upfront this group showed that specifically ablat ing E‑cad expression dramatically inhibited reprogramming, suggesting that the MET required for induced pluripotency required the activity of E‑cad, and that it was not just a marker of fate change (Li et al, 2010).
…the spatial and mechanical input provided by E‑cad has an important role in altering cell fate The new study by Redmer and colleagues takes this work a step further and shows that E‑cad has an even more significant role than was previously appreciated. The authors confirm previous findings implicating a functional role for E‑cad in reprogramming by showing that the loss of E‑cad expression drives pluripotent stem cells to differen tiate. In addition, they show that the forced expression of a Yamanaka cocktail including E‑cad is able to drive induced pluripotency, even in the absence of OCT4. This is impor tant because although many groups have shown that most members of the Yamanaka cocktail of reprogramming factors—OCT4, SOX2, KLF4 and c‑MYC—can be omitted or replaced by various manipulations, there have been few instances of reprogramming to the pluripotent state proceeding in the absence of OCT4. The results of Redmer and co-workers emphasize that the spatial and mechanical input provided by E‑cad has an important role in altering cell fate. Perhaps these results should not be sur prising, as previous work has demonstrated that forced expression of E‑cad in cells facili tates the transition of an alternative type of pluripotent stem cell (FAB-SC) to become bona fide embryonic stem cells (Chou et al, 2008). Conversely, considering their differ ent biological activities, it is unlikely that E‑cad functionally replaced OCT4 in the experiments carried out by Redmer and col leagues. Perhaps the pertinent question is how the addition of E‑cad to the Yamanaka cocktail was able to drive reprogramming in the experiments by the Besser group. The most likely scenario, considering the work of Li and colleagues (2010), is that E‑cad and KLF4 acted to drive and maintain a MET early in reprogramming. Then, as SOX2 and KLF4 are known to drive transcription of highly overlapping patterns of pluripotency genes (Sridharan et al, 2009), these two factors sufficed to initiate the pluripotency programme, despite the lack of OCT4. As 614 EMBO reports VOL 12 | NO 7 | 2011
h ot off th e press
c‑MYC is known to act on metabolic and proliferation pathways (Sridharan et al, 2009), this protein could have facilitated the conversion to a self-renewing state and prevented the onset of senescence pathways (model summarized in Fig 1). The data presented by Redmer and col leagues reinforce the idea that MET is nec essary for, but not sufficient to induce, pluripotency in mesenchymal cells such as fibroblasts. Other groups have shown that reprogramming epithelial cells such as keratinocytes that already express E‑cad to the pluripotent state is quicker and more efficient (Aasen et al, 2008); this is thought to be due to the lack of a requirement for MET, although this has not been formally tested. Should that be true, it could be illustrative to determine whether the trans criptional and epigenetic transitions that occur in mesenchymal target cells are con served or distinct from those in epithelial cells on their way to the pluripotent state. Finally, it is clear from these data and pre vious work (Chou et al, 2008; Li et al, 2010) that reinforcement of an epithelial state is crucial to the stability of pluripotent stem cells, but it does not address the reason that this might be the case. This work was per formed in murine cells, which are capable of surviving at clonal density, whereas human pluripotent stem cells are not. Do murine and human induced pluripotent stem cells
…considering their different biological activities, it is unlikely that E‑cad functionally replaced Oct4… maintain similar epithelial characters and do they need polarity signals from proteins such as E‑cad, or do they simply need to maintain the epithelial state? Considerable effort will be required to tease apart these possibili ties, but it is necessary to achieve an under standing of not only pluripotency, but also the process by which the Yamanaka factors induce a complete remodelling of cell fate. References Aasen T et al (2008) Nat Biotechnol 26: 1276–1284 Acloque H et al (2009) J Clin Invest 119: 1438–1449 Chou YF et al (2008) Cell 135: 449–461 Li R et al (2010) Cell Stem Cell 7: 51–63 Perez-Moreno M, Jamora C, Fuchs E (2003) Cell 112: 535–548 Redmer T et al (2011) EMBO Rep 12: 719–725 Samavarchi-Tehrani P et al (2010) Cell Stem Cell 7: 64–77 Sridharan R et al (2009) Cell 136: 364–377 Stadtfeld M et al (2008) Cell Stem Cell 2: 230–240 Takahashi K, Yamanaka S (2006) Cell 126: 663–676
William E. Lowry is in the Department of Molecular, Cell and Developmental Biology, UCLA, Los Angeles, California, USA E‑mail:
[email protected] Published online 24 June 2011
EMBO reports (2011) 12, 613–614. doi:10.1038/embor.2011.117
A new postal code for dendritic mRNA transport in neurons Carsten Drepper & Michael Sendtner
T
he controlled sorting and transport of proteins has long been accepted as the basic regulatory mechanism responsible for establishing polarized struc tures in most cell types, whereas the sub cellular sorting and transport of messenger RNAs (mRNAs) has been considered to be a special feature of highly polarized cells, such as the neurons of the central nervous system. The concomitant idea that proteins can be synthesized in specific cellular compartments—under local control in den dritic spines in response to the activation of neighbouring synapses—was thought to be
the exception to the rule. This view, how ever, has changed. It has been shown that the majority of mRNAs are not evenly dis tributed in the cells of Drosophila embryos (Lécuyer et al, 2007), indicating that the controlled subcellular transport and sorting of mRNAs is at least as common as protein sorting, even in early cells, long before they differentiate into highly polarized neurons or other specialized cell types. In a paper published in this issue of in EMBO reports, Subramanian et al (2011) enhanced our understanding of neuronal mRNA transport. They found that a guanine
©2011 European Molecular Biology Organization
upfront
hot of f the pre s s
INSET An
?
Cytoplasm
??
FMRP ? U U U U U
STOP
Nucleus
G G G
ORF
G G G
G G G
G G G
G G G
G G G
FXR1?
An
G G G
G G G
FXR2?
G G G
G G G
G G G
G G G
A A A A A
AAAAA
? ?
An Dendrite
Mic mGluR G uR u
rot
ub
KIF5
ule
s
An
Fig 1 | G‑quadruplex RNA structure in the 3´UTR of dendritically translocated mRNAs. Three– four G‑quadruplex structural elements are found in many dendritically transported mRNAs, including PSD95 and CaMKIIa (Subramanian et al, 2011). These structural elements are generally followed by an AU‑rich element, suggesting that other proteins are involved in the final transport complex. Once formed, the complex is transported in a synaptic-activity-dependent manner into the dendrites of cortical neurons, probably by recruiting additional proteins that mediate a link to KIF5 or other motor proteins for dendritic translocation. FMRP, fragile-X mental retardation protein; mGluR, metabotropic glutamate receptor.
(G)-quadruplex or G‑quartet RNA struc ture in the 3' untranslated region (UTR) of the mRNAs for PSD95 and CaMKIIa guides these mRNAs into the dendritic processes of neurons. This domain is also found in the 3'UTR of many other mRNAs that are known to be sorted into this cellular compartment. In addition, the authors provide evidence that the translocation of the mRNAs carrying this domain is dependent on synaptic activ ity and can be induced by metabotropic glutamate receptor activation (Fig 1). Research into the subcellular sorting of mRNAs has mainly been approached from the perspective of the proteins that bind to mRNAs at each stage of their lifespan, beginning at transcription and continu ing through nuclear processing, export and subcellular transport. Components of these protein complexes are also involved in regulating the stability and translational control of mRNAs. Most of the proteins involved have common features, includ ing a conserved structural domain called the RNA recognition motif (RRM; Cléry
et al, 2008), which interacts with RNAs. Protein complexes that are responsible for the dendritic translocation of mRNAs such as CaMKIIα and Arc, which are relevant for synaptic plasticity, have been characterized to consist of more than 40 proteins. These include many RNA-binding proteins (Kanai et al, 2004), including members of the fragile-X mental retardation protein (FMRP) family—FMR1, FXR1 and FXR2—and other proteins involved in transport processes and the regulation of local translation. However, components such as heterogeneous nuclear ribonucleoprotein U (hnRNP U) and other members of the hnRNP family (reviewed in Han et al, 2010), are not unique to these complexes, but are found in several com plexes involved in mRNA processing. Thus, the question arises how proteins in many complexes provide specificity such that indi vidual mRNAs are transported to dendrites and localized to specific ribosomes within those dendrites, where they are translated. The mechanisms that achieve this specificity and the mechanisms that initiate translation
©2011 European Molecular Biology Organization
in response to synaptic activity are far from being understood. Subramanian et al (2011) have taken a different approach to identifying these mech anisms. Instead of looking at mRNA trans port proteins, as has been done previously, the authors based their investigation on pre vious studies that identified G‑quadruplexes as structural elements within RNAs that are required for binding to FMRP family proteins (Darnell et al, 2001; Schaeffer et al, 2001). They looked for the G‑quadruplex motif (Fig 1) in 34 well-characterized exam ples of dendritically localized mRNAs and found the consensus structure in approximately 30% of them, located in the 3'UTR. These G‑quadruplex-containing mRNAs included CaMKIIa and PSD95, as well as Shank1, dendrin, glycine receptor A1 subunit and several others. Thus, in com parison to the ZIP code domain—which spe cifically guides neuritic targeting (Kislauskis et al, 1994; Bassell et al, 1998) and is found only in the 3'UTR of the β-actin mRNA—the G‑quadruplex structural domain seems to EMBO reports VOL 12 | NO 7 | 2011 615
upfront be more common, although it is not found in other well-characterized mRNAs that are sorted into dendrites, including Arc (Arg3.1) mRNA (Wallace et al, 1998). Through a series of biochemical experiments, the authors provide detailed evidence that the G‑quadruplex domain is necessary for the neuritic translocation of CaMKIIa mRNA and PSD95 mRNA in cultured mouse embryonic cortical neurons. The question remains, how ever, of what distinguishes G‑quadruplex domain mRNAs from other mRNAs that do not carry the domain, but are nevertheless transported into neurites. It is possible that both G‑quadruplex and other mRNAs are recognized by the same binding proteins and are thus transported and processed by similar mechanisms. In fact, the protein complex responsible for the dendritic translocation of CaMKIIa mRNA (Kanai et al, 2004). includes FMRPs (FMR1, FXR1 and FXR2), hnRNP U, staufen and many others. However, this complex also binds to Arc mRNA, which does not contain a G‑quadruplex domain, as shown by the authors of this study. There are several possible explanations for this observation. It could be that the mRNAs for CaMKIIa and Arc bind to different proteins within the same complex, one through its G‑quadruplex domain and the other through a different motif. Alternatively, it might be that the two mRNAs bind to different complexes that are composed of similar sets of proteins. Indeed, the complex responsible for the dendritic transport of CaMKIIa mRNA was purified using antibodies against KIF5, a motor pro tein that is involved in the dendritic transport of mRNAs. It is possible that KIF5 is present in several mRNA transport complexes, both in those containing G‑quadruplex bind ing proteins and in those with mRNAs from other groups. Thus, the mRNAs for Arc and CaMKIIa might not be hooked onto the same messenger ribonucleoprotein (mRNP) complex. To solve this open question, it will be important to use approaches that start by analysing cis-recognition motifs in dendritically transported mRNAs. This will allow for the identification of the proteins and protein complexes that interact with
616 EMBO reports VOL 12 | NO 7 | 2011
h ot off th e press
individual mRNAs. The identification of mRNA postal codes—such as the ZIP-code domain in β-actin (Kislauskis et al, 1994) and the G‑quadruplex reported in the study by the Subramanian group—are an essential contribution to this progression. Subramanian et al (2011) actually show that the dendritic transport of CaMKIIa and PSD95 mRNAs is disturbed when the G‑quadruplex structure is missing, indicat ing that it is important to the sorting of these mRNAs. However, dendritic transport is not severely altered in the absence of FMRP, indicating that FMRP is not the only protein responsible for transport of these species into dendrites. It is possible that other mem bers of the FMRP family can be functionally substituted for FMRP, but this remains to be demonstrated. The authors also provide evi dence that G‑quadruplexes act as synapticactivity-dependent dendritic-localization elements for mRNAs. This implies that cor responding elements exist that act as sensors for synaptic activity. The authors suggest the hypothesis that changes in intracellular ion levels due to neural activity could induce structural changes in the G‑quadruplex, and that this could enhance the efficacy of den dritic translocation. However, alternative models are possible, and it remains to be determined whether the G‑quadruplex itself or other proteins within the complex act as sensors for synaptic activity. The study by Subramanian et al opens the door to addressing burning questions in the field. The work was performed using embry onic cortical neurons at a stage when mature synapses are not yet formed. The ablation of the G‑quadruplex domain in the CaMKIIa gene by homologous recombination in mice could reveal whether this structural element is also necessary in vivo for den dritic mRNA translocation and/or trans lational control in dendrites when fully differentiated synapses are activated in the adult brain. Such studies could also help to identify common functional features of G‑quadruplex-containing mRNAs, and what distinguishes them at a functional level from dendritic mRNAs that do not carry this structural element. Following the idea that
the mRNA initiates the assembly of its own transport particle—instead of a preformed protein complex taking charge of the mRNA somewhere between the nucleus and the dendrite—it would be interesting to see how the G‑quadruplex motif recruits proteins into transport complexes, and whether addi tional domains within the 3'UTR contribute to the assembly of mRNP granules. It could be that the composition of these complexes differs between cell types, or even between different types of neurons, depending on which mRNAs are present and the way in which they need to be sorted. The identification of the G‑quadruplex structural motif as a postal code for the sub cellular sorting of mRNAs should have a broad impact on the field. The race is now on to explain the way in which the trans location and translation of mRNA, controlled at the subcellular level, guides neuronal function and how disturbing these processes leads to diseases that range from fragile-X mental retardation (Bassell & Warren, 2008) and spinal muscular atrophy (Sendtner, 2010), to late onset neurodegenerative dis orders, such as frontotemporal dementia (Barmada & Finkbeiner, 2010). References Barmada SJ, Finkbeiner S (2010) Rev Neurosci 21: 251–272 Bassell GJ, Warren ST (2008) Neuron 60: 201–214 Bassell GJ et al (1998) J Neurosci 18: 251–265 Cléry A et al (2008) Curr Opin Struct Biol 18: 290–298 Darnell JC et al (2001) Cell 107: 489–499 Han SP et al (2010) Biochem J 430: 379–392 Kanai Y et al (2004) Neuron 43: 513–525 Kislauskis EH et al (1994) J Cell Biol 127: 441–451 Lécuyer E et al (2007) Cell 131: 174–187 Schaeffer C et al (2001) EMBO J 20: 4803–4813 Sendtner (2010) Nat Neurosci 13: 795–799 Subramanian et al (2011) EMBO Rep 12: 696–703 Wallace CS et al (1998) J Neurosci 18: 26–35
Carsten Drepper and Michael Sendtner are at the Institute for Clinical Neurobiology in Würzburg, Germany. E‑mail:
[email protected] Published online 17 June 2011
EMBO reports (2011) 12, 614–616. doi:10.1038/embor.2011.119
©2011 European Molecular Biology Organization
