Hella Anna Bolck

Clear cell renal cell carcinoma exhibits striking intra-tumoral heterogeneity at morphological and genetic levels, complicating treatment and contributing to disease progression. CcRCCs with rhabdoid differentiation are highly aggressive tumors characterized by distinct histopathologies. However, the relationship between morphology, underlying molecular alterations, and tumor behavior remains largely unclear. Here, we present Deep Visual Multi-Omics, an approach integrating digital pathology, morphology-guided single-cell isolation, and ultra-sensitive multi-omics profiling to link cell morphologies to their molecular underpinnings. Across five tumors, we profiled ~40,000 AI-classified and expert-curated cells. We identified progressive molecular dysregulation across cells with increasing histopathological grade coexisting within heterogeneous tumors as well as distinct molecular alterations associated with aggressive rhabdoid ccRCC cells, including signatures consistent with enhanced FOXM1-driven proliferation, altered cell-matrix interactions, and a putative immunomodulatory phenotype. Notably, rhabdoid cells exhibited elevated expression of IFN-beta, PD-L1, CD38, ITGB2, and integrin signaling, suggesting that they themselves may act as a source of signals influencing the local immune microenvironment. Besides providing new insights into the biology of ccRCC and highlighting avenues for future translational studies, this illustrates the potential of Deep Visual Multi-omics to dissect cancer heterogeneity and characterize high-risk cell populations.

Clear cell renal cell carcinoma (ccRCC) remains the most lethal urological malignancy, with high metastatic rates, both at initial diagnosis and during disease progression, contributing to poor survival outcomes. Current diagnostic and prognostic approaches rely primarily on histopathology, limiting early detection of localized disease and relevant intervention for metastatic patients. Here, we performed the most extensive to-date mass spectrometry-based discovery profiling of longitudinal plasma samples collected across multiple clinical follow-up points spanning up to five years post-diagnosis., to characterize the circulating plasma proteome and identify biomarkers for localized and metastatic disease. Network analysis identified protein modules enriched in pathways involved in matrix remodeling and metabolic deregulation, perpetuating the ccRCC phenotype. A five-protein signature, comprising PRL, THBS1, ANGPT1, IGFBP1, and SRGN, demonstrated high diagnostic performance for localized ccRCC. Notably, PRL appeared as a promising stand-alone biomarker (AUC = 0.812), with independent validation confirming its utility as a diagnostic biomarker. Importantly, a six-protein signature (AMBP, C1S, C2, IGFBP3, RASGRP2, TFRC) stringently distinguished metastatic from high-grade non-metastatic ccRCC cases. Further validation of these signatures could inform clinical decision-making, enabling early detection of metastasis and minimal residual disease and real-time longitudinal monitoring for ccRCC patients.

The development of autonomous aerial robots capable of safely navigating complex real-world environments without or with little human intervention represents a major milestone in robotics and artificial intelligence (AI). While rapid advances in AI-enabled decision-making, sensing, and control systems are unlocking new capabilities for unmanned aerial vehicles (UAVs), their translation into safe and scalable real-life applications remains a major challenge. In this Perspective, we examine key AI technologies relevant to aerial autonomy and discuss early application scenarios in unmanned aviation and airspace management, with a focus on their assurance-relevant properties. We analyze regulatory obstacles that limit deployment, particularly for AI-enabled and beyond visual line of sight (BVLOS) operations, and highlight why traditional risk assessment and certification approaches are need to be updated to account for adaptive, data-driven systems. Building on this analysis, we argue that testing infrastructure must be understood as a core scientific instrument, enabling systematic evidence generation under realistic and safety-critical conditions, validating autonomous functions, ensuring safety, and building trust among regulators and the public. As a concrete example, we introduce LINA, a scientifically-grounded, integrated experimentation and validation platform in Switzerland designed to support iterative, regulator-aware development of autonomous systems across technology readiness levels. We highlight how LINA function as sandbox for system-level science, regulatory learning, and trust building, thereby enabling the responsible and societally acceptable integration of autonomous aerial systems and strengthening Switzerland’s role in advancing aerial robotics research and innovation.

Single-cell RNA sequencing is used for profiling gene expression differences between cells. It can be performed with short reads, which provide high-throughput and high-quality information at the gene level, or with long reads, which provide isoform resolution via preserving full-length transcripts. It is, however, unclear how comparable the data is between the two methods. We investigate the types of bias introduced at the library preparation and the bioinformatic processing steps on the transcripts recovered from long- and short-read sequencing, and the effects of filtering, enabled by sequencing of full-length transcripts, on gene expression. For each sample, we sequenced the same 10x Genomics 3′ complementary DNA (cDNA) using Illumina short reads and Pacific Biosciences long reads and cross-compared each molecule matched through cell barcode and unique molecular identifier. We find that both methods render highly comparable results and recover a large proportion of cells and transcripts. However, platform-dependent cDNA library processing and data analysis steps introduce biases. Short-read sequencing provided higher sequencing depth, but long-read sequencing allowed for retaining transcripts shorter than 500 bp and for removal of degraded cDNA contaminated by template switching oligos. Filtering of artefacts, identifiable only from full-length transcripts, reduces gene count correlation between the two methods.

Summary: Clear cell renal cell carcinoma (ccRCC) exhibits significant intra-tumoral heterogeneity (ITH) at both morphological and genetic levels, complicating treatment and contributing to disease progression. Among these, ccRCCs with focal rhabdoid differentiation stand out as highly aggressive tumors distinguished by cells with unique morphological features. However, the correlation between distinct morphological phenotypes, specific molecular alterations, and their influence on tumor behavior remains poorly understood. In this study, we integrated advanced AI-based image analysis with single-cell isolation and multi-omics profiling to dissect the link between clinically relevant morphological and molecular features of ccRCC cells. Using a novel digital pathology workflow, we quantified low-grade, high-grade, and rhabdoid morphologies in ccRCC diagnostic images with unprecedented precision. Subsequently, isolation of two sets of 1,000 morphologically distinct cells for detailed mRNA and protein expression analyses, revealed significant increasing dysregulation associating with higher histopathological grades. Rhabdoid ccRCC cells (grade 4) demonstrated unique molecular profiles, including upregulated FOXM1-driven proliferation, disrupted cell-matrix interactions, and enhanced immune evasion pathways. Despite high T-cell infiltration in rhabdoid areas, we identified a rhabdoid-specific immunosuppressive network driven by cytokines, IFN-beta, and integrin signaling, likely contributing to T-cell exhaustion. Rhabdoid ccRCC cells develop a distinct immunosuppressive signaling network, involving PD-L1 and novel immunomodulatory factors such as CD38 and ITGB2. These findings provide a basis for novel therapeutic strategies targeting these pathways in combination with immunotherapy to improve outcomes for patients with aggressive rhabdoid ccRCC. Key Points: ccRCC is characterized by well-established morphological heterogeneity but the correlation with the underlying molecular aberrations remained elusive. By integrating AI-based image analysis with single cell isolation and deep multi-omics profiling, we dissect the molecular intricacies of ccRCC, from targeted collection of 1,000 morphologically distinct cells. Our results demonstrate significant dysregulation of gene and protein expression correlating with higher histopathological grades in ccRCC. Aggressive ccRCC cells with rhabdoid differentiation (grade 4) display distinct molecular profiles, as they upregulate FOXM1-mediated proliferation, ECM remodeling and the immune evasion responses, suggesting new therapeutic avenues enhancing ICI efficacy in these patients.

Papillary renal cell carcinoma (pRCC) is the second most frequent renal cancer subtype but has no indicated targeted treatments. MET inhibition may be a treatment for MET‐driven pRCC, but there is a large group of non‐MET‐driven pRCC without targeted therapy. Activation of NRF2‐ARE pathway has been suggested to be involved in pRCC. To study the relevance of the NRF2‐ARE pathway, we characterized 60 pRCCs by copy number analysis and Whole Exome Sequencing. Because stabilisation of NRF2 results in enhanced expression of NQO1, a reductase that prevents production of reactive oxygen species, protein expression of NQO1 was analysed by immunohistochemistry (IHC) from tissue microarrays (TMAs) and by enzymatic activity assay. Finally, patient‐derived pRCC cells (PDCs) were applied for drug profiling with 18 NRF2‐ARE pathway inhibitors. We identified MET mutations in 5%, and mutations in four genes of NRF2‐ARE pathway (NFE2L2, KEAP1, CUL3 and BACH1) in 10% of 60 pRCC samples. IHC analysis of TMAs of 638 renal cancers showed the correlation of the expression of NQO1 with poor survival outcome (p < .001) and high tumour grade (p < .001) and stage (p < .001) in pRCC. NQO1 mRNA, protein levels and enzymatic activity were increased in 56% of matched pRCC tissue samples and patient‐derived cells (PDCs, n = 9). Drug screening revealed that Brusatol and Convallatoxin are potential novel drugs for pRCC. Inhibition of NRF2 represents a novel therapeutic approach for MET‐independent pRCC patients.

Summary: Omics profiling has proven of great use for unbiased and comprehensive identification of key features that define biological phenotypes and underlie medical conditions. While each omics profile assists characterization of specific molecular components relevant for the studied phenotype, their joint evaluation can offer deeper insights into the overall mechanistic functioning of biological systems. Here, we introduce an approach where starting from representative traits (e.g., differentially expressed elements) obtained for each omics profile, we construct and analyze joint interaction networks. The resulting networks rely on the existing knowledge of confident interactions among biological entities. We use these maps to identify and describe central elements, which connect multiple entities characteristic for the studied phenotypes and we leverage MONET network decomposition tool in order to highlight functionally connected network modules. In order to enable broad usage of this approach, we developed the NOODAI software platform, which enables integrative omics analysis through a user-friendly interface. The analysis outcomes are presented both as raw output tables as well as high-quality summary plots and written reports. Since the MONET tool enables the use of algorithms with strong performance in identifying disease-relevant modules, the NOODAI software platform can be of a high value for the analysis of clinical multi-omics datasets. Availability and Implementation: The platform is available as a web application freely accessible at https://omics-oracle.com . The source code is freely available from GitHub under the GPL3 license at: https://github.com/TotuTiberiu/NOODAI .

Numerous imaging techniques are available for observing and interrogating biological samples, and several of them can be used consecutively to enable correlative analysis of different image modalities with varying resolutions and the inclusion of structural or molecular information. Achieving accurate registration of multimodal images is essential for the correlative analysis process, but it remains a challenging computer vision task with no widely accepted solution. Moreover, supervised registration methods require annotated data produced by experts, which is limited. To address this challenge, we propose a general unsupervised pipeline for multimodal image registration using deep learning. We provide a comprehensive evaluation of the proposed pipeline versus the current state-of-the-art image registration and style transfer methods on four types of biological problems utilizing different microscopy modalities. We found that style transfer of modality domains paired with fully unsupervised training leads to comparable image registration accuracy to supervised methods and, most importantly, does not require human intervention.

Human CtIP is best known for its role in DNA end resection to initiate DNA double-strand break repair by homologous recombination. Recently, CtIP has also been shown to protect reversed replication forks from nucleolytic degradation upon DNA replication stress. However, still little is known about the DNA damage response (DDR) networks that preserve genome integrity and sustain cell survival in the context of CtIP insufficiency. Here, to reveal such potential buffering relationships, we screened a DDR siRNA library in CtIP-deficient cells to identify candidate genes that induce synthetic sickness/lethality (SSL). Our analyses unveil a negative genetic interaction between CtIP and BARD1, the heterodimeric binding partner of BRCA1. We found that simultaneous disruption of CtIP and BARD1 triggers enhanced apoptosis due to persistent replication stress-induced DNA lesions giving rise to chromosomal abnormalities. Moreover, we observed that the genetic interaction between CtIP and BARD1 occurs independently of the BRCA1-BARD1 complex formation and might be, therefore, therapeutical relevant for the treatment of BRCA-defective tumors.