Designing patient-oriented combination therapies for acute myeloid leukemia based on efficacy/toxicity integration and bipartite network modeling.

Acute myeloid leukemia (AML), a heterogeneous and aggressive blood cancer, does not respond well to single-drug therapy. A combination of drugs is required to effectively treat this disease. Computational models are critical for combination therapy discovery due to the tens of thousands of two-drug combinations, even with approved drugs. While predicting synergistic drugs is the focus of current methods, few consider drug efficacy and potential toxicity, which are crucial for treatment success. To find effective new drug candidates, we constructed a bipartite network using patient-derived tumor samples and drugs. The network is based on drug-response screening and summarizes all treatment response heterogeneity as drug response weights. This bipartite network is then projected onto the drug part, resulting in the drug similarity network. Distinct drug clusters were identified using community detection methods, each targeting different biological processes and pathways as revealed by enrichment and pathway analysis of the drugs' protein targets. Four drugs with the highest efficacy and lowest toxicity from each cluster were selected and tested for drug sensitivity using cell viability assays on various samples. Results show that ruxolitinib-ulixertinib and sapanisertib-LY3009120 are the most effective combinations with the least toxicity and the best synergistic effect on blast cells. These findings lay the foundation for personalized and successful AML therapies, ultimately leading to the development of drug combinations that can be used alongside standard first-line AML treatment.

Robust scoring of selective drug responses for patient-tailored therapy selection.

Most patients with advanced malignancies are treated with severely toxic, first-line chemotherapies. Personalized treatment strategies have led to improved patient outcomes and could replace one-size-fits-all therapies, yet they need to be tailored by testing of a range of targeted drugs in primary patient cells. Most functional precision medicine studies use simple drug-response metrics, which cannot quantify the selective effects of drugs (i.e., the differential responses of cancer cells and normal cells). We developed a computational method for selective drug-sensitivity scoring (DSS), which enables normalization of the individual patient's responses against normal cell responses. The selective response scoring uses the inhibition of noncancerous cells as a proxy for potential drug toxicity, which can in turn be used to identify effective and safer treatment options. Here, we explain how to apply the selective DSS calculation for guiding precision medicine in patients with leukemia treated across three cancer centers in Europe and the USA; the generic methods are also widely applicable to other malignancies that are amenable to drug testing. The open-source and extendable R-codes provide a robust means to tailor personalized treatment strategies on the basis of increasingly available ex vivo drug-testing data from patients in real-world and clinical trial settings. We also make available drug-response profiles to 527 anticancer compounds tested in 10 healthy bone marrow samples as reference data for selective scoring and de-prioritization of drugs that show broadly toxic effects. The procedure takes <60 min and requires basic skills in R.

Proteomic and Metabolomic Analysis of Bone Marrow and Plasma from Patients with Extramedullary Multiple Myeloma Identifies Distinct Protein and Metabolite Signatures.

Multiple myeloma (MM) is an incurable haematological malignancy of plasma cells in the bone marrow. In rare cases, an aggressive form of MM called extramedullary multiple myeloma (EMM) develops, where myeloma cells enter the bloodstream and colonise distal organs or soft tissues. This variant is associated with refractoriness to conventional therapies and a short overall survival. The molecular mechanisms associated with EMM are not yet fully understood. Here, we analysed the proteome of bone marrow mononuclear cells and blood plasma from eight patients (one serial sample) with EMM and eight patients without extramedullary spread. The patients with EMM had a significantly reduced overall survival with a median survival of 19 months. Label-free mass spectrometry revealed 225 proteins with a significant differential abundance between bone marrow mononuclear cells (BMNCs) isolated from patients with MM and EMM. This plasma proteomics analysis identified 22 proteins with a significant differential abundance. Three proteins, namely vascular cell adhesion molecule 1 (VCAM1), pigment epithelium derived factor (PEDF), and hepatocyte growth factor activator (HGFA), were verified as the promising markers of EMM, with the combined protein panel showing excellent accuracy in distinguishing EMM patients from MM patients. Metabolomic analysis revealed a distinct metabolite signature in EMM patient plasma compared to MM patient plasma. The results provide much needed insight into the phenotypic profile of EMM and in identifying promising plasma-derived markers of EMM that may inform novel drug development strategies.

Integrative phosphoproteomics defines two biologically distinct groups of KMT2A rearranged acute myeloid leukaemia with different drug response phenotypes.

Acute myeloid leukaemia (AML) patients harbouring certain chromosome abnormalities have particularly adverse prognosis. For these patients, targeted therapies have not yet made a significant clinical impact. To understand the molecular landscape of poor prognosis AML we profiled 74 patients from two different centres (in UK and Finland) at the proteomic, phosphoproteomic and drug response phenotypic levels. These data were complemented with transcriptomics analysis for 39 cases. Data integration highlighted a phosphoproteomics signature that define two biologically distinct groups of KMT2A rearranged leukaemia, which we term MLLGA and MLLGB. MLLGA presented increased DOT1L phosphorylation, HOXA gene expression, CDK1 activity and phosphorylation of proteins involved in RNA metabolism, replication and DNA damage when compared to MLLGB and no KMT2A rearranged samples. MLLGA was particularly sensitive to 15 compounds including genotoxic drugs and inhibitors of mitotic kinases and inosine-5-monosphosphate dehydrogenase (IMPDH) relative to other cases. Intermediate-risk KMT2A-MLLT3 cases were mainly represented in a third group closer to MLLGA than to MLLGB. The expression of IMPDH2 and multiple nucleolar proteins was higher in MLLGA and correlated with the response to IMPDH inhibition in KMT2A rearranged leukaemia, suggesting a role of the nucleolar activity in sensitivity to treatment. In summary, our multilayer molecular profiling of AML with poor prognosis and KMT2A-MLLT3 karyotypes identified a phosphoproteomics signature that defines two biologically and phenotypically distinct groups of KMT2A rearranged leukaemia. These data provide a rationale for the potential development of specific therapies for AML patients characterised by the MLLGA phosphoproteomics signature identified in this study.

CKS1 inhibition depletes leukemic stem cells and protects healthy hematopoietic stem cells in acute myeloid leukemia.

Acute myeloid leukemia (AML) is an aggressive hematological disorder comprising a hierarchy of quiescent leukemic stem cells (LSCs) and proliferating blasts with limited self-renewal ability. AML has a dismal prognosis, with extremely low 2-year survival rates in the poorest cytogenetic risk patients, primarily due to the failure of intensive chemotherapy protocols to deplete LSCs and toxicity of therapy toward healthy hematopoietic cells. We studied the role of cyclin-dependent kinase regulatory subunit 1 (CKS1)-dependent protein degradation in primary human AML and healthy hematopoiesis xenograft models in vivo. Using a small-molecule inhibitor (CKS1i), we demonstrate a dual role for CKS1-dependent protein degradation in reducing patient-derived AML blasts in vivo and, importantly, depleting LSCs, whereas inhibition of CKS1 has the opposite effect on normal hematopoiesis, protecting normal hematopoietic stem cells from chemotherapeutic toxicity. Proteomic analysis of responses to CKS1i in our patient-derived xenograft mouse model demonstrate that inhibition of CKS1 in AML leads to hyperactivation of RAC1 and accumulation of lethal reactive oxygen species, whereas healthy hematopoietic cells enter quiescence in response to CKS1i, protecting hematopoietic stem cells. Together, these findings demonstrate that CKS1-dependent proteostasis is a key vulnerability in malignant stem cell biology.

Growth Response and Differentiation of Bone Marrow-Derived Mesenchymal Stem/Stromal Cells in the Presence of Novel Multiple Myeloma Drug Melflufen.

Mesenchymal stem/stromal cells (MSCs) are self-renewing and multipotent progenitors, which constitute the main cellular compartment of the bone marrow stroma. Because MSCs have an important role in the pathogenesis of multiple myeloma, it is essential to know if novel drugs target MSCs. Melflufen is a novel anticancer peptide-drug conjugate compound for patients with relapsed refractory multiple myeloma. Here, we studied the cytotoxicity of melflufen, melphalan and doxorubicin in healthy human bone marrow-derived MSCs (BMSCs) and how these drugs affect BMSC proliferation. We established co-cultures of BMSCs with MM.1S myeloma cells to see if BMSCs increase or decrease the cytotoxicity of melflufen, melphalan, bortezomib and doxorubicin. We evaluated how the drugs affect BMSC differentiation into adipocytes and osteoblasts and the BMSC-supported formation of vascular networks. Our results showed that BMSCs were more sensitive to melflufen than to melphalan. The cytotoxicity of melflufen in myeloma cells was not affected by the co-culture with BMSCs, as was the case for melphalan, bortezomib and doxorubicin. Adipogenesis, osteogenesis and BMSC-mediated angiogenesis were all affected by melflufen. Melphalan and doxorubicin affected BMSC differentiation in similar ways. The effects on adipogenesis and osteogenesis were not solely because of effects on proliferation, seen from the differential expression of differentiation markers normalized by cell number. Overall, our results indicate that melflufen has a significant impact on BMSCs, which could possibly affect therapy outcome.

The Peptide-Drug Conjugate Melflufen Modulates the Unfolded Protein Response of Multiple Myeloma and Amyloidogenic Plasma Cells and Induces Cell Death.

Immunoglobulin light-chain (AL) amyloidosis is a rare disease caused by clonal plasma cell secretion of misfolded light chains that assemble as toxic amyloid fibrils, depositing in vital organs including the heart and kidneys, causing organ dysfunction. Plasma cell-directed therapeutics are expected to reduce production of toxic light chain by eliminating amyloidogenic cells in bone marrow, thereby diminishing amyloid fibril deposition and providing the potential for organ recovery. Melphalan flufenamide (melflufen) is a first-in-class peptide-drug conjugate that targets aminopeptidases and rapidly releases alkylating agents inside tumor cells. Melflufen is highly lipophilic, permitting rapid uptake by cells, where it is enzymatically hydrolyzed by aminopeptidases, resulting in intracellular accumulation of the alkylating agents, including melphalan. Previous data demonstrating sensitivity of myeloma cells to melflufen suggest that the drug might be useful in AL amyloidosis. We describe the effects of melflufen on amyloidogenic plasma cells in vitro and ex vivo, demonstrating enhanced cytotoxic effects in comparison to melphalan, as well as novel mechanisms of action through the unfolded protein response (UPR) pathway. These findings provide evidence that melflufen-mediated cytotoxicity extends to amyloidogenic plasma cells, and support the rationale for the evaluation of melflufen in patients with AL amyloidosis.

Identification of Protein Biomarker Signatures for Acute Myeloid Leukemia (AML) Using Both Nontargeted and Targeted Approaches.

Acute myeloid leukemia (AML) is characterized by an increasing number of clonal myeloid blast cells which are incapable of differentiating into mature leukocytes. AML risk stratification is based on genetic background, which also serves as a means to identify the optimal treatment of individual patients. However, constant refinements are needed, and the inclusion of significant measurements, based on the various omics approaches that are currently available to researchers/clinicians, have the potential to increase overall accuracy with respect to patient management. Using both nontargeted (label-free mass spectrometry) and targeted (multiplex immunoassays) proteomics, a range of proteins were found to be significantly changed in AML patients with different genetic backgrounds. The inclusion of validated proteomic biomarker panels could be an important factor in the prognostic classification of AML patients. The ability to measure both cellular and secreted analytes, at diagnosis and during the course of treatment, has advantages in identifying transforming biological mechanisms in patients, assisting important clinical management decisions.

S100 Calcium Binding Protein Family Members Associate With Poor Patient Outcome and Response to Proteasome Inhibition in Multiple Myeloma.

Despite several new therapeutic options, multiple myeloma (MM) patients experience multiple relapses and inevitably become refractory to treatment. Insights into drug resistance mechanisms may lead to the development of novel treatment strategies. The S100 family is comprised of 21 calcium binding protein members with 17 S100 genes located in the 1q21 region, which is commonly amplified in MM. Dysregulated expression of S100 family members is associated with tumor initiation, progression and inflammation. However, the relationship between the S100 family and MM pathogenesis and drug response is unknown. In this study, the roles of S100 members were systematically studied at the copy number, transcriptional and protein level with patients' survival and drug response. Copy number analysis revealed a predominant pattern of gains occurring in S100 genes clustering in the 1q21 locus. In general, gains of genes encoding S100 family members associated with worse patient survival. However, S100 gene copy number and S100 gene expression did not necessarily correlate, and high expression of S100A4 associated with poor patient survival. Furthermore, integrated analysis of S100 gene expression and ex vivo drug sensitivity data showed significant negative correlation between expression of S100 family members (S100A8, S100A9, and S100A12) and sensitivity to some drugs used in current MM treatment, including proteasome inhibitors (bortezomib, carfilzomib, and ixazomib) and histone deacetylase inhibitor panobinostat. Combined proteomic and pharmacological data exhibited significant negative association of S100 members (S100A4, S100A8, and S100A9) with proteasome inhibitors and panobinostat. Clinically, the higher expression of S100A4 and S100A10 were significantly linked to shorter progression free survival in patients receiving carfilzomib-based therapy. The results indicate an association and highlight the potential functional importance of S100 members on chromosome 1q21 in the development of MM and resistance to established myeloma drugs, including proteasome inhibitors.

Heterogeneous modulation of Bcl-2 family members and drug efflux mediate MCL-1 inhibitor resistance in multiple myeloma.

Antiapoptotic Bcl-2 family members have recently (re)emerged as key drug targets in cancer, with a tissue- and tumor-specific activity profile of available BH3 mimetics. In multiple myeloma, MCL-1 has been described as a major gatekeeper of apoptosis. This discovery has led to the rapid establishment of clinical trials evaluating the impact of various MCL-1 inhibitors. However, our understanding about the clinical impact and optimal use of MCL-1 inhibitors is still limited. We therefore explored mechanisms of acquired MCL-1 inhibitor resistance and optimization strategies in myeloma. Our findings indicated heterogeneous paths to resistance involving baseline Bcl-2 family alterations of proapoptotic (BAK, BAX, and BIM) and antiapoptotic (Bcl-2 and MCL-1) proteins. These manifestations depend on the BH3 profile of parental cells that guide the enhanced formation of Bcl-2:BIM and/or the dynamic (ie, treatment-induced) formation of Bcl-xL:BIM and Bcl-xL:BAK complexes. Accordingly, an unbiased high-throughput drug-screening approach (n = 528) indicated alternative BH3 mimetics as top combination partners for MCL-1 inhibitors in sensitive and resistant cells (Bcl-xL>Bcl-2 inhibition), whereas established drug classes were mainly antagonistic (eg, antimitotic agents). We also revealed reduced activity of MCL-1 inhibitors in the presence of stromal support as a drug-class effect that was overcome by concurrent Bcl-xL or Bcl-2 inhibition. Finally, we demonstrated heterogeneous Bcl-2 family deregulation and MCL-1 inhibitor cross-resistance in carfilzomib-resistant cells, a phenomenon linked to the MDR1-driven drug efflux of MCL-1 inhibitors. The implications of our findings for clinical practice emphasize the need for patient-adapted treatment protocols, with the tracking of tumor- and/or clone-specific adaptations in response to MCL-1 inhibition.

Aminopeptidase Expression in Multiple Myeloma Associates with Disease Progression and Sensitivity to Melflufen.

Multiple myeloma (MM) is characterized by extensive immunoglobulin production leading to an excessive load on protein homeostasis in tumor cells. Aminopeptidases contribute to proteolysis by catalyzing the hydrolysis of amino acids from proteins or peptides and function downstream of the ubiquitin-proteasome pathway. Notably, aminopeptidases can be utilized in the delivery of antibody and peptide-conjugated drugs, such as melflufen, currently in clinical trials. We analyzed the expression of 39 aminopeptidase genes in MM samples from 122 patients treated at Finnish cancer centers and 892 patients from the CoMMpass database. Based on ranked abundance, LAP3, ERAP2, METAP2, TTP2, and DPP7 were highly expressed in MM. ERAP2, XPNPEP1, DPP3, RNPEP, and CTSV were differentially expressed between relapsed/refractory and newly diagnosed MM samples (p < 0.05). Sensitivity to melflufen was detected ex vivo in 11/15 MM patient samples, and high sensitivity was observed, especially in relapsed/refractory samples. Survival analysis revealed that high expression of XPNPEP1, RNPEP, DPP3, and BLMH (p < 0.05) was associated with shorter overall survival. Hydrolysis analysis demonstrated that melflufen is a substrate for aminopeptidases LAP3, LTA4H, RNPEP, and ANPEP. The sensitivity of MM cell lines to melflufen was reduced by aminopeptidase inhibitors. These results indicate critical roles of aminopeptidases in disease progression and the activity of melflufen in MM.

HSV-1 ICP27 targets the TBK1-activated STING signalsome to inhibit virus-induced type I IFN expression.

Herpes simplex virus (HSV) 1 stimulates type I IFN expression through the cGAS-STING-TBK1 signaling axis. Macrophages have recently been proposed to be an essential source of IFN during viral infection. However, it is not known how HSV-1 inhibits IFN expression in this cell type. Here, we show that HSV-1 inhibits type I IFN induction through the cGAS-STING-TBK1 pathway in human macrophages, in a manner dependent on the conserved herpesvirus protein ICP27. This viral protein was expressed de novo in macrophages with early nuclear localization followed by later translocation to the cytoplasm where ICP27 prevented activation of IRF3. ICP27 interacted with TBK1 and STING in a manner that was dependent on TBK1 activity and the RGG motif in ICP27. Thus, HSV-1 inhibits expression of type I IFN in human macrophages through ICP27-dependent targeting of the TBK1-activated STING signalsome.

Monosodium urate activates Src/Pyk2/PI3 kinase and cathepsin dependent unconventional protein secretion from human primary macrophages.

Monosodium urate (MSU) is an endogenous danger signal that is crystallized from uric acid released from injured cells. MSU is known to activate inflammatory response in macrophages but the molecular mechanisms involved have remained uncharacterized. Activated macrophages start to secrete proteins to activate immune response and to recruit other immune cells to the site of infection and/or tissue damage. Secretome characterization after activation of innate immune system is essential to unravel the details of early phases of defense responses. Here, we have analyzed the secretome of human primary macrophages stimulated with MSU using quantitative two-dimensional gel electrophoresis based proteomics as well as high-throughput qualitative GeLC-MS/MS approach combining protein separation by SDS-PAGE and protein identification by liquid chromatography-MS/MS. Both methods showed that MSU stimulation induced robust protein secretion from lipopolysaccharide-primed human macrophages. Bioinformatic analysis of the secretome data showed that MSU stimulation strongly activates unconventional, vesicle mediated protein secretion. The unconventionally secreted proteins included pro-inflammatory cytokines like IL-1ß and IL-18, interferon-induced proteins, and danger signal proteins. Also active forms of lysosomal proteases cathepsins were secreted on MSU stimulation, and cathepsin activity was essential for MSU-induced unconventional protein secretion. Additionally, proteins associated to phosphorylation events including Src family tyrosine kinases were increased in the secretome of MSU-stimulated cells. Our functional studies demonstrated that Src, Pyk2, and PI3 kinases act upstream of cathepsins to activate the overall protein secretion from macrophages. In conclusion, we provide the first comprehensive characterization of protein secretion pathways activated by MSU in human macrophages, and reveal a novel role for cathepsins and Src, Pyk2, PI3 kinases in the activation of unconventional protein secretion.

Global secretome characterization of herpes simplex virus 1-infected human primary macrophages.

Herpes simplex virus 1 (HSV-1) is a common pathogen infecting the majority of people worldwide at some stage in their lives. The early host response to viral infection is initiated by the cells of the innate immune response, including macrophages. Here, we have characterized the secretome of HSV-1-infected human primary macrophages using high-throughput quantitative proteomics. We identified and quantified 516 distinct human proteins with high confidence from the macrophage secretome upon HSV-1 infection, and the secretion of 411 proteins was >2-fold increased upon beta interferon (IFN-ß) priming and/or HSV-1 infection. Bioinformatics analysis of the secretome data revealed that most of the secreted proteins were intracellular, and almost 80% of the proteins whose secretion increased more than 2-fold were known exosomal proteins. This strongly suggests that nonclassical, vesicle-mediated protein secretion is activated in IFN-ß-primed and HSV-1-infected macrophages. Proteins related to immune and inflammatory responses, interferon-induced proteins, and endogenous danger signal proteins were efficiently secreted upon IFN-ß priming and HSV-1 infection. The secreted IFN-induced proteins include interferon-induced tetratricopeptide protein 2 (IFIT2), IFIT3, signal transducer and activator of transcription 1 (STAT1), and myxovirus resistance protein A (MxA), implicating that these proteins also have important extracellular antiviral functions. Proinflammatory cytokine interleukin-1ß was not released by HSV-1-infected macrophages, demonstrating that HSV-1 can antagonize inflammasome function. In conclusion, our results provide a global view of the secretome of HSV-1-infected macrophages, revealing host factors possibly having a role in antiviral defense.