Subtle changes in synaptic transmission are likely responsible for learning and memory, but the molecular cast of characters at play in this complex process is only partly known. Two papers in this week's Nature Neuroscience online introduce two new players, and quite distinct ones at that. As it turns out, the transcription factor NF-κB and the neuronal calcium sensor-1 both can have profound effects on synaptic plasticity by modulating neuronal signaling.

NF-κB is famous for its involvement in immune responses, yet even in the early '90s, Kaltschmidt et al. spotted it in synaptosomes in a complex with its inhibitory subunit IκB. This finding raised the intriguing possibility that activation of this transcription factor at nerve endings leads to its delivery to the nucleus, where it could modify gene transcription. Now, principal investigator David Baltimore at the California Institute of Technology, Pasadena, with colleagues there and at the University of California at Los Angeles, shows that this is indeed what happens.

First author Mollie Meffert and coworkers followed the fate of activated NF-κB by coupling its p65 subunit to green fluorescent protein (GFP). Meffert expressed this chimera in cultured hippocampal neurons and used its fluorescence as a tracer. To see if the protein is retrogradely transported to the nucleus from dendrites, the authors bleached the fluorescence in a small section of a dendrite and then measured its recovery. She found that the green chimera diffused into the bleached section only from the distal, outer end of the dendrite. Importantly, this diffusion was rapidly accelerated by stimulating the neurons with glutamate or N-methyl-D-aspartate, indicating that neuronal signaling drives the transport of NF-κB toward the nucleus.

The authors then tested the physiological relevance of NF-κB in vivo by looking at the behavior of p65-deficient mice. In a radial arm maze, where animals are trained to find a food treat at the end of every arm, mice lacking the NF-κB subunit were more likely to waste time revisiting previously harvested arms. After two days of trials, the NF-κB-negative mice visited food-containing arms about 20 percent of the time, vs. 40 percent for wild-type animals. After eight days, the difference was no longer significant, indicating that the deficient mice do learn, albeit more slowly.

The researchers further report that NF-κB can be activated in neurons by Ca2+, and that this requires the activity of the calcium-dependent kinase CaMKII. In the second Nature Neuroscience paper, principal author Felix Schweizer, also at the University of California at Los Angeles, together with colleagues at UCLA and at Baylor College of Medicine, Houston, Texas, show that another calcium-binding protein, neuronal calcium sensor-1 (NCS-1), plays an important role in short-term synaptic plasticity.

Short-term synaptic plasticity is often measured as the change of postsynaptic transmission that follows repeated stimulation. This can be either an increase (facilitation) or a decrease (depression) in the strength of the postsynaptic current. By stimulating one of a pair of connected hippocampal neurons and measuring the current elicited in the other, first author Tanya Sippy and colleagues tested the effect of NCS-1 on synaptic plasticity. Sippy found that in normal cells, a second stimulus results in depression—the amplitude of the induced second current being about half that of the first. But when the authors transfected cells with low amounts of NCS-1, the plasticity reversed—now, the second stimulus resulted in facilitation, inducing almost twice as much current in the postsynaptic neuron as the first stimulus. Importantly, Sippy found that NCS-1 infected cells had the same basal level of synaptic transmission as wild-type cells.

These findings address one of the most debated questions about short-term plasticity, namely whether it is a function of preexisting synaptic strength. Sippy's findings suggest that it is not, as NCS-1 does not alter basal levels. But what exactly is the role of this calcium binder? Most models conceived to explain facilitation require that calcium released from previous stimulations remains in the presynaptic terminal, so that upon subsequent stimulation it will bolster the release of neurotransmitters at the synapse. But as NCS-1 binds calcium, it is not likely to contribute to the buildup of the free cation. More recently, Blatow et al. have suggested that buffers in the presynaptic terminal may be responsible for facilitation. Although such chelators could mop up Ca2+, Blatow suggests that they are easily saturated; thus, subsequent stimuli evoke more synaptic activity because the release of calcium overwhelms the buffering capacity.

Could NCS-1 be such a buffer? Probably not, suggest Sippy and colleagues, because then it would be expected to affect basal transmission, but the authors found it does not. Instead, they suggest that NCS-1 functions as a "release sensor," which, in cooperation with other unidentified molecules, can raise presynaptic calcium levels.

Perhaps most significant is the suggestion that neurons can alter their plasticity quite dramatically by merely modifying the expression of NCS-1. In this regard, it is worth noting that the protein has been shown to be upregulated in the prefrontal cortex of schizophrenic and bipolar patients.—Tom Fagan


  1. Meffert et al. have provided a fascinating report, but the data are not compelling. The most intriguing finding, that RelA-GFP diffuses in dendrites in a distal-->proximal direction, is represented by photomicrographs (and quantifications thereof) that show only the distal border of the photobleached area. More importantly, techniques applied to whole-culture lysates or extracts were performed with cultures that almost certainly contained large numbers of glia. We have documented in three publications that cultures of nearly pure cortical neurons do NOT show an induction of NF-κB DNA-binding activity in response to glutamate;(1-3) all the glutamate-evoked NF-κB activity detected by gel shifts can be attributed to glial contamination of cultures. The recalcitrant tendencies of neuronal NF-κB has since been confirmed by others (e.g., ref. 4). In addition to the gel shifts, luciferase reporter gene assays obviously would be confounded by glial contamination, as well.


    . Characterization of a neuronal kappaB-binding factor distinct from NF-kappaB. Brain Res Mol Brain Res. 1999 Apr 20;67(2):303-15. PubMed.

    . Inhibition of the activity of a neuronal kappaB-binding factor by glutamate. J Neurochem. 1999 Nov;73(5):1851-8. PubMed.

    . Neuronal kappa B-binding factors consist of Sp1-related proteins. Functional implications for autoregulation of N-methyl-D-aspartate receptor-1 expression. J Biol Chem. 2002 Nov 22;277(47):44911-9. PubMed.

    . Specific deficiency in nuclear factor-kappaB activation in neurons of the central nervous system. Lab Invest. 2001 Sep;81(9):1275-88. PubMed.

  2. Reply to comment by Steve Barger:

    Activation of NF-κB by basal synaptic activity (as well as glutamate, depolarization, and bicuculline) was observed in high-density neuronal cultures maintained with defined media in the absence of serum.(1) Under these conditions, glial contribution to the culture is minimal; staining for markers of glia and neurons (GFAP and neurofilament, respectively) shows the cultures to be roughly 90-95 percent hippocampal neurons. In addition, NF-κB activation was observed in physiologically active preparations of synaptosomes, the isolated synaptic subcompartment of neurons (Fig.1 c-f). Under our stimulation conditions, activation of glial NF-κB by glutamate was not observed when glia were cultured separately (EMSA Fig.1b) or when they were co-cultured with neurons and examined by microscopy (GFPp65 translocation, Sup.Fig.1). Multiple differences exist between our experimental systems and those of Barger(2), including, but not limited to, duration and magnitude of stimulation, neuronal cell type, age of culture, and method of extract preparation. The activation of neuronal NF-κB by glutamate or glutamate analogs has also been observed by other researchers in a variety of systems (see, for example, refs 3-6).

    Regarding our photobleaching experiments: Examining movement of the bleached border most proximal to the cell body initially sounds attractive. However, the difficulty with this approach is that fluorescence feeds into this bleached edge from unbleached dendrites branching off in the vicinity, making the edge indistinct and precluding quantification of border movement. By instead choosing to examine FRAP, we have been able to quantify GFPp65 movement (Fig.2a-d) and have used separate bleaching experiments to monitor nuclear accumulation of the GFPp65 from distal processes (Fig.2e,f).


    1. Meffert MK, Chang JM, Wiltgen BJ, Fanselow MS, Baltimore D. NF-kappaB functions in synaptic signaling and behavior. Nat Neurosci. 2003 Oct;6(10):1072-8. Epub 2003 Aug 31. Abstract

    2. Mao X, Moerman AM, Barger SW. Neuronal kappa B-binding factors consist of Sp1-related proteins. Functional implications for autoregulation of N-methyl-D-aspartate receptor-1 expression. J Biol Chem. 2002 Nov 22;277(47):44911-9. Epub 2002 Sep 18. Abstract

    3. Kaltschmidt C, Kaltschmidt B, Baeuerle PA. Stimulation of ionotropic glutamate receptors activates transcription factor NF-kappa B in primary neurons. Proc Natl Acad Sci U S A. 1995 Oct 10;92(21):9618-22. Abstract

    4. Cruise L, Ho LK, Veitch K, Fuller G, Morris BJ. Kainate receptors activate NF-kappaB via MAP kinase in striatal neurones. Neuroreport. 2000 Feb 7;11(2):395-8. Abstract

    5. Lipsky RH, Xu K, Zhu D, Kelly C, Terhakopian A, Novelli A, Marini AM. Nuclear factor kappaB is a critical determinant in N-methyl-D-aspartate receptor-mediated neuroprotection. J Neurochem. 2001 Jul;78(2):254-64. Abstract

    6. Pizzi M, Goffi F, Boroni F, Benarese M, Perkins SE, Liou HC, Spano P. Opposing roles for NF-kappa B/Rel factors p65 and c-Rel in the modulation of neuron survival elicited by glutamate and interleukin-1beta. J Biol Chem. 2002 Jun 7;277(23):20717-23. Epub 2002 Mar 23. Abstract

  3. I am not certain if this is the proper venue for this discussion or how Dr. Baltimore feels about continuing this dialog. But, I appreciate this opportunity to tell our story—a tale for which it has been difficult to find a receptive audience. I would like to emphasize that my goal here is not to be confrontational but to uncover an explanation for the discrepancies; therefore, I simply want to explain our findings and how they contrast with the literature. I would also like to clarify that I take issue only with the whole-cell EMSAs of NF-κB in glutamate-treated neurons; activation of NF-κB in synaptosomes is not something we have analyzed. Still, synaptic activation of NF-κB would have to be considered irrelevant to nuclear transcription if one cannot demonstrate that its DNA-binding activity reaches the nucleus.

    Let me simply reemphasize the fact that we do not see glutamate activation of NF-κB DNA binding in nuclear extracts made from nearly pure cultures of cerebral (i.e., neocortical or hippocampal) neurons. Furthermore, in our hands, glutamate does not activate a NF-κB-responsive reporter gene transfected into cerebral neurons. Failing to scour all our papers is forgivable for someone as busy as the president of a major university must be. Nevertheless, I feel compelled to respond that we did indeed perform experiments with glutamate applications (i.e., duration and magnitude) nearly identical to those used by Meffert et al. (One obvious difference was our omission of the cocktail of tetrodotoxin and glutamate receptor antagonists.) These data were reported in papers other than the one Dr. Baltimore cited. As regards the neuronal cell type, we have extensively tested hippocampal and neocortical neurons from rat and mouse. As for the age of the neurons, it is true that we established our cultures from fetuses (E18) rather than the neonates used by Meffert et al. We then allowed the neurons to mature in culture for approximately a week and a half before using them in experiments. By using fetuses, we can more efficiently limit astrocyte numbers by killing their progenitors while the numbers are still low (as the largest wave of astrogliogenesis occurs after E18). This approach also makes it possible to include the mitotic inhibitor only briefly, so that its potential inhibition of NF-κB (1) can be removed several days before the experiments. At the time of birth, astrocyte numbers are nearly as high as those of neurons, and both cell types survive quite well in the serum-free medium used by Meffert et al. Indeed, serum requirements for astrocyte proliferation in culture are generally overemphasized; clearly, astrocytes proliferate just fine in the absence of serum in vivo! Dr. Baltimore mentions GFAP staining to determine the glial contamination, a measure that would overlook non-astrocytic glia. Greg Brewer, a pioneer in culturing postnatal cerebrocortical cells, finds that adult rat brains maintained under conditions that appear similar to those used by Meffert et al. produce cultures that are five percent GFAP-positive yet still only 80 percent neurons (2). While I would be surprised if half of the glia died during the culture period, as Dr. Baltimore’s numbers suggest, I would not be surprised if all the NF-κB detected by EMSA came from the remaining glia, which he admits may have been as high as 10 percent. By comparison, our methods produce cultures that are less than one percent glia; omitting mitotic poisons results in nearly 30 percent astrocytes, even in serum-free medium.

    With respect to the extraction protocol, it should be noted that our extraction and assay conditions are perfectly capable of detecting NF-κB in other cell types or in mixed neuron/glia co-cultures. It is impossible to determine what extraction procedure was used by Meffert et al.; at the time of this writing, the online supplemental material to which the reader is referred contains no information about extraction procedures or EMSA conditions.

    Dr. Baltimore has mentioned the evidence for glutamate-evoked activation of neuronal NF-κB that was published by Kaltschmidt and others. In fact, the approaches used in those reports are dramatically different from those used by either Meffert et al. or ourselves. Namely, the other investigators either relied on immunocytochemically detected translocation of NF-κB proteins to the neuronal nucleus (Kaltschmidt, Cruise, Pizzi), a phenomenon that is not tantamount to DNA binding, or they used cerebellar granule cells (Lipsky, Pizzi), which do not show cerebral responses to excitatory amino acids.

    So, it would seem that the controversy persists. In my opinion, it is difficult to ascribe biological significance to an effect that would be so finicky as to depend on tetrodotoxin or small differences in age or extraction procedures. But, of course, there is still a caveat to our data, as well. By removing the glia, we may have changed the intrinsic phenotype of the neurons so that they no longer respond to glutamate in the same way. However, we can at least be confident that such an effect of the glia cannot be attributed to diffusible paracrine factors, as we have tested co-cultures in which the glia are separated from neuronal contact only by a permeable membrane (3).

    Regarding the movement of RelA-GFP into photobleached dendrites, Dr. Baltimore replies that “fluorescence feeds into this [proximal] bleached edge from unbleached dendrites branching off in the vicinity.” I appreciate this correction of the intial Alzforum News report stating that the authors “found that the green chimera diffused into the bleached section only from the distal, outer end of the dendrite.”


    . Mechanism of cytosine arabinoside-mediated apoptosis: role of Rel A (p65) dephosphorylation. Oncogene. 2003 Jul 10;22(28):4356-69. PubMed.

    . Age-related changes in neuronal glucose uptake in response to glutamate and beta-amyloid. J Neurosci Res. 2003 May 15;72(4):527-36. PubMed.

    . Characterization of a neuronal kappaB-binding factor distinct from NF-kappaB. Brain Res Mol Brain Res. 1999 Apr 20;67(2):303-15. PubMed.

Make a Comment

To make a comment you must login or register.


Paper Citations

  1. . Brain synapses contain inducible forms of the transcription factor NF-kappa B. Mech Dev. 1993 Oct;43(2-3):135-47. PubMed.
  2. . Ca2+ buffer saturation underlies paired pulse facilitation in calbindin-D28k-containing terminals. Neuron. 2003 Apr 10;38(1):79-88. PubMed.

Further Reading


  1. . Beta-amyloid binds to p57NTR and activates NFkappaB in human neuroblastoma cells. J Neurosci Res. 1998 Dec 15;54(6):798-804. PubMed.
  2. . Human neuronal calcium sensor-1 shows the highest expression level in cerebral cortex. Neurosci Lett. 2002 Feb 15;319(2):67-70. PubMed.

Primary Papers

  1. . NF-kappa B functions in synaptic signaling and behavior. Nat Neurosci. 2003 Oct;6(10):1072-8. PubMed.
  2. . Acute changes in short-term plasticity at synapses with elevated levels of neuronal calcium sensor-1. Nat Neurosci. 2003 Oct;6(10):1031-8. PubMed.