Related Articles

Myeloid neoplasms with PHF6 mutations: context-dependent genomic and prognostic characterization in 176 informative cases

Recent reports suggest a favorable prognosis for PHF6 mutation (PHF6MUT) in chronic myelomonocytic leukemia (CMML) and unfavorable in acute myeloid leukemia (AML). We accessed 176 consecutive patients with a spectrum of myeloid neoplasms with PHF6MUT, including AML (N = 67), CMML (N = 49), myelodysplastic syndromes (MDS; N = 36), myeloproliferative neoplasms (MPN; N = 16), and MDS/MPN (N = 8). PHF6 mutations were classified as nonsense (43%) or frameshift (30%) with the PHD2 domain being the most frequently (64%) affected region. Median follow-up was 25 months with 110 (63%) deaths and 44 allogenic transplants. Our top-line observations include (a) a distinctly superior overall survival (OS; 81 vs. 18 months; p < 0.01) and blast transformation-free survival (BTFS; “not reached” vs. 44 months; p < 0.01) in patients with CMML vs. those with other myeloid neoplasms, (ii) a higher than expected frequency of isolated loss of Y chromosome, in the setting of CMML (16% vs. expected 6%) and MDS (8% vs expected 2.5%), (iii) a significant association, in MDS, between PHF6MUT variant allele fraction (VAF) > 20% and inferior OS (HR 3.0, 95% CI 1.1–8.1, multivariate p = 0.02) as well as female gender and inferior BTFS (HR 26.8, 95% CI 1.9–368.3, multivariate p = 0.01), (iv) a relatively favorable median post-transplant survival of 46 months. Multivariable analysis also identified high-risk karyotype (HR 5.1, 95% CI 1.2–20.9, p = 0.02), and hemoglobin <10 g/dL (HR 2.7, 95% CI 1.0–7.2, p = 0.04), as independent predictors of inferior OS in patients with MDS. The current study provides disease-specific information on genotype and prognosis of PHF6-mutated myeloid neoplasms.

The design space of E(3)-equivariant atom-centred interatomic potentials

Molecular dynamics simulation is an important tool in computational materials science and chemistry, and in the past decade it has been revolutionized by machine learning. This rapid progress in machine learning interatomic potentials has produced a number of new architectures in just the past few years. Particularly notable among these are the atomic cluster expansion, which unified many of the earlier ideas around atom-density-based descriptors, and Neural Equivariant Interatomic Potentials (NequIP), a message-passing neural network with equivariant features that exhibited state-of-the-art accuracy at the time. Here we construct a mathematical framework that unifies these models: atomic cluster expansion is extended and recast as one layer of a multi-layer architecture, while the linearized version of NequIP is understood as a particular sparsification of a much larger polynomial model. Our framework also provides a practical tool for systematically probing different choices in this unified design space. An ablation study of NequIP, via a set of experiments looking at in- and out-of-domain accuracy and smooth extrapolation very far from the training data, sheds some light on which design choices are critical to achieving high accuracy. A much-simplified version of NequIP, which we call BOTnet (for body-ordered tensor network), has an interpretable architecture and maintains its accuracy on benchmark datasets.

Targeting of TAMs: can we be more clever than cancer cells?

With increasing incidence and geography, cancer is one of the leading causes of death, reduced quality of life and disability worldwide. Principal progress in the development of new anticancer therapies, in improving the efficiency of immunotherapeutic tools, and in the personification of conventional therapies needs to consider cancer-specific and patient-specific programming of innate immunity. Intratumoral TAMs and their precursors, resident macrophages and monocytes, are principal regulators of tumor progression and therapy resistance. Our review summarizes the accumulated evidence for the subpopulations of TAMs and their increasing number of biomarkers, indicating their predictive value for the clinical parameters of carcinogenesis and therapy resistance, with a focus on solid cancers of non-infectious etiology. We present the state-of-the-art knowledge about the tumor-supporting functions of TAMs at all stages of tumor progression and highlight biomarkers, recently identified by single-cell and spatial analytical methods, that discriminate between tumor-promoting and tumor-inhibiting TAMs, where both subtypes express a combination of prototype M1 and M2 genes. Our review focuses on novel mechanisms involved in the crosstalk among epigenetic, signaling, transcriptional and metabolic pathways in TAMs. Particular attention has been given to the recently identified link between cancer cell metabolism and the epigenetic programming of TAMs by histone lactylation, which can be responsible for the unlimited protumoral programming of TAMs. Finally, we explain how TAMs interfere with currently used anticancer therapeutics and summarize the most advanced data from clinical trials, which we divide into four categories: inhibition of TAM survival and differentiation, inhibition of monocyte/TAM recruitment into tumors, functional reprogramming of TAMs, and genetic enhancement of macrophages.

Inactivation of the SLC25A1 gene during embryogenesis induces a unique senescence program controlled by p53

Germline inactivating mutations of the SLC25A1 gene contribute to various human disorders, including Velocardiofacial (VCFS), DiGeorge (DGS) syndromes and combined D/L-2-hydroxyglutaric aciduria (D/L-2HGA), a severe systemic disease characterized by the accumulation of 2-hydroxyglutaric acid (2HG). The mechanisms by which SLC25A1 loss leads to these syndromes remain largely unclear. Here, we describe a mouse model of SLC25A1 deficiency that mimics human VCFS/DGS and D/L-2HGA. Surprisingly, inactivation of both Slc25a1 alleles results in alterations in the development of multiple organs, and in a severe proliferation defect by activating two senescence programs, oncogene-induced senescence (OIS) and mitochondrial dysfunction-induced senescence (MiDAS), which converge upon the induction of the p53 tumor suppressor. Mechanistically, cells and tissues with dysfunctional SLC25A1 protein undergo metabolic and transcriptional rewiring leading to the accumulation of 2HG via a non-canonical pathway and to the depletion of nicotinamide adenine dinucleotide, NAD+, which trigger senescence. Replenishing the pool of NAD+ or promoting the clearance of 2HG rescues the proliferation defect of cells with dysfunctional SLC25A1 in a cooperative fashion. Further, removal of p53 activity via RNA interference restores proliferation, indicating that p53 acts as a critical barrier to the expansion of cells lacking functional SLC25A1. These findings reveal unexpected pathogenic roles of senescence and of p53 in D/L-2HGA and identify potential therapeutic strategies to correct salient molecular alterations driving this disease.

A data-driven generative strategy to avoid reward hacking in multi-objective molecular design

Molecular design using data-driven generative models has emerged as a promising technology, impacting various fields such as drug discovery and the development of functional materials. However, this approach is often susceptible to optimization failure due to reward hacking, where prediction models fail to extrapolate, i.e., fail to accurately predict properties for designed molecules that considerably deviate from the training data. While methods for estimating prediction reliability, such as the applicability domain (AD), have been used for mitigating reward hacking, multi-objective optimization makes it challenging. The difficulty arises from the need to determine in advance whether the multiple ADs with some reliability levels overlap in chemical space, and to appropriately adjust the reliability levels for each property prediction. Herein, we propose a reliable design framework to perform multi-objective optimization using generative models while preventing reward hacking. To demonstrate the effectiveness of the proposed framework, we designed candidates for anticancer drugs as a typical example of multi-objective optimization. We successfully designed molecules with high predicted values and reliabilities, including an approved drug. In addition, the reliability levels can be automatically adjusted according to the property prioritization specified by the user without any detailed settings.

Responses

Your email address will not be published. Required fields are marked *