Acute myeloid leukemia (AML) is an aggressive and genetically heterogeneous hematological malignancy characterized by the accumulation of immature myeloid blasts that disrupt healthy hematopoiesis. Despite advances in molecular profiling and targeted therapies, overcoming drug resistance and relapse remains a significant clinical challenge, resulting in poor long-term outcomes. Crucially, disease persistence is sustained not merely by intrinsic genetic lesions but by a highly adaptive bone marrow microenvironment (BMME) that functions as a therapeutic barrier. While the healthy niche tightly regulates hematopoietic stem cell maintenance, leukemic blasts co-opt stromal, vascular, and immune components to establish a sanctuary that fuels proliferation and shields the disease from cytotoxic stress. However, dissecting these reciprocal dependency mechanisms uncovers critical vulnerabilities, presenting a vital opportunity to develop novel targeted therapies. In this review, we discuss the architecture of the healthy BMME and its pathological AML-driven remodeling. We describe the role of specific signaling axes that govern AML-BMME crosstalk and evaluate targeted therapeutic strategies designed to uncouple these protective interactions. Finally, we highlight that current preclinical models lack the complexity of the BMME stromal components and its spatial organization, a limitation that continues to hinder clinical translation and delay the development of effective combination therapies.

Combining mathematical modeling with experiments enables quantitative understanding of cell signaling, transcriptional regulation, and cell fate decisions. Here, we provide a systems biology approach to link signal transduction with B cells fate decisions, to enable quantitative prediction of B-cell proliferation, and differentiation. We describe methodology to run simulations that reveal how signal transduction regulates gene expression and predicts cell fate decision. We describe how to quantitively validate modeling predictions with wet-lab experiments.

Here we describe the design, synthesis and validation of the first NF-κB PROTAC degrader with demonstrable subunit selectivity for RelA/p65.

Decades of research into the molecular signalling determinants of B cell fates, and recent progress in characterising the genetic drivers of lymphoma, has led to a detailed understanding of B cell malignancies but also revealed daunting heterogeneity. While current therapies for diffuse large B-cell lymphoma are effective for some patients, they are largely agnostic to the biology of each individual's disease, and approximately one third of patients experience relapsed/refractory disease. Consequently, the challenge is to understand how each patient's mutational burden and tumour microenvironment combine to determine their response to treatment; overcoming this challenge will improve outcomes in lymphoma. This mini review highlights how data-driven modelling, statistical approaches and machine learning are being used to unravel the heterogeneity of lymphoma. We review how mechanistic computational models provide a framework to embed patient data within knowledge of signalling. Focusing on recurrently dysregulated signalling networks in lymphoma (including NF-κB, apoptosis and the cell cycle), we discuss the application of state-of-the-art mechanistic models to lymphoma. We review recent advances in which computational models have demonstrated the power to predict prognosis, identify promising combination therapies and develop digital twins that can recapitulate clinical trial results. With the future of treatment for lymphoma poised to transition from one-size-fits-all towards personalised therapies, computational models are well-placed to identify the right treatments to the right patients, improving outcomes for all lymphoma patients.

In Diffuse Large B-cell Lymphoma (DLBCL), elevated anti-apoptotic BCL2-family proteins (e.g., MCL1, BCL2, BCLXL) and NF-κB subunits (RelA, RelB, cRel) confer poor prognosis. Heterogeneous expression, regulatory complexity, and redundancy offsetting the inhibition of individual proteins, complicate the assignment of targeted therapy. We combined flow cytometry "fingerprinting", immunofluorescence imaging, and computational modeling to identify therapeutic vulnerabilities in DLBCL. The combined workflow predicted selective responses to BCL2 inhibition (venetoclax) and non-canonical NF-κB inhibition (Amgen16). Within the U2932 cell line we identified distinct resistance mechanisms to BCL2 inhibition in cellular sub-populations recapitulating intratumoral heterogeneity. Co-cultures with CD40L-expressing stromal cells, mimicking the tumor microenvironment (TME), induced resistance to BCL2 and BCLXL targeting BH3-mimetics via cell-type specific upregulation of BCLXL or MCL1. Computational models, validated experimentally, showed that basal NF-κB activation determined whether CD40 activation drove BH3-mimetic resistance through upregulation of RelB and BCLXL, or cRel and MCL1. High basal NF-κB activity could be overcome by inhibiting BTK to resensitize cells to BH3-mimetics in CD40L co-culture. Importantly, non-canonical NF-κB inhibition overcame heterogeneous compensatory BCL2 upregulation, restoring sensitivity to both BCL2- and BCLXL-targeting BH3-mimetics. Combined molecular fingerprinting and computational modelling provides a strategy for the precision use of BH3-mimetics and NF-κB inhibitors in DLBCL.

Background/Objectives: Acute myeloid leukemia (AML) is an aggressive neoplasm. Although most patients respond to induction therapy, they commonly relapse due to recurrent disease in the bone marrow microenvironment (BMME). So, the disruption of the BMME, releasing tumor cells into the peripheral circulation, has therapeutic potential. Methods: Using both primary donor AML cells and cell lines, we developed an in vitro co-culture model of the AML BMME. We used this model to identify the most effective agent(s) to block AML cell adherence and reverse adhesion-mediated treatment resistance. Results: We identified that anti-CD44 treatment significantly increased the efficacy of cytarabine. However, some AML cells remained adhered, and transcriptional analysis identified focal adhesion kinase (FAK) signaling as a contributing factor; the adhered cells showed elevated FAK phosphorylation that was reduced by the FAK inhibitor, defactinib. Importantly, we demonstrated that anti-CD44 and defactinib were highly synergistic at diminishing the adhesion of the most primitive CD34high AML cells in primary autologous co-cultures. Conclusions: Taken together, we identified anti-CD44 and defactinib as a promising therapeutic combination to release AML cells from the chemoprotective AML BMME. As anti-CD44 is already available as a recombinant humanized monoclonal antibody, the combination of this agent with defactinib could be rapidly tested in AML clinical trials.

Introduction: Multiple Myeloma (MM) is a heterogeneous disease making it difficult to accurately predict the disease course in individual patients. Staging systems in MM have evolved, firstly the International Staging System (ISS), then incorporation of genomic aberrations with revised-ISS. Recently, the IMWG agreed on consensus genomic staging of high-risk MM. However, no scoring system captures the nuances of the genetic landscape of each MM patient. In the era of genomic profiling, this information should be utilised to predict disease behaviour, which can be achieved through computational modelling. We recently developed and validated a model signalling network within B cells, which predicts pathway activation and protein abundance in B cells. We hypothesised that inclusion of genetic mutations from MM as parameter changes in this model could enable the creation of patient specific virtual cells that could improve disease stratification. Methods: Genetic data for 53 newly diagnosed MM patients, enrolled to the NCRI Myeloma XI was collected. The cohort were phenotypically high risk, relapsing within 30 months of maintenance randomisation. All patients achieved at least a partial response prior to relapse. Whole exome sequencing was conducted at presentation and relapse. All genetic mutations (SNPs, copy number variants, translocations) were verified for their oncogenic potential with OncoKB. Each mutation in each patient was mapped to a model parameter to create a set of 53 unique patient models in silico (e.g. gain of one copy of BCL2 increased the parameter representing BCL2 expression by 50%). To determine the apoptotic, and proliferative signalling state of the patient models 6-hour simulation was performed and the predicted abundance of cytoplasmic Smac, cytochrome C and cadherin-1 was stored. Patients were grouped according to their predicted signalling state, namely anti-apoptotic (AA, n=16), pro-proliferative (PP, n=11), anti-apoptotic and pro-proliferative (AAPP, n=15) and non-proliferative or apoptotic (NAP, n=11). We compared progression-free survival (PFS) in days between different groups using Cox regression for Hazard ratios (HR) and Log-Rank test for p-values. Results: As a benchmark we first evaluated the ISS in our cohort. ISS II versus ISS I was associated with a significantly worse PFS (HR 2.27 95% CI 1.07-4.8, p=0.032) as well as ISS III (HR 2.45 CI 1.13-5.31 p=0.024). However, ISS II was not significantly different from ISS III, likely due to the phenotypic high-risk cohort assessed. High-risk lesions del(17p), gain(1q), del(1p), (t(4;14), t(14;16), t(14;20) and neutral lesions, del(13q), HRD, t(6;14), t(11;14) and t(MYC), were all non-significant predictors of progression (p>0.05). Maintenance strategy (lenalidomide vs observation) was also not a significant predictor of outcome, consistent with the phenotypic behaviour. We then applied the model-driven signalling state stratification (AA, PP, AAPP, NAP) and found model stratification alone significantly predicted PFS (p=0.047). The PP group was associated with the longest PFS and the NAP group had the shortest (HR 3.33 CI 1.37-8.11, p=0.008). No HR genetic lesions were noted in the PP group, compared to 2/11 (18%), 4/16 (25%) and 7/15 (47%) for the NAP, AA and AAPP groups respectively. To incorporate modelling with existing prognostic tools we combined all good prognosis patient groups (ISS I [n=13], and PP patients [n=11]) into one group and compared it to the rest. This stratification method identified 9/53 (17%) patients with high ISS (II/III), but good prognosis predicted through modelling, and assigned them to the low-risk group. The remaining high risk patients (n=29) had a significantly higher HR of 2.21 for shorter PFS (CI 1.24-3.94, p=0.007). Conclusion:We show that computational modelling can be utilised to stratify MM patients into both good and poor prognostic groups in ways that improve on previous staging systems. Furthermore, this has been illustrated within a phenotypical high-risk cohort, emphasising its ability to uncover distinct subgroups in previously hard to classify patients. This approach provides the foundation for a novel approach to improving treatment decisions, developing personalised approaches such as treatment intensification based on the full genetic profile of an individual patient. Characterising the signalling state in a larger cohort and predicting response to treatment is underway.

Genetic heterogeneity and co-occurring driver mutations impact clinical outcomes in blood cancers, but predicting the emergent effect of co-occurring mutations that impact multiple complex and interacting signalling networks is challenging. Here, we used mathematical models to predict the impact of co-occurring mutations on cellular signalling and cell fates in diffuse large B cell lymphoma and multiple myeloma. Simulations predicted adverse impact on clinical prognosis when combinations of mutations induced both anti-apoptotic (AA) and pro-proliferative (PP) signalling. We integrated patient-specific mutational profiles into personalised lymphoma models, and identified patients characterised by simultaneous upregulation of anti-apoptotic and pro-proliferative (AAPP) signalling in all genomic and cell-of-origin classifications (8-25% of patients). In a discovery cohort and two validation cohorts, patients with upregulation of neither, one (AA or PP), or both (AAPP) signalling states had good, intermediate and poor prognosis respectively. Combining AAPP signalling with genetic or clinical prognostic predictors reliably stratified patients into striking prognostic categories. AAPP patients in poor prognosis genetic clusters had 7.8 months median overall survival, while patients lacking both features had 90% overall survival at 120 months in a validation cohort. Personalised computational models enable identification of novel risk-stratified patient subgroups, providing a valuable tool for future risk-adapted clinical trials.

Genetic heterogeneity and co-occurring driver mutations impact clinical outcomes in blood cancers. Grouping tumours into clusters based on genetic alterations is prognostically informative. However, predicting the emergent effect of co-occurring mutations that impact multiple complex and interacting signalling networks remains challenging. Here, we used mathematical models to predict the impact of co-occurring mutations on cellular signalling and cell fates in diffuse large B cell lymphoma (DLBCL) and multiple myeloma (MM). Simulations predicted adverse impact on clinical prognosis when combinations of mutations induced both pro-proliferative and anti-apoptotic signalling. So, we established a pipeline to integrate patient-specific mutational profiles into personalised lymphoma models. Using this approach, we identified a subgroup (19%) of patients characterised by simultaneous upregulation of anti-apoptotic and pro-proliferative (AAPP) signalling. AAPP patients have dismal prognosis and can be identified within all current genomic and cell-of-origin classifications. Combining personalised molecular simulations with mutational clustering enabled stratification of patients into clinically informative prognostic categories: good (80% progression-free survival at 120 months), intermediate (median progression-free survival of 93 months), and poor (AAPP, median progression-free survival of 26 months). This study shows that personalised computational models enable identification of novel high-risk patient subgroups, providing a valuable tool for future risk-stratified clinical trials.

Chronic lymphocytic leukemia (CLL) is the most prevalent type of leukemia in the western world. Despite the positive clinical effects of new targeted therapies, CLL still remains an incurable and refractory disease and resistance to treatments are commonly encountered. The Nuclear Factor-Kappa B (NF-κB) transcription factor has been implicated in the pathology of CLL, with high levels of NF-κB associated with disease progression and drug resistance. This aberrant NF-κB activation can be caused by genetic mutations in the tumor cells and microenvironmental factors, which promote NF-κB signaling. Activation can be induced via two distinct pathways, the canonical and non-canonical pathway, which result in tumor cell proliferation, survival and drug resistance. Therefore, understanding how the CLL microenvironment drives NF-κB activation is important for deciphering how CLL cells evade treatment and may aid the development of novel targeting therapeutics. The CLL microenvironment is comprised of various cells, including nurse like cells, mesenchymal stromal cells, follicular dendritic cells and CD4+ T cells. By activating different receptors, including the B cell receptor and CD40, these cells cause overactivity of the canonical and non-canonical NF-κB pathways. Within this review, we will explore the different components of the CLL microenvironment that drive the NF-κB pathway, investigating how this knowledge is being translated in the development of new therapeutics.

The nuclear factor κB (NF-κB) system is critical for various biological functions in numerous cell types, including the inflammatory response, cell proliferation, survival, differentiation, and pathogenic responses. Each cell type is characterized by a subset of 15 NF-κB dimers whose activity is regulated in a stimulus-responsive manner. Numerous studies have produced different mathematical models that account for cell type–specific NF-κB activities. However, whereas the concentrations or abundances of NF-κB subunits may differ between cell types, the biochemical interactions that constitute the NF-κB signaling system do not. Here, we synthesized a consensus mathematical model of the NF-κB multidimer system, which could account for the cell type–specific repertoires of NF-κB dimers and their cell type–specific activation and cross-talk. Our review demonstrates that these distinct cell type–specific properties of NF-κB signaling can be explained largely as emergent effects of the cell type–specific expression of NF-κB monomers. The consensus systems model represents a knowledge base that may be used to gain insights into the control and function of NF-κB in diverse physiological and pathological scenarios and that describes a path for generating similar regulatory knowledge bases for other pleiotropic signaling systems.

IntroductionImproving treatments for Diffuse Large B-Cell Lymphoma (DLBCL) is challenged by the vast heterogeneity of the disease. Nuclear factor-κB (NF-κB) is frequently aberrantly activated in DLBCL. Transcriptionally active NF-κB is a dimer containing either RelA, RelB or cRel, but the variability in the composition of NF-κB between and within DLBCL cell populations is not known.ResultsHere we describe a new flow cytometry-based analysis technique termed “NF-κB fingerprinting” and demonstrate its applicability to DLBCL cell lines, DLBCL core-needle biopsy samples, and healthy donor blood samples. We find each of these cell populations has a unique NF-κB fingerprint and that widely used cell-of-origin classifications are inadequate to capture NF-κB heterogeneity in DLBCL. Computational modeling predicts that RelA is a key determinant of response to microenvironmental stimuli, and we experimentally identify substantial variability in RelA between and within ABC-DLBCL cell lines. We find that when we incorporate NF-κB fingerprints and mutational information into computational models we can predict how heterogeneous DLBCL cell populations respond to microenvironmental stimuli, and we validate these predictions experimentally.DiscussionOur results show that the composition of NF-κB is highly heterogeneous in DLBCL and predictive of how DLBCL cells will respond to microenvironmental stimuli. We find that commonly occurring mutations in the NF-κB signaling pathway reduce DLBCL’s response to microenvironmental stimuli. NF-κB fingerprinting is a widely applicable analysis technique to quantify NF-κB heterogeneity in B cell malignancies that reveals functionally significant differences in NF-κB composition within and between cell populations.

In healthy cells, pro- and anti-apoptotic BCL2 family and BH3-only proteins are expressed in a delicate equilibrium. In contrast, this homeostasis is frequently perturbed in cancer cells due to the overexpression of anti-apoptotic BCL2 family proteins. Variability in the expression and sequestration of these proteins in Diffuse Large B cell Lymphoma (DLBCL) likely contributes to variability in response to BH3-mimetics. Successful deployment of BH3-mimetics in DLBCL requires reliable predictions of which lymphoma cells will respond. Here we show that a computational systems biology approach enables accurate prediction of the sensitivity of DLBCL cells to BH3-mimetics. We found that fractional killing of DLBCL, can be explained by cell-to-cell variability in the molecular abundances of signaling proteins. Importantly, by combining protein interaction data with a knowledge of genetic lesions in DLBCL cells, our in silico models accurately predict in vitro response to BH3-mimetics. Furthermore, through virtual DLBCL cells we predict synergistic combinations of BH3-mimetics, which we then experimentally validated. These results show that computational systems biology models of apoptotic signaling, when constrained by experimental data, can facilitate the rational assignment of efficacious targeted inhibitors in B cell malignancies, paving the way for development of more personalized approaches to treatment.

B‐cell progenitor fate determinant interferon regulatory factor 4 (IRF4) exerts key roles in the pathogenesis and progression of multiple myeloma (MM), a currently incurable plasma cell malignancy. Aberrant expression of IRF4 and the establishment of a positive auto‐regulatory loop with oncogene MYC, drives a MM specific gene‐expression program leading to the abnormal expansion of malignant immature plasma cells. Targeting the IRF4‐MYC oncogenic loop has the potential to provide a selective and effective therapy for MM. Here we evaluate the use of bromodomain inhibitors to target the IRF4‐MYC axis through combined inhibition of their known epigenetic regulators, BRD4 and CBP/EP300. Although all inhibitors induced cell death, we found no synergistic effect of targeting both of these regulators on the viability of MM cell‐lines. Importantly, for all inhibitors over a time period up to 72 h, we detected reduced IRF4 mRNA, but a limited decrease in IRF4 protein expression or mRNA levels of downstream target genes. This indicates that inhibitor‐induced loss of cell viability is not mediated through reduced IRF4 protein expression, as previously proposed. Further analysis revealed a long half‐life of IRF4 protein in MM cells. In support of our experimental observations, gene network modeling of MM suggests that bromodomain inhibition is exerted primarily through MYC and not IRF4. These findings suggest that despite the autofeedback positive regulatory loop between IRF4 and MYC, bromodomain inhibitors are not effective at targeting IRF4 in MM and that novel therapeutic strategies should focus on the direct inhibition or degradation of IRF4.

Background: Bcl-2 proteins are prominent regulators of mitochondrial-dependent apoptosis in healthy cells and lymphoma. In healthy cells, finely balanced interactions between family members within the Bcl-2 family of proteins ensure normal cell functions. However, in cancerous cells, anti-apoptotic Bcl-2 family members are often overexpressed, tipping the scale towards cell survival. In recent years, therapeutic drugs targeting anti-apoptotic Bcl-2 proteins (BH3-mimetics) have been developed. However, these BH3-mimetics have shown heterogeneous clinical activity in different cancer subtypes, despite Bcl-2 overexpression. Recent experimental studies have suggested that complex interactions between the Bcl-2 family of proteins mediate the response to BH3-mimetics. Therefore, assigning the correct BH3-mimetic to the DLBCL cells in which they will be effective requires predicting the response of the apoptotic molecular signalling network. This response is an emergent property of the abundance of the signalling network’s components along with the binding affinities, regulatory interactions and kinetic rates. Aims: Construct a computational model that enables BH3-mimetics to be tested in virtual DLBCL cell lines and accurately predicts cellular responses, as measured experimentally. Methods: By combining published data including protein abundances, binding affinities, mutational profiles and kinetic rates we construct a computational model of the distinct apoptotic signalling network in multiple DLBCL cell lines. The simulation incorporates experimentally-measured Bcl-2 expression profiles, generating a set of virtual DLBCL cell lines. Bcl-2 (ABT-199), Mcl-1 (S63845) and Bcl-xL (A1331852) inhibitors are simulated to predict cell viability and compared to experimental results. Results: Our virtual DLBCL cell lines recapitulate the Bcl-2 protein expression profiles in DLBCL cell lines. Our computational cell line models recreate how cells expressing distinct abundances of Bcl-2 proteins respond to BH3-mimetics through complex shifts in binding partners between pro-and anti-apoptotic Bcl-2 family proteins. The result of this work is a computational tool that, by simulating the regulatory interactions governing apoptosis, can predict the sensitivity of DLBCL cells to BH3-mimetics (Fig. 1). We validate predicted heterogeneous response of DLBCL cells to BH3-mimetics with experimental data from DLBCL cell lines. Image:Summary/Conclusion: The cellular response to BH3-mimetics is a complex emergent behaviour resulting from a combination of affinity, abundance and genetics, and is therefore challenging to predict. Computational systems biology provides a tool to overcome this challenge. Using a combination of in silico and in vitro experiments, we show that a computational model can predict the sensitivity of a cell to targeted therapies. Future work will test the ability of the model to predict patient responses to targeted therapies in order to enable a personalised medicine approach.

The retention and re-migration of Chronic Lymphocytic Leukemia cells into cytoprotective and proliferative lymphoid niches is thought to contribute to the development of resistance, leading to subsequent disease relapse. The aim of this study was to elucidate the molecular processes that govern CLL cell migration to elicit a more complete inhibition of tumor cell migration. We compared the phenotypic and transcriptional changes induced in CLL cells using two distinct models designed to recapitulate the peripheral circulation, CLL cell migration across an endothelial barrier, and the lymph node interaction between CLL cells and activated T cells. Initially, CLL cells were co-cultured with CD40L-expressing fibroblasts and exhibited an activated B-cell phenotype, and their transcriptional signatures demonstrated the upregulation of pro-survival and anti-apoptotic genes and overrepresentation of the NF-κB signaling pathway. Using our dynamic circulating model, we were able to study the transcriptomics and miRNomics associated with CLL migration. More than 3000 genes were altered when CLL cells underwent transendothelial migration, with an overrepresentation of adhesion and cell migration gene sets. From this analysis, an upregulation of the FAK signaling pathway was observed. Importantly, PTK2 (FAK) gene expression was significantly upregulated in migrating CLL cells (PTK2 Fold-change = 4.9). Here we demonstrate that TLR9 agonism increased levels of p-FAK (p ≤ 0.05), which could be prevented by pharmacological inhibition of FAK with defactinib (p ≤ 0.01). Furthermore, a reduction in CLL cell migration and invasion was observed when FAK was inhibited (p ≤ 0.0001), supporting a role for FAK in both CLL migration and tissue invasion. When taken together, our data highlights the potential for combining FAK inhibition with current targeted therapies as a more effective treatment regime for CLL.