Long-term potentiation is not one event. It is a phosphorylation relay

Strengthening a glutamatergic synapse takes seconds, and in those seconds a signal passes through at least four proteins, each of which changes state by gaining or losing a phosphate. None of those state changes alters how much protein is present. That is the central practical problem of synaptic plasticity research, and it is the reason a bench neuroscientist studying learning and memory owns more phospho-specific antibodies than total-protein ones.

The coincidence detector at the top of the chain

The NMDA receptor is the entry point, and it is unusual among ligand-gated channels because binding its agonist is not sufficient to open it. At resting membrane potential the pore is plugged by extracellular Mg²⁺. Two 1984 papers, working independently with patch-clamp and voltage-clamp recordings, showed that the receptor's strange voltage sensitivity is not an intrinsic gate at all but the voltage dependence of that Mg²⁺ block: remove Mg²⁺ from the bath and the current–voltage relationship straightens out ¹². The consequence is that the channel passes current only when glutamate is bound and the postsynaptic membrane is already depolarized. Presynaptic activity and postsynaptic activity have to coincide. That single biophysical fact is why the NMDA receptor sits at the top of nearly every model of associative plasticity.

What comes through the open pore is Ca²⁺, and Ca²⁺ is the message. Everything downstream is the cell's interpretation of it.

NR1, NR2A, NR2B: the subunit that decides what the signal means

Functional NMDA receptors are heteromers built around the obligatory NR1 subunit together with NR2 subunits. In situ hybridization across the developing rat CNS established the expression pattern that still frames the field: Grin1 transcript is present in virtually all neurons at all stages, while the NR2 transcripts diverge sharply. NR2B and NR2D appear prenatally; NR2A and NR2C are first detected near birth and then rise steeply, with most transcripts peaking around postnatal day 20. Recombinant NR1–NR2 channels built from different NR2 partners show comparable Ca²⁺ permeability but marked differences in voltage-dependent Mg²⁺ block and in offset decay time constants ³. The developmental NR2B-to-NR2A shift is therefore not cosmetic. A synapse that swaps NR2B for NR2A shortens its NMDA current and changes how much charge a given burst delivers.

The NR2B cytoplasmic tail does something NR2A's does not: it binds CaMKII. Autophosphorylated CaMKII binds with high affinity to a roughly 50-amino-acid domain in the NR2B tail, with little or no binding to the NR2A or NR1 tails ⁵. That interaction is not merely an anchor. Binding to NR2B generates sustained Ca²⁺/calmodulin-independent kinase activity by a mechanism independent of the kinase's own phosphorylation state, suppresses inhibitory autophosphorylation, and traps calmodulin ⁶. Knock-in mice carrying two point mutations that impair CaMKII binding to GluN2B show roughly 50% reduced LTP, impaired CaMKII phosphorylation of the AMPA receptor GluA1 subunit, and a recall deficit in the Morris water maze at 24 hours that is absent at one to two hours — a consolidation phenotype rather than an acquisition one ⁷. NR2B is thus both a channel subunit and a signaling platform, and the two roles are separable.

The NR2B tail is also a tyrosine-phosphorylation substrate. Screening the C-terminal cytoplasmic region, seven of 25 tyrosines were phosphorylated by Fyn in vitro; of those, Tyr-1252, Tyr-1336, and Tyr-1472 were phosphorylated in cells co-expressing active Fyn, with Tyr-1472 the major site. Tyr-1472 phosphorylation was greatly reduced in fyn mutant mice, rose over the developmental window in which hippocampal LTP appears, and increased after LTP induction in CA1 ⁴. The three sites have diverged since. Tyr-1472 has the deepest literature; Tyr-1336 is associated with a distinct function — Fyn-mediated phosphorylation at Tyr-1336, but not at Tyr-1472, controls calpain-mediated cleavage of the NR2B C-terminus in neurons, a process relevant to both plasticity and excitotoxicity ²⁴. Tyr-1252 is established as a Fyn site but is much less functionally characterized; a Tyr-1252 phospho-antibody is a discovery reagent, not a readout with an agreed meaning.

The molecular switch: CaMKII at Thr286

CaMKII is a dodecameric holoenzyme, and its behavior was the first clue that a kinase could store information. In 1986, Miller and Kennedy showed that Ca²⁺/calmodulin-stimulated autophosphorylation — incorporation of only 3 to 12 of a possible 30 phosphates per holoenzyme — renders kinase activity toward exogenous substrates independent of calcium, so that activity outlasts the calcium transient that started it ⁸. The site is Thr286. A knock-in point mutation that blocks Thr286 autophosphorylation without affecting calmodulin-dependent activity abolishes NMDA-receptor-dependent LTP in CA1 and abolishes spatial learning in the Morris water maze ⁹.

This is the cleanest available demonstration that a single phosphorylation event, not a change in protein abundance, carries the memory signal. Total CaMKII is one of the most abundant proteins in the postsynaptic density and barely moves. Phospho-Thr286 moves.

More recent work has made the switch model richer rather than simpler. Cross-regulation between Thr286 and the Thr305/306 sites within the twelve-subunit holoenzyme allows CaMKII to compute stimulus frequency, amplitude, and duration: Thr286 phosphorylation is required for both LTP and LTD, while Thr305/306 phosphorylation selectively promotes LTD and directs the kinase toward inhibitory rather than excitatory synapses ¹¹. CaMKII also acquires a structural role in the later phase of LTP, enlarging and strengthening the spine after it has finished its catalytic work ¹⁰.

The AMPA receptor is where the potentiation becomes electrical

CaMKII and PKA converge on the GluA1 (GluR1) subunit of the AMPA receptor, on two adjacent serines in its intracellular C-terminus. The sites were mapped in 1996: PKA specifically phosphorylates Ser-845 in transfected cells and in cultured neurons, and phosphorylation of that residue potentiates the peak current through GluR1 homomeric channels by about 40% ¹². Ser-831 is the other major site.

The two sites do different things, and this is the part worth being precise about.

Ser831 changes conductance. Coexpression of activated CaMKII with GluR1 does not alter glutamate affinity, desensitization kinetics, rectification, or gating. What it does is shift the receptor toward higher single-channel conductance states — an effect reproduced by mutating Ser-831 to aspartate ¹³. Potentiation here is the same number of channels carrying more current each.

Ser845 changes open probability and surface delivery. Including purified PKA catalytic subunit in the patch pipette raises the peak open probability of neuronal AMPA receptors to 0.92, versus 0.39 with the phosphatase calcineurin present; the S845A mutation eliminates the PKA effect while the S831A mutation does not ¹⁴. Separately, PKA phosphorylation of GluA1, together with CaMKII activity, is required for activity-driven synaptic incorporation of the receptor ¹⁶. Ser845 therefore reports both a gating change and a trafficking state.

Bidirectional plasticity uses these sites asymmetrically, and the asymmetry depends on synaptic history. LTD induction at naive synapses dephosphorylates the PKA site; at previously potentiated synapses it dephosphorylates the CaMKII site. LTP at naive synapses increases CaMKII-site phosphorylation, while LTP at depressed synapses increases PKA-site phosphorylation ¹⁵. Two identical stimulus protocols recruit different kinases depending on what happened at that synapse before. No total-protein measurement can see this. Mice carrying knock-in mutations at the GluR1 phosphorylation sites show deficits in both LTP and LTD and impaired retention in spatial learning tasks, which ties the biochemistry to behavior ¹⁷.

GluA2, Ser880, and the other direction

Depression has its own phospho-site. The GluA2 (GluR2) C-terminal sequence IESVKI ends in a PDZ ligand, and the serine within it — Ser880 — is a PKC substrate phosphorylated in vivo. Phosphorylation at Ser880 sharply decreases GluR2 binding to GRIP1 but leaves binding to PICK1 intact. In cultured hippocampal neurons, Ser880-phosphorylated GluR2 is enriched in dendrites and largely absent from excitatory synapses; PKC activation raises Ser880 phosphorylation, recruits PICK1 to synapses, and drives rapid internalization of surface GluR2 ¹⁸. Removing GluA2 from a synapse does more than weaken it. Because GluA2 is the subunit that makes an AMPA receptor calcium-impermeable, synapses that lose it gain calcium-permeable AMPA receptors, a plasticity state that has become central to addiction neuroscience: cocaine exposure increases calcium-permeable AMPA receptor levels in both the ventral tegmental area and the nucleus accumbens, with different time courses and different subunit dependencies in the two regions ¹⁹.

PSD-95 holds the result in place

None of this is stable without a scaffold. PSD-95 binds, through its second PDZ domain, the seven-residue C-terminal tSXV motif shared by NR2 subunits and certain NR1 splice forms, and it co-localizes with NR2B in cultured hippocampal neurons ²⁰. The broader PSD-95-like membrane-associated guanylate kinase family regulates basal synaptic AMPA receptor function and trafficking, acting both as an anchor for synaptic AMPA receptors and as a signaling scaffold that positions kinase complexes next to NMDA receptors ²¹. This is why PSD-95 is the default postsynaptic counterstain: in an immunofluorescence experiment it defines what counts as a synapse, so that a phospho-signal can be scored as synaptic or not rather than merely present.

Why the field buys these reagents

Learning and memory. The Thr286 and GluA1 knock-in mice are the field's cleanest causal links between a phosphate and a behavior ⁹¹⁷.

Alzheimer's disease. Amyloid-β promotes endocytosis of NMDA receptors in cortical neurons and depresses NMDA-evoked currents; dephosphorylation of NR2B at Tyr1472 tracks that receptor internalization ²⁵. The synaptic phenotype precedes and may drive the cognitive one.

Excitotoxicity and stroke. Synaptic and extrasynaptic NMDA receptor pools have opposing consequences — synaptic activity builds a neuroprotective transcriptional program, extrasynaptic activity promotes cell death — and the balance between them contributes to acute ischemic injury ²⁶.

Chronic pain. After L5 spinal nerve transection, NR2B phosphorylation at Tyr1472 appears in the superficial dorsal horn, is attenuated by an NR2B-selective antagonist, and is absent in mice lacking Fyn; electron microscopy localizes the phosphorylated receptor to the postsynaptic density ²³.

Fear learning. Y1472F knock-in mice show impaired fear-related learning, reduced amygdala LTP, impaired NMDA-receptor-mediated CaMKII signaling, and mislocalized NR2B at amygdala synapses ²².

Neurodevelopmental disorders. De novo variants in GRIN2B cluster in transmembrane segments and ligand-binding sites and are associated with intellectual disability, autism spectrum disorder, epilepsy, movement disorder, and cortical visual impairment ²⁷.

What these antibodies let you do, and what they demand

Phospho-specific antibodies of this class are used for immunoblot of synaptosomal and PSD fractions, immunofluorescence on cultured neurons or fixed sections, and immunohistochemistry on perfusion-fixed tissue. The discipline is unforgiving, and it is worth stating plainly.

Phosphatase inhibitors go in every buffer — homogenization, wash, and lysis — not just the first one; sodium fluoride and orthovanadate cover serine/threonine and tyrosine phosphatases respectively, and omitting either loses signal you will not know you lost. Tissue is snap-frozen or the animal is perfusion-fixed immediately; every minute of warm postmortem time is a minute of phosphatase activity, and phospho-epitopes are the first casualties. Run a lambda protein phosphatase control: a lysate aliquot treated with λ-PP should abolish the phospho band while leaving the total-protein band intact, and if it does not, the antibody is reading something other than the phosphate. Run a blocking-peptide control with the phosphopeptide immunogen and, where possible, the corresponding non-phosphopeptide; competition by the former but not the latter is the specificity argument. Finally, a phospho-signal is only interpretable normalized to total protein on the same samples. A doubling of phospho-GluA1 means nothing until you know GluA1 abundance did not double. Report the ratio, not the band.

Caveat — A change in phospho-signal is not a change in kinase activity. Phosphorylation state is the balance of kinase and phosphatase, and calcineurin, PP1, and STEP are all active at these synapses ¹⁴²⁵. Loss of phospho-Thr286 or phospho-Ser845 is equally consistent with kinase inhibition and with phosphatase activation, and distinguishing them requires a separate experiment. Equally, an unchanged phospho-signal does not mean nothing happened: a site that is dephosphorylated at naive synapses and phosphorylated at potentiated ones can give a flat population average across a mixed field ¹⁵.

Choosing and reporting the reagent

The International Working Group for Antibody Validation set out five conceptual pillars — genetic strategies, orthogonal strategies, independent antibody strategies, expression of tagged proteins, and immunocapture followed by mass spectrometry — to be applied in an application-specific manner ²⁸. For phospho-antibodies, the genetic pillar has an unusually clean form available: a phospho-null knock-in or point mutant should abolish the signal. Where such a line exists for the site you are studying, it is the strongest validation you can run. Record the catalog number, lot, host, dilution, and application in your methods section, and report the validation you performed rather than the one the datasheet claims.

Pel-Freez P-Series — Glutamatergic Synaptic Plasticity Panel

Catalog no.	Antibody	Host	Status
P43101	NMDA Receptor NR2B Subunit (10 µg)	Rabbit	In stock
P43301	Phospho-NR2B (Tyr1472)	Rabbit	In stock
P45101	Phospho-GluR1 (Ser845, PKA site)	Rabbit	In stock
P43201	NMDA Receptor NR2A Subunit (10 µg)	Rabbit	Currently out of stock — contact for availability
P40020	NMDA Receptor NR2A Subunit	Rabbit	Currently out of stock — contact for availability
P40021	NMDA Receptor NR2A Subunit, N-terminus	Rabbit	Currently out of stock — contact for availability
P40022	NMDA Receptor NR2B Subunit	Rabbit	Currently out of stock — contact for availability
P40023	NMDA Receptor NR2B Subunit, N-terminus	Rabbit	Currently out of stock — contact for availability
P80001	NMDA Receptor NR1 Subunit	Mouse	Currently out of stock — contact for availability
P40024	Phospho-NR2B (Tyr1252)	Rabbit	Currently out of stock — contact for availability
P40025	Phospho-NR2B (Tyr1336)	Rabbit	Currently out of stock — contact for availability
P40001	Phospho-CaMKII (Thr286)	Rabbit	Currently out of stock — contact for availability
P80000	GluR1 (AMPA receptor)	Mouse	Currently out of stock — contact for availability
P40013	Phospho-GluR1 (Ser831, CaMKII site)	Rabbit	Currently out of stock — contact for availability
P40014	GluR2 (AMPA receptor)	Rabbit	Currently out of stock — contact for availability
P40015	GluR2/3 (AMPA receptor)	Rabbit	Currently out of stock — contact for availability
P40016	Phospho-GluR2 (Ser880, PKC site)	Rabbit	Currently out of stock — contact for availability
P40026	PSD-95 (postsynaptic scaffold)	Rabbit	Currently out of stock — contact for availability

References

1. Nowak L, Bregestovski P, Ascher P, et al. Magnesium gates glutamate-activated channels in mouse central neurones. Nature. 1984;307:462–465. PMID 6320006. doi:10.1038/307462a0
2. Mayer ML, Westbrook GL, Guthrie PB Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones. Nature. 1984;309:261–263. PMID 6325946. doi:10.1038/309261a0
3. Monyer H, Burnashev N, Laurie DJ, et al. Developmental and regional expression in the rat brain and functional properties of four NMDA receptors. Neuron. 1994;12:529–540. PMID 7512349. doi:10.1016/0896-6273(94)90210-0
4. Nakazawa T, Komai S, Tezuka T, et al. Characterization of Fyn-mediated tyrosine phosphorylation sites on GluR epsilon 2 (NR2B) subunit of the N-methyl-D-aspartate receptor. J Biol Chem. 2001;276:693–699. PMID 11024032. doi:10.1074/jbc.M008085200
5. Strack S, Colbran RJ Autophosphorylation-dependent targeting of calcium/calmodulin-dependent protein kinase II by the NR2B subunit of the N-methyl-D-aspartate receptor. J Biol Chem. 1998;273:20689–20692. PMID 9694809. doi:10.1074/jbc.273.33.20689
6. Bayer KU, De Koninck P, Leonard AS, et al. Interaction with the NMDA receptor locks CaMKII in an active conformation. Nature. 2001;411:801–805. PMID 11459059. doi:10.1038/35081080
7. Halt AR, Dallapiazza RF, Zhou Y, et al. CaMKII binding to GluN2B is critical during memory consolidation. EMBO J. 2012;31:1203–1216. PMID 22234183. doi:10.1038/emboj.2011.482
8. Miller SG, Kennedy MB Regulation of brain type II Ca2+/calmodulin-dependent protein kinase by autophosphorylation: a Ca2+-triggered molecular switch. Cell. 1986;44:861–870. PMID 3006921. doi:10.1016/0092-8674(86)90008-5
9. Giese KP, Fedorov NB, Filipkowski RK, Silva AJ Autophosphorylation at Thr286 of the alpha calcium-calmodulin kinase II in LTP and learning. Science. 1998;279:870–873. PMID 9452388. doi:10.1126/science.279.5352.870
10. Lisman J, Yasuda R, Raghavachari S Mechanisms of CaMKII action in long-term potentiation. Nat Rev Neurosci. 2012;13:169–182. PMID 22334212. doi:10.1038/nrn3192
11. Cook SG, Buonarati OR, Coultrap SJ, Bayer KU CaMKII holoenzyme mechanisms that govern the LTP versus LTD decision. Sci Adv. 2021;7(16):eabe2300. PMID 33853773. doi:10.1126/sciadv.abe2300
12. Roche KW, O'Brien RJ, Mammen AL, et al. Characterization of multiple phosphorylation sites on the AMPA receptor GluR1 subunit. Neuron. 1996;16:1179–1188. PMID 8663994. doi:10.1016/s0896-6273(00)80144-0
13. Derkach V, Barria A, Soderling TR Ca2+/calmodulin-kinase II enhances channel conductance of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate type glutamate receptors. Proc Natl Acad Sci U S A. 1999;96:3269–3274. PMID 10077673. doi:10.1073/pnas.96.6.3269
14. Banke TG, Bowie D, Lee H, et al. Control of GluR1 AMPA receptor function by cAMP-dependent protein kinase. J Neurosci. 2000;20:89–102. PMID 10627585. doi:10.1523/JNEUROSCI.20-01-00089.2000
15. Lee HK, Barbarosie M, Kameyama K, et al. Regulation of distinct AMPA receptor phosphorylation sites during bidirectional synaptic plasticity. Nature. 2000;405:955–959. PMID 10879537. doi:10.1038/35016089
16. Esteban JA, Shi SH, Wilson C, et al. PKA phosphorylation of AMPA receptor subunits controls synaptic trafficking underlying plasticity. Nat Neurosci. 2003;6:136–143. PMID 12536214. doi:10.1038/nn997
17. Lee HK, Takamiya K, Han JS, et al. Phosphorylation of the AMPA receptor GluR1 subunit is required for synaptic plasticity and retention of spatial memory. Cell. 2003;112:631–643. PMID 12628184. doi:10.1016/s0092-8674(03)00122-3
18. Chung HJ, Xia J, Scannevin RH, et al. Phosphorylation of the AMPA receptor subunit GluR2 differentially regulates its interaction with PDZ domain-containing proteins. J Neurosci. 2000;20:7258–7267. PMID 11007883. doi:10.1523/JNEUROSCI.20-19-07258.2000
19. Wolf ME, Tseng KY Calcium-permeable AMPA receptors in the VTA and nucleus accumbens after cocaine exposure: when, how, and why? Front Mol Neurosci. 2012;5:72. PMID 22754497. doi:10.3389/fnmol.2012.00072
20. Kornau HC, Schenker LT, Kennedy MB, Seeburg PH Domain interaction between NMDA receptor subunits and the postsynaptic density protein PSD-95. Science. 1995;269:1737–1740. PMID 7569905. doi:10.1126/science.7569905
21. Xu W PSD-95-like membrane associated guanylate kinases (PSD-MAGUKs) and synaptic plasticity. Curr Opin Neurobiol. 2011;21:306–312. PMID 21450454. doi:10.1016/j.conb.2011.03.001
22. Nakazawa T, Komai S, Watabe AM, et al. NR2B tyrosine phosphorylation modulates fear learning as well as amygdaloid synaptic plasticity. EMBO J. 2006;25:2867–2877. PMID 16710293. doi:10.1038/sj.emboj.7601156
23. Abe T, Matsumura S, Katano T, et al. Fyn kinase-mediated phosphorylation of NMDA receptor NR2B subunit at Tyr1472 is essential for maintenance of neuropathic pain. Eur J Neurosci. 2005;22:1445–1454. PMID 16190898. doi:10.1111/j.1460-9568.2005.04340.x
24. Wu HY, Hsu FC, Gleichman AJ, et al. Fyn-mediated phosphorylation of NR2B Tyr-1336 controls calpain-mediated NR2B cleavage in neurons and heterologous systems. J Biol Chem. 2007;282:20075–20087. PMID 17526495. doi:10.1074/jbc.M700624200
25. Snyder EM, Nong Y, Almeida CG, et al. Regulation of NMDA receptor trafficking by amyloid-beta. Nat Neurosci. 2005;8:1051–1058. PMID 16025111. doi:10.1038/nn1503
26. Hardingham GE, Bading H Synaptic versus extrasynaptic NMDA receptor signalling: implications for neurodegenerative disorders. Nat Rev Neurosci. 2010;11:682–696. PMID 20842175. doi:10.1038/nrn2911
27. Platzer K, Yuan H, Schütz H, et al. GRIN2B encephalopathy: novel findings on phenotype, variant clustering, functional consequences and treatment aspects. J Med Genet. 2017;54:460–470. PMID 28377535. doi:10.1136/jmedgenet-2016-104509
28. Uhlen M, Bandrowski A, Carr S, et al. A proposal for validation of antibodies. Nat Methods. 2016;13:823–827. PMID 27595404. doi:10.1038/nmeth.3995

Sourcing note. Primary literature located through PubMed; every quantitative figure was taken from the source article’s own record. Findings are reported at the grain of the study that produced them and those grains are not interchangeable. Product specifications are from the current Pel-Freez datasheets.