.1 C2orf49 0.530 0.015 55218 - Cl4orf114 0.525 0.050 56853 - BRUNOL4 0.524 0.021 55183 NM 018151.3 RIF1 0.522 0.031 58493 NM 021218.1 C9orf80 0.521 0.040 1305 - COL13A1 0.520 0.001 55193 NM 018313.2 PB1 0.520 0.002 3164 NM 173158.1 NR4A1 0.519 0.013 56986 NM 020234.3 MDS009 0.519 0.050 3990 NM 000236.1 LIPC 0.517 0.035 10846 NM 006661.1 PDE10A 0.516 0.047 11282 NM 054013.1 MGAT4B 0.516 0.022 219771 NM 181698.1 C10orf9 0.514 0.014 150763 NM 207328.1 GPAT2 0.514 0.023 5888 NM 002875.2 RAD51 0.512 0.015 26287 NM 020349.2 ANKRD2 0.512 0.003 138724 NM 203299.1 C9orf131 0.512 0.009 168975 NM 173538.1 FU35802 0.511 0.017 11193 NM 007187.3 WBP4 0.511 0.035 59307 NM 021805.1 SIGIRR 0.510 0.020 80144 NM 025074.3 FRAS1 0.509 0.003 197350 XM 113871.4 LOC197350 0.507 0.029 203111 NM 173549.1 FU39553 0.506 0.011 387535 - HCRP1 0.505 0.044 3225 NM 006897.1 HOXC9 0.501 0.028 170482 NM 130441.2 CLEC4C 0.500 0.031 23224 NM 015180.3 SYNE2 0.496 0.006 170959 NM 133473.1 ZNF431 0.496 0.018 1770 NM 001372.2 DNAH9 0.493 0.016 9330 - GTF3C3 0.486 0.002 57404 - CYP20A1 0.486 0.012 7050 NM 173207.1 TGIF 0.485 0.009 3603 NM 172217.1 IL16 0.484 0.008 6872 - TAF1 0.482 0.005 81796 NM 030958.1 SLC05A1 0.481 0.010 1544 - CYP1A2 0.480 0.027 51601 NM 145197.1 LIPTl 0.478 0.047 57096 NM 020366.2 RPGRIP1 0.478 0.024 54961 NM 017857.2 SSH3 0.477 0.018 84870 NM 032784.2 THSD2 0.477 0.019 51253 - MRPL37 0.477 0.002 29967 - LRP12 0.477 0.023 8839 - WISP2 0.476 0.002 54964 NM 017860.3 C1orf56 0.475 0.003 1788 NM 022552.3 DNMT3A 0.475 0.017 8241 NM 152856.1 RBM10 0.473 0.025 23279 NM 015231.1 NUP160 0.473 0.002 79650 - FU13154 0.472 0.015 2139 NM 005244.3 EYA2 0.472 0.041 10896 NM 022375.2 OCLM 0.472 0.048 80086 NM 025019.1 TUBA4 0.472 0.034 85442 NM 033404.2 KNDC1 0.472 0.008 128025 NM 144625.2 WDR64 0.471 0.010 124540 - MSI2 0.471 0.031 65981 NM 001002259.1 C1QDC1 0.470 0.016 254013 - MGC50559 0.469 0.048 51199 - NIN 0.468 0.025 7932 NM 007160.2 OR2H2 0.467 0.018 7368 NM 003360.2 UGT8 0.467 0.022 55022 NM 017933.3 FU20701 0.467 0.024 11215 NM 144490.1 AKAPll 0.466 0.008 23658 NM 012322.1 LSM5 0.466 0.021 1956 NM 005228.3 EGFR 0.462 0.027 81491 NM 030784.1 GPR63 0.459 0.044 160728 NM 145913.2 SLC5A8 0.459 0.026 120406 - FAM55B 0.456 0.007 595 NM 053056.1 CCND1 0.455 0.049 57181 NM 020342.1 SLC39A10 0.453 0.018 7221 XR 000147.4 TRPC2 0.452 0.022 6718 NM 005989.2 AKR1D1 0.452 0.007 8701 NM 003777.2 DNAHll 0.452 0.013 3980 - LIG3 0.452 0.047 5939 NM 002898.2 RBMS2 0.449 0.024 11127 - KIF3A 0.448 0.014 255488 NM 182757.2 IBRDC2 0.447 0.008 26272 NM 012176.2 FBX04 0.445 0.022 158798 NM 178813.5 AKAP14 0.443 0.043 283298 NM 198474.2 OLFML1 0.443 0.012 22876 NM 198331.1 INPP5F 0.442 0.031 147841 NM 182513.1 SPBC24 0.441 0.007 5892 - RAD51L3 0.440 0.025 7442 - TRPV1 0.438 0.031 374659 NM 198527.1 MGC45386 0.437 0.040 51091 - SLA/LP 0.436 0.048 1770 NM 001372.2 DNAH9 0.435 0.048 3659 NM 002198.1 IRF1 0.435 0.034 27094 NM 171829.1 KCNMB3 0.435 0.010 7110 - TMF1 0.433 0.039 26154 NM 173076.2 ABCA12 0.433 0.015 57718 NM 058237.1 PPP4R4 0.432 0.007 2553 - GABPB1 0.430 0.016 56242 NM 021047.1 ZNF253 0.429 0.015 64106 NM 022146.1 GPR147 0.428 0.021 152024 XM 379183.1 LOC152024 0.428 0.036 9463 - PRKCABP 0.428 0.046 4481 NM 138716.1 MSR1 0.424 0.033 28512 NM 020345.3 NKIRAS1 0.424 0.010 89884 NM 033343.2 LHX4 0.423 0.048 7337 - UBE3A 0.422 0.006 51105 NM 016018.3 PHF20L1 0.422 0.016 9824 NM 014783.2 ARHGAP11A 0.421 0.040 54442 - KCTD5 0.420 0.005 255061 - TAC4 0.418 0.015 2334 NM 002025.1 AFF2 0.418 0.006 2678 NM 013421.1 GGTl 0.417 0.005 166979 NM 152623.1 FU37927 0.415 0.013 2257 - FGF12 0.415 0.016 5067 XM 039627.10 CNTN3 0.414 0.035 93654 - ST7-AS2 0.414 0.013 2957 - GTF2A1 0.414 0.028 6934 - TCF7L2 0.413 0.028 220074 NM 145309.1 LRRC51 0.413 0.040 5793 NM 002841.2 PTPRG 0.412 0.018 116496 NM 052966.1 C1orf24 0.412 0.005 284723 NM 207348.1 SLC25A34 0.412 0.030 9552 NM 004890.1 SPAG7 0.410 0.012 11078 NM 138632.1 HRIHFB2122 0.410 0.034 80013 NM 024948.2 C10orf97 0.409 0.022 10559 - SLC35A1 0.408 0.031 55607 XM 371933.1 PPP1R9A 0.408 0.046 154661 NM 138290.1 RPIB9 0.408 0.024 80352 NM 025236.2 RNF39 0.408 0.038 126205 NM 176811.2 NALP8 0.407 0.046 55245 NM 018244.3 C20orf44 0.406 0.039 339400 - FLG-AS1 0.406 0.040 127255 NM 145258.1 LRRC44 0.405 0.014 26115 XM 371074.2 TANC2 0.405 0.012 9374 - PPT2 0.405 0.017 80070 NM 025003.2 ADAMTS20 0.405 0.021 51542 NM 001005739.1 VPS54 0.405 0.028 27430 NM 182796.1 MAT2B 0.404 0.010 4902 NM 004558.2 NRTN 0.404 0.008 3909 NM 198129.1 LAMA3 0.403 0.029 79815 NM 024759.1 NIPAL2 0.403 0.011 1827 - RCAN1 0.403 0.045 146223 NM 181521.1 CKLFSF4 0.403 0.027 56899 - ANKS1B 0.402 0.026 84067 NM 032127.1 FAM160A2 0.402 0.049 353174 NM 180990.2 LGICZ 0.402 0.011 93594 NM 145647.1 WDR67 0.402 0.026 3909 NM 198129.1 LAMA3 0.401 0.025 636 NM 001714.2 BICD1 0.401 0.044 374907 NM 198540.2 B3GALT7 0.399 0.050 266977 NM 153840.2 GPRllO 0.399 0.048 65078 NM 023004.5 RTN4R 0.398 0.032 2900 NM 014619.2 GRIK4 0.397 0.031 9887 NM 173156.1 C1orf16 0.396 0.018 375601 - OR7E31P 0.396 0.026 168537 NM 153236.3 GIMAP7 0.394 0.005 152816 - C4orf26 0.394 0.021 58529 NM 021245.2 MYOZ1 0.394 0.030 639 NM 182907.1 PRDM1 0.392 0.017 114899 NM 181435.4 C1QTNF3 0.391 0.023 1912 - PHC2 0.390 0.006 219938 NM 174927.1 SPATA19 0.390 0.036 9043 NM 172345.1 SPAG9 0.389 0.009 116966 NM 181265.2 WDR17 0.389 0.027 1361 - CPB2 0.389 0.036 79686 NM 024633.2 Cl4orf139 0.388 0.028 7273 NM 133378.2 TIN 0.388 0.020 3773 NM 170742.1 KCNJ16 0.388 0.030 731 NM 000562.1 C8A 0.387 0.050 3755 - KCNG1 0.387 0.043 29062 NM 014149.2 HSPC049 0.387 0.032 114907 NM 148177.1 FBX032 0.386 0.016 55260 NM 018273.2 TMEM143 0.385 0.035 79675 NM 024622.2 FASTKD1 0.385 0.030 7273 NM 133378.2 TIN 0.384 0.022 80144 NM 025074.3 FRAS1 0.384 0.038 123775 NM 152337.1 C16orf46 0.383 0.042 79092 NM 024110.2 CARD14 0.383 0.012 54802 NM 017646.3 TRITl 0.383 0.038 1977 NM 001968.2 EIF4E 0.382 0.037 79258 NM 033467.2 MELL1 0.382 0.022 4617 NM 005593.1 MYF5 0.382 0.010 10570 - DPYSL4 0.381 0.013 1538 XM 088636.6 CYLC1 0.381 0.041 6758 NM 021015.3 SSX5 0.381 0.017 3805 NM 002255.3 KIR2DL4 0.381 0.044 3980 NM 002311.2 LIG3 0.380 0.036 9894 NM 016111.1 TEL02 0.380 0.035 158521 - FMR1NB 0.380 0.015 2140 NM 001990.2 EYA3 0.380 0.006 55758 NM 018254.2 RCOR3 0.379 0.013 23001 NM 178585.1 WDFY3 0.379 0.016 136371 NM 080871.2 ASB10 0.378 0.011 545 NM 001184.2 ATR 0.378 0.025 3841 NM 002269.2 KPNA5 0.378 0.011 27130 - INVS 0.377 0.030 11254 NM 007231.1 SLC6A14 0.377 0.039 22 NM 004299.3 ABCB7 0.377 0.046 148022 NM 182919.1 TICAM1 0.377 0.035 9177 NM 006028.3 HTR3B 0.376 0.033 152078 - C3orf55 0.375 0.032 4241 NM 033316.2 MFI2 0.375 0.032 375298 NM 201548.3 CERKL 0.374 0.016 79862 NM 024804.1 ZNF669 0.373 0.016 57409 NM 020679.2 AD023 0.372 0.032 4939 NM 002535.1 OAS2 0.371 0.026 7220 - TRPCl 0.371 0.026 4063 NM 002348.2 LY9 0.370 0.037 114814 - GNRHR2 0.370 0.029 2786 NM 004485.2 GNG4 0.369 0.013 3925 NM 203401.1 STMN1 0.369 0.045 55294 NM 018315.3 FBXW7 0.368 0.008 116092 NM 052951.2 DNTIIP1 0.368 0.009 8496 NM 003622.2 PPFIBP1 0.368 0.022 3431 - SP110 0.367 0.025 10599 NM 006446.2 SLC01B1 0.367 0.035 23234 NM 015190.3 DNAJC9 0.366 0.013 51599 NM 205834.2 LISCH7 0.366 0.039 1386 NM 001880.2 ATF2 0.365 0.011 1846 NM 057158.2 DUSP4 0.364 0.023 2626 NM 002052.2 GATA4 0.364 0.011 6772 NM 007315.2 STATl 0.364 0.021 1859 NM 130436.1 DYRK1A 0.364 0.023 64072 NM 022124.2 CDH23 0.364 0.021 84893 NM 032807.3 FBX018 0.363 0.030 22876 NM 014937.2 INPP5F 0.363 0.023 84617 - TUBB6 0.362 0.040 9092 NM 005146.3 SARTl 0.362 0.043 8853 - DDEF2 0.362 0.027 25843 - PREI3 0.361 0.017 79956 NM 024896.2 ERMP1 0.361 0.019 220594 NM 145809.1 USP32P2 0.360 0.028 23637 NM 012197.2 RABGAP1 0.360 0.020 81562 NM 030805.1 LMAN2L 0.359 0.006 1356 NM 000096.1 CP 0.358 0.044 10725 - NFAT5 0.357 0.049 286135 XM 379573.2 LOC286135 0.357 0.044 253143 NM 173566.1 PRR14L 0.357 0.024 132884 NM 147127.2 EVC2 0.357 0.035 257169 NM 152786.1 C9orf43 0.356 0.040 84069 NM 032129.1 PLEKHN1 0.356 0.010 8290 NM 003493.2 HIST3H3 0.356 0.032 9337 - CNOT8 0.356 0.049 974 NM 000626.1 CD79B 0.355 0.024 10782 NM 016325.2 ZNF274 0.355 0.047 80835 NM 138697.2 TAS1R1 0.355 0.032 10753 - CAPN9 0.355 0.044 26251 NM 012283.1 KCNG2 0.353 0.017 55527 NM 018708.1 FEM1A 0.352 0.011 55821 NM 018436.2 ALLC 0.352 0.035 8863 NM 016831.1 PER3 0.351 0.047 64062 NM 022118.3 C13orf10 0.351 0.034 6873 NM 003184.2 TAF2 -0.350 0.007 81617 NM 030925.1 CAB39L -0.351 0.021 9465 NM 016377.2 AKAP7 -0.351 0.037 5923 NM 002891.3 RASGRF1 -0.353 0.016 9770 NM 014737.1 RASSF2 -0.353 0.021 22876 NM 014937.2 INPP5F -0.353 0.036 4200 NM 002396.3 ME2 -0.354 0.012 8749 NM 014237.1 ADAM18 -0.355 0.031 2068 - ERCC2 -0.356 0.036 84109 NM 198179.1 GPR103 -0.357 0.028 7593 NM 198055.1 ZNF42 -0.357 0.012 3234 NM 019558.2 HOXD8 -0.357 0.040 63036 NM 033440.1 ELA2A -0.357 0.045 94239 NM 138635.2 H2AFV -0.357 0.035 91120 - ZNF682 -0.359 0.019 54949 NM 017841.1 FU20487 -0.359 0.029 1056 NM 001807.2 CEL -0.360 0.027 57502 NM 020742.2 NLGN4X -0.360 0.028 55148 NM 018108.2 C14orf130 -0.360 0.012 91289 NM 033200.1 BC002942 -0.360 0.023 7699 NM 003440.2 ZNF140 -0.360 0.014 163183 NM 153233.1 FU36445 -0.361 0.019 80243 NM 025170.4 DEPDC2 -0.361 0.027 83546 NM 031429.1 RTBDN -0.361 0.034 472 NM 000051.2 ATM -0.361 0.041 1038 NM 004065.2 CDR1 -0.361 0.036 54207 NM 138318.1 KCNK10 -0.361 0.013 55967 NM 018838.3 DAP13 -0.362 0.025 9258 - MFHAS1 -0.363 0.009 29106 NM 013243.2 SCG3 -0.363 0.016 9899 NM 014848.2 SV2B -0.364 0.036 55156 NM 018120.3 ARMCl -0.364 0.035 130560 NM 139073.2 SPATA3 -0.365 0.021 25 NM 007313.2 ABL1 -0.365 0.035 1907 NM 001956.2 EDN2 -0.366 0.010 2694 NM 005142.2 GIF -0.366 0.017 29765 NM 013353.1 TMOD4 -0.367 0.034 283373 XM 370696.2 FU34236 -0.367 0.040 9645 NM 014632.2 MICAL2 -0.369 0.028 5933 NM 002895.2 RBL1 -0.369 0.048 127255 - LRRC44 -0.369 0.034 63941 NM 031232.2 APBA2BP -0.370 0.041 758 NM 001585.2 C22orf1 -0.370 0.045 51105 NM 032205.2 PHF20L1 -0.372 0.035 845 NM 001232.1 CASQ2 -0.374 0.014 1848 NM 001946.2 DUSP6 -0.374 0.019 84295 NM 032458.2 PHF6 -0.374 0.036 55012 NM 017917.2 C14orf10 -0.375 0.044 7273 NM 133378.2 TIN -0.375 0.035 4756 - NE01 -0.375 0.017 23095 NM 015074.2 KIFlB -0.377 0.030 79658 NM 024605.3 ARHGAP10 -0.377 0.036 128646 NM 178460.1 PTPNS1L2 -0.378 0.044 27065 NM 014392.2 D4S234E -0.378 0.046 23060 XM 042833.8 ZNF609 -0.378 0.038 112802 NM 033448.1 KRT61RS -0.379 0.013 286046 NM 173683.2 C8orf21 -0.380 0.022 80741 NM 001002849.1 LY6G5C -0.381 0.011 60506 NM 022567.1 NYX -0.381 0.036 5222 NM 014224.1 PGA5 -0.381 0.045 64175 NM 022356.2 LEPRE1 -0.381 0.019 84239 NM 032279.2 ATP13A4 -0.382 0.013 146712 NM 194288.1 LOC146712 -0.382 0.033 143279 NM 182765.1 HECTD2 -0.382 0.034 399669 NM 203307.1 MGC35402 -0.382 0.044 64924 NM 022902.2 SLC30A5 -0.382 0.048 26154 NM 015657.3 ABCA12 -0.383 0.043 7052 NM 004613.2 TGM2 -0.383 0.048 55119 - PRPF38B -0.383 0.048 23624 NM 012116.2 CBLC -0.384 0.041 6932 NM 201633.1 TCF7 -0.384 0.037 90665 NM 134259.1 TBL1Y -0.384 0.043 3375 NM 000415.1 IAPP -0.385 0.048 63901 NM 198847.1 FU22794 -0.385 0.015 8100 NM 175605.2 TIClO -0.385 0.046 63027 NM 021945.4 C6orf85 -0.386 0.018 324 NM 000038.3 APC -0.386 0.010 114899 NM 181435.4 C1QTNF3 -0.386 0.027 55520 NM 018696.2 ELAC1 -0.387 0.014 55032 NM 017945.2 SLC35A5 -0.387 0.019 6772 NM 139266.1 STATl -0.387 0.023 51514 NM 016448.1 DTL -0.387 0.027 9941 - ENDOGL1 -0.387 0.007 22930 NM 012233.1 RAB3GAP -0.388 0.024 201163 NM 144997.4 FLCN -0.388 0.018 55773 - FU11046 -0.389 0.014 79712 NM 024659.2 GTDC1 -0.389 0.045 51703 NM 016234.3 ACSL5 -0.389 0.028 80339 NM 025225.2 ADPN -0.390 0.017 84436 NM 032423.2 ZNF528 -0.390 0.041 4058 NM 002344.3 LTK -0.390 0.045 6252 NM 021136.2 RTN1 -0.390 0.045 81035 - COLEC12 -0.391 0.010 8621 NM 031267.1 CDC2L5 -0.392 . 0.020 5897 NM 000536.1 RAG2 -0.393 0.029 3547 NM 001555.2 IGSF1 -0.394 0.035 30819 - KCNIP2 -0.394 0.025 80709 - AKNA -0.394 0.006 79019 NM 001002876.1 C22orf18 -0.395 0.023 80178 - FU13909 -0.397 0.016 3781 NM 021614.2 KCNN2 -0.397 0.018 1656 NM 004397.3 DDX6 -0.398 0.011 361 NM 004028.3 AQP4 -0.398 0.050 202134 XM 371783.2 LOC389347 -0.399 0.032 10214 NM 021014.2 SSX3 -0.400 0.019 54798 NM 017639.2 DCHS2 -0.400 0.031 51562 - MBIP -0.402 0.037 2299 NM 012188.3 FOXI1 -0.405 0.029 56956 - LHX9 -0.405 0.048 11066 - U1SNRNPBP -0.405 0.007 93624 XM 291105.1 TADA2B -0.405 0.014 89231 XM 496860.1 DPY19L1P1 -0.406 0.033 157680 NM 017890.3 VPS13B -0.407 0.004 54581 NM 022050.2 SCAND2 -0.407 0.021 130120 NM 198448.2 REG3G -0.407 0.006 375719 - AQP7P1 -0.407 0.006 9057 NM 003983.3 SLC7A6 -0.408 0.041 3547 NM 001555.2 IGSF1 -0.408 0.028 4771 NM 181825.1 NF2 -0.408 0.010 9637 - FEZ2 -0.408 0.026 9963 NM 152685.2 SLC23A1 -0.409 0.006 26268 NM 033481.2 FBX09 -0.409 0.028 26122 NM 015630.2 EPC2 -0.409 0.025 55181 NM 018149.5 SMG8 -0.410 0.040 8287 - USP9Y -0.410 0.015 448835 - LCE6A -0.410 0.021 1841 NM 012145.2 DTYMK -0.412 0.029 51338 NM 024021.2 MS4A4A -0.413 0.012 27198 NM 032554.2 GPR81 -0.413 0.039 54906 XM 374765.2 C10orf18 -0.414 0.022 54904 NM 023034.1 WHSC1L1 -0.416 0.033 55075 NM 001008224.1 UACA -0.417 0.021 9156 NM 130398.2 EX01 -0.417 0.012 5886 NM 005053.2 RAD23A -0.417 0.042 10349 NM 080282.2 ABCA10 -0.417 0.030 10903 NM 181873.1 MTMR11 -0.418 0.035 6891 NM 000544.2 TAP2 -0.418 0.030 6872 NM 138923.1 TAF1 -0.419 0.040 5519 NM 181699.1 PPP2R1B -0.419 0.009 84106 NM 032152.3 PRAM1 -0.421 0.023 64601 - VPS16 -0.421 0.026 283820 - NOM02 -0.421 0.035 83850 NM 031913.1 FAM62C -0.421 0.034 2618 NM 000819.3 GART -0.421 0.022 80018 NM 024953.2 NAA25 -0.423 0.029 84132 - USP42 -0.423 0.032 2744 NM 014905.2 GLS -0.423 0.048 23192 - APG4B -0.424 0.019 286205 NM 173690.1 C9orf126 -0.424 0.039 6815 - STYX -0.426 0.021 10842 NM 006658.2 C7orf16 -0.427 0.029 123263 NM 139242.2 MtFMT -0.428 0.039 54991 NM 017891.2 FU20584 -0.428 0.019 25852 NM 015396.3 ARMC8 -0.429 0.018 89777 NM 080474.1 SERPINB12 -0.429 0.026 51513 NM 016135.2 ETV7 -0.431 0.018 27022 NM 012183.1 FOXD3 -0.431 0.028 79442 NM 024512.2 LRRC2 -0.432 0.045 222962 NM 153247.1 SLC29A4 -0.432 0.006 117155 - CATSPER2 -0.432 0.043 10368 NM 006539.2 CACNG3 -0.432 0.011 79041 - TMEM38A -0.433 0.021 10933 NM 206839.1 MORF4L1 -0.434 0.028 10771 NM 212479.1 ZMYND11 -0.434 0.016 115362 NM 052942.2 GBP5 -0.435 0.027 925 NM 171827.1 CD8A -0.437 0.033 286103 XM 497002.1 C8orf77 -0.437 0.014 284340 NM 198477.1 CXCL17 -0.438 0.020 85465 NM 033505.1 SEU -0.438 0.031 5473 NM 002704.2 PPBP -0.438 0.009 5255 NM 002637.1 PHKA1 -0.439 0.016 3889 NM 002282.2 KRTHB3 -0.440 0.005 9986 - RCE1 -0.441 0.012 26628 NR 002163.1 OR7E37P -0.442 0.012 6752 NM 001050.2 SSTR2 -0.443 0.010 202020 NM 152684.1 FU39653 -0.444 0.002 25911 NM 015448.1 DPCD -0.444 0.048 27434 NM 013284.1 POLM -0.445 0.015 3269 NM 000861.2 HRH1 -0.445 0.017 6493 NM 005069.2 51M2 -0.446 0.017 411 NM 000046.2 ARSB -0.447 0.025 80758 NM 030567.2 PRR7 -0.447 0.048 56659 - KCNK13 -0.447 0.019 79022 - TMEM106C -0.448 0.006 288 NM 020987.2 ANK3 -0.448 0.013 22954 NM 012210.2 TRIM32 -0.450 0.006 4477 NM 002443.2 MSMB -0.450 0.029 7201 NM 003301.1 TRHR -0.452 0.030 5373 NM 000303.1 PMM2 -0.452 0.003 7020 - TFAP2A -0.455 0.017 256356 NM 152776.1 GK5 -0.456 0.012 115548 NM 138782.1 FCH02 -0.457 0.036 2100 NM 001437.1 ESR2 -0.457 0.024 55103 - RALGPS2 -0.459 0.024 55758 - RCOR3 -0.460 0.034 1545 NM 000104.2 CYP1B1 -0.463 0.026 51184 NM 016301.2 GPN3 -0.464 0.019 84654 NM 032567.2 SPZ1 -0.465 0.044 26031 NM 015550.2 OSBPL3 -0.469 0.021 865 NM 022845.2 CBFB -0.470 0.021 51277 NM 016544.1 RBJ -0.470 0.015 60676 NM 020318.1 PAPPA2 -0.474 0.023 8829 NM 003873.3 NRP1 -0.475 0.017 1723 NM 001361.3 DHODH -0.476 0.006 143686 NM 144665.2 SESN3 -0.476 0.025 1111 - CHEK1 -0.476 0.048 1594 NM 000785.2 CYP27B1 -0.477 0.003 23549 - DNPEP -0.477 0.023 168002 NM 214462.1 DACT2 -0.478 0.014 144717 NM 144671.2 FAM109A -0.481 0.004 147699 NM 178494.2 PPM1N -0.482 0.009 22894 - 0153 -0.482 0.030 152687 NM 182524.1 ZNF595 -0.483 0.008 8556 NM 003672.2 CDC14A -0.484 0.006 83451 NM 148912.2 ABHD11 -0.486 0.050 114899 - C1QTNF3 -0.487 0.002 25809 - TTLL1 -0.488 0.026 767 NM 004056.4 CA8 -0.489 0.005 2322 NM 004119.1 FLT3 -0.489 0.038 10040 NM 005486.1 TOM1L1 -0.489 0.021 160857 NM 144974.1 CCDC122 -0.489 0.014 159989 - CCDC67 -0.490 0.012 203228 NM 018325.1 C9orf72 -0.494 0.010 5611 - DNAJC3 -0.494 0.007 10249 NM 201648.1 GLYAT -0.497 0.048 23001 NM 014991.3 WDFY3 -0.497 0.008 54845 NM 017697.2 ESRP1 -0.500 0.011 9833 NM 014791.2 MELK -0.506 0.027 51118 NM 016037.2 UTP11L -0.506 0.007 5276 NM 006217.3 SERPINI2 -0.508 0.026 81848 NM 030964.2 SPRY4 -0.510 0.008 150290 NM 152511.3 DUSP18 -0.510 0.003 54112 NM 022049.1 GPR88 -0.513 0.021 2888 - GRB14 -0.513 0.019 23504 - RIMBP2 -0.517 0.016 7013 - TERF1 -0.519 0.026 9847 - KIAA0528 -0.520 0.017 51061 NM 015914.5 TXNDC11 -0.526 0.007 201140 - DHRS7C -0.526 0.018 134957 NM 139244.2 STXBP5 -0.526 0.003 3785 NM 172109.1 KCNQ2 -0.527 0.012 54108 NM 017444.3 CHRACl -0.530 0.017 260434 NM 152901.1 PYDC1 -0.531 0.036 84874 NM 032788.1 ZNF514 -0.531 0.013 79666 NM 024613.2 PLEKHF2 -0.532 0.016 131450 - CD200R1 -0.533 0.014 4211 NM 002398.2 MEIS1 -0.533 0.036 124626 NM 198844.1 ZPBP2 -0.538 0.016 26275 - HIBCH -0.544 0.024 2618 NM 000819.3 GART -0.546 0.001 1636 NM 000789.2 ACE -0.548 0.003 3688 NM 033668.1 ITGB1 -0.549 0.019 4771 NM 016418.4 NF2 -0.555 0.005 7083 NM 003258.1 TK1 -0.556 0.014 10655 NM 006557.3 DMRT2 -0.557 0.037 84141 NM 032181.1 FAM176A -0.559 0.020 114134 NM 052885.1 SLC2A13 -0.561 0.003 36 NM 001609.2 ACADSB -0.564 0.010 85440 NM 033407.2 DOCK7 -0.567 0.001 4684 NM 181351.1 NCAM1 -0.571 0.015 285704 NM 173670.2 RGMB -0.572 0.012 2568 NM 014211.1 GABRP -0.574 0.006 7319 NM 181777.1 UBE2A -0.585 0.003 10396 NM 006095.1 ATP8A1 -0.588 0.023 7556 NM 015394.4 ZNF10 -0.588 0.003 152330 NM 175607.1 CNTN4 -0.590 0.017 51283 NM 016561.1 BFAR -0.590 0.036 89866 - LZTR2 -0.591 0.015 2257 - FGF12 -0.593 0.049 166378 NM 145207.1 SPATA5 -0.594 0.005 79672 NM 024619.2 FN3KRP -0.595 0.005 3196 NM 001534.2 TLX2 -0.595 0.002 80309 - SKIP -0.600 0.008 2235 NM 001012515.1 FECH -0.603 0.015 51444 NM 198128.1 RNF138 -0.605 0.021 145173 - B3GTL -0.606 0.002 9718 NM 014693.2 ECE2 -0.610 0.034 50839 NM 023921.1 TAS2R10 -0.618 0.003 11080 NM 007034.3 DNAJB4 -0.618 0.040 140803 NM 017662.3 TRPM6 -0.619 0.026 387837 - CLEC12B -0.626 0.002 7499 NM 175569.1 XG -0.627 0.006 1956 NM 005228.3 EGFR -0.632 0.033 8291 NM 003494.2 DYSF -0.638 0.008 221756 XM 376463.2 MGC39372 -0.640 0.011 54810 NM 017655.4 GIPC2 -0.650 0.021 26127 - FGFR10P2 -0.652 0.024 54751 NM 001024216.1 FBLIM1 -0.659 0.044 10863 - ADAM28 -0.665 0.007 284402 XM 378794.1 SCGB2B2 -0.668 0.013 23483 NM 014305.1 TGDS -0.673 0.005 9923 NM 014870.2 ZBTB40 -0.696 0.001 114879 NM 020896.2 OSBPL5 -0.708 0.028 57827 NM 021184.3 C6orf47 -0.710 0.019 143425 NM 175733.2 SYT9 -0.736 0.025 91807 NM 182493.1 MLCK -0.739 0.034 63892 NM 022065.3 THADA -0.746 0.016 1757 NM 007101.2 SARDH -0.813 0.017 253018 - HCG27 -0.829 0.048 80019 NM 024954.3 UBTD1 -0.867 0.029 9923 - ZBTB40 -1.023 0.017 23163 NM 138619.1 GGA3 -1.112 0.036 199699 NM 152654.1 DANOS -1.131 0.049 285033 XM 379111.2 LOC285033 -1.135 0.031 ANEXO IV

ARTÍCULO ENVIADO A CANCER RESEARCH

Low abundance transcríptome analyses identifíes novel markers, specific intracel/ular pathways and target genes in advanced human gastrointestinal cancer

224 Manuscript Home Author Instructions Revlewer Instructions Help Tips Logout Joumal Home

Detailed Status lnfonnation

IManuscript # jcAN-12-3474 ¡o !j:l012-09-lll0:46:23------_.. _ ------lsubmission Date lcurrent Stage

·¡ i'Low abundance transcriptome analyses identifies novel markers1 specific . Title _ _ _ 'intracellular pathways and target genes in human advanced human ...... _ gastrointestinal cancer . _ __ IRunning Title ;jLo~abund~nce-tr~nscriptome in gast;i-cand col~n cancers ·¡-l~~ot_a_n-_u_s_c-=.!-_ipt-.-.. _-T-y~p-e~.---,.[Res~arch Article_ _ __ .. _ . .

IC:CSteg()ry :jTheraPE!!Jtics1 Jarget?l_élf1d_C:t1ef11ical l3iQiogy Carolina Bizama and Felipe Benavente contributed equally to this work. Manuscript Comment Osvaldo Podhajcer and Manuel Gidekel both are corresponding authors.

Dr. Carolina Bizama 1 Dr. Felipe Benavente 1 Dr. Jaime A. Espinoza 1 Dr.

Contributing Authors Edgardo Salvatierra 1 Dr. Hilda A. Gutiérrez 1 Dr. Elmer Fernández, Mr.

Eduardo A. Sagrec!g 1 Mr.}vár:tRoa, Qr. Manuel. Gidekel_ . . ______. Studies on the low abundance transcriptome are of paramount importan ce to identify the intimate mechanisms of tumor progression. By using subtractive hybridization followed by transcriptome analysis of human samples obtained ~ from gastric and colon cancer and paired adjacent non-cancerous tissues we

: identified novel cancer biomarkers such as LAMA3 and TfN 1 highly specific intracellular pathways and gene targets such as IR52 IL17 IFNy VEGF-C, Abstract 1 1 1 WISP1 1 FZD5 and CfBPl that were not detectable by direct transcriptomic analyses. Knocking down the expression of CTBPl sensitized gastric cancer cells to 5-FU, cisplatin and epirubicin that are part of the mainstay repertoire for gastric cancer treatment. The use of these affordable combined platforms might have important implications in the understanding and treatment of : cancer. 'In search for novel markers and therapeutic targets we applied a combined · strategy of subtractive hybridization followed by transcriptomic microarrays : analysis to study the low abundance transcriptome of gastric and colon cancer , Précis : samples obtained from patients at advanced stages of the disease. Cancer type specific- signaling pathways and unique genes were identified that could i serve as targets for potential clinical application. •tu~~;:=o~fi~~cf ~t!y;~;cfs'ILo~ ~b~ ~dan_ce:_t~~;_s~;¡pto rT1~~. s~btr~~ti~~ ~U[>P~t!~~i o~ h'J'~¿i~i-~éltiQF1;:6~f'i _Gastrointestinal cancers: colorectal, __Gastrointestinal cancers: stomach, __Signa! transduction, CB CELLULAR, MOLECULAR, AND TUMOR iiKeywo

PhD. Osvaldo L. Podhajcer. Member number 75447 .

....

225 Do you confírm that neíther this paper nor any similar paper, in whole orín part, other than abstract or prelíminary communication, has been or will be submítted to or published elsewhere?

Dual Publication Yes

If no, please explain why below.:

Do you declare that authorization has been given to use any information conveyed by either personal communication or release of unpublíshed experimental data? Information Yes Authorization

f no, please explaín be!ow.:

Disclosure of Chemical Structures ell Line uthentication

Awaiting Submíssíon PayrT}er¡~ [i()i?~()~~Q~j~ji:42 )M

14anusr.:ript Heme 1 Author !nstructions 1 Reviewer Jnstrvctions 1 Help 1 Logout 1Joumal Home Terms of Serví ce UCensed under Patent #US 7,520,55561

226 Page 1 Low abundance transcriptome analyses identifies novel markers, specific intracellular pathways and

target genes in human advanced human gastrointestinal cancer

1 1 7 1 Authors and Affiliations: Carolina Bizama, ,7 Felipe Benavente, ' Jaime A. Espinoza, Edgardo

Salvatierra,2 Hilda A. Gutiérrez, 1 Elmer A. Femández,3 Eduardo A. Sagredo,4 Iván Roa, 5 Guillermo

6 1 2 Mazzolini, Manuel Gidekel, ' * and Osvaldo L. Podhajcer ,*.

1Applied Cellular and Molecular Biology PhD Program, Agricultura! and Forestry Sciences Faculty. La

Frontera University, Temuco, 4811230, Chile.

2Laboratory of Molecular and Cellular Therapy, Fundación Instituto Leloir-CONICET, Buenos Aires,

C1405BWE Argentina.

3School of Engineering, Intelligent Data Analysis Group, Catholic University of Córdoba, X5016DHK

Córdoba, Argentina.

4Biotechnology Program, Agricultura! and Forestry Sciences Faculty, La Frontera University, Temuco,

4811230, Chile.

5Pathology Service, Clínica Alemana de Santiago, Faculty of Medicine, Universidad del Desarrollo,

Santiago, 7650568, Chile.

6Gene Therapy Laboratory, School of Medicine, Austral University, Pilar-Buenos Aires, B1664INZ ,

Argentina.

7These authors contributed equally to this work.

Running Title: Low abundance transcriptome in gastric and colon cancers

Keywords: Low abundan ce transcriptome, subtractive suppression hybridization, CTBP l.

*Corresponding authors: Osvaldo L. Podhajcer, Laboratory of Molecular and Cellular Therapy,

Fundación Instituto Leloir, Buenos Aires, Argentina, C 1405BWE. Phone: +54-11-52387500 int31 07; Fax:

+54-11-52387501; E-mail: [email protected]. Manuel Gidekel, Av. Alvaro Casanova 4090,

Peñalolén, Santiago, 7941068. Phone: +56-999970216; E-mail: [email protected]. Page2 Word account (excluding references): 4977. This research article includes six principal figures and two tables. The supplementary information includes four figures, three tables and supplemental experimental procedures. Page 3 Abstract Studies on the low abundan ce transcriptome are of paramount importan ce to identizy the intimate mechanisms of tumor progression. By using subtractive hybridization followed by transcriptome analysis of human samples obtained from gastric and colon cancer and paired adjacent non-cancerous tissues we identified novel cancer biomarkers such as LAMA3 and TIN, highly specific intracellular pathways and gene targets such as IRS2, IL17, IFNy, VEGF-C, WISPI, FZD5 and CTBPI that were not detectable by direct transcriptomic analyses. Knocking down the expression of CTBPI sensitized gastric cancer cells to

5-FU, cisplatin and epirubicin that are part ofthe mainstay repertoire for gastric cancer treatment. The use of these affordable combined platforms might have important implications in the understanding and treatment of cancer. Page4 Introduction

Each year worldwide, more than 2 million people are diagnosed with gastric and colon cancer and

more than 1,3 million people die of both diseases, representing ~ 17% of cancer mortality globally (1 ).

Despite advances in diagnostic imaging that improved early detection of gastric and colon cancer advanced

stages ofboth diseases have still a poor prognosis (2). Recent advances in stratified medicine has improved

response in advanced colorectal carcinoma patients treated with the EGFR targeted antibody cetuximab or

panitumumab and in gastric cancer patients expressing the HER2 receptor treated with trastuzumab (3);

but e ven in those cases resistance develops rapidly, the benefit is transient and most of these individuals do

progress eventually (4).

Increasing evidence suggests that low abundance transcripts may play fundamental roles in biological processes. A challenge in functional genomics applied to human diseases and in particular to cancer research is to identify the expression profile of the low abundance transcriptome in cancerous tissues as a potential source of tumor-specific genes with potential diagnostic or therapeutic uses.

Microarrays platforms were very helpful in the identification of differentially expressed genes between cancer and non-cancerous adjacent tissue. However, direct screening of human tissues highlighted mainly transcripts related to cytoskeleton, cell-ECM interaction, cell cycle, focal adhesion contacts and other gene ontology groups that clearly represented highly abundant transcripts (5-6). It became clear that microarray analysis did not provide sufficient sensitivity to measure low abundance transcripts reproducibly (7). PCR­ based suppressive subtractive hybridization technique is a highly sensitive platform identifying variations in gene expression. Subtractive hybridization equalizes the abundance of cDNA within the target samples enriching low abundance and rare transcripts (8-9). Transcript detection and coverage can be significantly increased by coupling the initial subtraction to gene expression microarrays platforms. This approach has been applied and methodologically validated to cancer research to systematically identify tumor-specific transcripts that are not identified by conventional microarrays analysis (1 0-12). Page 5 Identification of low abundan ce transcripts can be a so urce of potential diagnostic and therapeutic

targets. Here we applied subtractive hybridization followed by microarray analysis to identify tumor

specific, low abundance transcripts that were differentially expressed in human gastric and colon

adenocarcinomas compared with their paired adjacent non-cancerous tissues. The vast majority ofthe low

abundant differentially expressed transcripts the cancer type-specific signaling pathways were not detected

when a direct microarray analysis was performed, leading to the identification of novel biomarkers and

gene targets aberrantly expressed in cancer samples. Further functional studies identified CTBPl as a

novel target to sensitize gastric cancer cells to chemotherapeutic drugs.

Material and Methods

Clinical Samples

Samples were obtained from the Hospital Temuco Tumor previous approval of the Ethics

Committee Bank and corresponded to patient that signed informed consent or following a previous step of

anonymization in deceased patients. Cancerous and adjacent non- cancerous tissues were obtained from

twelve patients with advanced gastric adenocarcinoma and ten patients with advanced colon

adenocarcinoma who did not receive adjuvant therapy prior. Collected tissues were preserved immediately

in RNALater (Ambion Inc, Austin Tx, USA) and stored at -80°C until used. Befare RNA extraction

samples were histologically analyzed to confirm malignancy.

Suppression subtractive hybridization (SSH)

Total RNA was extracted with Trizo! reagent (Invitrogen, Carlsbad, CA, USA), followed by RNA purification using RNeasy mini kit columns (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Human Universal Reference RNA and normal gastric and colon RNAs were purchased from

Clontech (Palo Alto, CA, USA). Total RNA (l¡..tg) was used for first strand synthesis using Super SMART

PCR cDNA Synthesis kit (Clontech, Palo Alto, CA, USA) and SuperScript III Reverse Transcriptase Page 6 (Invitrogen, Carlsbad, CA, USA) following the supplier' s protocol. Subtractive hybridization was

performed with the aid of the Clontech PCR-Select cDNA Subtraction Kit (Clontech, Palo Alto, CA,

USA). A customized primer 1 was designed to keep the T7-promoter region in the 5'end ofthe subtractive

amplicon (5'- CTAATACGACTCACTATAGGGCTCGAGCGGCC-3') in the secondary PCR. The in

vitro transcription of the secondary product was generated a single antisense RNA of the differentially

expressed genes that can be subsequently analyzed with the oligonucleotide arrays. Details are provided in

the Supplemental Experimental Procedures.

Microarray gene expression analysis

We used the 48.5K Exonic Evidence Based Oligonucleotide (HEEBO) arrays purchased from

Microarray Inc. (Nashville, TN, USA) based on a probe set designed by Illumina (San Diego,

http://www.illumina.com) and Stanford University. A detailed description of these arrays and protocols

can be found (http://microarray.org/sfgf/heebo.do). Normalization was performed in R statistical

environment using Limma pachage (www.r-proyect.org). The raw data have been deposited in the Gene

Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/projects/geo/) under accession number

GSE38940 and subseries GSE38932, GSE38939. Differentially expressed gene lists were tested for enrichment using the NIH database annotation, visualization and integrated discovery (DAVID) v6.7 analysis (13). The expanded list obtained from the indirect strategy was analyzed with PANTHER (protein analysis through evolutionary relationships) Classification System for pathways enrichment (14). Page 7 Results

Validation ofthe subtractive procedure before microarrays analysis .

The aim of this study was to identify differentially expressed genes between paired cancerous and

adjacent non-cancerous tissues using suppression subtractive hybridization followed by microarrays

analysis. This indirect strategy was compared with the differentially expressed genes obtained after direct

microarrays analysis of the same samples. HEEBO microarrays were used that contain 44,544 70-mer

oligonucleotide probes, representing approximately 30,718 unique genes. Por subtraction experiments,

cDNAs from each cancer sample or its paired adjacent non-cancerous tissue were used separately as tester while cDNAs obtained from normal gastric or colon tissue RNA were used as driver. Twenty four independent runs in gastric tissues and twenty in colon tissues were performed on the twelve gastric cancer samples and the ten colon cancer cDNAs and their paired adjacent non-cancerous tissue, respectively.

Comparison of both approaches was possible because the transcripts with or without previous subtractive hybridization were prepared individually and hybridized in a similar HEEBO microarray against an aRNA obtained from a Universal Reference (Fig.l).

The subtraction efficiency was confirmed by an average decrease of 50% in the expression levels of

100 housekeeping selected probes following the subtraction step (data not shown). Additionally, we confirmed the decreased expression ofG3PDH, QARS, TBP and UBE2D2 by qRT-PCR (data not shown).

Most of the genes that were identified in the indirect strategy showed expression values near zero in the direct one confirming the enrichment of low abundance transcripts using a previous subtractive hybridization (data not shown). The differentially expressed genes corresponding to gastric and colon cancer as well as the function and the biological process in which they are involved are depicted

(Supplementary Table S 1). Page 8 Identification oflow abundance cancer-associated transcripts

In gastric cancer, a total of 119 differentially expressed genes were detected by direct microarray

analysis, 81 up-regulated and 38 down-regulated (absolute fold change > 1.2, P value < 0.05). Whereas,

149 differentially expressed genes, 59 up-regulated and 90 down-regulated genes (absolute fold change >

1.5 and P value < 0.05) were identified using the indirect strategy. Only three down-regulated genes were

detected by both strategies (Fig. 2A and data not shown). In colon cancer, 103 differentially expressed

genes, 78 of them up-regulated and 25 down-regulated (absolute fold change > 1.2, P value < 0.05) were

detected with the direct strategy; whereas 67 differentially expressed genes were obtained with the indirect

strategy, 40 up-regulated and 27 down-regulated (absolute fold change > 1.3, P value < 0.002). Only 7 up­

regulated genes and 2 down regulated genes were detected by both strategies (Fig. 2B and data not shown).

Following the direct strategy, we found 13 transcripts shared by colon and gastric cancer including

TGFBI, THIL, CSEIL, CLDNI, XRN2, EST_AA911832, PMEPAI, LAMP2, LACTB2, NAT5, PPAI and

ZF that were up-regulated in the cancer samples and P KIB that was down-regulated. Only two transcripts

were shared between the two cancer types following the indirect approach, FCGBP and CA2, indicating

that the low abundance transcriptome is slightly more specific of each cancer type. Clustering analysis

demonstrated that the sets of differentially expressed genes ( obtained either from the direct or indirect

strategies) were able to segregate cancer from non-cancerous samples with perfect accuracy

(Supplementary Fig. S 1).

Further qRT-PCR analyses validated 88% (22/25) and 90% (27/30) ofthe genes obtained with the

direct strategy in gastric and colon cancer, respectively (Fig. 2C andE). qRT-PCR analyses also validated

70% (14/20) and 68.4% (13/19) of the genes obtained with the indirect strategy in gastric and colon cancer, respectively (Fig. 2D and F). Overall, these results indicate that the data obtained from the indirect strategy is robust since the vast majority of differentially expressed genes were validated by qRT-PCR. Page 9 Tissue Microarrays validation of differentially expressed transcripts

In order to further validate the direct microarrays data we performed tissue microarrays analysis

(TMA) of RCC2 (regulator of chromosome condensation 2) that was differentially expressed in gastric

cancer and JPH1 (Junctophilin 1) that was differentially expressed in colon cancer. RCC2 protein

expression was assessed in a cohort of 39 paired primary gastric cancer samples with adjacent non­

cancerous gastric tissue and 5 normal gastric tissues. Most gastric cancer samples showed high to moderate

staining of RCC2 protein (Fig. 3A) compared to the moderate to low staining intensity in adjacent non­

cancerous tissue (Fig. 3B). Normal gastric tissue exhibited low to complete absence of RCC2 expression

(Fig. 3C). Almost 70% of gastric cancer samples (28/39) showed increased expression levels of RCC2

compared to the adjacent non-cancerous tissues (Supp1ementary Fig. S2A and data not shown). As a

who1e, cancer samples exhibited more than 2-fo1d increased RCC2 expression compared to adjacent non­

cancerous samples (P < 0.0001; Fig. 3D). In addition, more than 30% of cancerous samples exhibited al so

nuclear staining compared to on1y 7% ofthe adjacent tissues (Supp1ementary Fig. S2B). Interestingly, the

nuclear staining in adjacent non ma1ignant epithe1ia1 cells was observed in the highest proliferative region

of the gastric mucosa and in signet-ring cells present in the diffuse subtype of gastric cancer

(Supplementary Fig. S2C and S2D).

JPH1 analysis was performed in 46 paired primary colon cancer tissues, their respective non­

cancerous adjacent tissues and 8 normal colon tissues. Seventy percent of cancer samples (32/46) exhibited

higher expression 1evels of JPH1 compared to their non-cancerous paired tissue (Fig. 3E and F;

Supplementary Fig. S2E and data not shown). Moreover, a1most 95% of adjacent non-cancerous samples

and all normal colon tissues exhibited none to low JPH1 expression 1eve1s (Fig. 3F and 3G; Supplementary

Fig. S2E). As a who1e, colon cancerous samples exhibited in average 2-fold increased expression 1evels of

JPH-1 compared to adjacent non-cancerous tissues (P < 0.0001; Fig. 31!). Interestingly, the two cases of signet-ring cell tumor availab1e showed negative JPH1 staining (Supplementary Fig. S2F). Page 10 To validate the information obtained by the indirect strategy, we performed TMA analysis on two

previously unexplored genes in gastric cancer: LAMA3 (Laminin alpha 3) a subunit of laminin 5 with

essential roles in cell adhesion and motility (15) and TTN (Titin), also known as connectin, that is

responsible for the passive elasticity of muscle and has been reported as a potential melanoma biomarker

(16-17). The expression ofLAMA3 was analyzed in 37 paired primary gastric cancerous tissues with their

respective adjacent non-cancerous tissues and 5 normal gastric tissues. Eighty four percent of malignant

samples (31/37) showed increased LAMA3 intracytoplasmic staining compared to its non-cancerous

counterpart (Fig. 4A and B; data not shown). Of note, 95% of the non-cancerous samples and all the

normal gastric samples exhibited negative staining while almost 60% of cancerous tissue exhibited

moderate to high Ievels (Fig. 4C). As a whole, gastric cancer samples exhibited in average 17-fold

increased expression levels ofLAMA3 compared to adjacent non-cancerous tissues (P < 0.0001; Fig. 4D).

The expression of TTN was studied in 35 paired primary gastric cancer tissues compared to their

respective adjacent non-cancerous one and in 5 gastric normal samples. Moderate to high intensity ofTTN

staining was observed in more than 60% of cancer samples (Fig. 4G) compared to less than 40% of

samples in non-cancerous adjacent tissues (Fig. 4E, F, G and H). In addition 5/6 samples ofnormal tissues showed low or none TTN staining (data not shown). Moreover, 63% ofmalignant gastric samples (22/35) expressed increased TTN expression compared to their respective adjacent non-cancerous tissues (data not shown). Mostly important, the moderate and strong staining intensity ofTTN in non-cancerous tissues was observed in areas of intestinal metaplasia, a pre-malignant lesion involved in gastric carcinogenesis (Fig.

41).

Both strategies enrich different gene ontologies terms and pathways

The differentially expressed genes were further analyzed for enrichment of gene ontology terms

(GO) and pathways, using the functional annotation clustering classification tool ofDAVID v6.7 database.

The main functional categories enriched in gastric cancer using the direct strategy included processes Page 11 associated to cell interaction with the surrounding stroma such as collagen; extracellular matrix; integrin binding; cell adhesion; and growth factor binding (Supplementary Table S2). Interestingly, in the indirect strategy the mostly enriched groups corresponded to genes involved in detection of biotic stimulus, regulation of cell migration, ion binding, cell recognition and adaptive immune response (Supplementary

Table S2). Using the DAVID Pathway Viewer, a short list of significantly enriched pathways was obtained that strongly differed between the direct and the indirect strategies. In close coincidence with the enriched functional categories, the differentially expressed genes detected by the direct strategy in gastric tissues were enriched in four pathways related to cell interaction with the ECM: integrin signaling pathway, ECM- receptor interaction, Gap junction and Focal adhesion. Interestingly, the genes detected by the indirect strategy were enriched along two main pathways: PB kinase pathway and calcium signaling pathway that were not detected by the direct strategy (Supplementary Table S2).

We performed a similar analysis in colon cancer that, with the exception of certain groups such as regulation of cell adhesion, rendered data clearly different from that obtained with gastric cancer. The enriched functional categories obtained with the direct strategy in colon cancer involved mainly purine biosynthetic process, regulation of protein ubiquination, cell death and cell cycle (Supplementary Table

S2). On the other hand, the indirect strategy highlighted genes involved in mitosis, response to organic substance, protein transport and extracellular region. Furthermore, the colon cancer genes detected by the direct strategy were enriched in three pathways: cell cycle, de novo purine biosynthesis and DNA replication. In contrast, genes detected by the indirect strategy were enriched along two pathways:

Parkinson's disease and antigen processing and presentation (Supplementary Table S2). Thus, we found substantial differences in the biological functions and signaling pathways represented in the differentially expressed genes detected by both strategies indicating that the major pathways involved in these two cancer types appear essentially different (see below). Page 12 The indirect strategy highlighted cancer type-specific signaling pathways

We performed an additional pathway enrichment analysis with the PANTHER database using an

expanded list of differentially expressed genes obtained with the indirect strategy in gastric cancer. For this

aim, we selected genes with > 1.3 absolute fold change and p-value < 0.05. Following this criterion,

enrichment analysis of the 408 up-regulated and 335 down-regulated genes in gastric cancer highlighted

discrete intracellular signaling pathways. U sing a similar cutoff of p-value for enriched pathways, analysis

of colon cancer samples highlighted intracellular pathways that differed from those observed in gastric

cancer tissue (Table 1). The differences between gastric and colon cancer-enriched signaling pathways

were seen even at the level of single gene analysis; despite the fact that the integrin, apoptosis and

hedgehog signaling pathways were highlighted both in gastric and colon cancer, no single gene was

shared between both tumor types (Table 2).

For a more detailed analysis of the enriched pathways, we conducted gene expression analysis on

custom designed PCR arrays consisting ofgenes involved in the Wnt/Hedgehog, PBK/AKT, Angiogenesis

and BIT cell activation pathways. Four samples of gastric cancer and their paired adjacent non-cancerous tissue were used to assess mRNA expression levels of the different genes associated with the specific pathways. The relative fold change of each gene in the cancer tissue was expressed in relation to their paired adjacent tissue; only those genes with an average differential fold expression value >2.0 were included. The data demonstrate that most of the genes in the different pathways were overexpressed with only few genes showing down-regulated expression (Fig. 5A and D).

It was interesting to see an increased expression ofwnt/hedgehog pathway-associated genes such as

Wnt-1 induced secreted protein 1 (WJSPJ), protein patched homolog 1 (PTCH), e-terminal of E1A binding protein (CTBP 1) and secreted frizzled-related protein 4 (SFRP4); moreover, among all the family of FZD receptors we observed a remarkable expression of frizzled family receptor 5 (FZD5) and its ligand

Wnt5 (Fig. 5A). In the PBK pathway we observed a striking overexpression of insulin receptor substrate 2

(IRS2) and the down regulation of IRSJ and IRS4. We could also observe a clear overexpression ofprotein Page 13 kinases such as PIK3C2B, PIK3CD, PRKCA, PRKCG and AKTJ, the protooncogen ABL1 and the insulin-

like growth factors IGF1, IGF2 and epidermal growth factor receptor (EGFR) (Fig. 5B). Regarding

angiogenesis, the lymphoangiogenesis promoter vascular endothelial growth factor C (VEGF-C), the

interleukin 8 (IL8), the inhibitor of DNA bind 2 (ID2) and the neurophylin 1 (NRP 1) were notably

upregulated in gastric cancer samples compared to adjacent non-cancerous tissue (Fig. 5C). It was also

remarkable that the two genes that exhibited the largest levels of expression in the T cell and B cell

activation pathway in gastric cancer corresponded to two inflammatory cytokines such as IL17B and IFN-

gamma (Fig. 5D). Thus, the overall data identified only few specific genes that might be responsible for

leading the aberrant activity of each signaling pathway.

One of the overexpressed genes, CTBP1 was further analyzed since no previous evidence

associated its hyper-expression with gastric cancer. CTBP is a nuclear protein that associates with histone

deacetylases and binds to chromatin but may also function as a transcriptional co-repressor that interacts

with adenoviral E1A (18). In both cases it is involved in the regulation ofthe transcriptional status ofthe

cell. CTBP1 was found to be expressed by the six gastric and colon cancer celllines analyzed (Fig. 6A and

Supplementary Fig. S3). Targeting CTBP1 expression with a specific siRNA reduced mRNA and protein

levels in gastric cancer cell lines by almost 80% (Fig. 6B and C). Decreased expression of CTBP1

following transient siRNA expression in gastric cancer cells inhibited their clonogenic and migration

capacity (Fig. 6D andE).

In addition to the regulation of the transcriptional activity, CTBP1 has been shown to sensitize

certain malignant cell lines to the genotoxic effects of chemotherapeutic drugs such as 5-FU through

mechanisms associated either with apoptosis or modulation ofMDR1 levels (19). Therefore, we decided to target gastric cancer cells with the siRNA to CTBP1 followed by cells exposure to chemotherapeutic drugs

under current use in gastric cancer treatment. We observed a highly remarkable chemosensitizing effect when gastric cancer cells expressing reduced levels of CTBP1 dueto siRNA expression were treated with the different drugs. 5-FU had an IC50 of 29.4 ~M and 11.8 ~M in the presence of control siRNA in AGS Page 14 and MKN45 gastric cancer cells, respectively (Fig. 6F). Treatment with the specific anti-CTBP1 siRNA followed by the addition of 5-FU reduced significantly the IC50 to 5.6 flM and 1.9 flM in AGS and

MKN45 cells, respectively (Fig. 6F). The genotoxic agent cisplatin and the anthracycline epirubicin are also part of the mainstay treatment in gastric cancer. Interestingly, treatment of gastric cancer cells with cisplatin significantly reduced the IC50 of MKN45 cells from 10.7 J.!M to 2.5 flM and that of AGS cells from 11.2 flM to 2.6 J.!M while epirubicin treatment reduced the IC50 of AGS cells from 0.07 flM to 0.005 flM and that ofMKN45 from 0.06 flM to 0.005 flM (Fig. 6F).

Discussion

One of the main limitations of global gene expression analysis is the identification of the low abundan ce transcriptome. This is of paramount importan ce in cancer research sin ce subtle changes in gene activity of few genes could lead to malignant transformation and tumor dissemination. In this work we demonstrate that the combined use of subtractive hybridization followed by microarrays analysis identified novel genes that could serve as markers of the disease. Mostly important, these combined strategy led to the identification of specific intracellular signaling pathways, their leading genes and potential novel targets for improving treatment of advanced gastric and colon cancer. To our knowledge this is the first study that makes use of the combination of subtractive hybridization and microarrays to identizy the low abundance transcriptome in advanced human gastric and colon cancer.

Previous studies that combined suppressive subtractive hybridization with microarrays used the cDNA clones obtained after subtraction as probes printed in the array to screen malignant tissues such as breast (20-22), prostate (23), lung (12, 20, 24) and colon (25) cancer. These studies identified novel biomarkers and potentially new targets that were not identified by common gene expression analysis.

However, the main drawback was that the results were directly dependent on the signal level of the target sequences rather than the probes and the strategy was quiet ineffective for detecting genes with low expression levels. In this work we used the subtracted product as the target sequence, thus avoiding losing Page 15 the original molecular complexity of tissue samples and eliminating the need for further library

construction for the subsequent identification by sequencing of differentially expressed genes (10). Thus,

by using suppressive subtractive hybridization followed by microarrays analysis we were able to found

differentially expressed low abundance transcripts that would have not been detected as differential by

direct microarrays analysis.

Consistent with prevwus data (5-6, 26) direct transcriptome analysis identified genes with

biological functions mainly associated with cell-ECM interaction and immune response, and genes

associated with cell cycle in colon cancer. We validated by TMA two novel genes that were not previously

associated with gastric and colon cancer, RCC2 and JPH1, respectively. RCC2 role was associated with

fibronectin-dependent cell adhesion and cytokinesis (27-28). Consistent with the assigned roles for RCC2

we observed its expression both in the cytoplasm and nucleous of gastric cancer samples. Further

preliminary studies demonstrated that knockdown of RCC2 expression in AGS gastric cancer cells using a

specific siRNA led to almost 50% inhibition of tumor cell migration (Supplementary Fig. S4) while cell

proliferation and clonogenicity were not affected. The present data also showed that JPH1 Gunctophilin 1)

a gene associated with Ca2+ signaling, was overexpressed by more than 70% of colon cancer samples;

interestingly, JPH1 appeared in a profile of genes highly related to sensitivity to the bcr-abl tyrosine kinase

inhibitor dasatinib recently approved for the treatrnent of chronic myeloid leukemia (29).

The use of subtractive hybridization as a previous step identified a novel set of genes that shared

with the direct strategy only 3 genes in gastric cancer and 9 genes in colon cancer. Further analysis by

qRT-PCR ofthe differentially expressed genes after the previous subtraction step validated clase to 70% of

the differentially expressed genes demonstrating the robustness of the method. Moreover, TMA studies

validated the overexpression oflaminin a3 (LAMA3) one ofthe three subunits ofLaminin-322 (Ln-332).

There is no evidence of ln-332 involvement in human gastric cancer; however, more recent studies demonstrated that gastric cancer cell lines exhibit transcriptional silencing of LAMA3 due to promoter Page 16 methylation (30). Interestingly, the present data demonstrated more than 15-fold overexpression of

LAMA3 in gastric cancer samples compared to the adjacent non-malignant tissue while normal gastric

samples exhibited no expression. Moreover, LAMA3 staining was located in the cytoplasm of malignant

epithelial cells of gastric cancer samples showing no evidence of expression silencing. In addition, the

possibility that TTN might become a marker of a premalignant les ion warrants further investigation

In clear contrast to the direct microarray analysis, the enriched transcripts that appeared after

previous subtraction in gastric cancer were associated with pathways related to different signa!

transduction pathways and immune response. In addition, gene expression analysis of colon cancer

samples after previous subtractive hybridization highlighted signaling pathways different from those

observed in gastric cancer; both tumor types shared only the hedgehog, apoptosis and integrin signaling

pathway although no common gene was observed. Among the intracellular signaling pathways that we

selected for validation, we observed up to 50 genes out of 250 that exhibited at least 20-fold higher

expression in gastric cancer tumors compared to the adjacent non-cancerous tissue.

Interestingly, only 2, 10, 6 and 5 genes exhibited more than 100-fold overexpression in the wntlhedgehog, the PBK/Akt, angiogenesis and T/B cell activation intracellular pathways, respectively.

These genes can be considered as leading the aberrant activities of the enriched signaling pathways in gastric cancer. The largest overexpression levels in gastric cancer samples associated with the angiogenesis pathway was observed with VEGF-C. Recent studies have demonstrated that VEGF-C is associated with lymphatic spread and invasion of many cancer ce lis including gastric carcinoma (31-32). The present data confirmed that gastric adenocarcinomas showed elevated expression levels ofVEGF-C in 70% ofthe cases compared with the paired adjacent non-cancerous tissue (data not shown). However, we found no significant association between the expression levels of VEGF-C and the number of lymph nodes metastasis, most probably dueto the small amount of samples used in this study. Page 17 The largest differential expression between gastric cancer samples and non-cancerous tissue

associated with the PBK/AKT signaling pathway was observed with IRS2 in coincidence with the

simultaneous upregulation of IGF1, both IGF2 and its receptor IGF2R, and INS, all ofthem components

ofthe insulin/IGF signaling pathway. IRS2 is an intracellular signaling adaptor protein that binds to IGF-

IR and IR; thus, resulting in the activation ofPBK and the ERK pathway. Otherwise, IRS can interact with

other signaling pathways in a noncanonical manner and numerous cytokines and interferons have been

shown to induce downstream IRS signaling (33). Increased expression level of IRS-2 was associated with

lymph node metastasis in gastric cancer (34-35). IRS-2 plays a critica! role in determining the cellular

response to IGF stimulation anda recent report linked type 2 diabetes and incidence or mortality in gastric

cancer (36-37).

IL-17 and IFNy expression was associated with the activation of cytotoxic T and NK cells

pathways in response to Helicobacter pylori infection (38). Interleukin-17 is a CD4 T cell-derived

proinflammatory cytokine that plays a potential role in inflammatory response and has been shown to

stimulate the production of severa! other cytokines, including IL6, IL8 and TNF-a (39). In coincidence

with the present data, IL-17 expression has been observed in gastric cancer samples (40-41 ). IFNy is

another proinflammatory cytokine secreted by NK cells and CD8 T -cells that was associated with different tumor lineages (42-43).

Among the large family of FZD receptors, FZD5 exhibited almost 1,000-fold overexpression in gastric cancer samples. Interestingly, its ligand, Wnt5a, also exhibited the largest differential expression between cancer and adjacent non-cancerous samples. Wnt5a is in volved in the activation of the canonical and the non-canonical Wnt cascade in cells at tumor-stromal interface in invasion and metastasis (44).

Interestingly, increased IL-6 and TNFa levels due to Helicobacter Pylori infection upregulated Wnt5a levels in gastric cancer (45) and this overexpression has been correlated with advanced stages of the disease and poor prognosis (46). The largest overexpression in the Wnt pathway was associated with Page 18 WISPl. Although the expression of WISP1 was initially observed in wntl-overexpressing cells, and is

known at present as a wntfP-catenin effector, additional studies have demonstrated that its expression

could be explained by the coordinated effect ofwnt5a with antagonists ofthe canonical wnt pathway (47).

In order to functionally validate the PCR Arrays studies we decided to perform additional studies

with CTBP1 whose transcriptional activity was upregulated more than 70 fold in gastric cancer samples

compared with non-malignant adjacent tissue. We found that knocking down the expression of CTBP1

inhibited cell clonogenic and migration capacity that is consistent with CTBP1 role as mediator of

hypoxia-induced tumor cell migration (48). CTBP1 seems to contribute also to epithelial-mesenchymal

transition by repressing the expression of epithelial cell adhesion molecules (E-cadherin) and proapoptotic

genes (e.g. PERP, p21, Bax, Noxa) (49). However, the most interesting role ofCTBP1 is as sensitizer of

chemotherapeutic drugs (19). Here we show for the first time that knocking down the expression of

CTBP1 sensitized two different gastric cancer cell types AGS and MKN to three different

chemotherapeutic drugs, 5-FU, cisplatin and epirubicin decreasing their IC50 by 5-6 fold, 4-6 fold 12-14

fold respectively. A plausible explanation of the broad chemosensitizing effect of CTBP1 is the recent

evidence that CTBP1 can promote drug resistance by increasing the expression of MDR1 and in fact

downregulation of CTBP1 expression in breast cancer cells rendered malignant cells more sensitive to 5-

FU and other genotoxic agents such as cisplatin and etoposide (19, 50).

Molecular-targeted drug development has become an inflection point for cancer treatment.

However, even the most successful biological drugs are effective only in a minority of patients, sometimes even less that 10% of them can benefit from treatment. What has emerged from this highly sensitive microarray analysis is that we could identizy potential novel biomarkers and leading genes of signaling pathways aberrantly activated in these gastrointestinal carcinomas that might become novel targets. For advanced gastric cancer no effective drug has been developed so far; thus, it is tempting to speculate that this kind of affordable approach (subtractive hybridization + microarrays) might help identizy novel highly Page 19 specific targets that can provide the basis for increasingly specific targeted therapies and 1mprove

treatment efficacy.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Juan Carlos Roa MD, for tissue contribution and collaboration and Johannes Rainer, PhD

for CARMAweb support. We acknowledge the excellent technical support and help ofSoledad Lantadilla,

Gabriela Bascuñan, Tamara Sánchez, Javier Elorza, Germán González, Mónica Ramírez, Alejandra

Sandoval and Francisco Gamín.

Grant Support

This work was funded by PIA CTE-06, World Bank CONICYT Project. Carolina Bizama was

funded by PhD CONICYT Fellowship grant N°21070552 and CONICYT Internship grant N°28070019,

Felipe Benavente was funded by PhD PIA CTE-06 Fellowship and Jaime Espinoza was funded by PhD

CONICYT Fellowship grant No 21080240. Page 20 References

l. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J

Clin. 2011;61:69-90.

2. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of

cancer in 2008: GLOBOCAN 2008. Int J Cancer. 2010;127:2893-917.

3. Fomaro L, Lucchesi M, Caparello C, Vasile E, Caponi S, Ginocchi L, et al. Anti-HER agents in

gastric cancer: from bench to bedside. Nat Rev Gastroenterol Hepatol. 2011 ;8:369-83.

4. Jorgensen JT, Hersom M. HER2 as a Prognostic Marker in Gastric Cancer- A Systematic Analysis

ofData from the Literature. J Cancer. 2012;3:137-44.

5. Cui J, Chen Y, Chou WC, Sun L, Chen L, Suo J, et al. An integrated transcriptomic and

computational analysis for biomarker identification in gastric cancer. Nucleic Acids Res. 2011;39:1197-

207.

6. Yap YL, Zhang XW, Smith D, Soong R, Hill J. Molecular gene expression signature pattems for

gastric cancer diagnosis. Comput Biol Chem. 2007;31:275-87.

7. Cao W, Epstein C, Liu H, DeLoughery C, Ge N, Lin J, et al. Comparing gene discovery from

Affymetrix GeneChip microarrays and Clontech PCR-select cDNA subtraction: a case study. BMC

Genomics. 2004;5:26.

8. Diatchenko L, Lau YF, Campbell AP, Chenchik A, Moqadam F, Huang B, et al. Suppression

subtractive hybridization: a method for generating differentially regulated or tissue-specific cDNA probes

and libraries. Proc Natl Acad Sci U S A. 1996;93:6025-30.

9. Pan YS, Lee YS, Lee YL, Lee WC, Hsieh SY. Differentially profiling the low-expression

transcriptomes ofhuman hepatoma using a novel SSH/microarray approach. BMC Genomics. 2006;7:131.

10. Liu BH, Goh CH, Ooi LL, Hui KM. Identification ofunique and common low abundance tumour­

specific transcripts by suppression subtractive hybridization and oligonucleotide probe array analysis.

Oncogene. 2008;27:4128-36. Page 21 11. Liu Y, Zhu X, Zhu J, Liao S, Tang Q, Liu K, et al. Identification of differential expression of genes

in hepatocellular carcinoma by suppression subtractive hybridization combined cDNA microarray. Onco1

Rep. 2007;18:943-51.

12. Wang T, Hopkins D, Schmidt e, Silva S, Houghton R, Takita H, et al. Identification of genes

differentially over-expressed in lung squamous cell carcinoma using combination of cDNA subtraction and

microarray analysis. Oncogene. 2000; 19:1519-28.

13. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists

using DAVID bioinformatics resources. Nat Protoc. 2009;4:44-57.

14. Thomas PD, Kejariwal A, eampbell MJ, Mi H, Diemer K, Guo N, et al. PANTHER: a browsable

database of gene products organized by biological function, using curated protein family and subfamily

classification. Nucleic Acids Res. 2003;31:334-41.

15. Hamill KJ, Palier AS, Jones JC. Adhesion and migration, the diverse functions of the laminin

alpha3 subunit. Dermatol el in. 201 0;28:79-87.

16. Labeit D, Watanabe K, Witt e, Fujita H, Wu Y, Lahmers S, et al. ealcium-dependent molecular

spring elements in the giant protein titin. Proc Natl Acad Sci U S A. 2003;100:13716-21.

17. Pfohler e, Preuss KD, Tilgen W, Stark A, Regitz E, Fadle N, et al. Mitofilin and titin as target

antigens in melanoma-associated retinopathy. Int J eancer. 2007;120:788-95.

18. Boyd JM, Subramanian T, Schaeper U, La Regina M, Bayley S, ehinnadurai G. A region in the e- terminus of adenovirus 2/5 E1a protein is required for association with a cellular phosphoprotein and important for the negative modulation of T24-ras mediated transformation, tumorigenesis and metastasis.

EMBO J. 1993;12:469-78.

19. Birts eN, Harding R, Soosaipillai G, Halder T, Azim-Araghi A, Darley M, et al. Expression of etBP family protein isoforms in breast cancer and their role in chemoresistance. Biol eell. 2010;103:1-19. Page 22 20. Amatschek S, Koenig U, Auer H, Steinlein P, Pacher M, Gruenfelder A, et al. Tissue-wide

expression profiling using cDNA subtraction and microarrays to identifY tumor-specific genes. eancer

Res. 2004;64:844-56.

21. Barraclough DL, Sewart S, Rudland PS, Shoker BS, Sibson DR, Barraclough R, et al. Microarray

analysis of suppression subtracted hybridisation libraries identifies genes associated with breast cancer

progression. eell Oncol. 2010;32:87-99.

22. Yang GP, Ross DT, Kuang WW, Brown PO, Weigel RJ. eombining SSH and cDNA microarrays

for rapid identification of differentially expressed genes. Nucleic Acids Res. 1999;27: 1517-23.

23. Xu J, Stolk JA, Zhang X, Silva SJ, Houghton RL, Matsumura M, et al. Identification of

differentially expressed genes in human prostate cancer using subtraction and microarray. eancer Res.

2000;60: 1677-82.

24. Bangur es, Switzer A, Fan L, Marton MJ, Meyer MR, Wang T. Identification of genes over-

expressed in small cell lung carcinoma using suppression subtractive hybridization and cDNA microarray

expression analysis. Oncogene. 2002;21 :3 814-25.

25. ehen Y, Zhang YZ, Zhou ZG, Wang G, Yi ZN. Identification of differently expressed genes in

human colo rectal adenocarcinoma. World J Gastroenterol. 2006; 12:1025-32.

26. Bertucci F, Salas S, Eysteries S, Nasser V, Finetti P, Ginestier e, et al. Gene expression profiling of

colon cancer by DNA microarrays and correlation with histoclinical parameters. Oncogene. 2004;23:1377-

91.

27. Humphries ID, Byron A, Bass MD, eraig SE, Pinney JW, Knight D, et al. Proteomic analysis of integrin-associated complexes identifies Ree2 as a dual regulator of Rac1 and Arf6. Sci Signal.

2009;2:ra51.

28. Mollinari e, Reynaud e, Martineau-Thuillier S, Monier S, Kieffer S, Garin J, et al. The mammalian passenger protein TD-60 is an Ree1 family member with an essentia1 role in prometaphase to metaphase progression. Dev eell. 2003;5:295-307. Page 23 29. Huang F, Reeves K, Han X, Fairchild C, Platero S, Wong TW, et al. Identification of candidate

molecular markers predicting sensitivity in solid tumors to dasatinib: rationale for patient selection. Cancer

Res. 2007;67:2226-38.

30. Ii M, Yamamoto H, Taniguchi H, Adachi Y, Nakazawa M, Ohashi H, et al. Co-expression of

laminin beta3 and gamma2 chains and epigenetic inactivation oflaminin alpha3 chain in gastric cancer. Int

J Oncol. 2011;39:593-9.

31. Stacker SA, Achen MG, Jussila L, Baldwin ME, Alitalo K. Lymphangiogenesis and cancer

metastasis. Nat Rev Cancer. 2002;2:573-83.

32. Feng LZ, Zheng XY, Zhou LX, Fu B, Yu YW, Lu SC, et al. Correlation between expression of

S100A4 and VEGF-C, and lymph node metastasis and prognosis in gastric carcinoma. J Int Med Res.

2011 ;39: 1333-43.

33. Dearth RK, Cui X, Kim HJ, Hadsell DL, Lee AV. Oncogeni~ transformation by the signaling

adaptar proteins insulin receptor substrate (IRS)-1 and IRS-2. Cell Cycle. 2007;6:705-13.

34. Zuo QS, Yan R, Feng DX, Zhao R, Chen C, Jiang YM, et al. Loss of imprinting and abnormal

expression ofthe insulin-like growth factor 2 gene in gastric cancer. Mol Carcinog. 2011;50:390-6.

35. Lu Y, Lu P, Zhu Z, Xu H, Zhu X. Loss of imprinting of insulin-like growth factor 2 is associated

with increased risk of lymph node metastasis and gastric corpus cancer. J Exp Clin Cancer Res.

2009;28:125.

36. Gong Y, Yang YS, Zhang XM, Su M, Wang J, Han ID, et al. ABO blood type, diabetes and risk of

gastrointestinal cancer in northem China. World J Gastroenterol. 2012;18:563-9.

37. Tian T, Zhang LQ, Ma XH, Zhou JN, Shen J. Diabetes mellitus and incidence and mortality of gastric cancer: a meta-analysis. Exp Clin Endocrino! Diabetes. 2012;120:217-23.

38. Algood HM, Gallo-Romero J, Wilson KT, Peek RM, Jr., Cover TL. Host response to Helicobacter pylori infection befare initiation of the adaptive immune response. FEMS Immunol Med Microbio!.

2007;51 :577-86. Page 24 39. Jovanovic DV, Di Battista JA, Martel-Pelletier J, Jolicoeur FC, He Y, Zhang M, et al. IL-17

stimulates the production and expression ofproinflammatory cytokines, IL-beta and TNF-alpha, by human

macrophages. J Immunol. 1998;160:3513-21.

40. Zhang B, Rong G, Wei H, Zhang M, Bi J, Ma L, et al. The prevalence of Th17 cells in patients

with gastric cancer. Biochem Biophys Res Commun. 2008;374:533-7.

41. Iida T, Iwahashi M, Katsuda M, Ishida K, Nakamori M, Nakamura M, et al. Tumor-infiltrating

CD4+ Th 17 cells produce IL-17 in tumor microenvironment and promote tumor progression in human

gastric cancer. Oncol Rep. 2011;25:1271-7.

42. Dunn GP, Koebel CM, Schreiber RD. Interferons, immunity and cancer immunoediting. Nat Rev

Immunol. 2006;6:836-48.

43. Nishikawa H, Kato T, Tawara I, Ikeda H, Kuribayashi K, Allen PM, et al. IFN-gamma contro1s the

generation/activation of CD4+ CD25+ regulatory T cells in antitumor immune response. J Immunol.

2005; 175:4433-40.

44. Katoh M. Dysregulation of stem cell signaling network due to germline mutation, SNP,

Helicobacter pylori infection, epigenetic change and genetic alteration in gastric cancer. Cancer Biol Ther.

2007;6:832-9.

45. Katoh M. STAT3-induced WNT5A signaling loop in embryonic stem cells, adult normal tissues,

chronic persistent inflammation, rheumatoid arthritis and cancer (Review). Int J Mol Med. 2007;19:273-8.

46. Kurayoshi M, Oue N, Yamamoto H, Kishida M, Inoue A, Asahara T, et al. Expression of Wnt-5a is correlated with aggressiveness of gastric cancer by stimulating cell migration and invasion. Cancer Res.

2006;66: 10439-48.

47. Price RM, Tulsyan N, Dermody JJ, Schwalb M, Soteropoulos P, Castronuovo JJ, Jr. Gene expression after crush injury of human saphenous vein: using microarrays to define the transcriptional pro file. J Am Coll Surg. 2004; 199:411-8. Page 25 48. Zhang Q, Wang SY, Nottke AC, Rocheleau JV, Piston DW, Goodman RH. Redox sensor CtBP mediates hypoxia-induced tumor cell migration. Proc Natl Acad Sci U S A. 2006;103:9029-33.

49. Grooteclaes M, Deveraux Q, Hildebrand J, Zhang Q, Goodman RH, Frisch SM. C-terminal-binding protein corepresses epithelial and proapoptotic gene expression programs. Proc Natl Acad Sci U S A.

2003;100:4568-73.

50. Jin W, Scotto KW, Hait WN, Yang JM. Involvement ofCtBP1 in the transcriptional activation of the MDR1 gene in human multidrug resistant cancer cells. Biochem Pharmacol. 2007;74:851-9. Page 26 Figure legends

Figure l. Workflow procedures of the direct and the indirect strategies for the identification of

differentially expressed transcripts. Total RNA obtained from gastric and colon samples were processed

through a direct strategy, which included cDNA synthesis, aRNA amplification, microarray hybridization

and a indirect strategy, that included a previous PCR-based suppressive subtractive hybridization. Data

was separately obtained from each strategy and analyzed as described in experimental procedures.

Figure 2. The use of a previous subtractive hybridization step led to the identification of novel cancer­

associated transcripts by microarrays analysis. A-B, Venn diagrams showing overlapping genes obtained

by both strategies for the two types of cancer studied. C-F, quantitative real- time PCR (qRT-PCR)

analysis was performed on a selected list of differentially expressed genes in gastric and colon cancer

obtained by the direct and indirect strategies. mRNA levels were assessed in six samples of gastric and

colon cancer samples and their respective paired non-cancer tissue and results were normalized using three

housekeeping genes. Red color represent up-regulation and green color represent down-regulation. qRT­

PCR significance values: * P <0.05, ** P <0.01, *** P <0.001. ns, non significant by Student's t test.

Figure 3. Tissue microarray analyses confirmed the augmented expression of RCC2 and JPH1 in gastric

and colon cancer, respectively. A-C, RCC2 immunohistochemical staining in gastric cancer (A), adjacent

non-cancerous sample (B) and normal sample (C). D, quantification of RCC2 staining in gastric cancer

and adjacent non-cancerous tissue samples. Mean± SD (n=39). *** P < 0.001 by Student's t test.

E-G, JPHI immunohistochemical staining of JPH1 in colon cancer (E), non-cancer sample (F) and normal

colon tissue (G). H, quantification of JPH1 staining in colon cancer and adjacent non-cancerous counterpart. Data are expressed as mean± SD (n=46). *** P < 0.001 by Student's t test. Scale bars, 50 ¡..tm. Page 27 Figure 4. Tissue microarray analyses confirmed the augmented expression of LAMA3 and TTN in gastric

cancer. A and B, LAMA3 immunohistochemical staining in gastric cancer (A) and adjacent non-cancerous

sample (B). e and D, quantification of LAMA3 expression in gastric cancer and adjacent non-cancerous

samples. D, data are expressed as mean± SD (n=37), *** P < 0.001 by Student's t test.

E-1, TlN immunohistochemical staining in gastric cancer (E) adjacent non-cancerous sample (F) and

intestinal metaplasia (I). G and H, quantification of TlN expression in gastric cancer and adjacent non-

cancerous counterpart. H, data are expressed as mean± SD (n=35). * P < 0.05. 1, the red arrows show the

strong staining ofTlN in intestinal metaplasia. Scale bars, 100 Jlm.

Figure 5. PeR-Array analysis highlights overexpressed genes in the different pathways. A-D, the

expression profiles of genes relevant to the Wnt/Hedgehog (A); PBK/AKT (B); Angiogenesis (C); T and

B cell activation (D). Each PeR-Array contained 87 genes relevant to each pathway as well as 9

housekeeping genes and the expression values represent the average fold change of gastric cancer samples

relative to non-cancerous tissues. The data correspond to genes showing fold changes > 2.0.

Figure 6. Knockdown of eTBP1 expression using a siRNA sensitizes gastric cancer cells to

chemotherapeutic drugs. A, relative expression of eTBP1 in six gastric cancer cell lines (SNU16, AGS,

N87, SNU1, MKN45, Kato 111). Quantification of eTBP1 was performed by qRT-PeR using QARS and

TFeP2 as an interna! control. B and e, validation of eTBP1 knockdown by assessing mRNA and protein levels. AGS and MKN45 gastric cancer cells were transfected with eTBP1 and control siRNA. QARS and

TFeP2 were used as an interna! control, while a-tubulin was used as an interna! control for protein loading. D and E, MKN45 following pre-incubation for 48 hr with eTBP1 and control siRNA. D, migration analysis. Representative photographs of Giemsa-stained cells are shown. E, clonogenic capacity following pre-incubation for 48 hr with an siRNA against eTBP1. Representative photographs of crystal violet-stained colonies are shown. Data are expressed as mean ± SD (n=3). ** P <0.01, * P <0.05 by Page 28 Student's t test. F, gastric cancer cells were transfected with a siRNA against CTBPl. After 48 hours, cells were treated with varying concentration of the different chemotherapeutic agents for an additional 72 hours. Cell viability was evaluated with MTS. Data is expressed as mean± SD (n=3). 12 gastric cancer and 12 adjacent 10 colon cancer and 10 adjacent non-cancerous samples non-cancerous samples

• Total RNA extraction and i ntegrity analysis

Direct strategy

------.. 1·--·---·-··--·---~·-·---~~ • First strand c.DNA synthesis Superscript:IIJ cONA synthesis • cONA amplification S MART technology and [ • RSA 1Digestion amplífication

~------~------DNA 1 • First strand~DNA synthesís by .,) 1thesís Superscript 111 laRNA • Second strand synthesis lification • In vifro transcription and aRNA labelmg Suppressíon Subtractive 3ene • Hybrídizatíon on HEEBO mícroarray Hybrí dization

• In vitro transcription and aRNA labeling Gene • Hybridization in HEEBO microarray expressíon r• Scanning and data extraction ---- microarray • Bloinformatic analysls • Data validation with qRT-PCR and • Gene ontology ~------:r A Gastric cancer B Colon cancer

Direct lndirect Direct lndirect strategy strategy strategy strategy 8" E vtJ. <1$: ** ADAT2 ns TOP2A * CTSC * ILSBP ns PPAT * TGFB1 * SH3TC1 ns CLDN1 * PGM3 * * MACC1 IL8 * MMP7 ns * SYNE2 ** PPM1H * CSE1L * ** * ns ANX13 * AXIN2 * FAM498 ** ns * SRGAP1 * MMP7 ns ZNF410 ns TMPRSS4 * CKS2 ** NXT2 * * SAM09 ns CSE1L ** XRN2 * * GPR68 ns XPOS * KPNA2 * ns STAP2 ** USPL1 * FLNB * * TRIM26 * TARBP1 ** SUT1A1 * * IGJ * TGFB1 ** NFKB1A * ** FOXA1 ** JPH1 * CA12 * * AKR1C3 * GTPBP4 *** HLA-A ns * MAST3 ns RACGAP1 * ZNF292 ns ** XYLT2 * XRN2 * FUCA1 * * AGR2 * TOMM34 ** SELENBP1 * ns CD209 * IARS * CA2 * * GKN2 * NIT2 * ** COX7A1 * * PADI2 * Log2 Fold change LMOD1 ns Log2 Fold change ** SLC02A1 * *** CA12 -3 o 3 * -3 o 3 ** SULT1A1 * USP2 * CA4 ns PKIB * SCIN * RCG2 Jt-JH1

U) o:S S.. G) u e um

U) e:S G) (.) e m u e o z

...... m E oS.. z

e: *** ~.... c. *** No o oIX: ~ o u 111 en .5 e: "ñi -U)

Non-cancerous Cancerous Non-cancerous Cancerous c::J Non·canearous • Cancerous

Non~anc:ero\11 Cancerous c::J Non·caneetous • Cancero11s

Non-canc:ero111 Cancero11s Wnt/Hedgehog 8 Pl3KIAKT e Angiogenesis D T cell and B cell activation IIP1 IRS4 U:B CBL CCLZ ~LK KIT PTGS1 <3A CAP1 ADAMTS1 IYC BTK EFNA1 IRS1 KOR T9A CXCR4 FIGF 2A1 LCK CXCL6 ~H1 TGFBR1 PIKJCA SERPINF1 '300 SRC ANGPT1 1A1 SHC1 NRP2 IOU CDKN1B TNFAIP2 X17 E2F1 COL4A2 Z07 IFNA1 ERBB2 TIMP2 JFU NFKB1 FGF1 T'2B ACACA TIMP3 823 HRAS SPHK1 ¡LI2 CREB1 FGF2 ENG IG1 GRB2 PROK2 EF1 INS PDGFA 1L1 AEBP1 ANGPTL4 Z08 PIK3R1 PLXOC1 PTEN IL6 Z02 JAGt :o1 CD40 BAI1 'C1 NFKBIA FLTt ~T3 RAF1 CXCL11 tOZ COKN2A ITGAV JAK2 EFNB2 m1 NOTCH1 tL4 IGF2 TGFB2 :et ITGB1 PGF ANG PLAU :W2 CCR5 ITGB3 ~06 IGF2R MDK SL1 MAPK1 CDHS IC2 LAMAS PIK3R4 IFNG IR1 PRKCE TGFB lOA SPP1 ECGFt ws RHOA CXCL9 (86 ACOX1 MMP9 ~P2 SOS1 TEK EFNA3 rsA ABL1 LTA (N1 COKN1A 1 TNF 181 GSK3B 1 THBS2 ¡Qf COH1 TIMPt CDC42 • EDG1 102 MMP2 103 PRKCA • ANPEP rc1 AKT1 • PEF4 iHB EGFR • COL18A1 MAPK8 IU !P4 THBS1 :H1 PIK3C2B PRKCG NRPt !P1 - 1011000 120000 102 IGF1 - ~os VEGFC ¡pf IRS2 - ~~ -20 o 20 40 250 500 7501000 -10 o 10 20 30 ·20 20 6() 200- 600 1000 20 40 &O 80 2000 Fold Change Exprenion Values Fold Change Expression Valuu Fold Change Expression Values Fold Changtl Expression Yalues 8~1.0 o 100

...~ 0.8 a. - 80 ....al t:. "* (,) 0.6 ! 60 .. T 'O u Cll e .5!"" o.4 40 es T"'" ~l! 8 ..E f ~0.2"' 20 j .!! ~ O.QI.LL-¡.....-'-.L-.....,..-L.. o s•RNA control CTBPf siRNA c:ontrol CTBP1

CTBP1 ,_ CTBP1,.. __., ~~ .....~ .-~: , ...... ,.; . : ••, .... a-Tubulinl- -lo-Tubulinl- _, ).:_ ;~_: :· ,\;~1· siR NA siR NA CTBP1 CTBP1 Control Control 5-Fiuoracil Cisplatin Epirubicin

120 ICSO (¡o M) 120 IC50(¡oM) 120 IC50(¡oM) WT 3UI Á WT 13 8 4WT 0.09 a siRNA control 14.9 100 siRNAcontrol29.4 100 100 a siRNA control 0.07 o CTBP1 5.6 G CTBP1 2.6 o CTBP1 0.005 80 80 80 \ g g \T\J¡ 60 aso g 60 (,) (,) ~ ~ ~\ 4(1 40 40 \ ;;:,\ 20 20 20 \ );_ ....._,~~ o o 1! • o 3 4 ~ 6 7 8 9 10 4 5 6 7 8 9 10 2 3 4 5 6 7 8 Log (M]5-FU Log [MJ Cisplatln Log [M) Epirubicin

120 120 IC50(¡tM) 120 IC50{¡tM) ICSO {!.o M¡ 4WT 0.09 AWT 12.1 Á WT 10.7 T a siRNAcontrol 0.07 100 a siRNA control11 .6 100 a siRNA control11.2 100 o CTBP1 1.9 o CTBP1 2.6 "\! o CTBP1 0.005 80 80 80 \ g g \T 60 5 60 f$6 ~ u u ~ ~ 40 40 -\T 40 \ '¡ ') 20 20 1\¡ 20 o o '-1._ o ~--- 3 4 5 6 7 8 9 10 4 5 6 7 8 9 10 2 3 4 S 6 7 8 Log [M}5-FU Log [M] Clsplalln Log [M) Epirubicln Table 1. Subtractive hybridization followed by microarray analysis identified mostly non-overlapping intracellular pathways in gastric and colon cancer Gastric cancer principal pathways Number of genes P- value P00012:Cadherin signaling pathway 11 2.18E-03 P00056:VEGF signaling pathway 7 4.23E-03 P0001 0:8 cell activation 7 6.76E-03 P00036:1nterleukin signaling pathway 10 1.17E-02 P00006:Apoptosis signaling pathway 8 1.79E-02 P00053:T cell activation 7 2.01E-02 P00046:0xidative stress response 5 2.27E-02 P00034:1ntegrin signaling pathway 10 2.40E-02 P00054:Toll receptor signaling pathway 5 2.56E-02 P00025:Hedgehog signaling pathway 2 2.94E-02 P00026:Heterotrimeric G-protein signaling pathway-Gi alpha and Gs alpha 3 3.41 E-02 mediated pathway P00048:PI3 kinase pathway 7 3.50E-02 P00057:Wnt signaling pathway 14 4.52E-02 Colon cancer principal pathways Number of genes P- value P00017:DNA replication 4 3.11 E-03 P00016:Cytoskeletal regulation by Rho GTPase 4 7.14E-03 P00004:Aizheimer disease-presenilin pathway 9 8.37E-03 P00034:1ntegrin signaling pathway 11 1.46E-02 P002738:De novo purine biosynthesis 4 1.61 E-02 P00029:Huntington disease 10 2.12E-02 P00006:Apoptosis signaling pathway 8 2.44E-02 P00025:Hedgehog signaling pathway 2 3.41E-02 P00024:Giycolysis 3 3.41 E-02 P00059:P53 pathway 7 4.22E-02 Principal pathways were identified by PANTHER analysis using 743 and 651 genes differentially expressed in gastric and colon cancers. Table 2. Single gene analysis of common pathways enriched in gastric and colon cancers Pathway Gastric genes Colon genes

lntegrin signaling COL8A 1*, CUZD1, TRIM23, c14orf133, COL5A2, COL1A1, ITGB8, RND3, ITGAS, ARPC2, RAPGEF1, ITGAX, CDSL, FCRL3, SLC40A 1, ARPC3,GGT6, RAP2B, HSP90AB1, COL 12A1 RAPGEF1, LIMS1

Apoptosis signaling LARP2, ADAT2, IMPDH1, BAX, BIRC3, HSPA8, XIAP, TNFRSF10D, NFKBIA, HIST1H1E, CHUK, NFKB2, ATF2 CASP7, CCNB11P1, TNFSF10

Hedgehog signaling STK36, FBXW11 SAP, PROZ

* L1st of genes 1ncluded 1n the ennched pathway. Pnnc1pal pathways were 1dent1fied by PANTHER analys1s using 743 and 651 genes differentially expressed in gastric and colon cancers. ANEXO V

ARTÍCULO PUBLICADO (Co-autoría)

Reversa/ of gastrointestinal carcinoma-induced immunosuppression and induction of antítumoural immunity by a combination of cyc/ophosphamide and gene transfer of/L-12

263 l'v!OLECULAR ONCOLOGY XXX (2011) 1-14

available at www.sciencedirect.com _,, • i "'i .. ... =..,_•: ~~ • o • ·;;, ScienceDirect • • • www .e lsevi e r .e o m/1 o cate/m o 1o n e Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12

Mariana Malvicinia, Mariana Ingolottia, Flauia Piccionia, Mariana Garciaa,b, Juan Bayoa, Catalina Atorrasagastia, Laura Alaniza,b, jorge B. Aquinoa,b, Jaime A. Espinozad,e, Manuel Gidekeld, O. Graciela Scharovskyc, Pablo Matarb,c,**, Guillermo Mazzolinia,b,* aGene Therapy Laboratory, Liver Unit, School of Medicine, Austral University, Av. Presidente Perón 1500, (B16290DT) Derqui-Pilar, Buenos Aires, Argentina bCONICET (Consejo Nacional de Investigaciones Científicas y Técnicas), Argentina cinstitute of Experimental Genetics, School of Medica! Sciences, National University of Rosario, Santa Fe 3100, 2000 Rosario, Argentina ctventureLab, Adolfo Ibáñez University, Santiago, Chile eApplied Cellular and Molecular Biology Program, De la Frontera University, Temuco, Chile

ARTICLE INFO ABSTRACT

Article history: Immunotherapy-based strategies for gastrointestinal carcinomas (GIC) have been exploited Received 30 December 2010 so far, but these approaches haveto fa ce strong mechanisms of immune escape induced by Received in revised form tumours. We previously demonstrated that sub-therapeutic doses of an adenovirus ex­ 29 March 2011 pressing IL-12 genes (AdiL-12) mediated a potent antitumour effect against subcutaneous Accepted 30 March 2011 (s.c.) colorectal carcinomas (CRC) in mice pre-treated with low doses of cyclophosphamide Available online • (Cy). In our study we used this combination to assess its impact on the immunosuppressive microenvironment. In s.c. CRC model we demonstrated that non-responder mice failed to Keywords: decrease Tregs in tumour, spleen and peripheral blood. Reconstitution of Tregs into Gastrointestinal carcinoma tumour-bearing mice treated with combined therapy abolished the antitumoural effect. Immunosuppression In addition, Cy + AdiL-12 modified Tregs functionality by inhibiting the in vitro secretion Cyclophosphamide of IL-10 and TGF-~ and their ability to inhibit dendritic cells activation. Combined treat­ IL-12 ment decreased the number of myeloid-derived suppressor cells (MDSCs) in comparison Tregs to non-treated mice and, interestingly, administration of Tregs restored splenic MDSCs population. Furthermore, combined therapy potently generated specific cytotoxic IFN-y­ secreting CD4+ T cells able to eradicate established CRC tumours after adoptive transfer. Finally, we evaluated the combination on disseminated CRC and pancreatic carcinoma (PC). Cy + AdiL-12 were able to eradicate liver metastatic CRC (47%) and PC tumour nodules (40%) and to prolong animal survival. The results of this study support the hypothesis that Cy + AdiL-12 might be a valid immunotherapeutic strategy for advanced GIC. © 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. Al! rights reserved.

• Corresponding author. Gene Therapy Laboratory, Liver Unit, School of Medicine, Austral University. Av. Presidente Perón 1500, (B16290DT) Derqui-Pilar, Buenos Aires, Argentina. Tel.: +54 2322 482618. •• Corresponding author. Institute of Experimental Genetics, School of Medica! Sciences, National University of Rosario, Santa Fe 3100, 2000 Rosario, Argentina. Tel.: +54 341 4804558x244; fax: +54 341 4804569. E-mail addresses: [email protected] (P. Matar), [email protected] (G. Mazzolini). 1574-7891/$ - see front matter © 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. Al! rights reserved. doi:10.1016/j .molonc.2011.03.007

Please cite this article in press as: Malvicini, M. et al., Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j.molonc.2011.03.007 2 MOLECULAR ONCOLOGY XXX (2011) 1-14

l. Introduction this cytokine with other procedures, such as chemotherapy, immunostimulatory monoclonal antibodies, dendritic cells, Potent mechanisms of immunosuppression are active during and other immunotherapy approaches (Gomez et al., 2001; tumour growth limiting the effectiveness of immunotherapy Melero et al., 2001). Cyclophosphamide (Cy) is a widely known strategies (Zou, 2005). Mechanisms used by malignant chemotherapeutic agent that displays either immunosup­ tumours to elude immune recognition include tumour­ pressive or immunopotentiating effects, depending on the induced impairment of antigen presentation, activation of dosage and the duration of administration of this drug negative co-stimulatory signals and secretion ofimmunosup­ (Brodsky et al., 1998; Colvin, 1999). Severa! mechanisms have pressive factors (Almand et al., 2001; Bell et al., 1999; been proposed to explain Cy immunomodulatory activity in­ Ghiringhelli et al., 2005; Nestle et al., 1997). In addition, cancer cluding depletion of tumour-induced regulatory T cells, pro­ cells may induce the expansion anci/or recruitment of regula­ duction of soluble growth factors, Th2/Th1 shift in cytokine tory cell populations that may promete and sustain this im­ production, and others (Ghiringhelli et al., 2004; Matar et al., munosuppressive network; these populations include 2002; Proietti et al., 1998). However, the precise mechanism regulatory T cells (Tregs), myeloid-derived suppressor cells is not yet clarified. (MDSCs) and distinct subsets of immature and mature regula­ Recently, we have shown a novel synergistic combination tory dendritic cells (DCs) (Croci et al., 2007). of Cy and AdiL-12 for the eradication of CRC in mice. Cur­ Colorectal carcinoma (CRC) is the second most common rently, we explored the mechanisms of action between Cy cause of cancer mortality in westem countries and the 9.7% and AdiL-12 combination and observed that the potent anti­ of the most commonly diagnosed cancers worldwide (Feria y cancer effect obtained in advanced GIC is based on their capa­ et al., 2010). Twenty five percent of the patients have meta­ bility to block Tregs secretion ofiL-10 and TGF-13 cytokines, as static disease at diagnosis, and their predicted 5-year survival well as on the loss of their inhibitory activity exerted on DCs. among non-surgical patients is less than 10% (Lorenz et al., In addition, we observe a decrease in the number ofMDSCs af­ 2000). Pancreatic cancer is the fourth leading cause of ter Tregs inhibition. Finally, combined treatment showed syn­ cancer-related death and surgery is the only curative option; ergistic activity not only at the induction of the immune however, more than 80% of the patients showed local or dis­ response but also at the effector phase by generating potent seminated recurrence (Lorenz et al., 2000). Therefore, there IFN-y-secreting CD4+ T cells that were able to eradicate estab­ is an urgent need for new therapeutic strategies for advanced lished CRC tumour nodules. GIC. During the last two decades multiple immunotherapy­ based strategies for GIC have been developed with promising results not only at preclinical stage but also in the clinic 2. Materials and methods (Elkord et al., 2008; Shapira et al., 2010). The loss/down regula­ tion of MHC class I antigens, the lack of co-stimulatory mole­ 2.1. Animals and cell Iines cules, detective death receptor signalling, apoptosis of activated T cells, immunosuppressive cytokines, and activa­ Six- to eight-week-old male BALB/c mice and C57BU6 mice tion of suppressor T cells have been described as responsible were purchased from CNEA (National Atomic Energy Commis­ for the failure of anticancer treatment in mice and humans sion, Ezeiza, Argentina). Animals were maintained at our An­ (Mazzolini et al., 2007). imal Resource Facilities (School of Biomedical Sciences, Interleukin 12 (IL-12), a cytokine with multiple biological Austral University) in accordance with the experimental ethi­ effects, is one of the most potent antitumour cytokines cal committee and the NIH guidelines on the ethical use of an­ (Mazzolini et al., 2003; Trinchieri, 2003). However, high doses imals. The CT26 (colorectal carcinoma), BNL (hepatocellular of IL-12 are needed to achieve significan! antitumour effects carcinoma) and Panc02 (pancreatic carcinoma) tumour cell that results in severe toxicity (Leonard et al., 1997). In recent lines were used (kindly provided by Prof. Prieto, University of years, gene therapy has emerged as a therapeutic too! to allow Navarra, Spain). Cells were maintained in DMEM or RPMI desired levels ofiL-12 or other cytokines into tumoural/peritu­ 1640 supplemented with 10% heat-inactivated FCS, 2 mmol/ moural milieu avoiding undesired side effects (Colombo and L L-glutamine, 100 U/mL streptomycin, and 100 mg/mL peni­ Fomi, 1994). The efficacy of IL-12 gene transfer for advanced cillin and incubated at 37 oc in a 5% co2 humidified GIC in animal models has been shown consistently by differ­ atmosphere. ent research groups including ours (Barajas et al., 2001; Caruso et al., 1996; Mazzolini et al., 1999; Putzer et al., 2001). Intratumoural administration of an adenovirus expressing 2.2. Drugs IL-12 genes (AdiL-12) in patients with advanced GIC in a phase Itria! was a feasible and well-tolerated procedure that exerted Cyclophosphamide (Filaxis, Argentina) was dissolved in ster­ only mild antitumoural activity (Sangro et al., 2004). Immuno­ ile water at a concentration of 20 mg/mL and injected i.p. at suppressive activity induced by tumoural and peritumoural the indicated dose. cells seem to be critica! and might be responsible for the lack of clinical success in immune- and non immune-based 2.3. Adenouiral uectors therapies (Rabinovich et al., 2007; Zwimer et al., 2010). Seeking for immunotherapeutic synergies to fight against cancer, it Construction of a recombinant adenovirus encoding for IL-12 was possible to increase the efficacy of IL-12, by combining (AdiL-12) was previously described (Mazzolini et al., 1999). r~dli! · · ~ ·· · -- ~· · 81. Reversalof · testiDa1 · induced · · l ancr MOLECULAR ONCOLOGY XXX (2011) 1-14 3

2.4. In vivo experiments obtained bymechanical disruption and then treated with RBC ly­ sis buffer (0.15 mol/L NH4Cl, 1 mmol/L KHC03, 0.1 mmol/L Na2- 2.4.1. Subcutaneous CRC model EDT A) and washed with PBS 1% bovine serum alburnin, before CT26 cells were injected at a dos e of 5 x 105 cells s.c. into the flow cytometry analysis. For MDSCs analysis, tumour-bearing right flank of BALB/c rnice (day 0). Tumours were allowed to animals were distributed into different groups and treated reach approximately 85 mm3 in size befare treatment was with saline, Cy, AdiL-12 or Cy + AdiL-12. The mice were sacri­ started. Animals were distributed in different groups and ficed on day 15, spleens were excised and single cell suspensions then treated with saline i.p.; Cy (50 mg/kg i.p., day 8); AdiL- were prepared for flow cytometry analysis as described below. 12 (109 TCID50 intratumourally (i.t.), day 9); Cy (50 mg/kg MDSCs were phenotypically characterized as CD11b+ Gr1+. i.p.) + AdiL-12 (109 TCID50 i.t.). Adenovirus were diluted in sa­ line (final volume 50 11L) and i.t. injected in a single site; no 2.6. Generation of bone marrow-derived DCs leakage of material was observed after inoculations. DCs were generated from BALB/c rnice as described elsewhere 2.4.2. Adoptive transfer of CD4+CD25+ T cells (Tregs) (Alaniz et al., 2009). Flow cytometric analysis for CD11c and CT26 tumour-bearing rnice (approximately 85 mm3 in size) the capacity to produce IL-12 after LPS activation was per­ were i.p. treated with saline or Cy (50 mglkg i.p, day 8) + AdiL- formed to certify the presence and functional status of gener­ 12 (109 TCID50 i.t, day 9). Forty-eight and 96 h later, rnice were ated DCs at the end of procedure. injected i.v with 1 x 106 CD4+CD25+ T lymphocytes isolated by magnetic cell sorting from spleen of non-treated tumour­ 2.7. Cell purification by magnetic cell sorting bearing rnice excised at day 14 of tumour growth evolution. Spleen cell suspensions were obtained by mechanical disrup­ 2.4.3. Adoptive transfer of CD4+CD25- T cells tion of spleen from CT26-bearing mice treated with saline, Cy, AdiL-12 or Cy + AdiL-12. CD4+CD25+ and CD4+CD25- cells were purified by magnetic cell sorting using a mouse T regula­ 2.4.3.1. Preventive model. BALB/c rnice were simultaneously tory cell isolation kit and LD plus MS columns according to the inoculated with 5 x 105 CT26 cells s.c. into the right flank, and manufacturers' instructions (Miltenyi Biotec, Auburn, CA, with 2.5 x 106 CD4+CD25- cells i.v., isolated by magnetic cell USA). CD4+CD25+ cells (Tregs) obtained were >95% in purity sorting and in vitro re-stimulated, from spleen of saline, Cy, by flow cytometric analysis of intracellular Foxp3. AdiL-12 or Cy + AdiL-12-treated mice (see Section 2.13 below). 2.8. Expansion of CD4+CD25+ T cells and ca-culture 2.4.3.2. Therapeutic model. CT26 tumour-bearing mice (appr­ with DCs oximately 65 mm3 in size) were adoptively transferred with 2.5 x 106 CD4+CD25- cells i.v., isolated by magnetic cell sort­ Purifi.ed CD4+CD25+ T cells were cultured for 48h with rmiL-2 ing, from saline, Cy, AdiL-12 or Cy + AdiL-12-treated rnice. (10 UI!ml) (Peprotech, Rocky Hill, NJ, USA). Supematants were collected and used for allogeneic splenocytes proliferation as­ 2.4.4. Pancreatic tumour modei say and cytokines quantification. Tregs and DCs (ratio = 1:1) C57BU6 mice were inoculated in the right flank with 5 x 105 were co-cultured for 24 h and then LPS (Sigma Chernicals, St Panc02 cells s.c. When tumours reached 85 mm3 in size, ani­ Louis, MO, USA) was added (1 11g/ml) for additional 24 h. mals were treated with saline, Cy (50 mglkg i.p., day 8), AdiL-12 (109 TCID50 i.t., day 9) or Cy + AdiL-12. Tumour 2.9. Celllabelling growth was assessed twice a week by calliper. Tregs were labelled with the fluorescent dye Fast DiO (Molec­ 2.4.5. Liver metastatic CRC model ular Probes, Eugene, USA) according to the manufacturer in­ BALB/c mice received an intra-hepatic inoculation of 5 x 105 structions. Briefly, cells were incubated with 3 mM Fast DiO CT26 cells (da y 0). Then, rnice were distributed in experimen­ cell solution for 5 rnin at 37 oc in 5% COThumidifi.ed atmo­ tal groups and treated with saline; Cy (50 mglkg i.p., day 8); sphere and for 15 rnin at 4 oc and then washed with PBS. After AdiL-12 (109 TCID50 i.t., day 9) or Cy + AdiL-12. At day 20, an­ labelling Fast DiO-stained Tregs were diluted in saline and i.v. imals were sacrifi.ced and the volume of metastatic nodules injected into mice. Five days later, peripheral blood was col­ were measured with calliper. lected and rnice were sacrifi.ced. Spleen, liver, tumour and re­ gionallymph nodes were excised from each animal and single 2.5. Sample preparationfor Tregs and MDSCs cytometry cell suspensions were obtained by digestion with collagenase I analysis (Calbichem®, Merck KGaA, Darmstadt, Germany). The cell suspension was treated with RBC lysis buffer, washed with BALB/c rnice were injected with S x 105 CT26 cells s.c. into the PBS 1% bovine serum alburnin, fi.xed in 1% paraformaldehyde right flank (day O) and tumours were allowed to reach 85 mm3 and then analysed by flow cytometry (FACSAria, BD). befare the beginning of the treatment. For Tregs analysis, ani­ mals were distributed and treated with saline or Cy + AdiL-12. 2.10. Flow cytometry At day 15 and 22 (7 and 14 days post-treatment), peripheral blood was collected by cardiac puncture and anticoagulated with Peripheral blood mononuclear cells, tumour cells or spleno­ EDT A. Single splenic and tumoural cell suspensions were cytes were stained with the following conjugated antibodies: "Please cite tl:ús article"in presa as: Yalvic:ini,JL etal., Reversal of gastrointestinal can:inoma-induced immunosuppression and indncrion of antituíDomal immunit;y by a combination of cyclophosphamide and gene transfer of D.-12, Molecular Oncology ,.(2.011), ti10.1016/j.molori.c.2011.03.007 · ·· · · · 4 MOLECULAR ONCOLOGY XXX (2011) 1-14

phycoeritrin-anti-Foxp3 (eBioscience), PEeyS-anti-eD4 (BD a synergistic antitumoural effect on s.c. CRC model (CT26) Biosciences), allophycocyanin-anti-eDllb (BD Biosciences), (Malvicini et al., 2009). Two well defined groups were found in PE-anti-Grl (BD Bioscience) and their respective isotypes. Cy + AdiL-12-treated rnice. The two categories were identified Des were stained with APe-anti-eDllc (kindly provided by as responders (R) and non-responders (NR) mice, depending on Dr. Gabriel Rabinovich, IByME, Argentina), PE-anti-MHeii (BD whether they showed a >50% in tumour size reduction or not. biosciences), FITe- anti-eD40 (BD biosciences) and their re­ Post-treatrnent tumour sizes in NR mice at days 7 and 14 post­ spective isotypes. eel!s were fixed with 1% paraformaldehyde treatrnent were significantly higher (566 ± 143 and 3 and subjected to flow cytometry (FAeSAria, BD). Data were 614 ± 92 mm , respectively) than in R mice (58 ± 19 and 3 analysed using WinMDI software. 126 ± 58 mm , respectively) (p < 0.01) (Fig. lA). To determine whetherimmunosuppressive T cell-mediated mechanisms could 2.11. eytokine quantificationby ELISA be involved in the observed differential antitumoural response obtained after treatrnent with Cy + AdiL-12, we first analysed Transforrning growth factor-¡) (TGF-¡3), IL-10 and IL-12 concen­ the prevalence of Tregs in peripheral blood and spleens using trations in culture supernatants were measured by specific triple-staining flow cytometry. While the levels of CD4+ T cel!s ELISA kits (OptEIA, BD Biosciences Pharrningen for IL-10 and were similar for either untreated mice or treated with the combi­ IL-12 and R&D Systems for TGF-¡3). Assays were carried out nation (Supplementaryfigure#l), the percentage ofTregs was sig­ according to the instructions provided by the manufacturers. nificantly higherin NR than in R mice ( p < 0.05), in both peripheral blood and spleens (Fig. lB and C). In addition, we observed an in­ 2.12. Allogeneic splenocytes proliferation assay crease in the percentage of tumour-infiltrating CD4+ cells in R mice compared to NR and satine groups ( p < 0.05; Fig. lD, left 5 Borre marrow-derived Des (1 x 10 ) from BALB/c rnice were panel). The tumour-infiltrating CD4+eD25+Foxp3+/CD4+ pro­ treated for 20 min with 50 ¡.tg/ml mitomycin e (Sigma ehemi­ portian were evaluated in the same experiment. The percentage cal,St. Louis, MO, USA), washed and used as stimulators. Sple­ of Tregs in R was significantly lower than in untreated mice 6 nocytes (1 x 10 ) from naive e57BU6 mice were plated in (P < 0.05, Fig. 1D, right panel). On day 14, the percentage ofintra­ U-bottom 96-well plates ata ratio of 1:10 (Des: splenocytes). tumoural Tregs in R was lower than in NR and satine-treated mice Assays were carried out in the presence ofTregs supernatants ( p < 0.05), in agreement with the results obtained in peripheral derived from saline, ey, AdiL-12 or Cy + AdiL-12-treated rnice. blood and spleen. These results showed an association between Cell proliferation was evaluated after 4 days in culture, as de­ increased circulating and tumour-infiltrating Tregs with tumour scribed elsewhere (Malvicini et al., 2009). resistance to combined ey + AdiL-12 therapy. These results indi­ cate that depletion ofTregs is critica! to achieve a therapeutic ben­ 2.13. In vitro re-stimulation of CD4+CD25- T cells efit in CT26 CRC-bearing mice. To confirm the inhibitory effect ofTregs on the antitumoural BALB/c rnice were injected with CT26 cells and treated with efficacy of combined therapy, we adoptively transferred Cy, AdL-12 or Cy + AdiL-12 as described above. Splenocytes CD4+CD25+Foxp3+ T cells into CT26 tumour-bearing rnice were isolated 14 days after treatment and CD4+CD25- T cells that were treated with the combination. Thus, 1 x 106 were obtained by immunomagnetic selection. Then, purified eD4+CD25+Foxp3+ T cel!s derived from untreated tumour­ CD4+eD25- T cells were pooled, and 4 x 106cel!s/ml were bearing mice were administered i. v. on days 11 and 13 after tu­ co-cultured with rnitomycin C-treated CT26 cells (4 x 105/ml) mour inoculation (days 2 and 4 after the combined treatment and mitomycin C-treated splenocytes (4 x 105/ml) in a 24- with ey + AdiL-12). As a result, adoptive transfer ofTregs signif­ well plate (1 mi/well) with 10 UI/ml rmiL-2. Seven days later, icantly abolished the antitumoural effect achieved with the 6 3 viable cells were harvested and washed, adjusted to 4 x 10 / combined treatment (mean tumour volume ± SE [mm ] at day mL, and co-cultured again with rnitomycin e-treated eT26 28, Cy + AdiL-12: 144 ± 81 vs. ey + AdiL-12+ Tregs: 3191 ± 394, cells and splenocytes. On day 14, CD4+CD25- T cel!s were p < 0.01; Fig. 2A). Moreover, animal survival was strongly re­ used for in vivo adoptive T cells transfer experiments. duced after adoptive transfer of Tregs, showing a similar val u e in saline group (Fig. 2B). The trafficking behaviour of adminis­ 2.14. Statistical analysis tered Tregs was also analysed. Interestingly, FastDio-stained Tregs injected i.v. revealed that Tregs were distributed in differ­ Mann Whitney, Kruskal-Wallis tests and Kaplan-Meier, log ent organs, with higher levels in peripheral blood,lymph nades rank (InStat, GraphPad Software) were used to statistically ex­ and tumours (Fig. 2C). Altogether, these results strongly suggest amine the differences between groups. P < 0.05 was consid­ that, at least in this tumour model, CD4+eD25+Foxp3+ T cel!s ered statistically significant. should be inhibited in arder to achieve a potent antitumoural immune response.

3. Results 3.2. Combined therapy with Cy + AdiL-12 reverts the inhibitory effects of Tregs on DCs activation 3.1. Tregs depletion is required to generate an effective antitumoural response after treatment with Cy +AdiL-12 We investigated the effects of Cy + AdiL-12 on the immunosup­ pressive functions ofTregs derived from eT26 tumour-bearing We previously demonstrated that the sequential adrninistration rnice. Thus, we decided to assess the effects of combined ther­ of Cy followed by adenoviral gene transfer of IL-12 induced apy on the ability ofTregs to modulate not onlythe number but 'Please cite tbis arti<;1e in pmss as: Malvicini, M. et al, Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of U.-12, Molecular Oncology (2Q1~!lrJJ~Q:12!M1!PJl'()nc.2Q).1.03,gJ7 lv!OLECULAR ONCOLOGY XXX (2011) 1-14 5

A 8 perlpheral blood e spleen Saline 1000 ·~Salíne , C,.""--1> NiC,>M. IL-12 R l 21.2% 17.7% 1 .1. 13.8% ' 2.9% 1 'i aoo Day7 .§. ! --- !~! . 111..---"- ~ 600 o ... .;.. ~ ·~·~ .. -~ -~.. i 400 • ..:. •· .• . -'• FoxpJ -· Foxp3 ~ 200 '¡ ___:_ ·~ ·¡ ¿ ~ - '1 "' 1 ... '/..___-: ""'"~ '__ .-: '_e_ '14~ l~~-. -_ L. ------Foxp3 Foxp3 D tumour Saline í .1 2.3% ·1 4 R Oay7 -li~ :¡¡~~1>-NR ¡_Ad_:_~_1_: :~(J(-A-'~-~~%-·1_'_"~~ ;M: 1> R

• ... a!ofl'eC)'S ·..- '"coc~ ... _. .L-."""'" ·...... 0111 ...... • ~ .-.. .. ·~ ~-~~~-.. - - .. ------C04 ------Foxp3

~·"it_ Lt;_ ·ll:. ·l.L ~l ·~ . :~ C04 -· -· •· • ..:. Foxp3

Fig. 1 - Assessment ofTregs in miceR (responder) and NR (non-responders) to combined treatment Cy + AdiL-12 in s.c CT26 tnmour model. 5 (A) Tunzour volunze in R and NR nzice. BALE/e mice (n = 35) were s.c. inoculated with 5 X 10 CT26 cells into the right flank (day O) and tumours were allowed to reach 85 mm3 before the treatment began. Animals were treated with Cy (50 mglkg i.p., day 8) + AdiL-12 (109 TCID50 i.t., day 9) (n = 12) or with saline (n = 6). Tumour growth was assessed twice a week by calliper. Tumour volumes at days 7 and 14 after treatment are expressed as mean (bars, SEM). The experiment was performed four times. Mann Whitney test, Cy + AdiL-12 NR vs. Cy + AdlL-12 R: (##p < 0.01). Percentage ofCD4+CD25+Foxp3+1CD4+ cells in peripheral blood (B) or spleen (C) ofR and NR nzice. Mann Whitney test, Peripheral blood, day 7 and day 14: Cy + AdiL-12 R vs. saline (**p < 0.01) and vs. Cy + AdiL-12 NR (#p < 0.05). Spleen, day 7: Cy + AdiL-12 R vs. saline and Cy + AdlL-12 NR (#p < 0.05); Cy + AdiL-12 NR vs. saline (*p < 0.05); day 14: Cy + AdiL-12 R vs. saline and Cy + AdiL-12 NR (*and #p < 0.05). (D) Percentage oftunzour-infiltrating CD4+ Tcells {lift) and CD4+CD25+Foxp3+1CD4+ cells (right). Mann-Whitney test, CD4+ Tcells, day 7: Cy + AdiL-12 R vs. saline, (*p < 0.05); day 14: Cy + AdiL-12 R vs. saline and Cy + AdiL-12 NR (#p < 0.05); Cy + AdiL-12 NR vs. saline (*p < 0.05). CD4+CD25+Foxp3+1CD4+ cells, day 7: Cy + AdlL-12 R vs. saline, (*p < 0.05); day 14: Cy + AdiL-12 R vs. saline and Cy + AdiL-12 NR (#p < 0.05). The results shown represent four independent experiments.

also the maturation and/or functional status of DCs. To this untreated or treated with Cy, AdiL-12 or with a combination end, nalve DCs were co-cultured with Tregs isolated from dif­ of both. We found that Tregs from untreated mice (saline ferent experimental groups. When DCs were cultured with group) produced high levels of IL-10 and TGF-[3 (mean± SEM 3 saline-derived Tregs there was a significant reduction in the [mm ]: 172 ± 8, and 3820 ± 213 pg/ml, respectively; (Fig. 3C production ofiL-12 as well as in the expression ofMHC-11 and and D). In contrast, we observed that Tregs of mice treated the CD40 co-stimulatory molecules induced by LPS ( p < 0.05, with a single agent (Cy or AdiL-12) produced significantly Fig. 3A and B). In contrary, when DCs were cultured in the pres­ lower amounts ofboth cytokines (IL-10: 41 ±Sor 62 ± 11 pg/ ence of Cy +AdiL-12-treated mice-derived Tregs, they showed ml, respectively [p < 0,05 vs. saline]; TGF-[3: 926 ± 35 or a similar pattem of activation, IL-12 production and MHC-II­ 866 ± 42 pg/ml, respectively [p < 0,01 vs. saline]). Importantly, CD40 expression phenotype than the observed in control DCs Cy + AdiL-12 induced further inhibition ofiL-10 and TGF- [3 pro­ after LPS stimulation. duction by Tregs, in comparison to saline and single treat­ ments (11 ± 8 pg/ml; p < 0,05 and 647 ± 43 pg/ml; p < 0,01, 3.3. Combined therapy with Cy + AdiL-12 reverts Tregs­ respectively). Consistently with these results, supematants immunosuppressive phenotype by reducing IL-10 and TGF-{J of Tregs from untreated tumour-bearing mice inhibited DCs production stimulated allogeneic splenocytes proliferation (Fig. 3E). How­ ever, when supematants of Tregs derived from CT26-bearing We investigate in uitro the secretion of IL-10 and TGF-[3 by animals treated with Cy + AdiL-12 were assayed, a strong pro­ Tregs derived from spleens of CT26 tumour-bearing mice, liferation activity was detected. Tregs cell viability from all Please cite this article in press as: Malvicini, M. et al., Reversal of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j.molonc.2011.03.007 6 l'l'! OLE e U LAR O N e O LO G Y XXX ( 2 O11) 1-14

A 4000 -ti- Saline B 1oo ~¡- Saline plus Tregs 80 3500 _,_. Cy+Adll-12 Cy+Adll-12 plus Tregs kl1~ 3000 J. . ** M' '¡! ~ 2500 ., 20 E Ll· ::> 2000 0+------· -----~-----, o> o 20 40 60 L ::>o Days after tumour inoculation E 1500 ::> /) 1- .,.,/ / e 1000 + /ti / o i5.... 500 ./"' ,f< .. !:·J .. ~4t"~ !:. o ..C) o 5 10 15 20 25 30 "' ++ Oays after tumor lnoculatlon ,~ Tregs t ,._.."' ..1: ,._!:; o o~

Fig. 2- Re-infusion ofTregs in CT26-bearing mice treated with the combination Cy + AdiL-12. (A) Tumour growth: BALE/e mice were s.c. 5 inocnlated with 5 X 10 CT26 cells into the right flank (day O) and tumours were allowed to reach 85 mm3 before the treatrnent began. Animals (n = 6/group) were treated with saline or Cy (50 mglkg i.p., day 8) + AdiL-12 (109 TCID50 i.t., day 9). In addition,saline or Cy + AdiL-12- 6 treated mice were inoculated i.v. on days 11 and 13 with 1 X 10 CD4+CD25+ T cells (Tregs) isolated from untreated tumour-bearing mice. Tumour volume was assessed twice a week by calliper. The experiment was performed four times. Data are expressed as mean (bars, SEM). Kruskal-Wallis test, day 28: Cy + AdiL-12 vs. saline and Cy + AdlL-12+Tregs (*p < 0.05). B) Survival: Kaplan-Meier, log rank test, (**p < 0.01), C) In vivo distribution pattem of re~infitsed Tregs: Cy + AdiL-12-treated mice were inoculated with 1 X 106 fastDiO-dyed CD4+ CD25 + T cells (i.v; days 11 and 13) and biodistribution was analysed by flow cytometry. The results shown represent three independent experiments.

groups was confirmed by trypan blue exclusion test (not splenic MDSCs ( p < 0.01; Fig. 4A). Interestingly, these reduced shown); no evidence of apoptosis was observed by TUNEL as­ levels of MDSCs induced after Cy + Ad!L-12 treatment say in all groups (data no shown). Altogether, these results returned to baseline when Tregs from untreated tumour­ suggest that treatment of CT26-bearing mice with bearing mice were adoptively transferred (Fig. 4B). These Cy + Ad!L-12 reverts the inhibitory activity of Tregs on DCs, results suggest that the combined treatment affect the pro­ probably by reduction of IL-10 and/or TGF-13 secretion. portian of MDSCs. Also, the restitution of MDCSs levels after adoptive transfer of Tregs in Cy + Ad!L-12-treated mice 3.4. Depletion of MDSCs induced by the combination of suggests a cross-regulation between both cell types in our Cy with AdiL-12 is reverted after re-infusion ofTregs tumour modeL

It has been demonstrated that MDSCs contribute to generate 3.5. Adoptive transfer of in vitro expanded speciftc an immunosuppressive microenvironment leading to tumour IFN-r secreting CD4+ T lymphocytes derived from progression (Huang et aL, 2006). We decided to investigate Cy + AdiL-12-treated mice has potent antitumoural effects whether Cy + Ad!L-12 could affect the number of splenic MDSCs. The prevalence of CD11b+Gr1+ cells (Bronte et aL, We have previously found that sequential treatment of mice 1998) was determined in spleens derived from CT26 bearing CT26 with Cy followed by AdiL-12 significantly tumour-bearing mice treated with Cy, Ad!L-12 or the combi­ increased IFN-y secretion by CD4+ T lymphocytes (Malvicini nation by flow cytometry. We observed that Cy or Ad!L-12 et aL, 2009). Here, we decided to further assess whether in vitro as a single treatment induced a slight decrease in the amount expanded CD4+CD25- T cells (from here on CD4+) might have of MDSCs in comparison to the saline group. However, a direct therapeutic activity in vivo after i.v. injection in two Cy + Ad!L-12 induced a 3-fold decrease in the percentage of models. In a therapeutic model (Fig. SA), CD4+ T cells from Please cite this article in press as: Malvicini, M. et aL, Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j.molonc.2011.03.007 lVIOLECULAR ONCOLOGY XXX (2011) 1-14 7

DC+Tregs (sal... ) DC+Tregs (Cy) DC+Tregs (Adll·12) DC+Tregs (Cy+Adll·12) oc 1 1 8 l 1 1 CD11e '11 l\; - 32.%.

:l/b-:;--~-~-~ \'" .J \.___...-...... c:lllc,~t

1 37% MHCII ·¡li ! l \--~~~~·-=-

1 .L __:c,.:__, __ _ ~-----·--~ ~ ~ v< IPE ~ ~ ~ . "'""'•PE . ,.

¡, 1; ' ! ' 45% 24% 38% 35% 40% C040 ! 1 !1 !1 !¡ ! 1 ! .i .-~ ... ~.-~~-~- .,., ···~-----··--· .i '""""' ~,u.c '""""' e

50

Fig. 3 - Effect of combined tberapy Cy + AdiL-12 on tbe interaction DC/Tregs. A) Quantijication rif IL-12: Data are expressed as mean (hars, SEM). Kruskal-Wallis and Dunn's multiple comparisons test: DC vs. DC+Tregs (saline), *p < 0.05; DC+Tregs (saline) vs. DC+Tregs (Cy + AdiL-12), ***p < 0.001. B) Pbenotype of DCs (CD11c+MHC-II+CD40+). Mann Whitney test: DC vs. DC + Tregs (saline), *p < 0.05; DC + Tregs (saline) vs. DC+ Tregs (Cy + AdiL-12), *p < 0,05. The results shown represent tbree independent experiments. IL-10 (C)and TGF­ {3 (D) secretion by Tregs: Mann Whitney test, saline vs. Cy, AdiL-12, and Cy + AdiL-12, *p < 0.05; Cy + AdiL-12 vs. Cy and AdiL-12; #p < 0.05. (E) Effect ofTregs supernatants (SN) on allogeneic spleen cell (SC) proliferation under stimulation witb mitomycin C-treated DC. fH-Tbymidine). Mann Whitneytest, SC+DCvs. SC+DC+Tregs SN (saline), ***p < 0.001; SC+DC+Tregs SN (saline)vs. SC+DC+Tregs SN (Cy + AdiL-12), ***p < 0.001.

mice treated with Cy or Ad!L-12 as single agents were devoid of Ad!L-12) (Fig. 5C). The specificity of cytotoxic effect was con­ any significant effect on tumour growth and survival. On the firmed against BNL cells. contrary, CD4+ T cells from mice treated with Cy + Ad!L-12 showed a significant reduction of tumour volume {89% vs. sa­ 3.6. Cy + AdiL-12 is highly effective in two stringent line group at day 23, p < 0,05; 81% and 78% vs. Cy and Ad!L-12 gastrointestinal cancer models groups, respective!y; at day 28, p < 0,05) andan increase in ani­ mal survival ( p < 0.01). Similar result was obtained when CD4+ In order to examine whether combined treatment exert anti­ T cells from combined treatment were adoptively transferred in tumoural effects in aggressive tumour models we investigated a preventive tumour model {Fig. 5B). On the other hand, we in­ its therapeutic efficacy in a liver metastatic CRC model using vestigated the in vitro cytotoxic activity of CD4+ T cells used CT26 cells and also against an established pancreatic cancer for in vivo experiments and saw that cells derived from mice re­ model using Panc02 cells. For this purpose, we injected CT26 ceivingthe combined treatment displayed a significantly higher cells into the liver of mice (day 0), treated with Cy (day 10) lytic activity against CT26 cells ( p < 0.001 vs. saline; p < 0.01 vs. and 24 h later with Ad!L-12. Fig. 6A (left panel) shows that Cy or Ad!L-12) and produced higher levels ofiFN-y in compari­ no significant antimetastatic effect was observed when only son with control group ( p < 0.001 vs. saline; p < 0.01 vs. Cy or Cy and Ad!L-12 were administered, whereas the combined Please cite this article in press as: Malvicini, M. et al., Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j.molonc.2011.03.007 8 lv!OLECULAR ONCOLOGY XXX(2011) 1-14

Salina A 20 e..~ lt) ~ 15 u u "E CD g.1o AdiL-12 Cy+AdiL-12 + 1: C) .e+ ..... o o

~(:-e t::>~

B Salina Cy+AdiL-12 plus Trags .e e..~ ...... o

u "E CD

'iilf) 10 + 1: Cy+AdlL-12 C) + ** il .e ..... o o

Gr1

Fig. 4 - (A) E.ffect of combined treatment Cy + AdiL-12 on splenic MDSCs (CD11b+Gr1+): Mann Whitney test, saline vs. Cy + Ad-12 (**p < 0.01). Data are representative of three independent experiments (n = 5/group per experiment) B) E.ffect of in vivo administration oJTregs on splenic MDSCs in mice treated with the combination Cy + AdiL-12. Mann Whitney test, Cy + AdiL-12 vs. Saline, **p < 0,01 and Cy + AdiL-12 plus Tregs, #p < 0.05. Data are representative of two independent experiments (n = 6/group per experiment).

therapy showed a significant reduction in metastases growth (70% and 74% at day 54, respectively). However, no complete (98% vs. saline group, at day 20, p < 0.001). Therapeutic efficacy tumour regression was observed in mice treated with a single of the combination was superior to each individual agent therapy. In contrast, the combined therapy resulted in 3 (mean metastases volume±SE [mm ] at day 20: Cy + AdiL- a marked reduction in tumour volume (95% vs. saline group, 12: 15,6 ± 11,7; Cy: 191 ± 87,4; and Ad!L-12: 306 ± 167; at day 54; p < 0,05) and complete tumour regressions in 40% p < 0.01) and complete metastases regressions were observed of animals (4/10) (Fig. 6B, left panel). Importantly, the com­ in 7 out of 15 animals (47% vs. 0% in saline and single agents bined treatment produced a significant reduction in tumour groups). In addition, survival of mice receiving combined ther­ growth in comparison to the effects of each single agent 3 apy was significan tiy higher than the controls ( p < 0.001; (mean tumour volume ±SE [mm ] at day 54; Cy + Ad!L-12: Fig. 6A; right panel). Remarkably, Cy followed by Ad!L-12 209 ± 127; Cy: 1330 ± 358; Ad!L-12: 1177 ± 173; p < 0.05). Sur­ showed a potent antitumoural effect on Panc02 tumour nod­ viva! of mice receiving combined therapy was also signifi­ ules. Animals treated only with Cy or Ad!L-12 showed a signif­ cantly increased in comparison to mice receiving individual icant reduction in tumour volume with respect to saline group therapy or saline ( p < 0.001, Fig. 6B; right panel). Analysis of Please cite this article in press as: Malvicini, M. et al., Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j .molonc.2011.03.007 J\IOLECULAR ONCOLOGY XXX (2011) 1-14 9

A 35oo C04+CD25- (satine) 3000 ~ C04+CD25- (Cy) ...... - C04+CD25- (Adll-12) ~ C04+CD25- (Cy + Adll12) C04+CD25- (saline) -+- C04+CD25- (Adll-12) ~ 2500 §. -+ C04+CD25- (Cy) ~ C04+CD25- (Cy +Adll 12) G> 100,----...,.,--,-, E 2000 :::1 o 80 > 1500 5 -~ o !.... 60 E -¡¡¡ •• 1000 ·;;:> ~ .... 40 :::1 (/) 500 20 0+----r----.-, _...,.__.,.---r---. 10 20 30 o 10 20 30 40 50 60 Days after tumour inoculation Oays after tumour inoculation

B 3500 C04+CD25- (saline) -+ C04+CD25- (Cy) 3000 -+- C04+CD25- (Adll-12) ~ C04+CD25- (Cy + Adll 12) C04+CD25- (saline) -+- CD4+CD25- (Adll-12) ~ 2500 -+ C04+CD25- (Cy) ....,._ C04+CD25- (Cy +Adll.12) §. G> 100~------.---,.--, § 2000 o 80 ...> :::1 ~ o -¡¡¡ 60 > 40 ~ "E;:¡ (/) 20

0+---~--~-~~--~~~---, o 10 20 30 40 50 60 Days after tumour inoculation Days after tumour inoculation

80 e - CT26 C BNL ***" 1000 60 ..... 800

~ !1..... € ... 40 C) -¡¡ S: :J Lt ~ 20

o ~ ~ ·,$-0 (j. ~~ Ci~ t::J'Ii ~ ~ ":>~ .¡;""' ~~""' "'i-ti 'i-ti ej. " ej.

Fig. 5 - Antitumour effect of in vitro expanded CD4 + T -lymphocytes from CT26-bearing mice treated with Cy, AdiL-12 or Cy + AdiL-12. (A) 5 Therapeutic model: BALB/c mice were s.c. inoculated with 5 X 10 CT26 cells into the right flank (day O) and when tumours reached 65 mm3 in size 6 (day 6) the mice were treated with vitro expanded CD4+ cells (2,5 X 10 i.v) from different experimental groups (n = 4/group) {lift panel). Data are expressed as mean (bars, SEM). Mann Whituey test, days 23 and 28: CD4+CD25- (Cy + AdiL-12) vs. CD4+CD25- (saline, Cy and AdiL-12), *p < 0.05. Survival (right panel). Kaplan-Meier, log rank test, **p < 0.01. Data are representative of three independent experiments. 5 B) Preventive model: BALB/c mice (n = 4/group) were simultaneously inoculated with 5 X 10 CT26 cells s.c. into the right flank and adoptive 6 transfer of 2,5 X 10 CD4 + cells at the same time (day O) {lift panel). Data are expressed as mean (bars, SEM). Mann Whituey test, days 23 and 26:CD4+CD25- (Cy + AdiL-12) vs. CD4+CD25- (saline, Cy andAdiL-12), •p < 0.05. Survival (rightpanel). Kaplan-Meier, log rank test, **p < 0.01. Data are representative of three independent experiments. C) Specific CTL adivity {liftt panel): CD4+ T cells from different

Please cite this article in press as: Malvicini, M. et al., Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j .molonc.2011.03.007 10 MOLECULAR ONCOLOGY XXX (2011) 1-14

the in vivo interaction between both treatments was per­ adoptive transfer. Importantly, the levels of Tregs reached formed by the fractional product method (FTV) (Yokoyama with adoptive therapy in Cy + AdiL-12-treated rnice (Fig. 2C) et al., 2000). Fig. 6C summarises the relative tumour volume were similar to the percentages of Tregs found in NR rnice of different groups at 4 different time points. On day 29 after (Fig. 1B; 1D). Considering that Tregs are up regulated in cancer treatment, in the combination group there was a 1.3-fold im­ and that they induce a detrimental effect on the immune sys­ provement in the antitumour efficacy when compared to the tem, strategies aimed at reducing Tregs number may increase expected additive effect. On days 33 and 36, Cy + AdiL-12 the efficacy of any immunotherapy (Curiel et al., 2004; showed a 1.4-fold increase in the inhibition of tumour growth Orrnandy et al., 2005; Sakaguchi et al., 2001). A myriad of in­ over an additive effect (expected fractional tumour volume). hibitory mechanisms have been proposed to explain the im­ Moreover, on day 40 the increase was of 1.7-fold over the ad­ munosuppression induced by Tregs, including secretion of ditive effect. These results allow us to conclude that cytokine suppressors and the induction of apoptosis/cell cycle Cy + AdiL-12 has a synergistic effect on pancreatic carcinoma arrest on effector T cells (Tang and Bluestone, 2008). It has also growth inhibition. Similar doses of control adenovirus (Adi3- been suggested that Tregs could modulate the maturation Gal), alone or in combination with Cy, did not produce any sig­ and/or function of DCs (Larmonier et al., 2007; Onishi et al., nificant change of tumour growth in both experimental 2008). Indeed, intravital rnicroscopy studies have suggested models (data no shown). In both tumour models, the com­ that Tregs contact DCs more frequently than potential effector bined Cy plus AdiL-12 strategy was well tolerated with no T cell targets (Bluestone and Tang, 2005). Taking into account signs of toxicity (data not shown). Altogether, these results the current data regarding Tregs function in cancer and to the show that the antitumour effects of the combined treatment way tumour cells drive Tregs induction, we decided to inves­ can be achieved in very aggressive tumour models including tigate the immunosuppressive effects of Tregs derived from pancreatic cancer. tumour-bearing rnice and the effects of the combined Cy and AdiL-12 treatment on their regulatory capacity (Beyer and Schultze, 2006; Ghiringhelli et al., 2005). 4. Discussion Previously we demonstrated that in vivo treatment with Cy + AdiL-12 affected the maturation status ofDCs. Currently, A number of immunotherapy strategies for advanced GIC are we provide new evidence showing that this effect is mediated, currently under preclinical and clinical evaluation (Mazzolini at least in part, by Tregs. Ca-cultures of nai:ve DC with Tregs et al., 2007). However, there is a frustrating inconsistency in from untreated animal tumours resulted in a significant inhi­ the correlation between biological and clinical responses. bition of IL-12 production as well as in MHC-II and CD40 ex­ The reasons for the mentioned data discrepancies are mani­ pression by DCs. More importantly, the inhibitory activity of fold but the presence of strong mechanisms of immune es­ Tregs on the maturation/activation status of DCs was com­ cape induced by tumours appears to be important (Clark pletely abolished when Tregs derived from mice that received et al., 2009; Croci et al., 2007). Thus, it seemed reasonable to Cy + AdiL-12 were used. explore immunotherapy strategies aimed at inducing rever­ The mechanisms used by Tregs to generate immunosup­ sien of tolerogenic processes in tumour-bearing hosts, such pression remain conflictive and controversia!. Direct cell-cell as those induced by Tregs (Zou, 2006). interaction between Tregs and their target cells (effector We have previously demonstrated the synergistic antitu­ T cells or DCs) has been established as a prerequisite to exert moural effect of sequential systernic administration of sub­ their immunoregulatory activity (Vignali et al., 2008). On the optimal Cy doses followed by immuno-gene therapy with other hand, the role of soluble factors, such as IL-10 and AdiL-12 in rnice with CRC (Malvicini et al., 2009). One impor­ TGF-13 has also been investigated extensively (Dieckmann tant finding was that, regardless of the individual response et al., 2001; Takahashi et al., 1998; Thornton and Shevach, of each mouse, the synergistic antitumour activity was associ­ 1998). In the present work, we have detected high levels of ated with depletion of regulatory T cells. both cytokines in supernatants of Tregs isolated from spleen We analysed the levels of Tregs after treatment with the of untreated tumours. These supernatants containing high combination and classified them according to the treatment levels ofiL-10 and TGF-13 were able to inhibit allogeneic sple­ response in R and NR rnice. As a result, we observed that NR nocytes proliferation under stimulation with DCs. It is impor­ mice failed to significantly decrease Tregs significantly in tant to note that Cy + AdiL-12 significantly reduced the spleen, peripheral blood, and inside tumours. The relevance production of IL-10 and TGF-13 by Tregs and also abolished of Tregs depletion to achieve a therapeutic response in our their inhibitory effect on lymphocyte proliferation. Alto­ model was consisten! when the efficacy of Cy + AdiL-12 was gether, our results suggest that treatment of CRC-bearing completely abolished following reconstitution of Tregs by rnice with Cy + AdiL-12 not only induces CD4+CD25+Foxp3+

experimental groups (n = 4/group) were isolated and stimulated in vitro with mitomycin C-treated splenocytes and CT26 cells for 5 days. Specific CTL activity was evaluated against CT26 and BNL cells. The percentage of specific cytotoxicity was quantified whit the LDH Cytotoxicity Detection Kit and calculated according to the following formula: ([abs 492 nm experimental- abs 492 nm background])/[abs 492 nm maximum­ abs492 nm background]) X 100. Mano Whitney test, Cy + AdiL-U vs. saline, -p < 0.001; Cy + AdiL-12 vs. Cy and AdiL-12, #:ll#p < 0.01. Data represent the mean of triplicate cultures. Quantification rif IFNy (right panel}: The amount of IFN-y secreted by CD4+ T cells was evaluated after its isolation and in vitro stimulation as described above. Data are expressed as mean (bars, SEM). Kruskal-Wallis and Dunn's multiple comparisons test: Cy + AdiL-U vs. saline, -p < 0.001; Cy + AdiL-12 vs. Cy and AdiL-12, ##p < 0.01.

7Please cite this anide in press as: Yalvicini, M. et al., Reversa) of gastrointestinal can::inoma-induced ímmunosuppression and ·· induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of n.-12, Molecular Oncology J?011}, doi:10.t!ll§ñ{m.qt.qn~.20-ll~~-~Q07 :MOLECULAR ONCOLOGY XXX(2011) 1-14 11

A 14oo

Saline ..... Cy-+- AdiL-12 ....,_ Cy + Ad IL-12 ~ 100 .§.., .. 80 ~., .f! ~ .. ~ 60 E -¡;¡ .... > ..> >.... ::::¡ :;, 40 en

20

Days after tumour inoculation

8 4000 Saline -+- Ad1L12 3500 -+-Cy ...... Cy+Adll12 3000 ¡;;- Salín e ..... Cy -+- Adll-12 ....,_ Cy + Adll-12 E .§. 2500 ..E :;, o 2000 > :;,.... o 1500 -¡;¡ E > :;, 1- 1000 1::;, 40 en 500 20

o o~~~.~~~~~~~~~ o 10 20 30 40 50 60 o 20 40 60 80 100 Days after tumour lnoculation Days after tumour inoculation e

cr •.Yn.-1! E:q>

Fig. 6- Effect of combined treatment Cy + AdiL-12 on aggressive GIC models. A) Experimentalliver metastases r:fCRC (CT26). Leftpanel, Liver 5 metastases volume: BALB/c mice received an intra-hepatic injection of 5 X 10 CT26 cells (day O). Mice were distributed in different experimental groups and treated with saline (n = 6); Cy (50 mglkg i.p; day 8, n = 6); AdiL-12 (109 TCID50 i.t; day 9, n = 6); or Cy + AdiL-12 (n = 8). Metastases volume was measured at day 20. Data are representative of three independent experiments. Mann Whitney test: Cy + AdiL-12 vs. saline (***p < 0.001); Cy + AdiL-12 vs. AdiL-12 or Cy (##p < 0.01). Right panel, Survival: Kaplan-Meier, log rank test, (-p < 0.001). B) 5 Pancreatic tumour model {Panc02}. Left Panel, Tumour growth: C57BL/6 mice were inoculated s.c. in the right flank with 5 X 10 Panc02 cells. Treatments began when tnmours reached 85 mm3 in size. Animals were treated with saline (n = 6), Cy (50 mglkg i.p; day 8, n = 5), AdiL-12 (109 TCID50 i.t; day 9, n = 5), or Cy + AdlL-12 (n = 6). Tumour growth was assessed twice a week by calliper. Data are expressed as mean (bars, SEM). Data are representative of three independent experiments. Mann Whitney test: Cy + AdiL-12 vs. Cy or AdlL-12 (*p < 0.05). Rightpanel, Survival: Kaplan-Meier, log rank test (***p < 0.001). C) Analysis of the in vivo interaction between Cy and AdlL-12 by the fractional product method (FTV) in Panc02 model. a ITV (experimental mean tnmour volume)/(control mean tumour volume); b Day after treatment onset; e (Cy mean FTV) X (AdiL-12 mean FTV); d R = Expected ITV/Observed ITV. A ratio > 1 indicates a synergistic effect, and a ratio < 1 indicates a less than additive effect.

T cell depletion but also modifies its functionality, at least by immunoregulatory mechanisms associated with the synergis­ reduction of IL-10 and TGF-~ production. Our present results tic antitumour effects achieved with Cy + AdiL-12. strongly suggest that depletion of Tregs and/or inhibition of Recently, myeloid-derived suppressor cells (MDSCs) have Tregs ability to induce immunosuppression are key been recognized as a cell population that can negatively Please cite this article in press as: Malvicini, M. et al., Reversal of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of IL-12, Molecular Oncology (2011), doi:10.1016/j.molonc.2011.03.007 12 MOLECULAR ONCOLOGY XXX (2011) 1-14

regulate T-cell functions (Gabrilovich et al., 2001). These cells treatment has also a synergistic effect for pancreatic carci­ can act by inhibiting both the innate and adaptive immune re­ noma. Thus, Cy + AdiL-12 based treatment is a promisingim­ sponses (Huang et al., 2006). MDSCs are a heterogeneous pop­ munotherapy among the limited therapeutic options ulation of myeloid cells that includes macrophages, suggesting the possibility of achieving high efficacy with IL- granulocytes and other cells that express Ly-6C/G (recognized 12 without toxicity in more malignant tumour models. This by Gr-1 antibody) and CD11b in mice and suppress immune strategy could be improved in patients with disseminated dis­ responses in vivo and in vitro (Gabrilovich and Nagaraj, 2009). ease by the use of tumour-targeted expression of IL-12 (Kerkar It has been observed that elimination of MDSC by antibodies et al., 2010). In conclusion, Cy + AdiL-12 demonstrated efficacy or cytotoxic agents such as gemcitabine in mouse tumour to block Tregs and MDSCs-mediated immunosuppressive tu­ models increased antitumour responses (Ko et al., 2010; mour environment and, simultaneously, to amplify the effec­ Suzuki et al., 2005) In our study, we analysed the phenotype tor phase of the immune response by the induction of a potent and frequency of CD11b+Gr-1+ MDSCs in spleen of mice antitumour CTL activity. Considering our previous clínica! ex­ with CT26 tumours treated with Cy, AdiL-12 ora combination perience of separate usage of!L-12 gene transfer and Cy in pa­ of both. While the administration of Cy or AdiL-12 as single tients with cancer, it will be of interest to evaluate this agents induced a slight decrease in the percentage of MDSCs combination as a potential therapeutic strategy in advanced in comparison with the saline group, the combination of GI carcinomas clinically. Cy + AdiL-12 produced a significan! reduction of MDSCs in the spleens of tumour-bearing mice. However, other immuno­ suppressive CD11b populations, such as F4/80+ and CD11c+ Acknowledgements cells, might be involved but were not analysed. In agreement with our findings in CT26 tumour-bearing We would like to thank, Soledad Arregui and Guillermo mice, recent data published by Medina-Echeverz et al. {2010), Gastón for their technical assistance. This work was sup­ demonstrated that Cy in combination with IL-12 depleted ported in part by Agencia Nacional de Promoción Científica y not only Tregs but also tumour-infiltrating monocytic-MDCs Tecnológica (ANPCyT) grant PICT-2005/34788 (PM, OGS and (Mo-MDSCs) in another colorectal carcinoma model (MC38 GM); PICTO-CRUP 2005/31179 (GM); AECID 2008 D/022066/08 cells). Using this approach, they observed that recruitment (GM); PIP-CONICET 2009-2011 (PM). MM, JB and CA are fellows of inflammatory monocytes/macrophages (IMC) into tumour from ANPCyT. FP is a fellow from CONICET. microenvironment after the elimination of Mo-MDSC by Cy + IL-12 has a key role for the observed antitumoural effect (Medina-Echeverz et al., 2010) In our tumour model the immune rejection of established Supplementary data CT26 tumours was clearly affected by negative regulatory mechanisms linked to an increase in number and function Supplementary data associated with this article can be found ofTregs and MDSCs. We demonstrated that these negative in­ in the online version at doi:10.1016/j.molonc.2011.03.007. fluences could be overcome by the combination of Cy with AdiL-12. More importantly, the combined therapy not only REFERENCES has reverted immunosuppressive mechanisms induced dur­ ing tumour progression but also generated a strong cytotoxic immune response against tumour cells mediated by specific Alaniz, L., Rizzo, M., Malvicini, M., Jaunarena, J., Avella, D., IFN-y secreting CD4+ T lymphocytes. Atorrasagasti, C., Aquino, J.B., Garcia, M., Matar, P., Silva, M., The majority of immune therapy strategies to treat cancer Mazzolini, G., 2009. Low molecular weight hyaluronan inhibits have directed to the generation of CD8+ T cell responses colorectal carcinoma growth by decreasing tumor cell proliferation and stimulating immune response. Cancer Lett. against tumour-associated antigens. CD4+T cells have the 278, 9-16. ability to sustain CTL responses but can also function as effec­ Almand, B., Clark, J.!., Nikitina, E., van Beynen, J., English, N.R., tor cells either by production of potent cytokines (such as IFN­ Knight, S.C., Carbone, O.P., Gabrilovich, D.!., 2001. lncreased y) or by exhibiting anticancer cytotoxic activity (Appay et al., production of immature myeloid cells in cancer patients: 2002; Haigh et al., 2008; Spaapen et al., 2010). In our work, a mechanism of immunosuppression in cancer. J. Immunol. the in vitro expanded specific IFN-y secreting CD4+ T lympho­ 166, 678-689. cytes generated by Cy + AdiL-12 have demonstrated to be ef­ Appay, V., Zaunders, J.]., Papagno, L., Sutton, J., Jaramillo, A., Waters, A., Easterbrook, P., Grey, P., Smith, D., McMichael, A.J., fective in both a preventive and a therapeutic model using Cooper, D.A., Rowland-Jones, S.L., Kelleher, A.D., 2002. CT26 tumour model. We assume that the induction ofhigh cy­ Characterization of CD4(+) CTLs ex vivo. J. Immunol. 168, tolytic activity and the potent Th1-type response in vivo might 5954-5958. be the cause for the profound antitumour effects observed and Barajas, M., Mazzolini, G., Genove, G., Bilbao, R., Narvaiza, l., could be also exploited in the future in combination with other Schmitz, V., Sangro, B., Melero, l., Qian, C., Prieto, J., 2001. immunostimulatory strategies. Gene therapy of orthotopic hepatocellular carcinoma in rats The use of low-dose Cy combined with sub-therapeutic using adenovirus coding for interleukin 12. Hepatology 33, 52-61. AdiL-12 induces significan! regressions in two different and Bell, D., Chomarat, P., Broyles, D., Netto, G., Harb, G.M., aggressive tumour models: the CT26 liver metastatic model Lebecque, S., Valladeau, J., Davoust, J., Palucka, K.A., and the pancreatic cancer model. In this study we have shown Banchereau, J., 1999. In breast carcinoma tissue, immature by the fractional product method that the combined dendritic cells reside within the tumor, whereas mature Please cite tbis article in press as: Malvicini, M. et al, Reversa! of gastrointestinal carcinoma-induced immunosuppression and induction of antitumoural immunity by a combination of cyclophosphamide and gene transfer of n.-12, Molecular Oncology .(?!UiJ•. ,!1.2ltlP~9J§t.i&ttm!!tll~Z2!1,~-007 MOLECULAR ONCOLOGY XXX (2011) 1-14 13

dendritic cells are located in peritumoral areas. J. Exp. Med. Gomez, G.G., Hutchison, R.B., Kruse, C.A., 2001. Chemo­ 190, 1417-1426. immunotherapy and chemo-adoptive immunotherapy of Beyer, M., Schultze, ].L., 2006. Regulatory T cells in cancer. Blood cancer. Cancer Treat. Rev. 27, 375-402. 108, 804-811. Haigh, T.A., Lin, X., Jia, H., Hui, E.P., Chan, A.T., Rickinson, A.B., Bluestone, ].A., Tang, Q., 2005. How do CD4+CD25+ regulatory Taylor, G.S., 2008. EBV latent membrane proteins (LMPs) 1 and Tcells control autoimmunity? Curr. Opin. Immunol. 17,638-642. 2 as immunotherapeutic targets: LMP-specific CD4+ cytotoxic Brodsky, R.A., Petri, M., Smith, B.D., Seifter, E.]., Spivak, J.L., T cell recognition of EBV-transformed B celllines. J. Immunol. Styler, M., Dang, C.V., Brodsky, l., ]ones, R.]., 1998. 180, 1643-1654. Immunoablative high-dose cyclophosphamide without stem­ Huang, B., Pan, P.Y., Li, Q., Sato, A.!., Levy, D.E., Bromberg, ]., cell rescue for refractory, severe autoimmune disease. Ann. Divino, C.M., Chen, S.H., 2006. Gr-1+CD115+ immature Intem. Med. 129, 1031-1035. myeloid suppressor cells mediate the development of tumor­ Bronte, V., Wang, M., Overwijk, W.W., Surman, D.R., Pericle, F., induced T regulatory cells and T -cell anergy in tumor-bearing Rosenberg, S.A., Restifo, N.P., 1998. Apoptotic death of CD8+ T host. Cancer Res. 66, 1123-1131. lymphocytes after immunization: induction of a suppressive Kerkar, S.P., Muranski, P., Kaiser, A., Boni, A., Sanchez-Perez, L., population ofMac-1+/Gr-1+ cells. ]. Immunol. 161, 5313-5320. Yu, Z., Palmer, D.C., Reger, R.N., Borman, Z.A., Zhang, L., Caruso, M., Pham-Nguyen, K., Kwong, Y.L., Xu, B., Kosai, K.!., Margan, R.A., Gattinoni, L., Rosenberg, S.A., Trinchieri, G., Finegold, M., Woo, S.L., Chen, S.H., 1996. Adenovirus-mediated Restifo, N.P., 2010. Tumor-specific CD8+ T cells expressing interleukin-12 gene therapy for metastatic colon carcinoma. interleukin-12 eradicate established cancers in Proc. Natl. Acad. Sci. U S A 93, 11302-11306. lymphodepleted hosts. Cancer Res. 70, 6725-6734. Clark, C.E., Beatty, G.L., Vonderheide, R.H., 2009. Ko, ].S., Rayman, P., Ireland, ]., Swaidani, S., Li, G., Bunting, K.D., Immunosurveillance of pancreatic adenocarcinoma: insights Rini, B., Finke, ].H., Cohen, P.A., 2010. Direct and differential from genetically engineered mouse models of cancer. Cancer suppression of myeloid-derived suppressor cell subsets by Lett. 279, 1-7. sunitinib is compartmentally constrained. Cancer Res. 70, Colombo, M.P., Fomi, G., 1994. Cytokine gene transfer in tumor 3526-3536. inhibition and tumor therapy: where are we now? Immunol. Larmonier, N., Marran, M., Zeng, Y., Cantrell, ]., Romanoski, A., Today 15, 48-51. Sepassi, M., Thompson, S., Chen, X., Andreansky, S., Colvin, O.M., 1999. An overview of cyclophosphamide development Katsanis, E., 2007. Tumor-derived CD4(+)CD25(+) regulatory T and clinical applications. Curr. Pharm. Des S, 555-560. cell suppression of dendritic cell function involves TGF-beta Croci, D.O., Zacarias Fluck, M.F., Rico, M.]., Matar, P., and IL-10. Cancer Immunol. Immunother. 56, 48-59. Rabinovich, G.A., Scharovsky, O.G., 2007. Dynamic cross-talk Leonard, ].P., Sherman, M.L., Fisher, G.L., Buchanan, L.J., between tumor and immune cells in orchestrating the Larsen, G., Atkins, M.B., Sosman, ].A., Dutcher, ].P., immunosuppressive network at the tumor microenvironment. Vogelzang, N.]., Ryan, ].L., 1997. Effects of single-dose Cancer Immunol. Immunother 56, 1687-1700. interleukin-12 exposure on interleukin-12-associated toxicity Curiel, T.]., Coukos, G., Zou, L., Alvarez, X., Cheng, P., Mottram, P., and interferon-gamma production. Blood 90, 2541-2548. Evdemon-Hogan, M., Conejo-Garcia, ].R., Zhang, L., Burow, M., Lorenz, M., Staib-Sebler, E., Hochmuth, K., Heinrich, S., Gog, C., Zhu, Y., Wei, S., Kryczek, l., Daniel, B., Gordon, A., Myers, L., Vetter, G., Encke, A., Muller, H.H., 2000. Surgical resection of Lackner, A., Disis, M.L., Knutson, K.L., Chen, L., Zou, W., 2004. liver metastases of colorectal carcinoma: short and long-term Specific recruitrnent of regulatory T cells in ovarían carcinoma results. Semin. Oncol. 27, 112-119. fosters immune privilege and predicts reduced survival. Nat. Malvicini, M., Rizzo, M., Alaniz, L., Pinero, F., Garcia, M., Med. 10, 942-949. Atorrasagasti, C., Aquino, ].B., Rozados, V., Scharovsky, O.G., Dieckmann, D., Plottner, H., Berchtold, S., Berger, T., Schuler, G., Matar, P., Mazzolini, G., 2009. A novel synergistic combination of 2001. Ex vivo isolation and characterization of CD4(+)CD25(+) cyclophosphamide and gene transfer of interleukin-12 e radica tes T cells with regulatory properties from human blood. ]. Exp. colorectal carcinoma in mice. Clin. Cancer Res. 15, 7256-7265. Med. 193, 1303-1310. Matar, P., Rozados, V.R., Gervasoni, S.!., Scharovsky, G.O., 2002. Elkord, E., Hawkins, R.E., Stem, P.L., 2008. Immunotherapy for Th2/Th1 switch induced by a single low dose of gastrointestinal cancer: current status and strategies for cyclophosphamide in a rat metastatic lymphoma model. improving efficacy. Expert Opin. Biol. Ther. 8, 385-395. Cancer Immunol. Immunother. 50, 588-596. Ferlay, ]., Shin, H.R., Bray, F., Forman, D., Mathers, C., Mazzolini, G., Murillo, 0., Atorrasagasti, C., Dubrot, ]., Tirapu, l., Parkin, D.M., 2010. Estimates of worldwide burden of cancer in Rizzo, M., Arina, A., Alfara, C., Azpilicueta, A., Berasain, C., 2008: GLOBOCAN 2008. In t. ]. Cancer. Perez-Gracia, ].L., Gonzalez, A., Melero, l., 2007. Gabrilovich, D.!., Nagaraj, S., 2009. Myeloid-derived suppressor Immunotherapy and immunoescape in colorectal cancer. cells as regulators of the immune system. Nat. Rev. Immunol. World ]. Gastroenterol. 13, 5822-5831. 9, 162-174. Mazzolini, G., Prieto, J., Melero, l., 2003. Gene therapy of cancer Gabrilovich, D.!., Velders, M.P., Sotomayor, E.M., Kast, W.M., with interleukin-12. Curr. Pharm. Des. 9, 1981-1991. 2001. Mechanism of immune dysfunction in cancer mediated Mazzolini, G., Qian, C., Xie, X., Sun, Y., Lasarte, ].]., Drozdzik, M., by immature Gr-1+ myeloid cells. ]. Immunol. 166, Prieto,]., 1999. Regression of colon cancer and induction of 5398-5406. antitumor immunity by intratumoral injection of adenovirus Ghiringhelli, F., Larmonier, N., Schmitt, E., Parcellier, A., expressing interleukin-12. Cancer Gene Ther. 6, 514-522. Cathelin, D., Garrido, C., Chauffert, B., Solary, E., Bonnotte, B., Medina-Echeverz,]., Fioravanti,J.• Zabala, M., Ardaiz, N., Prieto,]., Martin, F., 2004. CD4+CD25+ regulatory T cells suppress Berraondo, P., 2010. Successful colon cancer eradication after tumor immunity but are sensitive to cyclophosphamide chemoimmunotherapy is associated with profound which allows immunotherapy of established tumors to be phenotypic change of intratumoral myeloid cells. ]. Immunol. curative. Eur. ]. Immunol. 34, 336-344. 186, 807-815. Ghiringhelli, F., Puig, P.E., Roux, S., Parcellier, A., Schmitt, E., Melero, l., Mazzolini, G., Narvaiza, l., Qian, C., Chen, L., Prieto,]., Solary, E., Kroemer, G., Martin, F., Chauffert, B., Zitvogel, L., 2001. IL-12 gene therapy for cancer: in synergy with other 2005. Tumor cells convert immature myeloid dendritic cells immunotherapies. Trends Immunol. 22, 113-115. into TGF-beta-secreting cells inducing CD4+CD25+ regulatory Nestle, F.O., Burg, G., Fah, ]., Wrone-Smith, T., Nickoloff, B.]., 1997. T cell proliferation. J. Exp. Med. 202, 919-929. Human sunlight-induced basal-cell-carcinoma-associated

Please cite tbis article in press as: Malvicini, M. et al, Reversa} of gastrointestinal carcinoma-induced irnmunosuppression and induction of antitumoural irnmunity by a combination of cyclophosphamide and gene transfer of U.-12, Molecular Oncology cl2Ql!),gpi:10.1,llt§'J,molonc.20tl:m,Q97 14 MOLECULAR ONCOLOGY XXX (2011) 1-14

dendritic cells are deficient in T cell co-stimulatory molecules Spaapen, R.M., Groen, R.W., van den Oudenalder, K., Guichelaar, T., and are impaired as antigen-presenting cells. Am. ]. Pathol. van Elk, M., Aarts-Riemens, T., Bloem, A.C., Storm, G., 150, 641-651. Martens, A.C., Lokhorst, H.M., Mutis, T., 2010. Eradication of Onishi, Y., Fehervari, Z., Yamaguchi, T., Sakaguchi, S., 2008. medullary multiple myeloma by CD4+ cytotoxic human T Foxp3+ natural regulatory T cells preferentially form lymphocytes directed ata single minor histocompatibility aggregates on dendritic cells in vitre and actively inhibit their antigen. Clin. Cancer Res. 16, 5481-5488. maturation. Proc. Natl. Acad. Sci. U S A 105, 10113-10118. Suzuki, E., Ka peor, V., Jassar, A. S., Kaiser, L.R., Albelda, S.M., 2005. Ormandy, L.A., Hillemann, T., Wedemeyer, H., Manns, M.P., Gemcitabine selectively eliminates splenic Gr-1+/CD11b+ Greten, T.F., Korangy, F., 2005. Increased populations of myeloid suppressor cells in tumor-bearing animals and regulatory T cells in peripheral blood of patients with enhances antitumor immune activity. Clin. Cancer Res. 11, hepatocellular carcinoma. Cancer Res. 65, 2457-2464. 6713-6721. Proietti, E., Greco, G., Garrone, B., Baccarini, S., Mauri, C., Takahashi, T., Kuniyasu, Y., Tod¡¡, M., Sakaguchi, N., Itoh, M., Venditti, M., Carlei, D., Belardelli, F., 1998. Importance of lwata, M., Shimizu, J., Sakaguchi, S., 1998. Immunologic self­ cyclophosphamide-induced bystander effect on T cells for tolerance maintained by CD25+CD4+ naturally anergic and a successful tumor eradication in response to adoptive suppressive T cells: induction of autoimmune disease by immunotherapy in mice. J. Clin. Invest. 101, 429-441. breaking their anergic/suppressive state. Int. Immunol. 10, Putzer, B.M., Stiewe, T., Rodicker, F., Schildgen, 0., Ruhm, S., 1969-1980. Dirsch, 0., Fiedler, M., Damen, U., Tennant, B., Scherer, C., Tang, Q., Bluestone, ].A., 2008. The Foxp3+ regulatory T cell: a jack Graham, F.L., Roggendorf, M., 2001. Large nontransplanted of all trades, master of regulation. Na t. Immunol. 9, 239-244. hepatocellular carcinoma in woodchucks: treatrnent with Thomton, A.M., Shevach, E.M., 1998. CD4+CD25+ adenovirus-mediated delivery of interleukin 12/B7.1 genes. J. immunoregulatory T cells suppress polyclonal T cell Natl. Cancer lnst. 93, 472-479. activation in vitre by inhibiting interleukin 2 production. ]. Rabinovich, G.A., Gabrilovich, D., Sotomayor, E.M., 2007. Exp. Med. 188, 287-296. Immunosuppressive strategies that are mediated by tumor Trinchieri, G., 2003. Interleukin-12 and the regulation of innate cells. Annu. Rev. Immunol. 25, 267-296. resistance and adaptive immunity. Nat. Rev. Immunol. 3, Sakaguchi, S., Sakaguchi, N., Shimizu, J., Yamazaki, S., 133-146. Sakihama, T., ltoh, M., Kuniyasu, Y., Nomura, T., Toda, M., Vignali, D.A., Collison, L.W., Workman, C.]., 2008. How regulatory Takahashi, T., 2001. Immunologic tolerance maintained by T cells work. Nat. Rev. Immunol. 8, 523-532. CD25+ CD4+ regulatory T cells: their common role in Yokoyama, Y., Dhanabal, M., Griffioen, A.W., Sukhatrne, V.P., controlling autoimmunity, tumor immunity, and Ramakrishnan, S., 2000. Synergy between angiostatin and transplantation tolerance. Immunol. Rev. 182, 18-32. endostatin: inhibition of ovarian cancer growth. Cancer Res. Sangro, B., Mazzolini, G., Ruiz, J., Herraiz, M., Quiroga, J., 60, 2190-2196. Herrero, !., Benito, A., Larrache, J., Pueyo, J., Subtil, J.C., Zou, W., 2005. Immunosuppressive networks in the tumour Olague, C., Sola, J., Sadaba, B., Lacasa, C., Melero, !., Qian, C., environment and their therapeutic relevance. Nat. Rev. Prieto, J., 2004. Phase I tria! of intratumoral injection of an Cancer 5, 263-274. adenovirus encoding interleukin-12 for advanced digestive Zou, W., 2006. Regulatory T cells, tumour immunity and tumors. J. Clin. Once!. 22, 1389-1397. immunotherapy. Nat. Rev. Immunol. 6, 295-307. Shapira, S., Lisiansky, V., Arber, N., Kraus, S., 2010. Targeted Zwirner, N.W., Croci, D.O., Domaica, C.!., Rabinovich, G.A., 2010. immunotherapy for colorectal cancer: monoclonal antibodies Overcoming the hurdles of tumor immunity by targeting and immunotoxins. Expert Opin. Investig. Drugs 19 (Suppl1), regulatory pathways in innate and adaptive immune cells. S67-S77. Curr. Pharm. Des 16, 255-267.

Please cite this anide in press as: Malvícini, M. et al., Reversa) of gastrointestinal carcinoma -induced "ímmunosuppression and induction of antitumoural immunity by a combination of cyclophosphamíde and gene transfer of ll.-12, Molecular Oncology 1{6Qll,),"~Qi,:1Q:1Q16/j.molon~ ..2011.03.007