Research in our laboratory centers on a number of fundamental questions in cell biology, for which we seek a molecular understanding. First, we would like to understand how eukaryotic cells regulate the abundance of their organelles. Intracellular regulatory pathways that control organelle growth and turnover are crucial for cells to maintain organelles in proper amounts at homeostasis and to become specialized upon differentiation. Second, we study the mechanisms by which proteins become properly localized within a cell. Protein sorting is essential to establish and maintain order and compartmentalization in all living cells. Third, we wish to elucidate the molecular components and mechanisms that dynamically organize the plasma membrane, both during endocytosis and during cell-cell fusion.
Regulation of Organelle Abundance
Starting with genetic approaches in the yeast Saccharomyces cerevisiae, we unraveled the mechanism of an intracellular signal transduction pathway that mediates communication between the endoplasmic reticulum (ER) lumen and the nucleus. Folding, modification, and assembly of proteins entering the ER are mediated by a specific set of enzymes (some of which function as chaperones) whose amount is regulated according to need. A sensor in the ER lumen determines the need for more protein-folding capacity and sends a signal to the nucleus, where transcription of genes encoding ER resident proteins is induced. This pathway is termed the unfolded protein response (UPR) because it can be experimentally induced under conditions that cause aberrant protein folding in the ER. Under more physiological conditions the UPR serves as homeostatic control to adjust the protein-folding capacity of the ER according to need.
Three genes—IRE1, HAC1, and TRL1—collaborate in this unique signaling pathway. IRE1 encodes a transmembrane kinase that transmits the signal originating in the ER lumen across the ER. The three-dimensional structure of Ire1's ER luminal domain, which senses unfolded proteins in the ER, shows a deep groove reminiscent in its architecture to grooves found in MHC (major histocompatibility complex) molecules that are known to bind peptides and display them on the cell surface for immune surveillance. It is likely that unfolded protein chains similarly bind into the groove of Ire1 to activate it. Because of steric considerations, compactly folded protein domains cannot reach into the groove, thus providing an elegant solution to the problem of how Ire1 could distinguish folded from unfolded proteins. In a fashion similar to that of growth factor receptors in the plasma membrane of higher eukaryotic cells, ligand binding (that is, binding of unfolded protein) activates Ire1 by inducing its oligomerization and autophosphorylation.
HAC1 encodes a bZIP transcription factor that regulates transcription of genes controlled by the UPR. Intriguingly, the activity of Hac1 is controlled through the regulated splicing of its mRNA. Removal of an intron from the HAC1 mRNA is a prerequisite for its translation and results in the production of the active transcription factor. The HAC1 mRNA-splicing reaction occurs by an unprecedented mechanism that does not involve spliceosomes, which normally mediate mRNA splicing. Rather, splicing occurs by the sequential action of two enzymes, Ire1 and Trl1. Ire1 is a bifunctional enzyme that in addition to its kinase function also contains an endonuclease domain that cleaves HAC1 mRNA at both splice junctions. Surprisingly, the kinase function of Ire1 can be entirely bypassed, indicating that it primarily serves as a conformational module, providing an activation switch for the nuclease activity. The tRNA ligase (encoded by TRL1), an essential enzyme previously known exclusively for its role in pre-tRNA splicing, then joins the two exons. This reaction has been reconstituted in vitro from purified recombinant components.
We now have a comprehensive understanding of the cooperation of Ire1's kinase and RNase module in activating the enzyme. Both enzymatic properties and the crystal structure of the kinase/RNase domains demonstrate that the active form is an oligomer. The molecular interfaces that build the higher order structure allow one kinase to transphosphorylate its neighbor and organize the RNase active sites to afford their activation. Oligomerization can also be seen in living cells, where upon UPR activation Ire1 molecules coalesce in discrete foci of a few tens of molecules, to which HAC1 mRNA is recruited for splicing. A newly discovered targeting signal in the HAC1 mRNA serves to bring it to these "splicing factories," when the UPR is induced.
The salient features of the UPR signaling pathway, including the nonconventional splicing reaction, are conserved in mammalian cells, where it is essential for the development of dedicated secretory cells. The UPR is also co-opted by viruses as they enter cells and expand the ER for their own replication, required to expand the ER in rapidly growing cancer cells, and can trigger the death of cells that exceed their secretory capacity for a prolonged time. This latter example may be the reason why pancreatic beta cells die in type II diabetes. How cells make this life-or-death decision remains an exciting question. It now seems that parallel signaling pathways from the ER are turned off with different kinetics, thus allowing kinetic control by which cells can first try to fix the problem and then, if ER stress remains unmitigated despite efforts to remedy it, commit to apoptosis. It will be fascinating to test whether the switch from cytoprotective to apoptotic functions can be manipulated to our advantage by pharmacological intervention. It has been rewarding to see how lessons learned from yeast prove directly applicable to mammalian biology and disease, revealing fundamentally new insights into the mechanisms by which cells regulate their organellar composition according to need.
We are also studying the pathways that allow proteins to become selectively targeted to the ER. Our goal is a mechanistic understanding of the molecular adapters that couple protein synthesis to membrane translocation—the signal recognition particle (SRP) and its receptor.
The binding of SRP to its various partners—signal sequences, ribosomes, and SRP receptor—poses a variety of challenging problems. First, signal sequences are divergent. The binding pocket on SRP must therefore have remarkable plasticity. Second, SRP and its receptor are regulated by GTPase switches, which makes possible signal sequence loading in the cytosol and unloading at the membrane. Studies from our laboratory and others show that protein targeting involves an unprecedented cascade of three distinct, directly interacting GTPases that are part of SRP and its receptor. The current goal is to decipher how the GTPases are regulated and how the energy of GTP hydrolysis is used in the protein-targeting reaction, possibly to promote fidelity and unidirectionality. Recently obtained structures demonstrate that the SRP and SRP receptor GTPases belong to a novel subgroup in the GTPase superfamily. The structure of the complex of SRP and SRP receptor GTPases and kinetic studies reveal an intriguing symmetry by which both enzymes reciprocally activate each other by aligning their respective GTP substrates. The modus operandi and the general properties of the SRP-type GTPases are significantly different from those of canonical GTPases, and we have been able to trap many kinetic intermediates in the reaction cycle in engineered mutant proteins.
The signal sequence–binding domain contains an expansive hydrophobic groove that is lined in large part by methionine residues, supporting the previous hypothesis that the abundant flexible methionine side chains provide a fluid, hydrophobic environment for signal sequence binding. In apposition to the signal sequence–binding groove is a helix-turn-helix motif to which SRP RNA binds. We can now show that that binding of signal sequences to SRP54 leads to conformational changes that are transmitted to SRP RNA, which in turn accelerates the interaction with the SRP receptor. This model begins to explain why SRPs in all cells have obligate RNA subunits.
To accomplish this task, the UPR coordinately regulates expression not only of ER resident proteins but also of key enzymes in lipid biosynthesis, components of the protein degradation machine, and components that mediate ER-to-Golgi transport. The induction of ER contents is coupled to the biogenesis of new membrane, leading to a significant expansion of the organelle. In addition, cells make every effort to purge the ER lumen of misfolded proteins, by enhancing both retrograde transport to the cytosol for degradation and forward transport through the secretory pathway, and also by inducing an ER-specific form of autophagy (ER-phagy) to sequester and eventually degrade damaged ER containing hopelessly misfolded proteins.
Dynamics of Plasma Membrane Organization
The plasma membranes of all eukaryotic cells undergo constant remodeling. Exocytic and endocytic events respectively deliver and remove lipids and plasma membrane proteins to renew the membrane and to adjust its composition according to changing needs. Although much has been learned about the molecular events that carry out these processes, there are gaping holes in our current knowledge.
For example, we have considerable knowledge of the endocytic machineries that select cargo, shape membranes into vesicles, pinch them off from the plasma membrane, and propel them into the cell's interior. Yet it has remained a mystery what determines when and where a vesicle forms. We recently discovered new organelle-like structures, which we termed eisosomes (from the Greek eis, meaning "into" or "portal," and somes, meaning "body"). Eisosomes are huge protein complexes containing a few thousand copies of two predominant subunits, Pil1 and Lsp1, as well as the integral membrane protein Sur7. There are 25–40 eisosomes per cell, firmly anchored at positions underlying the plasma membrane. Microscopic and genetic analyses link eisosomes to endocytosis: both lipid and protein cargoes are internalized at sites colocalizing with eisosomes, and mutations tie eisosome components into an extensive network of genetic interactions with known endocytic effectors. Only a few of the eisosomes present in a cell are active at one time, suggesting that eisosomes function as regulated portals that govern both location and magnitude of membrane traffic into the cell.
A similar paucity exists regarding our knowledge of plasma membrane fusion during cell-cell fusion events. Many biologically important processes, such as fertilization and muscle or placental development, rely on spatially and temporally controlled plasma membrane fusion events. We have begun to explore the process of plasma membrane fusion, using yeast mating as the experimental system. With a combination of bioinformatic and genetic approaches, we have identified novel plasma membrane proteins involved in the fusion process. In mutant cells, the plasma membranes of the two mating partners become closely apposed yet frequently fail to fuse. Disruption of individual identified components only partially affects fusion, suggesting a functional redundancy in the pathway, which may explain why previous approaches have failed to identify the membrane fusion apparatus. We have also learned that the lipid composition of the membrane is important, consistent with the idea that membrane fusion components have to be recruited into specialized lipid domains at the site of fusion. None of the components identified to date structurally resemble fusion proteins characterized in other systems, such as the well-characterized viral or intracellular (SNARE) fusion proteins, suggesting that membrane fusion may occur by a novel mechanism.
This work is supported in part by grants from the National Institutes of Health.
As of May 30, 2012