New approaches to open problems in gene expression microarray data

In the past decade, the advent of efficient genome sequencing tools and high-throughput experimental biotechnology has lead to enormous progress in the life science. Among the most important innovations is the microarray tecnology. It allows to quantify the expression for thousands of genes simultaneously by measurin the hybridization from a tissue of interest to probes on a small glass or plastic slide. The characteristics of these data include a fair amount of random noise, a predictor dimension in the thousand, and a sample noise in the dozens. One of the most exciting areas to which microarray technology has been applied is the challenge of deciphering complex disease such as cancer. In these studies, samples are taken from two or more groups of individuals with heterogeneous phenotypes, pathologies, or clinical outcomes. these samples are hybridized to microarrays in an effort to find a small number of genes which are strongly correlated with the group of individuals. Eventhough today methods to analyse the data are welle developed and close to reach a standard organization (through the effort of preposed International project like Microarray Gene Expression Data -MGED- Society [1]) it is not unfrequant to stumble in a clinician's question that do not have a compelling statistical method that could permit to answer it.The contribution of this dissertation in deciphering disease regards the development of new approaches aiming at handle open problems posed by clinicians in handle specific experimental designs. In Chapter 1 starting from a biological necessary introduction, we revise the microarray tecnologies and all the important steps that involve an experiment from the production of the array, to the quality controls ending with preprocessing steps that will be used into the data analysis in the rest of the dissertation. While in Chapter 2 a critical review of standard analysis methods are provided stressing most of problems that In Chapter 3 is introduced a method to adress the issue of unbalanced design of miacroarray experiments. In microarray experiments, experimental design is a crucial starting-point for obtaining reasonable results. In a two-class problem, an equal or similar number of samples it should be collected between the two classes. However in some cases, e.g. rare pathologies, the approach to be taken is less evident. We propose to address this issue by applying a modified version of SAM [2]. MultiSAM consists in a reiterated application of a SAM analysis, comparing the less populated class (LPC) with 1,000 random samplings of the same size from the more populated class (MPC) A list of the differentially expressed genes is generated for each SAM application. After 1,000 reiterations, each single probe given a "score" ranging from 0 to 1,000 based on its recurrence in the 1,000 lists as differentially expressed. The performance of MultiSAM was compared to the performance of SAM and LIMMA [3] over two simulated data sets via beta and exponential distribution. The results of all three algorithms over low- noise data sets seems acceptable However, on a real unbalanced two-channel data set reagardin Chronic Lymphocitic Leukemia, LIMMA finds no significant probe, SAM finds 23 significantly changed probes but cannot separate the two classes, while MultiSAM finds 122 probes with score >300 and separates the data into two clusters by hierarchical clustering. We also report extra-assay validation in terms of differentially expressed genes Although standard algorithms perform well over low-noise simulated data sets, multi-SAM seems to be the only one able to reveal subtle differences in gene expression profiles on real unbalanced data. In Chapter 4 a method to adress similarities evaluation in a three-class prblem by means of Relevance Vector Machine [4] is described. In fact, looking at microarray data in a prognostic and diagnostic clinical framework, not only differences could have a crucial role. In some cases similarities can give useful and, sometimes even more, important information. The goal, given three classes, could be to establish, with a certain level of confidence, if the third one is similar to the first or the second one. In this work we show that Relevance Vector Machine (RVM) [2] could be a possible solutions to the limitation of standard supervised classification. In fact, RVM offers many advantages compared, for example, with his well-known precursor (Support Vector Machine - SVM [3]). Among these advantages, the estimate of posterior probability of class membership represents a key feature to address the similarity issue. This is a highly important, but often overlooked, option of any practical pattern recognition system. We focused on Tumor-Grade-three-class problem, so we have 67 samples of grade I (G1), 54 samples of grade 3 (G3) and 100 samples of grade 2 (G2). The goal is to find a model able to separate G1 from G3, then evaluate the third class G2 as test-set to obtain the probability for samples of G2 to be member of class G1 or class G3. The analysis showed that breast cancer samples of grade II have a molecular profile more similar to breast cancer samples of grade I. Looking at the literature this result have been guessed, but no measure of significance was gived before.

Depicting the role of CDKN2A/ARF alterations in adult BCR-ABL1-positive acute lymphoblastic leukemia patients: from genomic deletions to prognostic impact

This 9p21 locus, encode for important proteins involved in cell cycle regulation and apoptosis containing the p16/CDKN2A (cyclin-dependent kinase inhibitor 2a) tumor suppressor gene and two other related genes, p14/ARF and p15/CDKN2B. This locus, is a major target of inactivation in the pathogenesis of a number of human tumors, both solid and haematologic, and is a frequent site of loss or deletion also in acute lymphoblastic leukemia (ALL) ranging from 18% to 45% 1. In order to explore, at high resolution, the frequency and size of alterations affecting this locus in adult BCR-ABL1-positive ALL and to investigate their prognostic value, 112 patients (101 de novo and 11 relapse cases) were analyzed by genome-wide single nucleotide polymorphisms arrays and gene candidate deep exon sequencing. Paired diagnosis-relapse samples were further available and analyzed for 19 (19%) cases. CDKN2A/ARF and CDKN2B genomic alterations were identified in 29% and 25% of newly diagnosed patients, respectively. Deletions were monoallelic in 72% of cases and in 43% the minimal overlapping region of the lost area spanned only the CDKN2A/2B gene locus. The analysis at the time of relapse showed an almost significant increase in the detection rate of CDKN2A/ARF loss (47%) compared to diagnosis (p = 0.06). Point mutations within the 9p21 locus were found at very low level with only a non-synonymous substition in the exon 2 of CDKN2A. Finally, correlation with clinical outcome showed that deletions of CDKN2A/B are significantly associated with poor outcome in terms of overall survival (p = 0.0206), disease free-survival (p = 0.0010) and cumulative incidence of relapse (p = 0.0014). The inactivation of 9p21 locus by genomic deletions is a frequent event in BCR-ABL1-positive ALL. Deletions are frequently acquired at the leukemia progression and work as a poor prognostic marker.

Pre-clinical and clinical development of leukemia stem cell inhibitors in acute leukemias

In Leukemias, recent developments have demonstrated that the Hedgehog pathway plays a key-role in the peculiar ability of self renewal of leukemia stem cells. The aim of this research activity was to investigate, through a first in man, Phase I, open label, clinical trial, the role and the impact, mainly in terms of safety profile, adverse events and pharmacokinetics, of a Sonic Hedgehog inhibitor compound on a population of heavely pretreated patients affected by AML, CML, MF, or MDS, resistant or refractory to standard chemotherapy. Thirty-five patients have been enrolled. The drug was administered orally, in 28 days cycles, without rest periods. The compound showed a good safety profile. The half life was of 17-35 hours, justifying the daily administration. Significant signs of activity, in terms of reduction of bone marrow blast cell amount were seen in most of the patients enrolled. Interestingly, correlative biological studies demonstrated that, comparing the gene expression profyiling signature of separated CD34+ cells before and after one cycle of treatment, the most variably expressed genes were involved in the Hh pathway. Moreover, we observed that many genes involved in MDR (multidrug resistance)were significantly down regulated after treatment. These data might lead to future clinical trials based on combinatory approaches, including, for instance, Hh inhibitors and conventional chemotherapy.

Targeting the p53–MDM2 interaction by the small-molecule MDM2 antagonist Nutlin-3a: a new challenged target therapy in adult Philadelphia positive acute lymphoblastic leukemia patients

The human p53 tumor suppressor, known as the “guardian of the genome”, is one of the most important molecules in human cancers. One mechanism for suppressing p53 uses its negative regulator, MDM2, which modulates p53 by binding directly to and decreasing p53 stability. In testing novel therapeutic approaches activating p53, we investigated the preclinical activity of the MDM2 antagonist, Nutlin-3a, in Philadelphia positive (Ph+) and negative (Ph-) leukemic cell line models, and primary B-Acute lymphoblastic leukemia (ALL) patient samples. In this study we demonstrated that treatment with Nutlin-3a induced grow arrest and apoptosis mediated by p53 pathway in ALL cells with wild-type p53, in time and dose-dependent manner. Consequently, MDM2 inhibitor caused an increase of pro-apoptotic proteins and key regulators of cell cycle arrest. The dose-dependent reduction in cell viability was confirmed in primary blast cells from Ph+ ALL patients with the T315I Bcr-Abl kinase domain mutation. In order to better elucidate the implications of p53 activation and to identify biomarkers of clinical activity, gene expression profiling analysis in sensitive cell lines was performed. A total of 621 genes were differentially expressed (p < 0.05). We found a strong down-regulation of GAS41 (growth-arrest specific 1 gene) and BMI1 (a polycomb ring-finger oncogene) (fold-change -1.35 and -1.11, respectively; p-value 0.02 and 0.03, respectively) after in vitro treatment as compared to control cells. Both genes are repressors of INK4/ARF and p21. Given the importance of BMI in the control of apoptosis, we investigated its pattern in treated and untreated cells, confirming a marked decrease after exposure to MDM2 inhibitor in ALL cells. Noteworthy, the BMI-1 levels remained constant in resistant cells. Therefore, BMI-1 may be used as a biomarker of response. Our findings provide a strong rational for further clinical investigation of Nutlin-3a in Ph+ and Ph-ALL.

Role of the Transcription Factor Sox2 in the Osteogenic Lineage

The Sox2 transcription factor is modified by sumoylation at the K247 position although the addition of SUMO1 and Pias1 promotes the sumoylation of Sox2 at the additional K123 site. The role of sumoylation on Sox2 biological functions was analyzed by comparing the activity of WT and sumoylation mutants on the transcription of the FGF4 gene in HeLa cells and on the downregulation of the Wnt pathwayvin 293T cells. When SUMO1 and PIAS1 promote the sumoylation of WT Sox2, the transcriptional activity of the FGF4 promoter is inhibited showing that Sox2 sumoylation is necessary for the repression function. However, there is no effect of Sox2 sumoylation on β-Catenin activity. Since we were interested in osteoblast differentiation we set up an inducible system for Sox2 in primary osteoblasts. Following Sox2 doxycycline induction, 158 genes were differentially expressed: 120 up-regulated and 38 down-regulated. We annotated as direct Sox2 targets a number of genes involved in osteoblast biology and we further analyzed 3 of them involved in the BMP pathway. The results show that Sox2 regulates the BMP pathway without affecting SMAD phosphorylation, and that Sox2 sumoylation is not necessary for this function. We also found that genes involved in the Hippo pathway were direct Sox2 targets. As the Hippo pathway is activated by Sox2 and Sox2 interacts with the NF2 promoter, we checked the effect of Sox2 on the expression of NF2. We showed that Sox2 down-regulates the transcriptional activity of the NF2 promoter, allowing the transcription of the YAP/TEAD genes in osteoblasts, thus acting as an upstream regulator of the Hippo pathway. We conclude that Sox2 induction in osteoblasts triggers FGF dependent inhibition of the BMP, Wnt and Hippo pathways.

Molecular and functional characterization of the interplay between malignant and stromal cells in acute myeloid leukemia and myelodysplastic syndrome

In this thesis, we studied the cross-talk between malignant cells and stromal cells, with the aim to elucidate the respective contribution to myeloid neoplasm onset and progression. First, we characterized and compared mesenchymal stromal cells (MSCs) isolated from myelodysplastic syndrome (MDS-MSCs) and acute myeloid leukemia (AML-MSCs) patients. We demonstrated that, despite some unaltered functions, patient-derived MSCs show also intrinsic, distinct functional abnormalities, which could all potentially favor a leukemia-protective bone marrow (BM) niche in vivo. Second, we investigated the ability of AML cells to modulate the AML-MSC functions. In a GEP-screening, we found that 40% of BM-derived AML samples show a higher IFN-γ expression, compared to the mean IFN-γ expression in healthy BM-derived cells. We demonstrated that in co-culture experiments, IFN-γ+ AML cells modify AML-MSC gene expression and function, inducing the up-regulation of IDO1, and consequently the generation of T regulatory cells. Finally, we wondered if the transcriptome of stromal cells could be influenced by the hematopoietic-specific alterations, i.e. Dnmt3a and Asxl1 mutations, which occur early in MDS/AML patients. We found that Dnmt3a- and Asxl1-null BM cells, when transplanted in wild-type mice, induce profound and deletion-specific modifications in the transcriptome of wild-type BM stromal cells, suggesting the ability of Dnmt3a- and Asxl1-null BM cells to shape the niche. Furthermore, we compared the transcriptome of wild-type BM stromal cells, obtained from transplantation experiments, with that of MSCs isolated from low-risk MDS patients with DNMT3A and ASXL1 mutations, and we highlighted some common modifications, which could be potentially relevant for human disease and specific for DNMT3A/ASXL1 mutations. In conclusion, this thesis pointed out that there is a bi-directional cross-talk, in which stromal cells can influence malignant cells, and in turn malignant/pre-malignant cells can alter stromal cell gene expression and function. Both mechanisms could potentially contribute to the pathogenesis of myeloid malignancies.

Exploring novel targeted therapeutic strategies against Acute Myeloid Leukemia cells based on inhibition of bromodomain proteins and CDC20 activity

Acute myeloid leukemia (AML) is a haematological malignancies arising from the accumulation of undifferentiated myeloid progenitors with an uncontrolled proliferation. The genomic landscape of AML revealed that the disease is characterized by high level of heterogeneity and is subjected to clonal evolution driven by selective pressure of chemotherapy. In this study, we investigated the therapeutic effects of the inhibition of BRD4 and CDC20 in vitro and ex vivo. We demonstrated that inhibition of BRD4 with GSK1215101A in AML cell lines was effective under hypoxia. It induced the activation of antioxidant response both, at transcriptomic and metabolomic levels, driven by enrichment of NRF2 pathway under normoxic and hypoxic condition. Moreover, the combined treatment with Omaveloxolone, a drug inducing NRF2 activation and NF-κB inhibition, potentiated the effects on apoptosis and colony forming capacity of stem progenitor cells. Lastly, gene expression profiling data revealed that combination treatment induced major changes in genes related to cell cycle, together with enrichment of cell differentiation pathways and negative regulation of WNT, in normoxia and hypoxia. Regarding CDC20, we observed its up-regulation in AML patients. Treatment with two different inhibitors, Apcin and proTAME, was effective in primary AML cells and in AML cell lines, through induction of apoptosis and mitotic arrest. The lack of correlation between proliferation markers and CDC20 levels in AML cell subpopulations supports the idea of alternative CDC20 functions, independent from its essential role during mitosis. CDC20-KD experiments conducted in AML cell lines revealed a mild effect on apoptosis induction, but no significant change in cell cycle progression. In summary, these results allowed the identification of a new strategy combination to improve the effects of BRD4 inhibition on LSC residing in the BM hypoxic niche, and provide some new evidence regarding the potential role of CDC20 as a new target for AML treatment.

Research of predictive biomarkers to anti-CD22 antibody-drug conjugate treatment in B-cell acute lymphoblastic leukemia

Background: The treatment of B-cell acute lymphoblastic leukemia (B-ALL) has been enriched by novel agents targeting surface markers CD19 and CD22. Inotuzumab ozogamicin (INO) is a CD22-calicheamicin conjugated monoclonal antibody approved in the setting of relapse/refractory (R/R) B-ALL able to induce a high rate of deep responses, not durable over time. Aims: This study aims to identify predictive biomarkers to INO treatment in B- ALL by flow cytometric analysis of CD22 expression and gene expression profile. Materials and methods: Firstly, the impact on patient outcome in 30 R/R B-ALL patients of baseline CD22 expression in terms of CD22 blast percentage and CD22 fluorescent intensity (CD22-FI) was explored. Secondly, baseline gene expression profile of 18 R/R B-ALL patient samples was analyzed. For statistical analysis of differentially expressed genes (DEGs) patients were divided in non-responders (NR), defined as either INO-refractory or with duration of response (DoR) < 3 months, and responders (R). Gene expression results were analyzed with Ingenuity pathway analysis (IPA). Results: In our patient set higher CD22-FI, defined as higher quartiles (Q2-Q4), correlated with better patient outcome in terms of CR rate, OS and DoR, compared to lower CD22-FI (Q1). CD22 blast percentage was less able to discriminate patients’ outcome, although a trend for better outcome in patients with CD22 ≥ 90% could be appreciated. Concerning gene expression profile, 32 genes with corrected p value <0.05 and absolute FC ≥2 were differentially expressed in NR as compared to R. IPA upstream regulator and regulator effect analysis individuated the inhibition of tumor suppressor HIPK2 as causal upstream condition of the downregulation of 6 DEGs. Conclusions: CD22-FI integrates CD22-percentage on leukemic blasts for a more comprehensive target pre-treatment evaluation. Moreover, a unique pattern of gene expression signature based on HIPK2 downregulation was identified, providing important insights in mechanisms of resistance to INO.

Dissecting and resetting SETD2/H3K36ME3 deficiency in chronic myeloid leukemia

Chronic myeloid leukemia (CML) is characterized by the presence of the BCR::ABL1 fusion gene, leading to a constitutively active tyrosine kinase that drives the disease. Genomic instability is a hallmark of CML, contributing to disease progression and treatment resistance. A study identified SETD2, a histone methyltransferase, as frequently dysfunctional in advanced-phase CML, resulting in reduced trimethylation of Histone H3 at lysine 36 (H3K36Me3). This loss is associated with poor prognosis and increased genetic instability. Investigations revealed that SETD2 dysfunction is caused by post-translational modifications mediated by Aurora kinase A and MDM2, leading to proteasome-mediated degradation. Aurora kinase A phosphorylates SETD2, while MDM2 ubiquitinates it, targeting it for degradation. Inhibition of MDM2 and Aurora kinase A restored SETD2 expression and activity, suggesting potential therapeutic targets. Loss of SETD2 and H3K36Me3 impairs DNA repair mechanisms, favoring error-prone repair pathways over faithful ones, exacerbating genetic instability. Reintroduction of SETD2 into deficient cells restored DNA repair pathways, preserving genomic integrity. Analysis of CD34+ progenitor cells from CML patients showed reduced SETD2 levels compared to healthy individuals, correlating with decreased clonogenic capacity. Notably, SETD2 loss is not detectable at diagnosis but emerges during disease progression, indicating its role as an early indicator of CML advancement. Therapeutically, inhibitors targeting Aurora kinase A, MDM2, and the proteasome showed efficacy in cells expressing SETD2, particularly in those with low SETD2 levels. Proteasome inhibitors induced apoptosis and DNA damage in SETD2-deficient cells, highlighting their potential for CML treatment. In conclusion, SETD2 acts as a tumor suppressor in CML, with its dysfunction contributing to genetic instability and disease progression. Targeting the mechanisms of SETD2 loss presents promising therapeutic avenues for controlling CML proliferation and restoring genomic integrity.