京都大学 / 0048 / 新制・課程博士 / 博士(医科学) / 甲第22006号 / 医科博第104号 / 新制||医科||7(附属図書館) / 京都大学大学院医学研究科医科学専攻 / (主査)教授 井上 治久, 教授 伊佐 正, 教授 妻木 範行 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Deep sequencing of mRNAs (RNA-Seq) is now the preferred method for transcriptome-wide quantification of gene expression. Yet many mRNA isoforms, such as those eliminated by nonsense-mediated decay (NMD), are inherently unstable. Thus a significant drawback of steady-state RNA-Seq is that it provides marginal information on the flux through alternative splicing pathways. Measurement of such flux necessitates capture of newly made species prior to mRNA decay. One means to capture nascent mRNAs is affinity purifying either the exon junction complex (EJC) or activated spliceosomes. Late-stage spliceosomes deposit the EJC upstream of exon-exon junctions, where it remains associated until the first round of translation. As most mRNA decay pathways are translation-dependent, these EJC- or spliceosome-associated, pre-translational mRNAs should provide an accurate record of the initial population of alternate mRNA isoforms. Previous work has analyzed the protein composition and structure of pre- translational mRNPs in detail. While in the Moore lab, my project has focused on exploring the diversity of mRNA isoforms contained within these complexes. As expected, known NMD isoforms are more highly represented in pre-translational mRNPs than in RNA-Seq libraries. To investigate whether pre-translational mRNPs contain novel mRNA isoforms, we created a bioinformatics pipeline that identified thousands of previously unannotated splicing events. Though many can be attributed to “splicing noise”, others are evolutionarily-conserved events that produce new AS-NMD isoforms likely involved in maintenance of protein homeostasis. Several of these occur in genes whose overexpression has been linked to poor cancer prognosis.

No description available.

No description available.

The overall objective of this study has been to address some of the longstanding questions concerning functional organization of the interphase nucleus. This was achieved by using recently developed high-resolution fluorescence in situ hybridization techniques for a precise localization of specific DNA and RNA sequences in conjunction with immunocytochemistry and biochemical fractionation. This study is based on the philosophy that new insights may be gained by an approach that attempts to interrelate genomic organization, spatial arrangement of RNA metabolism, and nuclear substructure within the mammalian cell nucleus. The nuclear distribution of an exogenous, viral RNA (Epstein-Barr Virus, EBV) within nuclear matrix preparations was studied by developing an approach which couples in situhybridization with biochemical fractionation procedures. EBV RNA molecules accumulate in highly localized foci or elongated tracks within the nucleus of lymphoma cells. These RNA tracks were retained with spatial and quantitative fidelity in nuclear matrix preparations even after biochemical fractionation which removes 95% of cellular protein, DNA, and phospholipid. This provided direct evidence that the primary transcripts are localized via their binding to, or comprising part of, a non-chromatin nuclear substructure. Then the nuclear distribution of RNA from an endogenous gene, fibronectin, was investigated using fluorescence techniques modified for more sensitive detection of endogenous RNAs within nuclear morphology. A series of in situhybridization experiments were performed using different combinations of intron, cDNA, and genomic probes for RNA/RNA or RNA/DNA analysis in intact cells. Fibronectin RNAs were highly localized in the nucleus, forming foci or tracks. Both intron and exon sequences were highly concentrated at the same site within the nucleus, indicating the presence of primary unspliced transcripts. Double-color hybridization using a nontranscribed 5' flanking sequence probe and a genomic DNA probe showed that the gene and RNA track for fibronectin were spatially overlapped, with the gene consistently towards one end of the track. These results provided evidence that the accumulation of RNA molecules occurs directly at or near the site of transcription, and further indicated a structural polarity to the RNA track formation with the gene towards one end. It was further discovered that within a single cell, cDNA probes produced longer tracks than those formed with intron probes, i.e. intron signals were generally confined to a smaller part of the track than the exon signals, indicating that splicing occurs within the RNA track. Additional experiments using poly(A) RNA hybridization or anti-SC-35 antibody staining combined with fibronectin RNA hybridization have shown that the fibronectin tracks were associated with recently discovered transcript domains enriched in poly(A) RNA and splicing factors. To further determine whether other specific genes and RNAs are functionally organized within the nucleus, the nuclear distribution of several active or inactive genes was analyzed in terms of their spatial relationship to transcript domains. The results indicated that in addition to fibronectin, the genes or their primary transcripts from two other active genes, collagen and actin, were also closely associated with the domains. For both of these, over 90% of the gene/RNA sites were either overlapping or directly contacting the domains. In contrast. for two inactive genes, cardiac myosin heavy chain and neurotensin, it was found that both genes were separated from the domains in the majority of nuclei. Histone genes, which have several unique features, showed a relatively complex result with about half of the gene signals extremely close to the domains. Therefore, three actively expressed genes were demonstrated to be tightly associated with the domains and, moreover, their RNAs showed distinct and characteristic spatial relationships with the domains. In contrast, two inactive genes were not associated with the domains. One potential implication of these finding is that active genes may be preferentially localized in and around these transcript domains. The nuclear localization of another RNA, XIST, standing for X-inactivation specific transcript, was studied because of its potentially unique biological role. XIST is the only gene which is known to be expressed from the inactive human X chromosome but not from the active X chromosome, and was believed to be important in X inactivation. Using fluorescence in situhybridization, it was found that XIST RNA was highly localized within the nucleus and always completely overlapped the Barr body which is the condensed, inactive X chromosome. The different fine distribution pattern of XIST RNA within the nucleus as compared to other protein coding RNAs suggested a unique function for this RNA, possibly involving a structural role in inactivating the X chromosome. The final area of my thesis research was to study and acquire expertise in the applications of fluorescence in situ hybridization in gene mapping and cancer genetics. A retinoblastoma (RB)-related putative tumor suppressor gene, p107, was mapped to human chromosome 20 in band q11.2. Localization of p107 to 20q11.2 was of particular interest because of the correlation of breakpoints in this area with specific myeloid disorders such as acute nonlymphocytic leukemia and myelodysplastic syndrome. Other applications of in situ hybridization including the search for unknown genes at a known chromosomal breakpoint, detection of deletions, translocations or other chromosomal rearrangements associated with specific tumors were also explored and reviewed.

La plupart des gènes humains codants génèrent des transcrits alternatifs (isoformes) par épissage alternatif (alternative splicing, AS) et polyadénylation alternative (APA) en général dans la région codante et la région 3’ non traduite (3’UTR), respectivement. Le rôle de l’AS et la 3’UTR-APA est de plus en plus reconnu dans l’oncogenèse. En particulier, des réseaux d’AS connectant des facteurs d’épissage et des variants d’épissage ont récemment été identifiés. L’AS est aussi largement régulé par les agents anticancéreux génotoxiques, tel que la doxorubicine et le cisplatine (induisant des différents types de lésions sur l’ADN), qui sont régulièrementt utilisés dans les traitements du cancer du sein et du poumon non-à-petites-cellules (non-small-cell lung cancer, NSCLC), respectivement. Étant donné l’apparition fréquente de résistances aux chimiothérapies, comprendre les mécanismes sous-jacents est crucial pour surmonter ce problème clinique. Il existe des exemples d’évènements d’AS associés à la résistance aux agents anticancéreux, mais l’implication des facteurs d’épissage et des réseaux d’AS est très peu connue. De plus, une étude précédente a démontré que la doxorubicine réprime un grand groupe d’exon terminaux alternatifs (alternative last exons, ALE), qui correspondent à l’utilisation de sites de polyadénylation introniques (intronic polyadenylation, IPA). Les ALEs ont un rôle émergent dans le cancer, mais on ne sait encore que très peu sur leur régulation par d’autres agents anticancéreux, tel que le cisplatine. Afin de mieux comprendre le rôle des régulations d’AS et d’APA dans la réponse et la résistance cellulaire à la chimiothérapie, mon projet de thèse avait deux objectifs principaux : 1) déterminer l’étendue, les réseaux régulateurs, et les fonctions des régulations d’AS dans la résistance à la doxorubicine des cellules de cancer du sein, et 2) déterminer l’étendue, les mécanismes, et l’impact des régulations d’ALE en réponse au cisplatine dans des cellules de NSCLC. Dans la première partie, j’ai identifié par RNA-seq des milliers d’évènements d’AS et des dizaines de facteurs d’épissage régulés dans un modèle cellulaire de cancer du sein ER+ résistant à la doxorubicine. Par un miniscreen siARN, j’ai identifié deux facteurs, ZRANB2 et SYF2, impliqués dans la résistance à la doxorubicine. D’autres analyses RNA-seq ont révélé les évènements d’AS régulés par ces deux facteurs peu étudiés, ainsi que leur convergence vers l’exon 5 alternatif de l’oncogène ECT2. La déplétion de ZRANB2, SYF2, et du variant ECT2-ex5 réduit l’arrêt en phase S induit par la doxorubicine et la résistance des cellules. De plus, un niveau élevé d’inclusion de l’exon 5 d’ECT2 corrèle avec une mauvaise survie spécifiquement de patientes ER+ traitées par chimiothérapie. Dans la deuxième partie, j’ai identifié par 3’-seq que le traitement cisplatine (mais pas oxaliplatine) induit des ALEs/IPAs dans des milliers de gènes enrichis en gènes de cycle et de mort cellulaire. Cet effet est lié à une inhibition de la processivité de l’élongation dans les longs gènes. Une analyse 3’-seq sur polysomes m’a permis de montrer que ces régulations d’ALEs impactent le traductome, et a révélé un groupe d’isoformes particulièrement courtes peu efficacement traduites, dont un transcrit connu avec une fonction non-codante. En conclusion, j’ai pu identifier durant ma thèse un nouveau réseau d’AS impliqué dans la résistance à la doxorubicine des cancers du sein ER+, et une importante régulation d’ALEs impactant le traductome en réponse au cisplatine dans des cellules NSCLC. Ces travaux améliorent notre compréhension du rôle de l’AS et des ALE/IPA dans la réponse et la résistance cellulaire à la chimiothérapie anticancéreuse. Au plus long terme, les transcrits alternatifs et les régulateurs identifiés constituent des biomarqueurs candidats de chimiorésistance. / Most human coding genes generate alternative transcripts (isoforms) through alternative splicing (AS) and alternative polyadenylation (APA), most often within the coding region and the 3’ untranslated region (3’UTR), respectively. Both AS and 3’UTR-APA regulations have been increasingly involved in oncogenesis. In particular, AS networks connecting oncogenic splicing factors and oncogenic splicing variants have been recently identified. AS is also widely regulated by genotoxic anticancer drugs, like doxorubicin and cisplatin that induce different types of DNA lesions and are widely used in breast cancer and non-small-cell lung cancer (NSCLC) therapy, respectively. Given the frequent occurrence of resistance to chemotherapy, understanding the underlying mechanisms is crucial to overcome this major issue. There are examples of AS events associated with anticancer drug resistance, but very little is known about the splicing factors and therefore the AS networks involved. In addition, a previous study showed that doxorubicin represses a large set of alternative last exons (ALE) corresponding to the use of intronic polyadenylation (IPA) sites. ALEs have an emerging role in cancer, but little is known about its regulation by other anticancer drugs, like cisplatin. In order to better understand the role of AS and APA regulation in cell response and resistance to chemotherapy, my PhD project had two main aims: 1) determine the extent, regulatory networks and function of AS regulation in breast cancer cell resistance to doxorubicin, and 2) determine the extent, mechanism and impact of ALE regulation in response to cisplatin in NSCLC cells. In the first part, I identified by RNA-seq thousands of AS events and dozens of splicing factors regulated in a cell model of acquired resistance to doxorubicin in ER+ breast cancer. Through an siRNA miniscreen, I found two splicing factors, ZRANB2 and SYF2, involved in doxorubicin resistance. Further RNA-seq analyses revealed the AS events regulated by depletion of these poorly characterized splicing factors, and their convergence on the alternative exon 5 of the oncogene ECT2. Depletion of ZRANB2, SYF2 and the ECT2-Ex5 variant reduces doxorubicin-induced S phase arrest and doxorubicin resistance. In addition, high inclusion levels of ECT2-Ex5 correlate with poor survival specifically in ER+ breast cancer treated with chemotherapy. In the second part, I found by 3’-seq that in NSCLC cell treatment with cisplatin (but not oxaliplatin) induces ALE/IPA in thousands of genes enriched in cell cycle and cell death. This effect is linked to an inhibition of transcription elongation processivity in long genes. 3’-seq analysis on polysomes showed that this ALE regulation impacts the translatome, and revealed a set of particularly short isoforms that were inefficiently translated, including a transcript with a non-coding function. In conclusion, during my thesis, I could identify a novel AS network involved in doxorubicin resistance in ER+ breast cancer, and widespread ALE regulation impacting the translatome in lung cancer cisplatin response. This work increases our understanding of AS and IPA role in cell response and resistance to anti-cancer chemotherapy. In the longer term, the identified alternative transcripts and regulators constitute candidate biomarkers of chemoresistance.

Le cancer colorectal (CRC) est l'un des cancers les plus agressifs. Afin de rechercher de nouveaux marqueurs de progression de ce cancer, des puces tout transcriptome et épissage, ainsi que des puces microARNs, ont été précédemment utilisées au laboratoire.Dans une première partie, l’analyse des données transcriptomiques a permis de trouver un ensemble de transcrits dérégulés dans les adénomes et dans le CRC, dont la surexpression de TIMP1, associée à la diminution de rétention de l’intron 3, par rapport à la muqueuse normale. Ce transcrit aberrant n’est pas sujet au mécanisme de dégradation active des ARNm («Nonsense-Mediated mRNA Decay »). Nous avons alors analysé les mécanismes d’épissage de TIMP1, ce qui nous a conduits à identifier hnRNPA1 comme un régulateur important, à la fois in vitro et in vivo, de la rétention de l’intron 3 de TIMP1, via sa fixation au début de l’exon 4. Le rôle de TIMP1i3 (+) dans la progression du CRC reste à identifier.Dans une seconde partie, l’analyse des données de la puce microARNs (miRs) a permis de trouver un ensemble de miRs dérégulés dans les CRC. Nous avons sélectionné des gènes codant des facteurs d’épissage dont l’expression était modifiée au cours de la progression cancéreuse, en tant que cibles possibles de certains de ces miRs. Six interactions ont été montrées (PRMT5/miR145; SRSF6/miR375; RBMX/miR23a; RBMX/miR24; RBMX/miR125a5p; SRSF11/miR143).En utilisant la technique de « Luciferase Reporter Gene Assay » couplée à la mutagenèse dirigée, nous avons montré que le miR145 régule négativement l’expression de PRMT5 par interaction avec la région 3’ non traduite du gène. Du fait du contrôle de la machinerie d’épissage des ARN prémessagers par PRMT5, nous proposons que miR145 pourrait être un régulateur général de l’épissage. / Colorectal cancer (CRC) is one of the most vaggressive cancers in the world. In order to look for new markers in progression of this cancer, all transcriptome and splice chips, as well as microRNA chips, have been previously used in the laboratory.In the first part of my thesis, transcriptomic data analysis revealed a set of deregulated transcripts in adenoma and in CRC, including overexpression of TIMP1, associated with the decrease of intron 3 retention, compared to normal mucosae. This aberrant transcript is not subject to the active mechanism of mRNAs degradation ("Nonsense-Mediated mRNA Decay"). Then we analysed the splicing mechanisms of TIMP1, which led us to identify hnRNPA1 as a major regulator, both in vitro and in vivo, of TIMP1 intron 3 retention via its binding in the beginning of exon 4. TIMP1i3 (+) role in the progression of CRC remains to be identified.In a second part, the analysis of microRNA data (miRs) revealed a set of deregulated miRs in CRC. The aim was to identify target genes for these miRs.We selected genes encoding splice factors whose expression was modified during cancer progression as possible targets for some of these miRs. Six interactions were shown (PRMT5/miR145; SRSF6/miR375; RBMX/miR23a; RBMX/miR24; RBMX/miR125a5p; SRSF11/miR143).Using the "Luciferase Reporter Gene Assay" technique coupled with site directed mutagenesis, we have shown that miR145 negatively regulated the expression of PRMT5 by interaction with its 3’untranslated translated region. Due to the control of the premessenger RNA splicing machinery by PRMT5, we propose that miR145 could be a general splicing regulator.

Filip Lhota: Study of alternatively spliced variants of estrogen receptor alpha in breast cancer cell lines Abstract: Estrogen receptor α (ER-α) is a transcription factor responsible for mediation of the activities of its natural ligand 17-β-estradiol (E2), the hormone that together with progesterone belongs to the key regulators of mammary epithelial as well as breast cancer cells proliferation. Except to the major gene product consisting of all eight coding exons of ER-α, numerous qualitatively and quantitatively different spliced variants originated from primary transcript by activity of alternative splicing is expressed. Despite that some of these spliced variants have been functionally characterized, their precise role on final ER-α cellular activity remains to be elucidated. The functional characterization of individual alternative forms of ER-α and description of its participation on the overall ER-α activity is important for our understanding of their biogenesis and is also critical for the delineation of molecular bases for ER-α regulation during anti cancer chemotherapy. This work aimed to study the influence of alternatively spliced ER-α variants on the growth characteristics of clones constructed from stable mammary tissue cell lines in regulation to cultivation conditions and cellular...

U5 small ribonucleoprotein particle (U5 snRNP) is a crucial component of the spliceosome, the complex responsible for pre-mRNA splicing. Despite the importance of U5 snRNP, not much is known about its biogenesis. When we depleted one of the core U5 components, protein PRPF8, the other U5-specific proteins do not associate with U5 snRNA and the incomplete U5 was accumulated in nuclear structures known as Cajal bodies. To further clarify the role of PRPF8 in U5 snRNP assembly, we studied PRPF8 mutations that cause an autosomal dominant retinal disorder, retinitis pigmentosa (RP). We prepared eight different PRPF8 variants carrying RP-associated mutations and expressed them stably in human cell culture. We showed that most mutations interfere with the assembly of snRNPs which consequently leads to reduced efficiency of splicing. The mutant PRPF8 together with EFTUD2 are stalled in the cytoplasm in a form of U5 snRNP assembly intermediate. Strikingly, we identified several chaperons including the HSP90/R2TP complex and ZNHIT2 as new PRPF8's interactors and potential U5 snRNP assembly factors. Our results further imply that these chaperons preferentially bind the unassembled U5 complexes and that HSP90 is required for stability of...

This paper investigates the use of deep Convolutional Neural Networks for modeling the intronic regulation of Alternative Splicing on the basis of DNA sequence. By training the CNN on massively parallel synthetic DNA libraries of Alternative 5'-splicing and Alternatively Skipped exon events, the model is capable of predicting the relative abundance of alternatively spliced mRNA isoforms on held-out library data to a very high accuracy (R2 = 0.77 for Alt. 5'-splicing). Furthermore, the CNN is shown to generalize alternative splicing across cell lines efficiently. The Convolutional Neural Net is tested against a Logistic regression model and the results show that while prediction accuracy on the synthetic library is notably higher compared to the LR model, the CNN is worse at generalizing to new intronic contexts. Tests on non-synthetic human SNP genes suggest the CNN is dependent on the relative position of the intronic region it was trained for, a problem which is alleviated with LR. The increased library prediction accuracy of the CNN compared to Logistic regression is concluded to come from the non-linearity introduced by the deep layer architecture. It adds the capacity to model complex regulatory interactions and combinatorial RBP effects which studies have shown largely affect alternative splicing. However, the architecture makes interpreting the CNN hard, as the regulatory interactions are encoded deep within the layers. Nevertheless, high-performance modeling of alternative splicing using CNNs may still prove useful in numerous Synthetic biology applications, for example to model differentially spliced genes as is done in this paper. / Den här uppsatsen undersöker hur djupa neurala nätverk baserade på faltning ("Convolutions") kan användas för att modellera den introniska regleringen av Alternativ Splicing med endast DNA-sekvensen som indata. Nätverket tränas på ett massivt parallelt bibliotek av syntetiskt DNA innehållandes Alternativa Splicing-event där delar av de introniska regionerna har randomiserats. Uppsatsen visar att nätverksarkitekturen kan förutspå den relativa mängden alternativt splicat RNA till en mycket hög noggrannhet inom det syntetiska biblioteket. Modellen generaliserar även alternativ splicing mellan mänskliga celltyper väl. Hursomhelst, tester på icke-syntetiska mänskliga gener med SNP-mutationer visar att nätverkets prestanda försämras när den introniska region som används som indata flyttas i jämförelse till den relativa position som modellen tränats på. Uppsatsen jämför modellen med Logistic regression och drar slutsatsen att nätverkets förbättrade prestanda grundar sig i dess förmåga att modellera icke-linjära beroenden i datan. Detta medför dock svårigheter i att tolka vad modellen faktiskt lärt sig, eftersom interaktionen mellan reglerande element är inbäddat i nätverkslagren. Trots det kan högpresterande modellering av alternativ splicing med hjälp av neurala nät vara användbart, exempelvis inom Syntetisk biologi där modellen kan användas för att kontrollera regleringen av splicing när man konstruerar syntetiska gener.