Search

Concentration of creatine (A, B) and 2-Aminoadipate (C, D) in human prostate tissue samples. The concentration values by untargeted metabolomics (normalized values) are shown in (A) and (C). The concentration values obtained by absolute quantification (µg/g tissue) are shown in (B) and (D).

Table S1. List of gene expression assays (Applied Biosystems).

Metabolomics dataset

Metabolomics dataset

Metabolomics dataset

Metabolomics dataset

SUMMARYThe significance of biophysical modulators of the Epithelial to Mesenchymal Transition (EMT) is demonstrated by experiments that document full EMT on stiff, nano-patterned substrates in the absence of biochemical induction. Yet, current models focus on biochemical triggers of EMT without addressing its mechanosensitive nature. Here we built a Boolean model of EMT triggered by mechanosensing – mitogen crosstalk. Our model reproduces epithelial, hybrid E/M and mesenchymal phenotypes, the role of autocrineTGFβsignaling in maintaining mesenchymal cells in the absence of external drivers, inhibition of proliferation byTGFβ, and its apoptotic effects on soft ECM. We offer testable predictions on the density-dependence of partial EMT, its molecular drivers, and the conflict between mitosis and hybrid E/M stability. Our model opens the door to modeling the effects of the biomechanical environment on cancer cell stemness linked to the hybrid E/M state, as well as the mutually inhibitory crosstalk between EMT and senescence.

How do cells make efficient collective decisions during tissue morphogenesis? Humans and other organisms use feedback between movement and sensing known as ‘sensorimotor coordination’ or ‘active perception’ to inform behaviour, but active perception has not before been investigated at a cellular level within organs. Here we provide the first proof of concept in silico/in vivo study demonstrating that filopodia (actin-rich, dynamic, finger-like cell membrane protrusions) play an unexpected role in speeding up collective endothelial decisions during the time-constrained process of ‘tip cell’ selection during blood vessel formation (angiogenesis). We first validate simulation predictions in vivo with live imaging of zebrafish intersegmental vessel growth. Further simulation studies then indicate the effect is due to the coupled positive feedback between movement and sensing on filopodia conferring a bistable switch-like property to Notch lateral inhibition, ensuring tip selection is a rapid and robust process. We then employ measures from computational neuroscience to assess whether filopodia function as a primitive (basal) form of active perception and find evidence in support. By viewing cell behaviour through the ‘basal cognitive lens' we acquire a fresh perspective on the tip cell selection process, revealing a hidden, yet vital time-keeping role for filopodia. Finally, we discuss a myriad of new and exciting research directions stemming from our conceptual approach to interpreting cell behaviour. This article is part of the theme issue ‘Basal cognition: multicellularity, neurons and the cognitive lens’.

How do cells make efficient collective decisions during tissue morphogenesis? Humans and other organisms utilize feedback between movement and sensing known as ‘sensorimotor coordination’ or ‘active perception’ to inform behaviour, but active perception has not before been investigated at a cellular level within organs. Here we provide the first proof of concept in silico/in vivo study demonstrating that filopodia (actin-rich, dynamic, finger like cell-membrane protrusions) play an unexpected role in speeding up collective endothelial decisions during the time-constrained process of ‘tip cell’ selection during blood vessel formation (angiogenesis).We first validate simulation predictions in vivo with live imaging of zebrafish intersegmental vessel growth. Further simulation studies then indicate the effect is due to the coupled positive feedback between movement and sensing on filopodia conferring a bistable switch-like property to Notch lateral inhibition, ensuring tip selection is a rapid and robust process. We then employ measures from computational neuroscience to assess whether filopodia function as a primitive (‘basal’) form of active perception and find evidence in support. By viewing cell behaviour in tissues through the ‘basal cognitive lens’ we acquire a fresh perspective on not only the well-studied tip cell selection process, revealing a hidden, yet vital, time-keeping role for filopodia, but on how to interpret and understand cell behaviour in general, opening up a myriad of new and exciting research directions.

Ubiquitous biological noise—random variation in the rate of molecular processes—is known to result in variable gene expression among isogenic cells. Such heterogeneity can synergize with phenotypic plasticity of the regulatory system to generate functionally distinct, dynamic phenotype-mosaics. Here we discuss parallels between phenotypic plasticity in response to environmental versus noise-driven variation. In a case study of in vivo mosaic heterogeneity, we detail evidence that in mice the endothelial von Willebrand factor (vWF) gene is expressed in a mosaic ON/OFF pattern in several capillary beds. This vWF mosaicism is dynamic; clonal analysis of cultured endothelial cells indicates that it is generated by random transitions of a noise-sensitive bistable switch. Finally, the hearts of vWF-null mice show abnormal capillary structure and cardiac dysfunction. Our findings suggest that stochastic phenotype switching may be a homeostatic strategy in the adult vasculature, and more generally, a requirement of healthy tissue function.

Graphical abstract, Epithelial cells respond to their physical neighborhood with mechano-sensitive behaviors required for development and tissue maintenance. These include anchorage dependence, matrix stiffness-dependent proliferation, contact inhibition of proliferation and migration, and collective migration that balances cell crawling with the maintenance of cell junctions. While required for development and tissue repair, these coordinated responses to the microenvironment also contribute to cancer metastasis. Predictive models of the signaling networks that coordinate these behaviors are critical in controlling cell behavior to halt disease. Here we propose a Boolean regulatory network model that synthesizes mechanosensitive signaling that links anchorage to a matrix of varying stiffness and cell density sensing to contact inhibition, proliferation, migration, and apoptosis. Our model can reproduce anchorage dependence and anoikis, detachment-induced cytokinesis errors, the effect of matrix stiffness on proliferation, and contact inhibition of proliferation and migration by two mechanisms that converge on the YAP transcription factor. In addition, we offer testable predictions related to cell cycle-dependent anoikis sensitivity, the molecular requirements for abolishing contact inhibition, and substrate stiffness dependent expression of the catalytic subunit of PI3K. Moreover, our model predicts heterogeneity in migratory vs. non-migratory phenotypes in sub-confluent monolayers, and co-inhibition but semi-independent induction of proliferation vs. migration as a function of cell density and mitogenic stimulation. Our model serves as a stepping-stone towards modeling mechanosensitive routes to the epithelial to mesenchymal transition, capturing the effects of the mesenchymal state on anoikis resistance, and understanding the balance between migration versus proliferation at each stage of the epithelial to mesenchymal transition.

We perform logic-based network analysis on a model of the mammalian cell cycle. This model is composed of a Restriction Switch driving cell cycle commitment and a Phase Switch driving mitotic entry and exit. By generalizing the concept of stable motif, i.e., a self-sustaining positive feedback loop that maintains an associated state, we introduce the concept of a conditionally stable motif, the stability of which is contingent on external conditions. We show that the stable motifs of the Phase Switch are contingent on the state of three nodes through which it receives input from the rest of the network. Biologically, these conditions correspond to cell cycle checkpoints. Holding these nodes locked (akin to a checkpoint-free cell) transforms the Phase Switch into an autonomous oscillator that robustly toggles through the cell cycle phases G1, G2 and mitosis. The conditionally stable motifs of the Phase Switch Oscillator are organized into an ordered sequence, such that they serially stabilize each other but also cause their own destabilization. Along the way they channel the dynamics of the module onto a narrow path in state space, lending robustness to the oscillation. Self-destabilizing conditionally stable motifs suggest a general negative feedback mechanism leading to sustained oscillations.

The PI3K/AKT signaling pathway plays a role in most cellular functions linked to cancer progression, including cell growth, proliferation, cell survival, tissue invasion and angiogenesis. It is generally recognized that hyperactive PI3K/AKT1 are oncogenic due to their boost to cell survival, cell cycle entry and growth-promoting metabolism. That said, the dynamics of PI3K and AKT1 during cell cycle progression are highly nonlinear. In addition to negative feedback that curtails their activity, protein expression of PI3K subunits has been shown to oscillate in dividing cells. The low-PI3K/low-AKT1 phase of these oscillations is required for cytokinesis, indicating that oncogenic PI3K may directly contribute to genome duplication. To explore this, we construct a Boolean model of growth factor signaling that can reproduce PI3K oscillations and link them to cell cycle progression and apoptosis. The resulting modular model reproduces hyperactive PI3K-driven cytokinesis failure and genome duplication and predicts the molecular drivers responsible for these failures by linking hyperactive PI3K to mis-regulation of Polo-like kinase 1 (Plk1) expression late in G2. To do this, our model captures the role of Plk1 in cell cycle progression and accurately reproduces multiple effects of its loss: G2 arrest, mitotic catastrophe, chromosome mis-segregation / aneuploidy due to premature anaphase, and cytokinesis failure leading to genome duplication, depending on the timing of Plk1 inhibition along the cell cycle. Finally, we offer testable predictions on the molecular drivers of PI3K oscillations, the timing of these oscillations with respect to division, and the role of altered Plk1 and FoxO activity in genome-level defects caused by hyperactive PI3K. Our model is an important starting point for the predictive modeling of cell fate decisions that include AKT1-driven senescence, as well as the non-intuitive effects of drugs that interfere with mitosis., Author summary Complex diseases such as cancer often alter more than one facet of a cell’s function. In addition to breakdown in individual functions, cancer progression leads to unhealthy combinations of cellular behaviors. For example, cancer cells rely on non-physiological combinations of cell functions drawn from an arsenal that includes proliferation, resistance to apoptosis, migration, and blood vessel recruitment. These functions are all critical to health or development, often in a different tissue than that of the tumor. Building predictive models that reproduce this coordination of functions could greatly boost our ability to combat complex disease. Here, we develop a large network model of the processes that control a mammalian cell’s life and death. Our model reproduces a non-intuitive oscillation in a key cell division pathway (PI3K/AKT1), along with the cell-cycle altering effect of its oncogenic activation. To do this, we incorporate the role of Polo-like kinase 1 (mitotic driver, chemotherapy target) and model mitotic failure when Plk1 is blocked. Finally, we offer testable predictions on the unexplored drivers of PI3K oscillations, their timing with respect to division, and the mechanism by which hyperactive PI3K leads to genome-level defects. Thus, our work can aid development of powerful models that cover most processes that go awry when cells transition into malignancy.

Beekeepers providing pollination services for California almond orchards have reported observing dead or malformed brood during and immediately after almond bloom&mdash, effects that they attribute to pesticide exposure. The objective of this study was to test commonly used insecticides and fungicides during almond bloom on honey bee larval development in a laboratory bioassay. In vitro rearing of worker honey bee larvae was performed to test the effect of three insecticides (chlorantraniliprole, diflubenzuron, and methoxyfenozide) and three fungicides (propiconazole, iprodione, and a mixture of boscalid-pyraclostrobin), applied alone or in insecticide-fungicide combinations, on larval development. Young worker larvae were fed diets contaminated with active ingredients at concentration ratios simulating a tank-mix at the maximum label rate. Overall, larvae receiving insecticide and insecticide-fungicide combinations were less likely to survive to adulthood when compared to the control or fungicide-only treatments. The insecticide chlorantraniliprole increased larval mortality when combined with the fungicides propiconazole or iprodione, but not alone, the chlorantraniliprole-propiconazole combination was also found to be highly toxic to adult workers treated topically. Diflubenzuron generally increased larval mortality, but no synergistic effect was observed when combined with fungicides. Neither methoxyfenozide nor any methoxyfenozide-fungicide combination increased mortality. Exposure to insecticides applied during almond bloom has the potential to harm honey bees and this effect may, in certain instances, be more damaging when insecticides are applied in combination with fungicides.

How do cells make efficient collective decisions during tissue morphogenesis? Humans and other organisms use feedback between movement and sensing known as ‘sensorimotor coordination’ or ‘active perception’ to inform behaviour, but active perception has not before been investigated at a cellular level within organs. Here we provide the first proof of concept in silico / in vivo study demonstrating that filopodia (actin-rich, dynamic, finger-like cell membrane protrusions) play an unexpected role in speeding up collective endothelial decisions during the time-constrained process of ‘tip cell’ selection during blood vessel formation (angiogenesis). We first validate simulation predictions in vivo with live imaging of zebrafish intersegmental vessel growth. Further simulation studies then indicate the effect is due to the coupled positive feedback between movement and sensing on filopodia conferring a bistable switch-like property to Notch lateral inhibition, ensuring tip selection is a rapid and robust process. We then employ measures from computational neuroscience to assess whether filopodia function as a primitive (basal) form of active perception and find evidence in support. By viewing cell behaviour through the ‘basal cognitive lens' we acquire a fresh perspective on the tip cell selection process, revealing a hidden, yet vital time-keeping role for filopodia. Finally, we discuss a myriad of new and exciting research directions stemming from our conceptual approach to interpreting cell behaviour. This article is part of the theme issue ‘Basal cognition: multicellularity, neurons and the cognitive lens’.

Endothelial cells (ECs) exhibit dramatic plasticity of form at the single- and collective-cell level during new vessel growth, adult vascular homeostasis, and pathology. Understanding how, when, and why individual ECs coordinate decisions to change shape, in relation to the myriad of dynamic environmental signals, is key to understanding normal and pathological blood vessel behavior. However, this is a complex spatial and temporal problem. In this review we show that the multidisciplinary field of Adaptive Systems offers a refreshing perspective, common biological language, and straightforward toolkit that cell biologists can use to untangle the complexity of dynamic, morphogenetic systems.

Angiogenesis is a highly dynamic morphogenesis process; however, surprisingly little is known about the timing of the different molecular processes involved. Although the role of the VEGF-notch-DLL4 signaling pathway has been established as essential for tip/stalk cell competition during sprouting, the speed and dynamic properties of the underlying process at the individual cell level has not been fully elucidated. In this study, using mathematical modeling we investigate how specific, biologically meaningful, local conditions around and within an individual cell can influence their unique tip/stalk phenotype switching kinetics. To this end we constructed an ordinary differential equation model of VEGF-notch-DLL4 signaling in a system of two, coupled endothelial cells (EC). Our studies reveal that at any given point in an angiogenic vessel the time it takes a cell to decide to take on a tip or stalk phenotype may be drastically different, and this asynchrony of tip/stalk cell decisions along vessels itself acts to speed up later competitions. We unexpectedly uncover intermediate "partial" yet stable states lying between the tip and stalk cell fates, and identify that internal cellular factors, such as NAD-dependent deacetylase sirtuin-1 (Sirt1) and Lunatic fringe 1 (Lfng1), can specifically determine the length of time a cell spends in these newly identified partial tip/stalk states. Importantly, the model predicts that these partial EC states can arise during normal angiogenesis, in particular during cell rearrangement in sprouts, providing a novel two-stage mechanism for rapid adaptive behavior to the cells highly dynamic environment. Overall, this study demonstrates that different factors (both internal and external to EC) can be used to modulate the speed of tip/stalk decisions, opening up new opportunities and challenges for future biological experiments and therapeutic targeting to manipulate vascular network topology, and our basic understanding of developmental/pathological angiogenesis.

Sepsis is a systemic inflammatory response to infections associated with organ failure that is the most frequent cause of death in hospitalized patients. Exaggerated endothelial activation, altered blood flow, vascular leakage, and other disturbances synergistically contribute to sepsis-induced organ failure. The underlying signaling events associated with endothelial proinflammatory activation are not well understood, yet they likely consist of molecular pathways that act in an endothelium-specific manner. We found that LPS, a critical factor in the pathogenesis of sepsis, is internalized by endothelial cells, leading to intracellular signaling without the need for priming as found recently in immune cells. By identifying a novel role for retinoic acid–inducible gene-I (RIG-I) as a central regulator of endothelial activation functioning independent of TLR4, we provide evidence that the current paradigm of TLR4 solely being responsible for LPS-mediated endothelial responses is incomplete. RIG-I, as well as the adaptor protein mitochondrial antiviral signaling protein, regulates NF-κB–mediated induction of adhesion molecules and proinflammatory cytokine expression in response to LPS. Our findings provide essential new insights into the proinflammatory signaling pathways in endothelial cells and suggest that combined endothelial-specific inhibition of RIG-I and TLR4 will provide protection from aberrant endothelial responses associated with sepsis.

Intractable diseases such as cancer are associated with breakdown in multiple individual functions, which conspire to create unhealthy phenotype-combinations. An important challenge is to decipher how these functions are coordinated in health and disease. We approach this by drawing on dynamical systems theory. We posit that distinct phenotype-combinations are generated by interactions among robust regulatory switches, each in control of a discrete set of phenotypic outcomes. First, we demonstrate the advantage of characterizing multi-switch regulatory systems in terms of their constituent switches by building a multiswitch cell cycle model which points to novel, testable interactions critical for early G2/M commitment to division. Second, we define quantitative measures of dynamical modularity, namely that global cell states are discrete combinations of switch-level phenotypes. Finally, we formulate three general principles that govern the way coupled switches coordinate their function.

Endothelial cells display remarkable phenotypic heterogeneity. An important goal is to elucidate the scope and mechanisms of endothelial heterogeneity and to use this information to develop vascular bed–specific therapies. We reexamine our current understanding of the molecular basis of endothelial heterogeneity. We introduce multistability as a new explanatory framework in vascular biology. We draw on the field of nonlinear dynamics to propose a dynamical systems framework for modeling multistability and its derivative properties, including robustness, memory, and plasticity. Our perspective allows for both a conceptual and quantitative description of system-level features of endothelial regulation.

Vascular endothelial growth factor receptor 1 (VEGFR1) is a marker for endothelial-specific gene expression. We previously reported that the human VEGFR1 promoter (between −748 and +284) contains information for expression in the intact endothelium of transgenic mice. The objective of this study was to dissect the cis-regulatory elements underlying VEGFR1 promoter activity in vitro and in vivo. In primary endothelial cells, binding sites for E74-like factor 1 (ELF-1; between −49 and −52), cyclic adenosine monophosphate response element binding (CREB; between −74 and −81), and early growth response factor 1/3 (EGR-1/3; between −16 to −25) were shown to play a positive role in gene transcription, whereas a putative E26 transformation-specificsequence (ETS) motif between −36 and −39 had a net negative effect on promoter activity. When targeted to the Hprt locus of mice, mutations of the ELF-1 binding site and the CRE element reduced promoter activity in the embryonic vasculature and resulted in a virtual loss of expression in adult endothelium. Postnatally, the EGR binding site mutant displayed significantly reduced promoter activity in a subset of vascular beds. In contrast, mutation of the −39 ETS site resulted in increased LacZ staining in multiple vascular beds. Together, these results provide new insights into the transcriptional regulatory mechanisms of VEGFR1.

Roundabout4 (Robo4) is an endothelial cell-specific gene that plays an important role in endothelial cell stability. We previously identified a 3-kb Robo4 promoter and demonstrated the importance of its proximal region in regulating Robo4 gene expression. To investigate the role of the upstream promoter in Robo4 gene regulation, we searched evolutionarily conserved promoter regions by phylogenetic footprinting and identified three conserved promoter regions. The most upstream region included a conserved AP-1 binding motif at position -2875. A mutation in the AP-1 motif significantly decreased Robo4 promoter activity in a transient reporter assay. An electrophoretic mobility shift assay and a chromatin immunoprecipitation assay demonstrated binding of a c-Jun/c-Jun complex and a c-Jun/Fra-1 complex to the AP-1 motif. Knockdown experiments using siRNA revealed that both c-Jun/c-Jun and c-Jun/Fra-1 complexes regulate Robo4 gene expression, and that the c-Jun/c-Jun complex is essential for maximum promoter activation. Collectively, these results indicate that AP-1 complexes regulate Robo4 gene expression in endothelial cells.

Previous studies have shown that biological noise may drive dynamic phenotypic mosaicism in isogenic unicellular organisms. However, there is no evidence for a similar mechanism operating in metazoans. Here we show that the endothelial-restricted gene, von Willebrand factor (VWF), is expressed in a mosaic pattern in the capillaries of many vascular beds and in the aorta. In capillaries, the mosaicism is dynamically regulated, with VWF switching between ON and OFF states during the lifetime of the animal. Clonal analysis of cultured endothelial cells reveals that dynamic mosaic heterogeneity is controlled by a low-barrier, noise-sensitive bistable switch that involves random transitions in the DNA methylation status of the VWF promoter. Finally, the hearts of VWF-null mice demonstrate an abnormal endothelial phenotype as well as cardiac dysfunction. Together, these findings suggest a novel stochastic phenotype switching strategy for adaptive homoeostasis in the adult vasculature., Spontaneous phenotypic heterogeneity confers a population-level advantage to cells that are exposed to fluctuating environments. Here the authors show that the endothelium of some organs displays a dynamic mosaicism in expression of Von Willebrand factor, suggesting bet hedging as a strategy for adaptive homeostasis.

Among unicellular organisms, stochastic phenotype switching is a documented strategy for survival. These populations "hedge their bets": while the majority of their cells are adapted to their present environment, a minority remains poised to thrive under drastically different conditions. Bet hedging has also been described in metazoan cells, primarily in vitro. However, its role in tissue homeostasis has yet to be established. Here, we show that von Willebrand factor (vWF) is expressed in a spatially heterogeneous manner in a small fraction of capillary endothelial cells in the heart, skeletal muscle, lung and brain. Moreover, these mosaic patterns are dynamic, in that vWF expression stochastically toggles ON/OFF over time. By contrast, expression of vWF in the aorta and liver is static in time. In cultured primary endothelial cells, biological noise resulted in mosaic vWF heterogeneity through a promoter-level DNA methylation switch. Finally, vWF-/- mice demonstrated extensive endothelial cell damage in capillaries of the heart and impaired cardiac function, but not kidney or aorta. Taken together, these findings suggest that dynamic mosaicism of vWF expression is functionally relevant and that bet hedging represents a previously unrecognized strategy for adaptive, organ-specific homeostasis.

Tie2 is predominantly expressed by endothelial cells and is involved in vascular integrity control during sepsis. Changes in Tie2 expression during sepsis development may contribute to microvascular dysfunction. Understanding the kinetics and molecular basis of these changes may assist in the development of therapeutic intervention to counteract microvascular dysfunction.To investigate the molecular mechanisms underlying the changes in Tie2 expression upon lipopolysaccharide (LPS) challenge.Studies were performed in LPS and pro-inflammatory cytokine challenged mice as well as in mice subjected to hemorrhagic shock, primary endothelial cells were used for in vitro experiments in static and flow conditions. Eight hours after LPS challenge, Tie2 mRNA loss was observed in all major organs, while loss of Tie2 protein was predominantly observed in lungs and kidneys, in the capillaries. A similar loss could be induced by secondary cytokines TNF-alpha and IL-1 beta. Ang2 protein administration did not affect Tie2 protein expression nor was Tie2 protein rescued in LPS-challenged Ang2-deficient mice, excluding a major role for Ang2 in Tie2 down regulation. In vitro, endothelial loss of Tie2 was observed upon lowering of shear stress, not upon LPS and TNF-alpha stimulation, suggesting that inflammation related haemodynamic changes play a major role in loss of Tie2 in vivo, as also hemorrhagic shock induced Tie2 mRNA loss. In vitro, this loss was partially counteracted by pre-incubation with a pharmacologically NF-(DB)-B-0 inhibitor (BAY11-7082), an effect further substantiated in vivo by pre-treatment of mice with the NF-(DB)-B-0 inhibitor prior to the inflammatory challenge.Microvascular bed specific loss of Tie2 mRNA and protein in vivo upon LPS, TNF alpha, IL-1 beta challenge, as well as in response to hemorrhagic shock, is likely an indirect effect caused by a change in endothelial shear stress. This loss of Tie2 mRNA, but not Tie2 protein, induced by TNF alpha exposure was shown to be controlled by NF-(DB)-B-0 signaling. Drugs aiming at restoring vascular integrity in sepsis could focus on preventing the Tie2 loss.

A region of the human von Willebrand factor (VWF) gene between −2812 and the end of the first intron (termed vWF2) was previously shown to direct expression in the endothelium of capillaries and a subset of larger blood vessels in the heart and skeletal muscle. Here, our goal was to delineate the DNA sequences responsible for this effect. A series of constructs containing deletions or mutations of vWF2 coupled to LacZ were targeted to the Hprt locus of mice, and the resulting animals were analyzed for reporter gene expression. The findings demonstrate that DNA sequences between −843 and −620 are necessary for expression in capillary but not large vessel endothelium in heart and skeletal muscle. Further, expression of VWF in capillaries and larger vessels of both tissues required the presence of a native or heterologous intron. In vitro assays implicated a role for ERG-binding ETS motif at −56 in mediating basal expression of VWF. In Hprt-targeted mice, mutation of the ETS consensus motif resulted in loss of LacZ expression in the endothelium of the heart and skeletal muscle. Together, these data indicate that distinct DNA modules regulate vascular bed–specific expression of VWF.

Accurate and efficient community detection in networks is a key challenge for complex network theory and its applications. The problem is analogous to cluster analysis in data mining, a field rich in metric space-based methods. Common to these methods is a geometric, distance-based definition of clusters or communities. Here we propose a new geometric approach to graph community detection based on graph Voronoi diagrams. Our method serves as proof of principle that the definition of appropriate distance metrics on graphs can bring a rich set of metric space-based clustering methods to network science. We employ a simple edge metric that reflects the intra- or inter-community character of edges, and a graph density-based rule to identify seed nodes of Voronoi cells. Our algorithm outperforms most network community detection methods applicable to large networks on benchmark as well as real-world networks. In addition to offering a computationally efficient alternative for community detection, our method opens new avenues for adapting a wide range of data mining algorithms to complex networks from the class of centroid- and density-based clustering methods.