Alternative splicing of messenger RNAs can result in the production of protein isoforms with different functions, but why would a cell need alternative mRNAs that encode the same protein? The mRNA for brain-derived neurotrophic factor (BDNF), for example, exists in brain with two alternative 3’ untranslated regions, a short form and a long form. In today’s issue of Cell, Baoji Xu and colleagues from Georgetown University in Washington, DC, provide an explanation: the alternative mRNAs end up in different cellular locations, with the short form sent to the soma and the long form to dendrites. Dendritic BDNF, they show, has definite subcellular functions, controlling dendritic morphology and synaptic plasticity, both of which are affected in Alzheimer disease (AD).

Splicing that generates long and short protein isoforms is also implicated in AD by work that appears in the July 2 Journal of Biological Chemistry online. Sic Chan and colleagues at the University of Central Florida in Orlando report how different isoforms of the endocytic adaptor protein Numb have distinct effects on the trafficking and processing of the amyloid precursor protein (APP) and generation of amyloid-β in neurons. Importantly, the researchers suggest that the splicing of Numb is affected by cellular stress to result in the promotion of Aβ production. Together, the two studies demonstrate the role of splicing in regulating protein function in a way that is complex, barely understood, and potentially important for disease.

In the BDNF work, first authors Juan Ji An, Kusumika Gharami, and Guey-Ying Liao led the effort to determine whether the 3’ alternative splicing of BDNF, which has no effect on the protein coding region, was instead a way to target the message to soma versus dendrite. To do this, they first looked at localization of the splice forms and found the long form was preferentially located in dendrites in cultured rat cortical neurons. The long 3’ sequence was sufficient to target a green fluorescent protein reporter mRNA to dendrites, and that the message was translated there.

To follow up these findings in vivo, the researchers used a previously generated BDNF knock-in mouse that produces only a truncated 3’UTR message, and no long form. These mice have normal levels of total BDNF mRNA and protein, but the investigators found that cortical neurons cultured from the mice lack dendritic BDNF, and have less activity-stimulated BDNF release.

With no change in overall BDNF levels, the mutant mice offered a unique opportunity to explore the function of dendritic BDNF. When the researchers examined CA1 pyramidal neurons from young mice, they found the morphology of dendritic arbors looked normal, but the spines on distal dendrites were thinner and more numerous than in wild-type mice. Initial growth of spines was not affected in younger mice, but the lack of BDNF seemed to impair subsequent pruning and enlargement. The morphological changes had a functional counterpart, as the mutant mice displayed impairment of long-term potentiation at dendritic synapses, but not at somatic synapses.

While the study strongly implicates dendritically targeted BDNF in the normal formation and function of spines, the data beg the question of how BDNF acts. In their discussion, the authors favor an autocrine mechanism involving activation of the TrkB receptor. In addition, they point out that the strategy of using alternative 3’UTRs to target the same protein to different subcellular localization may not be unique to BDNF. The authors cite the calmodulin-stimulated kinase subunit CamKIIa as another mRNA that has the potential to be targeted this way (Blichenberg et al., 2001).

The second paper presents a variation on the splicing theme, where a switch in protein isoforms has the potential to affect APP processing. The protein in question is Numb, a putative endocytic adaptor protein that regulates processing of the γ-secretase substrate Notch. All four known human isoforms of Numb have been shown to associate with APP (Roncarati et al., 2002). In the new report, first author George Kyriazis and colleagues demonstrate that overexpression of a Numb isoform with a shorter PTB protein-protein interaction domain, but not one with a longer domain, leads to enhanced levels of APP protein in PC12 cells, and increased processing as indicated by accumulation of C-terminal fragments and Aβ. The accumulation of APP was not due to changes in APP mRNA or in activity of any of the secretases, but instead seemed to result from the isoform-specific targeting of APP to endosomes after its internalization from the plasma membrane. The alternative, longer isoform promoted APP delivery to lysosomes, and cells overexpressing the longer form had lower overall levels of APP and its products.

To support the physiological relevance of the Numb interaction, the researchers cultured primary cortical neurons under conditions of trophic factor withdrawal, and found that the expression of the short Numb isoform increased while the long form decreased, coincident with an increase in APP levels, and Aβ production. “Taken together, these results indicate that APP trafficking differs strikingly in the clones stably expressing the Numb proteins and raise the intriguing possibility that alternative splicing of Numb could alter the trafficking of APP and, concomitantly, its processing fate,” the authors conclude. In a previous paper, Chan and co-author Mark Mattson had shown that total Numb protein is increased in Aβ-containing regions in AD mouse brain (Chan et al., 2002). Chan told ARF in an e-mail that they are in the process of identifying those isoforms.—Pat McCaffrey


  1. The study by An and colleagues provides one of the first clues to an ongoing riddle. Why the already very complex pattern of multiple BDNF transcripts is further complicated by attaching the diverse mRNAs to two distinct polyadenylation signals. Making use of very elegant mouse models that were provided by colleagues in the field, the authors provide convincing evidence that mRNA versions with the longer 3'UTR are transported more efficiently into dendrites than those with the short 3'UTR tail.

    Given that BDNF is an important mediator of activity-dependent synaptic plasticity, elucidating these subtle details of dendritic BDNF production and action are crucial to better understand the role of this growth factor in dementias such as Alzheimer disease. Several interesting questions arise from their important findings:

    1. Data are required to show that regulated release of BDNF is affected selectively in dendrites and not in the soma, as the authors suggest.

    2. The lack of dendritic BDNF protein surprisingly has a gain-of-function phenotype (increased number of dendritic filopodia), raising the question whether dendritic TrkB receptors can exert an influence on the growth of filopodia in the absence of ligand (compare e.g., Hartmann et al., 2004). Furthermore, it would be exciting to see what would happen to synaptic function after acute, selective knockdown of dendritic BDNF mRNA using, for example, RNA interference.

    3. The LTP phenotype in their BDNF (klox/klox) mice is striking. However, how can early, protein synthesis independent LTP (which is known to depend on the availability of BDNF; see e.g., Korte et al., 1995; Patterson et al., 1996) be affected in the BDNF (klox/klox) mice, if translation of dendritic BDNF mRNA to functional protein is not required for this type of plasticity? This raises the interesting question, whether the observed BDNF effect on LTP in the study by An et al. depends on reduced basal extracellular levels of BDNF at the dendrites (being in favor of a permissive effect of BDNF for LTP), or rather on reduced availability of BDNF vesicles at post-synaptic sites for activity-dependent release (stressing an instructive effect of BDNF for LTP) after tetanic stimulation (compare Hartmann et al., 2001).

    4. Last, but not least, it seems possible that dendritic targeting of specific mRNAs could constitute the synaptic tag that has been postulated to steer synapse-specific modulation of LTP (compare Reymann and Frey, 2007).

    Future studies that can now take advantage of these new clues reported by An et al. will hopefully help to further elucidate the effects of dendritic BDNF protein on input-specific synaptic plasticity.


    . Synaptic secretion of BDNF after high-frequency stimulation of glutamatergic synapses. EMBO J. 2001 Nov 1;20(21):5887-97. PubMed.

    . Truncated TrkB receptor-induced outgrowth of dendritic filopodia involves the p75 neurotrophin receptor. J Cell Sci. 2004 Nov 15;117(Pt 24):5803-14. PubMed.

    . The late maintenance of hippocampal LTP: requirements, phases, 'synaptic tagging', 'late-associativity' and implications. Neuropharmacology. 2007 Jan;52(1):24-40. PubMed.

    . Recombinant BDNF rescues deficits in basal synaptic transmission and hippocampal LTP in BDNF knockout mice. Neuron. 1996 Jun;16(6):1137-45. PubMed.

    . Hippocampal long-term potentiation is impaired in mice lacking brain-derived neurotrophic factor. Proc Natl Acad Sci U S A. 1995 Sep 12;92(19):8856-60. PubMed.

    View all comments by Volkmar Lessmann

Make a Comment

To make a comment you must login or register.


Paper Citations

  1. . Identification of a cis-acting dendritic targeting element in the mRNA encoding the alpha subunit of Ca2+/calmodulin-dependent protein kinase II. Eur J Neurosci. 2001 May;13(10):1881-8. PubMed.
  2. . The gamma-secretase-generated intracellular domain of beta-amyloid precursor protein binds Numb and inhibits Notch signaling. Proc Natl Acad Sci U S A. 2002 May 14;99(10):7102-7. PubMed.
  3. . Numb modifies neuronal vulnerability to amyloid beta-peptide in an isoform-specific manner by a mechanism involving altered calcium homeostasis: implications for neuronal death in Alzheimer's disease. Neuromolecular Med. 2002;1(1):55-67. PubMed.

Further Reading

Primary Papers

  1. . Distinct role of long 3' UTR BDNF mRNA in spine morphology and synaptic plasticity in hippocampal neurons. Cell. 2008 Jul 11;134(1):175-87. PubMed.
  2. . Numb endocytic adapter proteins regulate the transport and processing of the amyloid precursor protein in an isoform-dependent manner: implications for Alzheimer disease pathogenesis. J Biol Chem. 2008 Sep 12;283(37):25492-502. PubMed.