Francesca Cesari

Development: Brown fat: muscle undercover?

undercover’, I would take it as an excuse not to join me in the gym. However, two new studies in Nature now reveal that brown fat and muscle have much more in common than was previously thought. In mammals, white adipose tissue stores energy in the form of fat and is associated with obesity, whereas the brown variant burns fat to generate heat, thereby counteracting obesity. Stimulating an increase in brown fat in humans might therefore be a potential treatment for obesity, but this first requires a full understanding of the developmental origin of the two fat tissues. Seale et al. used short hairpin RNA in precursor cells to knock down PRDM16, a transcriptional regu lator of brown fat cells, and observed that brown-fat-cell markers were repressed, whereas myogenic genes were activated. Furthermore, the cells turned into bona fide muscle cells with fused myotube morphology. Intrigued by these findings, they performed lineage-tracing experiments in mice that were genetically modified to mark the expression of myogenic factor-5 (Myf5), a skeletal muscle-specific gene. Myf5-expressing cells formed skeletal muscle and brown, but not white, fat cells. This suggests that brown fat cells arise from skeletal muscle precursors. Forced expression of PRDM16 in a muscle progenitor cell line blocked myogenesis and induced brown adipogenesis. To investigate the mechanism of action of PRDM16, Seale et al. purified a PRDM16 protein complex from fat cells and analysed it by mass spectrometry. They identified peroxisome pro liferator-activated receptor-γ (PPARγ), an essential regulator of adipogenic differentiation, as a binding partner of PRDM16. PRDM16 enhanced the expression of a PPARγ reporter gene and was unable to promote adipogenesis in PPARγ-deficient fibroblasts. So, PRDM16 stimulates brown adipogenesis by inducing PPARγ. In an independent study, Tseng et al. treated fat-cell precursors with bone morphogenetic proteins (BMPs), which are important for numerous processes during embryonic development. They found that BMP7 alone could instruct brown-fat-cell precursors to become mature cells that express brown-fat-cell markers and genes that are involved in the biogenesis and function of mitochondria. Interestingly, Prdm16 was among the BMP7-activated genes. BMP7 activates the p38 mitogenactivated protein kinase (MAPK) signalling pathway in brown precursor cells, and specific inhibitors of this pathway blocked the BMP7-induced expression of brown-fat-cell markers. Furthermore, BMP7-deficient embryos and newborn mice had reduced brown fat tissue, although other organs were normal. Similarly, Seale et al. observed abnormal morphology of brown fat cells and activation of muscle-specific genes in embryos that lacked PRDM16. By virally delivering BMP7 into wild-type mice, Tseng et al. induced an increase in brown fat mass, which led to higher energy expenditure and a reduction in weight gain compared to control animals. These studies suggest that treating humans with BMP7 or increasing the levels of PRDM16, which activate brown fat differentiation and increase energy expenditure, might be a new way to combat obesity. Francesca Cesari

Autophagy: Autophagy takes an alternative route

and ATG7 are thought to be essential for mammalian autophagy (also known as macroautophagy) — the lysosomal breakdown of organelles, proteins and other components of the cytoplasm to sustain metabolism during starvation and metabolic stress. Now, a study in Nature has uncovered an ATG5and ATG7-independent pathway that controls autophagy. Nishida et al. observed autophagic structures, such as autophagosomes (double membrane-bound vesicles that sequester materials to be degraded and deliver them to lysosomes), in Atg5 and in Atg7 mouse embryonic fibroblasts (MEFs) that were treated with etoposide (a cellular stress-inducing agent) or starved. This suggests that an ATG5and ATG7-independent autophagy system exists. As some markers of the ‘conventional’ autophagy pathway were not seen in Atg5 or Atg7 MEFs undergoing this process, the authors sought to identify the mechanism of ATG5and ATG7-independent autophagy. Gene expression analysis and silencing experiments in etoposide-treated and untreated Atg5 MEFs showed that the alternative autophagy pathway requires some components of the conventional autophagy pathway, such as UNC51-like kinase 1 and the phosphoinositide 3-kinase complexes (which act upstream to initiate autophagy), but not other components, such as proteins in the ubiquitin-like protein conjugation system (which is downstream in the pathway). Furthermore, RAB9, which is involved in trafficking from late endosomes to the trans-Golgi, is crucial for the formation of autophagosomes in the ATG5-independent route, but not during conventional autophagy. Next, the authors found that autophagic vacuoles are present in several Atg5 embryonic tissues. Notably, the mitochondria of maturing erythrocytes are still digested in the vacuoles of Atg5 mice, thereby suggesting that the alternative autophagy pathway is involved in the elimination of mitochondria during erythroid terminal differentiation. Together these results show that mammalian autophagy can occur through a canonical ATG5and ATG7-dependent pathway and through an alternative ATG5and ATG7-independent pathway. Whether these two pathways have different physiological roles and/or are activated by different stimuli in different cell types remains to be investigated. Francesca Cesari

Gene expression: A new signature for splicing

Alternative splicing of mRNA precursors, which occurs largely cotranscriptionally, allows individual genes to generate multiple mature mRNA isoforms that are translated into functionally different proteins. Despite the important role of alternative splicing in gene expression, its regulation is poorly understood. Now, Misteli and colleagues report that histone marks regulate alternative splicing by recruiting splicing regulators to splice sites through chromatin-binding adaptor proteins. The authors set out to study the regulation of alternative splicing using the human fibroblast growth factor receptor 2 (FGFR2) gene. This is a model gene for studying this process, as exons IIIb and IIIc undergo mutually exclusive and tissue-specific alternative splicing. Exon IIIb is included in FGFR2 in epithelial PNT2 prostate cells, whereas exon IIIc is included in FGFR2 in mesenchymal stem cells (MSCs). Also, binding of the splicing regulator polypyrimidine tract-binding protein (PTB) to splicing silencing elements upstream of exon IIIb is known to repress the inclusion of exon IIIb in FGFR2. The authors found that trimethylated Lys36 of histone H3 (H3K36me3) and H3K4me1 were enriched over FGFR2 in MSCs (in which exon IIIb is excluded), whereas H3K27me3, H3K4me3 and H3K9me1 were enriched over FGFR2 in PNT2 cells (in which exon IIIb is included). Notably, several PTB-dependent alternatively spliced exons in other genes showed similar splicing-specific histone modification patterns. Overexpression of a H3K36 methyl transferase (which catalyses the transfer of a methyl group to H3K36) in PNT2 and other cell types reduces the inclusion of exon IIIb in FGFR2, whereas its downregulation promotes the inclusion of this exon in FGFR2 in MSCs. Notably, overexpression and downregulation of a H3K4 methyltransferase has the opposite effect in these cell lines. These results suggest that histone modifications influence splice site selection, but how is this achieved? The chromatin-binding protein MRG15 (also known as MORF4L1) is known to bind to H3K36me3 and recruit the retinoblastoma-binding protein 2 (RBP2)–H3K4 demethylase complex to chromatin and could thereby generate a chromatin signature similar to that found in PTB-repressed alternatively spliced regions on FGFR2. Indeed, overexpression of MRG15 is sufficient to exclude the PTB-dependent exons from alternatively spliced precursor RNAs, whereas downregulation of MRG15 increases the inclusion of PTB-dependent exons, suggesting that MRG15 recruits PTB to splice sites, which have a specific chromatin signature. Importantly, endogenous PTB co-immunoprecipitates with MRG15, and overexpression of MRG15 in PNT2 cells forces the recruitment of PTB to exon IIIb in FGFR2, causing its exclusion from FGFR2. This study shows a role for histone modifications in the control of alternative splicing. The authors propose that chromatin-binding proteins read histone marks and transmit epigenetic information to the precursor mRNA processing machinery by interacting with splicing regulators. Francesca Cesari

DNA damage response: Change of guard at the checkpoint

to-ATR switch caused by DSB resection? Double-stranded DNA breaks (DSBs) activate two major DNAdamage checkpoint kinases, ataxia telangiectasia (A-T) mutated (ATM) and A-T and RAD3-related (ATR), which orchestrate the DNA damage response to delay cell cycle progression and allow the repair of DNA damage. However, the structural determinants of DNA that activate ATM and ATR and how the kinase activities are coordinated are poorly understood. Reporting in Molecular Cell, Shiotani and Zou provide new mechanistic insights. Using human cell extracts and DNA molecules with defined structures to assay ATM activation in vitro, the authors found that long double-stranded DNA (dsDNA) with blunt ends induces ATM phosphorylation (a marker of activation) more efficiently than short dsDNA with blunt ends. By contrast, dsDNA with long single-stranded overhangs (SSOs) exhibits reduced ability to activate ATM, whereas dsDNA with short SSOs activates ATM efficiently. So, the activation of ATM by DSBs is regulated by the length of both dsDNA and SSOs. DNA fragments with SSOs contain two types of ends: those of the dsDNA region (the junctions between the dsDNA and the ssDNA) and the ends of the SSOs. By chemically blocking these different dsDNA ends, Shiotani and Zou found that only the dsDNA–ssDNA junctions are crucial for ATM activation. Notably, the dsDNA–ssDNA junctions are also known to activate ATR. So, how are the activities of ATM and ATR coordinated at DSBs? D N A DA M AG E R E S P o N S E

Membrane trafficking: IFT proteins play a new game.

Primary cilia are present on most eukaryotic cells, where they function as sensory organelles to relay information from the external environment into the cell. Cilia are assembled by means of intraflagellar transport (IFT) — a process carried out by multimeric IFT particles and molecular motors. Finetti et al. now reveal an unexpected new role of IFT in cells lacking cilia, by showing that IFT is part of the membrane trafficking pathway that orchestrates signalling at the immune synapse — a platform that can integrate, fine-tune and terminate signalling.

Cell cycle: Destruct and arrest

(SAC) blocks chromosome segregation until all sister chromatids are properly attached to the mitotic spindle. It does so by inactivating CDC20 (also known as Slp1 and Fizzy) — an essential activator of the ubiquitinligase anaphase-promoting complex or cyclosome (APC/C), which promotes the degradation of proteins that arrest the cell cycle. Two studies now provide insights into how the SAC inhibits CDC20 and reveal a new role for CDC20 in APC/C-mediated protein degradation. Using nocodazole (a microtubuledepolymerizing agent) to activate the SAC in live-imaging studies of human cells, Nilsson et al. observed that a fluorescent CDC20 fusion protein was degraded in response to SAC. Degradation of CDC20 started at the time of nuclear envelope breakdown, which coincides with the time when APC/C recognizes its early mitotic substrates. This implies that CDC20 might be a target of APC/C. Indeed, the use of small interfering RNAs (siRNAs) against APC/C or the SAC protein MAD2 blocked CDC20 degradation, which suggests that CDC20 is degraded by APC/C in response to the SAC. Size-exclusion chromatography and quantitative immunoblotting experiments in nocodazole-treated cells revealed that CDC20 primarily interacts with the SAC component BUBR1 and not with MAD2, as previously thought. A mutant version of CDC20 that could not be ubiquitylated could bind to the SAC protein complex and be released from it when the SAC was chemically inhibited. However, it was not degraded by APC/C and was unable to maintain the SACinduced mitotic arrest. These findings contradict a previous model that suggests that ubiquitylation of CDC20 is needed to release it from the SAC complex and to activate APC/C. By contrast, Nilsson et al. show that ubiquitylation of CDC20 causes its degradation, which is required to maintain the SAC-associated mitotic arrest. CDC20-family proteins recognize APC/C substrates through a carboxyterminal WD40 repeat domain and recruit them to APC/C for degradation. Kimata et al. found that Nek2A, an APC/C substrate that interacts directly with APC/C independently of Cdc20, was not degraded in Xenopus laevis egg extracts that were depleted of Cdc20. This suggests that Cdc20 has an additional role besides substrate recruitment. By adding mutated versions of Cdc20 back to the egg extracts, the authors found that the amino-terminal C-box domain of Cdc20 facilitates Nek2A degradation. By contrast, the WD40-repeat domain was required for the degradation of ‘canonical’ substrates that are recruited to APC/C by Cdc20. The ability of purified APC/C from Cdc20-depleted egg extracts to ubiquitylate Nek2A in vitro was restored by adding the C-box domain of Cdc20 to the reaction. By contrast, the ubiquitylation of canonical substrates required the WD40 repeat domain. However, canonical substrates that were directly fused to the C-box domain of Cdc20 were efficiently ubiquitylated by APC/C, thereby revealing a new role for Cdc20 in promoting substrate ubiquitylation by APC/C. These studies represent important steps in understanding how the activity of APC/C is regulated during the cell cycle. Francesca Cesari

Protein degradation: Catching ubiquitin

ubiquitin-mediated protein degradation is well established, but little is known about how the proteasome recognizes its substrates. Now two studies have taken our understanding of this process one step further, by showing that the proteasome subunit Rpn13 is a high-affinity receptor for ubiquitin and recognizes its substrates through a new ubiquitinbinding domain. Using a yeast two-hybrid assay, Husnjak et al. found that a conserved N-terminal region of human RPN13 interacts with ubiquitin. To test whether Rpn13 binds ubiquitin chains in the context of intact proteasomes, the authors first purified proteasomes from a yeast rpn13∆ mutant and assayed ubiquitin-chain binding. They observed a reduction in ubiquitin-binding by rpn13∆ proteasomes, which is comparable to that observed in mutants of Rpn10, the only other proteasomal component known to function as ubiquitin receptor. Addition of recombinant Rpn13 to the rpn13∆ proteasome rescued the binding defect, thereby confirming that Rpn13 is a ubiquitin receptor. Analysis of the protein structure of yeast Rpn13 revealed that the N terminus of Rpn13, which binds to ubiquitin, has a configuration similar to the pleckstrin-homology domain (PHD). Therefore the authors named this domain pleckstrin-like receptor for ubiquitin (Pru). This is the first example of a PHD structure being found within the proteasome. In a separate study, Schreiner et al. found that murine Pru adopts a PHD fold structure and binds Lys48-linked diubiquitin with high affinity. In contrast to all other ubiquitin-binding proteins, RPN13 does not use α-helices to bind to ubiquitin, but instead uses the loop regions of the Pru. Mutants with amino-acid substitutions in the Pru loops did not bind to ubiquitin, which confirms that the loop regions of Pru are responsible for this interaction. Husnjak et al. determined the ubiquitin-binding residues of yeast Pru and showed that their mutation impairs binding. To function as a proteasomal ubiquitin receptor, Rpn13 has to bind to ubiquitin and proteasome components simultaneously. So, is this the case? Schreiner et al. found that the proteasome subunit Rpn2 and the ubiquitin-binding surfaces of Rpn13 are largely independent; Rpn13 binds to Rpn2 through the Pru domain without disrupting the Rpn13 loops that bind to ubiquitin. Husnjak et al. showed that Rpn13, like Rpn10, binds to ubiquitin-like (UBL)/ubiquitin-associated (UBA) proteins, which bind to and deliver ubiquitylated targets to the proteasome. Based on these findings, the authors propose that ubiquitin conjugates might bind to UBL/UBA proteins, which dock them to the proteasome and pass them to Rpn13 and Rpn10. Alternatively, UBL/UBA proteins and the intrinsic receptor might simultaneously bind to the targets. Because the C terminus of Rpn13 binds to deubiquitylating enzymes, both studies propose that Rpn13 might couple the ubiquitin-chain recognition and disassembly at the proteasome, thereby linking two essential steps of selective protein degradation. Francesca Cesari

Stem cells: Harvest in the right season

topoietic stem cells (HSCs) from the bone marrow into the bloodstream is the basis for bone marrow transplantation procedures. However, the molecular mechanisms that regulate this process are not well understood. A study in Nature now shows that the mobilization of HSCs follows a physiologically regulated circadian rhythm. While investigating the mechanism of enforced HSC mobilization by the growth factor granulocyte colony-stimulating factor (G–CSF), Paul Frenette and colleagues observed that the continuous exposure of mice to light could significantly affect HSC mobilization. Thus, a light-induced signal could influence the release of HSCs from the bone marrow. This unexpected finding led the authors to investigate the circadian pattern of HSC release and its role in homeostasis. Under normal conditions, HSCs do not circulate steadily or randomly but undergo precise circadian fluctuations, which can be altered by changes in the light–dark cycle, such as a ‘jet lag’ or the exposure of animals to continuous light. The authors also observed that expression of the Cxcl12 gene, which encodes a chemokine that attracts HSCs to the bone marrow, mirrored the oscillations in HSCs. These findings suggest that the decrease in Cxcl12 expression could be necessary for the rhythmic mobilization of HSCs. How is the light–dark cycle converted into fluctuations of Cxcl12 expression and release of HSCs into the bloodstream? By performing chemical and surgical ablation experiments, Frenette and colleagues showed that local signals from the sympathetic nervous system (SNS), mediated by the β3-adrenergic receptor, regulate the rhythmic expression of Cxcl12 and mediate the circadian exit of stem cells from bone marrow. Next, the authors showed that core genes of the circadian clock, such as Bmal1, Per1 and Per2, orchestrate Cxcl12 expression and the trafficking of stem cells, probably by regulating the rhythmic secretion of noradrenaline (which activates the β3-adrenergic receptor) from the nerve terminals. But what is the molecule that regulates Cxcl12 expression in response to SNS stimuli? Previous studies have shown that the transcription factor SP1 can bind to the promoter of Cxcl12. SP1 phosphorylation by the cAMP-dependent protein kinase (PKA) enhanced its DNA-binding activity. Furthermore, PKA is activated in response to β3adrenergic signalling. Using specific inhibitors, Frenette and colleagues showed that local stimulation of β3-adrenergic signalling causes the degradation of SP1 and a reduction in Cxcl12 expression, which results in increased HSC release. These findings suggest that choosing the right time for harvesting HSCs could increase the yields of stem cells for clinical applications. Francesca Cesari

Chromosome segregation: Tension rules

from being prematurely segregated at the first meiotic division in mammals has remained elusive. New research by Yoshinori Watanabe and colleagues shows that, in mammals, a family of proteins called shugoshins regulate centromeric cohesion in mitosis and meiosis in a tensiondependent manner. Meiosis is a specialized type of cell division that generates non-identical haploid gametes from diploid progenitor cells. This is achieved by a single round of DNA replication followed by two chromosome segregation phases: segregation of the homologues (meiosis I) is followed by the dissolution of centromere cohesion and the segregation of sister chromatids (meiosis II). Previous studies in yeast have shown that shugoshin associates with protein phosphatase-2A (PP2A) throughout meiosis I and prevents phosphorylation of centromeric cohesin — a multiprotein complex that maintains tight association of sister chromatids. This prevents cleavage of cohesin by the protease separase and subsequent sister chromatid separation. To investigate the role of mammalian shugoshins SGO1 and SGO2 during meiosis, the team led by Watanabe used mouse oocytes that were cultured in vitro. First, the authors showed that SGO1 and SGO2 localize to the centromeres during the two chromosome segregation phases of meiosis as well as during mitosis, but SGO2 is much more abundant than SGO1 in oocytes. Furthermore, REC8, a meiosis-specific cohesin component, colocalizes with SGO2 during meiosis I and is protected from cleavage by the PP2A pathway. Next, they depleted oocytes of either one or both shugoshins by RNA interference and found that SGO2, but not SGO1, is important for protecting centromeric cohesion in meiosis I. By contrast, SGO1 appears to be more important during mitosis than SGO2. So, why are centromeric cohesins not protected from cleavage during meiosis II and mitosis despite the presence of SGO2? The authors proposed that the tension generated by the spindle microtubules when they pull the kinetochores to opposite directions — which occurs in mitosis and meiosis II but not in meiosis I — could affect the localization of SGO2, and thereby affect the protection of centromeric cohesin. To test this hypothesis, Watanabe and colleagues first confirmed previous observations that SGO2 relocates from the centromeres towards the kinetochores during metaphase II. At this stage, REC8 is retained at its original centromeric position and can be cleaved, thereby triggering sister chromatid separation. Next, by artificially removing tension from the centromeres at the metaphase– anaphase transition in mitosis, the colocalization of shugoshin and cohesin was preserved and centromeric cohesin was protected from separase cleavage. Together, these observations suggest that shugoshins are involved in all instances of centromeric cohesion protection, both during mitosis and meiosis. In addition, the relocalization of SGO2 (due to the tension generated by the spindle microtubules when sister kinetochores are pulled to opposite poles) triggers the events that lead to sister chromatid separation in mitosis and meiosis II. How tension causes the relocation of shugoshins has yet to be determined. Francesca Cesari

Cell signalling: It's good to talk

How signals are processed by interacting cells is largely unknown. In a new study in Science, Pawson, Linding and colleagues have dissected cell-specific signalling networks during the interaction between cells expressing transmembrane Eph receptor Tyr kinases (EphRs) and cells expressing their membrane-bound ephrin ligands. They found that the receptorand ligand-expressing cells use different Tyr kinases and phosphorylation targets to process signals induced by cell–cell contacts. To study bidirectional EphR– ephrin signalling in the context of direct cell–cell interactions, the authors labelled cells expressing EphB2 or ephrin B1 using different amino acid isotopes, which enable relative quantification of Tyr phosphorylation peptides (and therefore signalling events) by mass spectrometry in a cell line-specific manner. They found that global changes in Tyr phosphorylation induced by cell–cell contact differ between the two cell types. The comparison of cell-specific modulation of the 100 Tyr phosphorylation sites present in both cell types identified asymmetric regulation of proteins that have a wide range of molecular and cellular functions. For example, the phosphorylation of adaptor proteins is preferentially increased in EphB2-expressing cells, suggesting that they have cell-specific regulatory roles. When cells expressing EphB2 and ephrin B1 contact each other in cell culture they segregate and form distinct colonies with well-defined boundaries. To study the role of selected signalling proteins during this process, a small interfering RNA (siRNA) screen was performed. Among the 200 targets that affected EphB2–ephrin B1 cell segregation, 37 were significantly modulated in EphB2-expressing cells and 26 were significantly modulated in ephrinexpressing cells. The siRNA data were combined with the quantified modulation of Tyr phosphorylation in the two cell types. The authors then used the NetPhorest and NetworKIN algorithms to derive a computational model of cell-specific dynamic signalling networks during cell segregation. The analysis revealed that the asymmetric regulation of Tyr phosphorylation events in EphB2and ephrin B1-expressing cells is achieved through the alternative use of kinases and adaptor proteins. Cells expressing a variant of ephrin B1 that lacks the cytoplasmic region (ephrin B1ΔIC) elicit a unidirectional signal in EphB2-expressing cells but do not induce signalling in the cell in which they are expressed. Notably, the analysis of EphB2–ephrin B1ΔIC signalling revealed that the cytoplasmic region of ephrin B1 is required for cell sorting and also affects signalling in neighbouring cells that express EphB2. This shows that EphR–ephrin signalling is not entirely cell autonomous. Using proteomic and computational approaches, this study provides important new insights into cellspecific signalling networks in two populations of cells, while they contact each other. Similar integrative network biology approaches will enable more accurate studies of the effects of cell–cell interactions and signalling networks during normal and pathological processes. Francesca Cesari