As a first step to investigate the molecular composition of GluK2/GluK5 complexes, we performed single molecule subunit counting experiments (Ulbrich and Isacoff, 2007
) on homomeric GluK2 receptors. Upon fusion of mEGFP to the C-terminus of GluK2 (GluK2-mEGFP) and low density expression in Xenopus
oocytes we observed sparse, well-resolved and stationary spots of green fluorescence on the cell surface using TIRF microscopy (). The photobleaching of a single GFP is a discrete process; thus the fluorescence intensity of a protein complex with one or several GFP molecules drops in a stepwise fashion, and the number of steps reflects the number of GFP-tagged subunits in the complex. Fluorescence intensity trajectories (e.g. ) show that the majority of the GluK2-mEGFP spots bleached in three or four steps, with smaller numbers at one and two steps (, black bars). This distribution of one, two, three and four bleaching steps originates from the fact that not all subunits contain a fluorescent mEGFP. The distribution observed for GluK2-mEGFP agrees well with the binominal distribution expected for a tetramer, based on a probability of p
=0.80 for an individual mEGFP to be fluorescent (, red dashes). Similar values have been obtained on a variety of other membrane proteins (Tombola et al., 2008
; Ulbrich and Isacoff, 2007
; Yu et al., 2009
), and this value was used to predict distributions throughout this study. The bright fluorescence and immobility of the spots, along with the close agreement between the experimental results and the theoretical prediction, demonstrate that the method enables the investigation of the subunit composition of GluK2 containing receptors at the surface of living cells with high accuracy, and confirms that GluK2 forms homotetrameric receptors.
Next, we asked whether GluK5 alone can be detected at the cell surface. In contrast to GluK2-mEGFP, injection of RNA encoding GluK5-mEGFP did not yield bright fluorescent spots as typical for mEGFP-constructs located at the cell surface. Only diffuse dim fluorescence was observed, confirming GluK5-mEGFP expression, and consistent with the expectation that the subunit is retained intracellularly.
The distribution of GluK5 changed when it was co-expressed with GluK2. When GluK2-mEGFP was co-expressed with GluK5-mCherry at an RNA ratio of 1:3 many bright, clearly resolved and immobile red fluorescent spots from GluK5-mCherry were observed (, right image), which co-localized with green fluorescent spots of GluK2-mEGFP (, left image). Indeed, almost all of the red fluorescent spots (95.3 %, 341/358) also showed green fluorescence from GluK2-mEGFP. In contrast, a sizable fraction (41.4 %, 241/582) of the green fluorescent GluK2-mEGFP spots lacked a red-fluorescent GluK5-mCherry subunit. The results suggest that the cell membrane contained two populations of receptors: GluK2 homotetramers and GluK2/GluK5 heterotetramers. The small fraction of GluK5-mCherry for which no co-localization with GluK2-mEGFP was observed (4.7 %, 17/358) is fully accounted for by the number of complexes we predict to contain two subunits of each type, of which, based on a 0.8 probability of fluorescence, 4.0 % harbor two silent mEGFP molecules. This close agreement between observation and prediction is consistent with an absence of homomeric GluK5 complexes.
Co-expression of GluK2-mEGFP and GluK5-mCherry
Thus, this single molecule co-localization experiment supports earlier evidence that co-expression with GluK2 brings GluK5 to the surface, and that excess GluK2 subunits form homomeric complexes (Barberis et al., 2008
; Ma-Högemeier et al., 2010
). Importantly, the results also suggest a 2:2 stoichiometry in the GluK2/GluK5 heteromeric receptor.
To test rigorously the exact subunit composition of GluK2/GluK5 heteromeric receptors, we counted bleaching steps of GluK2-mEGFP in spots where GluK5-mCherry was co-localized (). The majority of red/green spots gave one or two mEGFP bleaching steps (, black bars), with a distribution closely following the prediction for a complex with two GluK2-mEGFP molecules (, red dashes), consistent with each heteromeric receptor containing two GluK2 subunits and two GluK5 subunits. The small number of spots with three and four green bleaching steps (7.3%, 9/124) that co-localized with red GluK5-mCherry is consistent with the predicted occurrence of two or more receptor complexes located within in one diffraction-limited spot. On average, ~280 green and red-green fluorescent signals were observed in movies from these experiments, corresponding to a spot density of 6.8 % (see Experimental Procedures). This density, in first approximation, equals the probability for random co-localization.
To further test the interpretation that GluK2/GluK5 heteromers have a defined 2:2 stoichiometry, we tagged the GluK5 subunit with EGFP and counted how many GluK5 subunits are found in complex with unlabeled GluK2. Since GluK5 homomers are not trafficked to the surface, co-injection of GluK5-mEGFP and unlabeled GluK2 was expected to only give fluorescence from heteromeric receptors. In the cells expressing the unlabeled GluK2, GluK5-mEGFP gave bright, immobile fluorescent spots. The majority of fluorescent spots bleached in one or two steps (, black bars), agreeing with the prediction for complexes consisting of two labeled GluK5 subunits and two unlabeled GluK2 subunits. The small fraction of spots with three and four bleaching steps (4.3%, 40/932) is consistent with the predicted occurrence of two or more receptor complexes located within in one diffraction-limited spot at the density of spots that was employed (4.9%, 200 spots/movie).
In summary, the results show that surface expressed GluK2/GluK5 receptors have a predominant, if not exclusive, 2:2 stoichiometry.
Previous experiments suggested that mutation of intracellular ER retention motifs regions promotes surface expression of GluK5 and that this receptor is homomeric (Hayes et al., 2003
; Nasu-Nishimura et al., 2006
; Ren et al., 2003
; Vivithanaporn et al., 2006
). These “released” GluK5 receptors are non-functional, but it is not known why. Transplanting the pore sequence of GluK5 into the background of GluK2 gives small currents, suggesting that GluK5 itself could be permeation competent (Villmann et al., 2008
). Furthermore, it is known that GluK5 subunits harbor a functional ligand binding domain, with high affinity for both kainate and glutamate (Barberis et al., 2008
; Herb et al., 1992
). One explanation for the lack of currents from GluK5 homomers could be that GluK5 subunits do not assemble into the correct tetrameric architecture. Indeed, in vitro
experiments have shown that the ATDs of GluK5 have an unusually low tendency to form homodimers, and that key contacts involved in tetramer formation are missing (Kumar and Mayer
; Kumar et al., 2011
). To check the assembly status of GluK5 homomers we generated GluK5ΔERret
-mEGFP with mutations rendering the two ER retention motifs and the endocytotic di-Leu motif inactive (), as described previously (Nasu-Nishimura et al., 2006
In contrast to wild type GluK5-mEGFP, expression of GluK5ΔERret-mEGFP gave bright immobile spots of green fluorescence at the cell surface. Bleaching analysis revealed mainly three and four bleaching steps per spot, consistent with assembly into homotetramers (). This demonstrates that, while heteromers are tightly regulated to contain two GluK2 and two GluK5 subunits, both GluK2 and GluK5 are intrinsically able to form tetramers. It also shows that a low stability of the ATD dimers and tetramers does not per se preclude tetramerization and trafficking. Our experiments, however, give no information, about whether the GluK5 homotetramers are fully and correctly assembled.