Publications

Motivated by the question of how biological systems maintain homeostasis in changing environments, Shinar and Feinberg introduced in 2010 the concept of absolute concentration robustness (ACR). A biochemical system exhibits ACR in some species if the steady-state value of that species does not depend on initial conditions. Thus, a system with ACR can maintain a constant level of one species even as the initial condition changes. Despite a great deal of interest in ACR in recent years, the following basic question remains open: How can we determine quickly whether a given biochemical system has ACR? Although various approaches to this problem have been proposed, we show that they are incomplete. Accordingly, we present new methods for deciding ACR, which harness computational algebra. We illustrate our results on several biochemical signaling networks.

Modelling liver disease requires in vitro systems that replicate disease progression1,2. Current tissue-derived organoids fail to reproduce the complex cellular composition and tissue architecture observed in vivo3. Here, we describe a multicellular organoid system composed of adult hepatocytes, cholangiocytes and mesenchymal cells that recapitulates the architecture of the liver periportal region and, when manipulated, models aspects of cholestatic injury and biliary fibrosis. We first generate reproducible hepatocyte organoids with functional bile canaliculi network that retain morphological features of in vivo tissue. By combining these with cholangiocytes and portal fibroblasts, we generate assembloids that mimic the cellular interactions of the periportal region. Assembloids are functional, consistently draining bile from bile canaliculi into the bile duct. Strikingly, manipulating the relative number of portal mesenchymal cells is sufficient to induce a fibrotic-like state, independently of an immune compartment. By generating chimeric assembloids of mutant and wild-type cells, or after gene knockdown, we show proof-of-concept that our system is amenable to investigating gene function and cell-autonomous mechanisms. Taken together, we demonstrate that liver assembloids represent a suitable in vitro system to study bile canaliculi formation, bile drainage, and how different cell types contribute to cholestatic disease and biliary fibrosis, in an all-in-one model.

Microtubule gliding assays provide a unique mechanism for molecular detection in which binding of analytes to the microtubule lattice reduces the microtubule gliding speed. The reduction in the gliding speed correlates with the density of the bound analytes, enabling its quantification. Although promising, this technique is still in the proof-of-concept stage. Improving the sensitivity and limit of detection of the assay could make the technique comparable to that of advanced molecular detection methods. This study demonstrates that reducing the ionic strength of the buffer increases the sensitivity of the assay by enhancing the interactions between kinesin and microtubules. When using a low ionic strength buffer (BRB10) compared with a standard buffer (BRB80), we observed a more pronounced reduction in microtubule gliding speed in the presence of analytes, improving the detection limit. Therefore, this approach offers a simple and scalable way to improve the sensitivity of motor-based detection assays.

Whereas alignment has been fundamental to sequence-based assessments of protein homology, it is ineffective for intrinsically disordered regions (IDRs) due to their lowered sequence conservation and unique sequence properties. Here, we present a web server implementation of SHARK (bio-shark.org), an alignment-free algorithm for homology classification that compares the overall amino acid composition and short regions (k-mers) shared between sequences (SHARK-scores). The output of such k-mer-based comparisons is used by SHARK-dive, a machine learning classifier to detect homology between unalignable, disordered sequences. SHARK-web provides sequence-versus-database assessment of protein sequence homology akin to conventional tools such as BLAST and HMMER. Additionally, we provide precomputed sets of IDR sequences from 16 model organism proteomes facilitating searches against species-specific IDR-omes. SHARK-dive offers superior overall homology detection performance to BLAST and HMMER, driven by a large increase in sensitivity to low sequence identity homologs, and can be used to facilitate the study of sequence-function relationships in disordered, difficult-to-align regions.

MEX-5 regulates the formation and dissolution of P granules in Caenorhabditis elegans embryos, yet the thermodynamic basis of its activity remains unclear. Here, using a time-resolved in vitro reconstitution system, we show that MEX-5 dissolves preassembled liquid-like PGL-3/RNA condensates by altering RNA availability and shifting the phase boundary. We develop a microfluidic assay to systematically analyze how MEX-5 influences phase separation. By measuring the contribution of PGL-3 to phase separation, we show that MEX-5 reduces the free energy of PGL-3, shifting the equilibrium toward dissolution. Our findings provide a quantitative framework for understanding how RNA-binding proteins modulate condensate stability and demonstrate the power of microfluidics in precisely mapping phase transitions.

Cytosolic aggregation of the nuclear protein TAR DNA-binding protein 43 (TDP-43) is associated with many neurodegenerative diseases, but the triggers for TDP-43 aggregation are still debated. Here, we demonstrate that TDP-43 aggregation requires a double event. One is up-concentration in stress granules beyond a threshold, and the other is oxidative stress. These two events collectively induce intra-condensate demixing, giving rise to a dynamic TDP-43-enriched phase within stress granules, which subsequently transition into pathological aggregates. Intra-condensate demixing of TDP-43 is observed in iPS-motor neurons, a disease mouse model, and patient samples. Mechanistically, intra-condensate demixing is triggered by local unfolding of the RRM1 domain for intermolecular disulfide bond formation and by increased hydrophobic patch interactions in the C-terminal domain. By engineering TDP-43 variants resistant to intra-condensate demixing, we successfully eliminate pathological TDP-43 aggregates in cells. We suggest that up-concentration inside condensates followed by intra-condensate demixing could be a general pathway for protein aggregation.

Neurodegenerative diseases, such as amyotrophic lateral sclerosis, are often associated with mutations in stress granule proteins. Aberrant stress granule condensate formation is associated with disease, making it a potential target for pharmacological intervention. Here, we identified lipoamide, a small molecule that specifically prevents cytoplasmic condensation of stress granule proteins. Thermal proteome profiling showed that lipoamide stabilizes intrinsically disordered domain-containing proteins, including SRSF1 and SFPQ, which are stress granule proteins necessary for lipoamide activity. SFPQ has redox-state-specific condensate dissolving behavior, which is modulated by the redox-active lipoamide dithiolane ring. In animals, lipoamide ameliorates aging-associated aggregation of a stress granule reporter protein, improves neuronal morphology and recovers motor defects caused by amyotrophic lateral sclerosis-associated FUS and TDP-43 mutants. Thus, lipoamide is a well-tolerated small-molecule modulator of stress granule condensation, and dissection of its molecular mechanism identified a cellular pathway for redox regulation of stress granule formation.

Upregulation of insulin mRNA translation upon hyperglycemia in pancreatic islet β-cells involves several RNA-binding proteins. Here, we found that G3BP1, a stress granule marker downregulated in islets of subjects with type 2 diabetes, binds to insulin mRNA in glucose concentration-dependent manner. We show in mouse insulinoma MIN6-K8 cells exposed to fasting glucose levels that G3BP1 and its paralog G3BP2 colocalize to cytosolic condensates with eIF3b, phospho-AMPKαThr172 and Ins1/2 mRNA. Glucose stimulation dissolves G3BP1+/2+ condensates with cytosolic redistribution of their components. The aldolase inhibitor aldometanib prevents the glucose- and pyruvate-induced dissolution of G3BP1+/2+ condensates, increases phospho-AMPKαThr172 levels and reduces those of phospho-mTORSer2448. G3BP1 or G3BP2 depletion precludes condensate assembly. KO of G3BP1 decreases Ins1/2 mRNA abundance and translation as well as proinsulin levels, and impaires glucose-stimulated insulin secretion. Further, other insulin secretagogues such as exendin-4 and palmitate, but not high KCl, prompts the dissolution of G3BP1+/2+ condensates. G3BP1+/2+/Ins mRNA+ condensates are also found in primary mouse and human β-cells. Hence, G3BP1+/2+ condensates represent a conserved glycolysis/aldolase-regulated compartment for the physiological storage and protection of insulin mRNA in resting β-cells.