1 
 
 
Nanoscale mobility of the apo state and TARP stoichiometry dictate 
the gating behavior of alternatively-spliced AMPA receptors 
 
 
 
G. Brent Dawe1,2,6, Md. Fahim Kadir3,6, Raminta Venskutonytė4,6, Amanda M. Perozzo1,2, 
Yuhao Yan1,2, Ryan P.D. Alexander1,2, Camilo Navarrete3,5, Eduardo A. Santander3, Marika 
Arsenault2, Christian Fuentes3,5, Mark R.P. Aurousseau2, Karla Frydenvang4, Nelson P. Barrera5, 
Jette S. Kastrup4,7, J. Michael Edwardson3,7, and Derek Bowie2,7,8,* 
 
 
1Integrated Program in Neuroscience, McGill University, Montréal, QC H3A 2B4, Canada 
2Department of Pharmacology and Therapeutics, McGill University, Montréal, QC H3G 1Y6, Canada 
3Department of Pharmacology, University of Cambridge, Cambridge CB2 1PD, UK 
4Department of Drug Design and Pharmacology, University of Copenhagen, 2100 Copenhagen, Denmark 
5Department of Physiology, Pontificia Universidad Católica de Chile, 8331150 Santiago, Chile 
6Co-first author  
7Senior Author 
8Lead Contact 
*Correspondence: derek.bowie@mcgill.ca   
  
Summary: 150 (of 150) words 
Character Count: 52,308 (of 45,000 with spaces) characters 
Pages: 42 Figures: 8 + 7 supplementary, Supplemental Tables: 5, Supplemental Movies: 4  
 
 
 
  
Manuscript
 2 
Summary 
Neurotransmitter-gated ion-channels are allosteric proteins that switch on and off in response 
to agonist binding. Most studies have focused on the agonist-bound, activated channel whilst 
assigning a lesser role to the apo or resting state. Here, we show that nanoscale mobility of 
resting AMPA-type ionotropic glutamate receptors (AMPARs) predetermines responsiveness to 
neurotransmitter, allosteric anions and auxiliary TARP subunits. Mobility at rest is regulated by 
alternative splicing of the flip/flop cassette of the ligand-binding domain which controls 
motions in the distant AMPAR amino-terminal domain (NTD). Flip variants promote moderate 
NTD movement which establishes slower channel desensitization and robust regulation by 
anions and auxiliary subunits. In contrast, greater NTD mobility imparted by the flop cassette 
acts as a master-switch to override allosteric regulation. In AMPAR heteromers, TARP 
stoichiometry further modifies these actions of the flip/flop cassette generating two 
functionally-distinct classes of partially- and fully-TARPed receptors typical of cerebellar stellate 
and Purkinje cells. 
 
 
 
Keywords 
Ion-channel; ionotropic glutamate receptor; electrophysiology; atomic force microscopy; X-ray 
crystallography; channel gating; patch clamp; synapse; alternative splicing; protein 
conformations 
  
 3 
Introduction 
α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-type ionotropic glutamate 
receptors (iGluRs) mediate most fast-excitatory neurotransmission in the mammalian brain 
(Dingledine et al., 1999; Traynelis et al., 2010). They form the hardwiring of glutamatergic 
circuits but also strengthen or weaken synaptic transmission during periods of sustained 
patterned activity or altered homeostasis (Herring and Nicoll, 2016; Turrigiano, 2017). AMPA 
receptors (AMPARs) are also implicated in numerous CNS disorders and thus are targeted for 
the development of clinically-relevant compounds (Bowie, 2008). Consequently, there has been 
a concerted effort to provide a full understanding of the structural and functional aspects of 
AMPAR signaling. 
 
AMPARs assemble as tetramers in either a homomeric (Sobolevsky et al., 2009) or heteromeric 
(Herguedas et al., 2016; Herguedas et al., 2019) subunit arrangement that may additionally 
include accessory subunits such as the transmembrane AMPA receptor regulatory proteins 
(TARPs) and cornichon families (Greger et al., 2017; Jackson and Nicoll, 2011) (Fig. 1A). The 
AMPAR subunit is composed of four functional domains that include: (i) a cytoplasmic C-
terminal domain (CTD, not shown in Fig. 1A) that directs receptor trafficking and synaptic 
anchoring (Shepherd and Huganir, 2007), (ii) a transmembrane domain (TMD) which forms a 
central ion channel pore that rapidly transports Na+ and Ca2+ ions in response to binding of the 
neurotransmitter, L-glutamate (L-Glu) (Dingledine et al., 1999; Traynelis et al., 2010), (iii) a 
clamshell-like ligand-binding domain (LBD, Fig. 1B) (Mayer and Armstrong, 2004) and (iv) an 
amino-terminal domain (NTD) which directs subunit assembly and receptor clustering at 
 4 
synapses (Garcia-Nafria et al., 2016). 
 
In addition to these four distinct regions of the overall tetrameric structure, the LBD dimer 
interface has been shown to be critical in determining the time course of AMPAR gating (Dawe 
et al., 2015). Specifically, recent work from our lab has identified a novel cation binding pocket 
that promotes channel activation by the formation of a network of electrostatic interactions at 
the apex of both the AMPAR (Dawe et al., 2016) and kainate receptor (KAR) (Dawe et al., 2013) 
LBD dimer interfaces. Interestingly, anions have also been shown to control the time course of 
AMPAR gating (Bowie, 2002); however, the structural basis of this mechanism has yet to be 
understood. Another unresolved issue related to AMPAR gating is the possible role of the NTD 
in channel gating. Recent work has highlighted the dynamic motions in the NTD of both AMPA- 
and kainate-type iGluRs (Dürr et al., 2014; Dutta et al., 2015; Matsuda et al., 2016; Meyerson et 
al., 2014; Nakagawa et al., 2005) and interactions with auxiliary proteins (Cais et al., 2014; 
Moykkynen et al., 2014; Shaikh et al., 2016) that may facilitate trans-synaptic contact formation 
(Garcia-Nafria et al., 2016) and permit AMPAR trafficking during synapse strengthening (Diaz-
Alonso et al., 2017; Watson et al., 2017). Since structural re-arrangements of the NTD 
accompany receptor desensitization (Dürr et al., 2014; Meyerson et al., 2014; Nakagawa et al., 
2005; Twomey et al., 2017a), it has been assumed that the underlying movement is triggered by 
agonist binding. However, it is also possible that the intrinsic thermodynamic mobility of the 
resting AMPAR determines NTD movement and its responsiveness to agonist.   
Here, we have designed experiments to distinguish between these two possibilities. Our data 
identify a novel allosteric anion binding pocket at the alternatively spliced flip/flop cassette that 
 5 
modifies NTD motions in the resting state prior to agonist binding and regulates channel gating 
in the presence and absence of auxiliary subunits. Ser775 of the GluA2flip isoform (GluA2i) 
renders AMPARs sensitive to anion modulation whereas Asn775 of the GluA2flop isoform 
(GluA2o) almost eliminates the effects of anions on the NTD and channel gating. Imaging by 
atomic force microscopy (AFM) reveals that the NTDs of GluA2o receptors are more mobile than 
those of GluA2i receptors, indicating differences in the intrinsic conformational flexibility of 
their resting states. This behavior is interchangeable via a single amino acid, the Ser775 residue, 
that operates as a molecular switch between flip and flop isoforms. TARP stoichiometry further 
modifies these actions of the flip/flop cassette on AMPAR heteromers generating two distinct 
classes of partially- and fully-TARPed GluA1/A2 receptors that match the functional profile of 
native AMPARs expressed by cerebellar stellate and Purkinje cells, respectively.  
 6 
Results 
Anions modulate AMPAR desensitization 
Although the structural and functional bases of anion and cation modulation of KARs have been 
studied extensively (Bowie, 2002, 2010; Dawe et al., 2013; Wong et al., 2006), much less is 
known about the effect of external ions on AMPARs. Recent work identified the structural 
mechanism of cation regulation (Fig. 1B) (Dawe et al., 2016); however, the nature of anion 
modulation of AMPARs (Bowie, 2002) remains unknown. The effect of external anions on 
AMPAR deactivation and desensitization was studied by recording agonist-evoked membrane 
currents in outside-out membrane patches excised from HEK293 cells expressing GluA2i(Q) (Fig. 
1C-H, see Methods). As observed previously for GluA1i receptors (Bowie, 2002), the time course 
of entry into desensitization for GluA2i was sensitive to external halides, accelerating 1.7- and 
4.6-fold in bromide (τ = 4.3 ± 0.2 ms; n = 7) and iodide (τ = 1.6 ± 0.1 ms; n = 7), respectively, 
compared to chloride (τ = 7.3 ± 0.3 ms; n = 15)  (Fig. 1C and F, Table S1). In contrast, 
deactivation rates were almost identical for all anions tested on GluA2i (Fig. 1D and G, Table 
S1). In keeping with this, recovery rates out of desensitization were also anion-dependent with 
GluA2i recovering from desensitization in external iodide (τrecovery = 62.5 ± 3.0 ms; n = 6) about 
3-fold more slowly than in external chloride (τrecovery = 24.1 ± 1.5 ms; n = 13) (Fig. 1E and H, 
Table S1). Interestingly, rates into and out of desensitization have a predictive relationship with 
the ionic radius of the external anion (Fig. 1F and H), with faster desensitization rates observed 
with anions of a larger radius. The relationship between ionic radius and channel 
desensitization is consistent with the existence of a specific anion binding pocket. Previous 
work has located cation binding pockets critical to AMPAR and KAR gating to the interface of 
 7 
LBD dimers (Dawe et al., 2013; Dawe et al., 2016) (Fig. 1B); consequently, we reasoned that 
anions may bind to this region too. Moreover, previous work has demonstrated the important 
role of the LBD dimer interface in regulating AMPAR gating, including desensitization (Dawe et 
al., 2015; Horning and Mayer, 2004; Sun et al., 2002). 
 
To determine the location of the anion binding pocket, two soluble constructs of the GluA2-LBD 
were crystallized in the presence of bromide ions (Fig. 2). Since bromide ions give anomalous 
scattering, we used this property to identify the position of the bound bromide ions in the 
structure and to distinguish them from other ions and water molecules in the X-ray diffraction 
data (Fig. 2A-D), as also previously done for localization of the anion binding site in kainate 
receptors (Plested and Mayer, 2007). Two X-ray structures were determined which correspond 
to the flop isoform (GluA2o-LBD; PDB code, 6GL4) and the flip-like mutant GluA2o-LBD N775S 
(PDB code, 6GIV). The structures are shown in Figs. 2 and S1 along with statistics of data 
collection and refinement in Table S2. The anomalous scattering data clearly indicate the 
location of two bromide ions near the base of the D1-D1 dimer interface in both structures (Fig. 
2C and D, Fig. S1A-C). More specifically, these anion binding sites are in a hydrophobic space, 
surrounded by Pro515 and Leu772 from one subunit, Ile502, Leu504 and Pro515 from the 
partner subunit, and capped by Lys514 (Fig. 2E and F, Fig. S1D and E, Fig. S2A and B). There are 
also several water molecules surrounding each bromide, separating the ion from Ser775 (Fig. 
2C-F). Interestingly, we also determined a structure of GluA2o-LBD in the presence of a high 
concentration of chloride ions and the electron density indicated that chloride ions bind to the 
same location as bromide in the LBD dimer interface (data not shown). 
 8 
 
Ser775 is notable for interacting with positive modulators of AMPARs (Fig. S1F) such as 
cyclothiazide (CTZ) (Partin et al., 1996), as well as being one of the residues that forms the 
alternatively spliced flip/flop cassette (Sommer et al., 1990). Given the high affinity for CTZ, we 
hypothesized that it would displace bromide ions from their bound positions. In agreement 
with this, electrophysiological responses observed in different external anions were non-
decaying in each case (Fig. S1G and H) consistent with the idea that CTZ out competes external 
anions for binding. Likewise, mutation of Leu504 to an Ala or Cys residue either reversed or 
eliminated anion regulation of GluA2i receptor decay kinetics, respectively (Fig. S2C-F), further 
validating the location of anion binding pocket as being at the LBD dimer interface. Taken 
together, these results identify an anion binding pocket which, when occupied by halide ions, 
may be responsible for regulating the rates of entry into and exit from AMPAR desensitization.  
 
Ser/Asn residue regulates anion effects 
To establish a causal relationship between the anion binding pocket and the functional effects 
of external anions on GluA2, we focused on position 775 which, as mentioned previously, is 
modified by alternative splicing and is also involved in all regulatory effects of the flip/flop 
cassette (see below). The flip isoform of the AMPAR contains a serine (Ser) at position 775 
whereas the flop isoform contains an asparagine (Asn) (Fig. 3A and B) (Sommer et al., 1990), 
which we hypothesized may differentially affect anion modulation of AMPARs. 
 
 9 
To test this, we first compared the effect of external anions on desensitization rates of GluA2i 
and GluA2o receptors whose amino acid sequences differ at nine residues (Fig. 3A) (Sommer et 
al., 1990). As anticipated, GluA2i and GluA2o receptors differed in their sensitivity to modulation 
by external ions (Fig. 3C and D), though the trend of faster desensitization in the presence of 
larger anions persisted (Fig. 3F). For example in GluA2o receptors, iodide accelerated rates into 
desensitization about 1.6-fold (τ = 0.8 ± 0.05 ms; n = 6) compared to chloride (τ = 1.3 ± 0.06 ms; 
n = 12) which was substantially less than the 4.6-fold difference observed on GluA2i receptors 
(Fig. 3C, D and F, Table S1). To examine whether this difference was due to residue 775, which 
is in close proximity to the bromide binding site (Fig. 2E), we repeated these experiments on 
the Ser775Asn GluA2i receptor (Fig. 3E and F, Table S1). As anticipated, the GluA2i S775N 
receptor was much less sensitive to modulation by external anions. For example, iodide (τ = 3.6 
± 0.5 ms; n = 6) accelerated desensitization compared to chloride (τ = 5.2 ± 0.4 ms; n = 12) by 
only 1.4-fold (Fig. 3E and F, Table S1). In contrast, mutation of the two more apical dimer 
interface residues that contribute to fast GluA2o desensitization (Fig. 3B) (Quirk et al., 2004) 
produced a receptor (T765N/P766A) that still exhibited robust anion sensitivity (Fig. S3). As a 
result, the near loss of anion modulation from flip- to flop-type GluA2 AMPARs can be primarily 
attributed to Asn775, consistent with our structural data placing bromide ions in the lower D1-
D1 LBD dimer interface. Interestingly, anion modulation of recovery from desensitization was 
still present in GluA2i S775N receptors (Fig. 3G and H) demonstrating that anion effects on rates 
into and out of desensitization have different mechanisms. 
 
 
 10 
Anions control resting and active AMPARs 
Anions may affect AMPARs by controlling the receptor’s resting and/or activated state(s). To 
delineate between these possibilities, we investigated whether external halides elicited global, 
conformational changes in protein structure in the absence and/or presence of agonist. 
Specifically, we measured protein height as an indicator of conformation using atomic force 
microscopy (AFM) since AMPAR activation and desensitization involve compression in the 
quaternary structure (Dürr et al., 2014; Herguedas et al., 2016; Meyerson et al., 2014; Twomey 
et al., 2017a, b) as in NMDA-type iGluRs (Balasuriya et al., 2014; Suzuki et al., 2013). To image 
the receptor, we purified and reconstituted individual AMPAR complexes into lipid bilayers (Fig. 
4 and Fig. S4, see Methods). 
 
The addition of L-Glu to NaCl-based external solution prompted a 0.69 ± 0.11 nm (n = 11) 
reversible reduction in GluA2i receptor height (Fig. 4A and C, Fig. S5) which was prevented by 
the competitive antagonist, CNQX, and the positive allosteric modulator, CTZ (Table S3, Fig. S5). 
AFM experiments were also repeated to determine if height changes could be induced by 
different anions (Fig. 4B and D, Fig. S5, Table S3). Unexpectedly, merely changing the main 
external anion species from NaCl to NaBr (0.74 ± 0.06 nm; n = 12) or NaI (0.87 ± 0.11 nm; n = 
13) produced a substantial and reversible vertical compression of individual GluA2i receptors 
(Fig. 4D and S5) comparable to L-Glu-evoked responses (Fig. 4C). Interestingly, much less 
additional compression (~ 0.2 nm) was observed when L-Glu (10 mM) was added to NaBr or NaI 
solutions (Fig. 4C), suggesting the principal role of anions is to prime the receptor. Since the 
 11 
action of L-Glu, however, was not entirely occluded, the structural rearrangements observed 
with anion substitution must differ from those elicited by agonist binding. 
 
To establish a relationship between anion-induced height changes (Fig. 4D) and their effect on 
desensitization (Fig. 1C and F), we repeated AFM measurements on GluA2o whose gating 
properties are relatively anion-insensitive. Remarkably, anion-induced height compression was 
almost completely absent from the GluA2o receptor when external solution containing NaCl 
was switched to either NaBr (0.03± 0.06 nm; n = 14) or NaI (0.08± 0.06 nm; n = 14) (Fig. 4D, 
Table S3). Importantly, agonist binding continued to elicit reductions in the height of all anion 
species (Fig. 4C) reaffirming that anion and agonist effects are different. In agreement with our 
electrophysiology data, iodide anions did not affect the AFM-reported height changes of GluA2i 
receptors containing the single S775N point mutation (Fig. 4D, Table S3), establishing a critical 
role for the Ser/Asn residue and causal relationship between the effects of anions on channel 
gating and height changes. Finally, AFM-reported height changes elicited by switches in 
external anions were absent from the GluA2i L504A receptor, whereas L-Glu persisted in 
reducing height in all anion species tested (Fig. S2G and H, Table S3). 
 
Taken together, these data indicate two important points. First, external anions prime the 
receptor prior to activation. Whether priming then determines the rate of AMPAR 
desensitization prior to agonist binding is investigated below. Second, anion modulation reveals 
a coupling between the LBD dimer interface and the NTD. This latter possibility was examined 
in additional AFM experiments on GluA2 lacking the NTD (i.e. GluA2 ΔNTD).   
 12 
 
Flip/flop cassette controls NTD motions 
Anion- and agonist-induced height changes were repeated using the truncated GluA2i ΔNTD 
receptor (Fig. 4E-G and Fig. S6). Contrary to wildtype behavior, virtually no height changes could 
be induced in the GluA2i ΔNTD receptor by replacing NaCl with NaI (-0.05 ± 0.08 nm; n = 15) or 
by agonist application (in NaCl, 0.03 ± 0.06 nm; n = 13) (Fig. 4E, Table S3), demonstrating that 
receptor compression requires the NTD. To explore this further, we re-engineered GluA2i and 
GluA2o receptors, replacing each NTD with an eGFP molecule (Fig. 4F and G). Importantly, the 
expressed GluA2 ΔNTD-eGFP constructs exhibited similar functional behavior in terms of anion 
sensitivity and responsiveness to agonist (data not shown). We reasoned that although anions 
and agonists bind to the LBD, the energy of binding may induce a conformational change that is 
more prominently observed in distant regions of the protein, even if the NTD is replaced by 
eGFP. Consistent with this, compression induced by external anions (NaCl to NaI, 0.32 ± 0.09 
nm; n = 18) and agonist application (in NaCl 0.28 ± 0.04 nm; n = 10) was restored in the GluA2i 
ΔNTD-eGFP construct, demonstrating that binding events in the LBD are transferred to the NTD. 
Since the GluA2i ΔNTD-eGFP construct retained the original linker sequences, we reasoned that 
the linkers might exert a similar downward pulling force on whatever is attached above. Since 
eGFP has a different size and likely does not form inter-subunit dimers, these distinctions may 
explain why height changes were about 2-fold less with GluA2i ΔNTD-eGFP. Finally, agonist but 
not anion binding induced compression of the GluA2o ΔNTD-eGFP construct (Fig. 4G) confirming 
that compression of the NTD is differentially controlled by the flip/flop cassette. Given this, we 
 13 
next tested if the intrinsic motions in resting GluA2 AMPARs may be a critical factor in 
determining its anion and agonist responsiveness.   
 
AMPARs have different resting states  
AFM visualization of the NTD movements of the GluA2 AMPAR (Fig. 5) revealed that both 
isoforms exist as two distinct globular structures, which show mobility even in the resting state 
(Fig. 5A, Movies S1 and S2). Unexpectedly, NTD movements at rest, expressed as cumulative 
squared displacement (CSD), were almost 3-fold greater for GluA2o (resting CSD, 15.53 ± 2.54 
nm2, n=7) compared to GluA2i (resting CSD, 5.58 ± 0.92 nm2, n=5) (Fig. 5B and D, Table S4) 
revealing that NTD motions are controlled by the flip/flop cassette. Bath application of a 
saturating concentration of L-Glu (10 mM) increased the NTD mobility of the GluA2i AMPAR 
(CSD, 13.90 ± 2.11 nm2, n=9) but not that of GluA2o (CSD, 17.12 ± 0.59 nm2, n=5) (Fig. 5C and D, 
Movies S3 and S4). As expected, the NTD mobility in the presence of L-Glu was reduced by 
CNQX for both GluA2i (5.33 ± 0.82 nm2, n=4) and GluA2o (9.50 ± 1.95 nm2, n=3) (Fig. 5D, Table 
S4) demonstrating a direct relationship between NTD mobility and agonist binding. NTD 
movement of GluA2i was agonist-concentration dependent (Fig. 5E, Table S4), whereas the 
mobility of GluA2o was not. Most importantly, the NTD mobility of the GluA2i S775N mutant 
(resting CSD, 14.00 ± 1.70, n=9) was also greater than that of wildtype GluA2i, but not of GluA2o 
(Fig. 5B, Table S4). This observation is important as it establishes a causal link between the 
effects of anions on channel gating (Figs. 1 and 3) and height changes (Fig. 4) with the resting 
state of the AMPAR. 
 
 14 
Given this, we reasoned that the lower conformational mobility of the GluA2i receptor at rest 
would favor more stable interactions in regions of the protein, such as the LBD dimer interface 
(Dawe et al., 2015; Dawe et al., 2016), that stabilize the activated state of the receptor. A more 
stable AMPAR structure would also favor anion binding to the LBD dimer interface and account 
for the greater effect of external halide ions on GluA2i receptors. Conversely, the higher 
mobility of the GluA2o receptor would disfavor anion binding (and its regulation) and 
destabilize the open state of the receptor, explaining its much more rapid desensitization 
kinetics and weaker anion sensitivity. Given this, we concluded that the flop cassette of the 
AMPAR receptor acts as a master-switch that overrides allosteric mechanisms, such as anion 
regulation that impacts the time AMPARs remain in the open state. We therefore hypothesized 
that GluA2o should also be similarly insensitive to regulation by auxiliary proteins, such as the 
prototypical TARP, stargazin (2). 
 
To test this hypothesis, we compared the electrophysiological responses of GluA2i, GluA2o and 
GluA2i S775N receptors expressed with 2 (Fig. 6A-D). As noted previously (Dawe et al., 2016), 
co-assembly of GluA2i receptors with 2 slowed desensitization rates (des, ms) and reduced 
equilibrium desensitization (Iequilibrium, %) by about 4-fold and 22-fold, respectively (Fig. 6A, C, 
and D, Table S1). Desensitization rates and equilibrium desensitization were also anion-
sensitive, exhibiting a similar rank order of potency as described for GluA2i receptors alone (Fig. 
6B-D, Table S1). In contrast, co-assembly of GluA2o receptors with 2 almost eliminated the 
effect of the TARP on desensitization rates, equilibrium desensitization and anion regulation 
(Fig. 6A-D, Table S1) in agreement with the hypothesis that the flop cassette overrides TARP 
 15 
regulation. Importantly, 2 also had a greatly attenuated effect on GluA2i S775N receptors, 
further demonstrating the pivotal role of the 775 residue as a molecular switch (Fig. 6A-E, Table 
S1). 
 
Native AMPARs are also sensitive to anion regulation 
To examine how these actions of the flip/flop cassette impact native receptors, we studied the 
gating properties of AMPARs in outside-out and nucleated patches excised from cerebellar 
Purkinje and stellate cells, respectively (Fig. 7, Table S5). Previous work has shown that AMPARs 
expressed by both cell types are regulated by 2 but that their gating properties are distinct 
(Barbour et al., 1994; Bats et al., 2012; Yamazaki et al., 2015). In keeping with this, rapid 
application (250 ms) of 10 mM L-Glu to excised patches from Purkinje cells elicited AMPAR 
responses that decayed at a slower rate and to a lesser extent than responses from stellate 
cells (Fig. 7A). There was a 6-fold difference in equilibrium desensitization between Purkinje 
and stellate cells corresponding to steady-state/peak values of 8.3 ± 0.6 % (n=21) and 1.4 ± 0.1 
% (n=29), respectively (Fig. 7B). Similarly, the decay kinetics of the responses from Purkinje cells 
were best fit with a weighted time constant of 7.5 ± 0.4 ms (n=20) compared to 2.9 ± 0.1 ms 
(n=30) for responses from stellate cells (Fig. 7C). These distinctions in decay kinetics and the 
degree of equilibrium desensitization of Purkinje and stellate cells are reminiscent of the 
functional differences between GluA2i/2 and GluA2o/2 receptors, respectively (Fig. 6A-D). 
However, two subsequent observations suggested that a more complicated explanation was 
required to account for the properties of AMPARs expressed by Purkinje and stellate cells. 
 
 16 
First, the positive allosteric modulator CTZ (100 M) eliminated macroscopic desensitization of 
AMPAR-mediated responses of Purkinje and stellate cells (Fig. 7A, insets) confirming that both 
cell types express flip-dominant AMPARs (Partin et al., 1995; Partin et al., 1994; Penn et al., 
2012). Flop-dominant AMPARs continue to desensitize in the presence of CTZ, albeit at a 
reduced rate (Partin et al., 1995; Partin et al., 1994), which is distinct from the responses 
observed in patches from Purkinje and stellate cells. Second, although AMPAR responses 
exhibited by Purkinje and stellate cells were sensitive to external anion modulation (Fig. 7D and 
E), their effect on decay kinetics were intermediate between GluA2i/2 and GluA2o/2 receptor 
responses (Fig. 6D, Table S1). AMPAR decay kinetics in both cell types slowed by about 2- to 3-
fold upon exchange of external Br- (Purkinje,  = 3.9 ± 0.5 ms, n=5; stellate,  = 2.0 ± 0.2 ms, 
n=8) with F- (Purkinje,  = 10.6 ± 0.9 ms, n=6; stellate,  = 5.0 ± 0.6 ms, n=5). These apparently 
inconsistent observations may be reconciled if Purkinje and/or stellate cells express AMPAR 
tetramers that contain both flip and flop variants. Consistent with this suggestion, previous 
work has shown that CTZ eliminates macroscopic desensitization of recombinant GluA1i/A2o or 
GluA1o/A2i heteromers (Partin et al., 1994), matching the responses from patches of Purkinje 
and stellate cells. Since the functional impact of the flip/flop cassette on GluA1/A2 heteromers 
has yet to be studied in terms of their modulation by TARP 2 and sensitivity to external anions, 
we performed additional experiments to better understand this relationship. 
 
TARP stoichiometry shapes the functional behavior of AMPAR heteromers  
To examine how alternative splicing impacts the function of GluA1/A2 heteromers, we initially 
focused on fully TARPed receptors by tethering the 2 auxiliary subunit to flip/flop variants of 
 17 
GluA1(Q) and GluA2(R) subunits. GluA1/A2 heteromerization was confirmed in each recording 
by testing for the loss of cytoplasmic polyamine block (Fig. S7, see Methods), as described 
previously (Partin et al., 1995). 
 
The flip/flop cassette had a profound and concomitant effect on the decay kinetics and 
equilibrium desensitization of fully-TARPed GluA1/A2 heteromers (Fig. 8A). Flip-only 
heteromers (A1i/2 + A2i/2) decayed with a slower rate ( = 10.6 ± 0.3 ms, n=6) and to a lesser 
extent (ss/peak = 18.5 ± 1.4 %, n=6) in response to agonist stimulation than flop-only receptors 
(A1o/2 + A2o/2), which had faster decay kinetics ( = 3.0 ± 0.2 ms, n=7) and more complete 
equilibrium desensitization (ss/peak = 3.5 ± 0.8 %, n=7) (Fig. 8B-D, Table S5). Although, 
GluA1/A2 heteromers containing both flip/flop variants had intermediate behavior, as might be 
expected, alternative splicing of GluA2 had the more dominant impact on channel gating. For 
example, GluA1i/A2o receptors (A1i/2 + A2o/2) exhibited faster ( = 3.3 ± 0.3 ms, n=7) and 
more complete desensitization (ss/peak = 7.1 ± 0.7 %, n=7) than GluA1o/A2i receptors (A1o/2 + 
A2i/2,  = 8.9 ± 0.5 ms, n=8; ss/peak = 14.3 ± 1.5 %, n=8) (Fig. 8B-D, Table S5). Analysis of anion 
effects on fully-TARPed GluA1/A2 heteromers revealed a similar relationship, where flip-only 
heteromers were more sensitive to anion-regulation than flop-only heteromers and alternative 
splicing of GluA2 had the more dominant effect (Fig. 8E, Table S5). Interestingly, the decay 
kinetics and equilibrium desensitization of fully-TARPed GluA1/A2 heteromers exhibited a linear 
relationship across all external anion conditions (Fig. 8F). Given this, we reasoned that this 
relationship could be used to interrogate TARP 2 stoichiometry of native AMPARs with the 
data already obtained from cerebellar Purkinje and stellate cells. 
 18 
 
In keeping with this, anion modulation of native AMPARs of cerebellar Purkinje cells was well fit 
by a linear relationship that was statistically indistinguishable from fully-TARPed recombinant 
GluA1/A2 heteromers (p = 0.10, ANOVA, Fig. 8F, red). The data taken from stellate cells, 
however, did not match the relationship: even though the stellate cell data were well fit by 
linear regression, the relationship had a different slope (Fig. 8G, cyan). Unlike fully-TARPed 
receptors and AMPARs from Purkinje cells, equilibrium responses elicited by AMPARs from 
stellate cells were only weakly sensitive to anion modulation (Fig. 7, Table S5). Given this, we 
reasoned that the response profile of AMPARs from stellate cells may be more consistent with 
a partially-TARPed receptor. 
 
To test this, we examined the functional behavior of flip/flop variants of GluA1/A2 heteromers 
where 2 was tethered to either the GluA1 or GluA2 subunit so that any GluA1/A2 tetramer 
combination would possess only two 2 auxiliary subunits. As anticipated, the relationship 
between decay kinetics and equilibrium desensitization of partially-TARPed GluA1/A2 receptors 
was different from that of fully-TARPed heteromers (Fig. 8G). For example, although flip-
containing heteromers exhibited slower desensitization kinetics (e.g. A1i/2 + A2i,  = 8.2 ± 0.8 
ms, n=10) than heteromers containing both flip and flop (e.g. A1i/2 + A2o,  = 3.4 ± 0.3 ms, 
n=8), the degree of equilibrium desensitization was similar in each case (A1i/2 + A2i, ss/peak = 
3.7 ± 0.9 %, n=10 versus A1i/2 + A2o, ss/peak = 1.5 ± 0.5 %, n=8) (Fig. 8B-C, Table S5). Although 
partially-TARPed heteromers containing both flip/flop variants had intermediate behavior, 
alternative splicing of GluA2 had the more dominant impact on desensitization kinetics. For 
 19 
example, GluA1i/A2o receptors exhibited faster desensitization kinetics (A1i/2 + A2o,  = 3.4 ± 
0.3 ms, n=8) than GluA1o/A2i receptors (A1o/2 + A2i,  = 7.0 ± 0.4 ms, n=6) (Fig. 8B-C, Table S5). 
Thus, the decay kinetics of partially-TARPed AMPARs varied according to subunit composition 
and the external anion type whereas equilibrium desensitization was relatively unchanged (Fig. 
8G, Table S5), much like the behavior of AMPARs from stellate cells (Fig. 7). In fact, linear 
regression plots of data from partially-TARPed AMPARs and stellate cells were statistically 
indistinguishable (p = 0.20, ANOVA, Fig. 8G), suggesting that stellate cells express partially-
TARPed AMPARs. Taken together, these data provide compelling evidence for the important 
role of TARP stoichiometry in dictating the functional behavior of recombinant and native 
AMPAR heteromers. Our results also provide a proof-of-principle approach for future enquiry 
interrogating the auxiliary subunit composition of native receptors.                   
  
 20 
Discussion 
The present study advances our understanding of a major neurotransmitter receptor in several 
important ways. First, it underlines the central importance of the apo state in priming the 
receptor prior to activation and dictating its responsiveness to channel activators, allosteric 
modulators and auxiliary proteins. Second, it uncovers an unappreciated and new role of 
alternative splicing of the flip/flop cassette which shapes AMPAR signaling by regulating the 
nature of the apo state. Third, it reveals the additional role of TARP stoichiometry in also 
dictating the functional behavior of AMPAR heteromers and its potential value in explaining the 
responsiveness of native AMPARs. Finally, it establishes that the LBD of AMPARs exerts a long-
range allosteric control on motions in the NTD. As discussed below, differences in the nanoscale 
mobility of the NTD may affect the trafficking and/or synapse strengthening of different 
subtypes of native AMPARs at glutamatergic synapses. 
 
The dynamic nature of the apo or resting state 
For decades, it has been assumed that the work performed by signaling proteins, such as ion 
channels, is initiated by the binding energy derived from a plethora of soluble activators, such 
as neurotransmitters. As a result, the role of the apo or resting state has been largely 
overlooked. However, some observations, particularly on nicotinic acetylcholine receptors 
(nAChRs), have challenged this orthodoxy. For example, nAChRs expressed by skeletal muscle 
can access the open or activated state of the channel in the absence of receptor agonists 
(Jackson, 1984, 1986) suggesting an underlying dynamic nature to the apo state. The probability 
that wildtype, unbound nAChRs enter into the open state is small, however many mutations 
 21 
throughout the protein can increase this probability almost as much as agonist binding (Jadey 
et al., 2011; Purohit and Auerbach, 2009). In keeping with this, some nAChR mutations 
associated with congenital myasthenic syndromes can be linked to selective changes in the apo 
state rather than the activated state (Engel et al., 2010; Jadey et al., 2011), highlighting the 
importance of the resting state in the context of human disease. 
 
The lurcher mutation (A654T) of the delta2 (2) iGluR subunit gives rise to cerebellar ataxia (Zuo 
et al., 1997) and was also concluded to reflect changes in the apo state since wildtype 2 iGluRs 
are unresponsive to neurotransmitter (Hansen et al., 2009; Kohda et al., 2000; Taverna et al., 
2000). The homologous mutation is also found in de novo missense mutations of GluA1 and 
GluA3 AMPAR subunits where it is associated with severe neurodevelopmental delay and 
autism (Geisheker et al., 2017). It was initially proposed that the lurcher mutation in GluA1 
channels also gives rise to constitutively active channels (Kohda et al., 2000; Schwarz et al., 
2001; Taverna et al., 2000); however, a more recent assessment concluded that the mutation 
has little or no effect on the apo state but rather increases sensitivity to the neurotransmitter L-
Glu (Klein and Howe, 2004). The present study uncovers the existence of functionally-distinct 
apo states of the AMPAR that are regulated through alternative splicing of the LBD. Rather than 
facilitating entry into the main open state, the flip/flop cassette establishes the mobility of the 
apo state and, in doing so, fine tunes the responsiveness of AMPARs to neurotransmitter, 
allosteric anions and auxiliary proteins. As explained below, the complex expression pattern of 
flip and flop isoforms in the vertebrate brain suggests that the apo or resting state plays a 
critical role in neuronal signaling of developing and adult neuronal circuits.  
 22 
A unifying role for the AMPA receptor flip/flop cassette 
We identify an entirely new and unifying role for the AMPAR flip/flop cassette through its 
regulation of the apo or resting state of AMPARs. Previous studies have linked the flip/flop 
cassette to several important and apparently disparate properties of AMPARs that include 
determining the rates into AMPAR desensitization (Koike et al., 2000; Mosbacher et al., 1994; 
Quirk et al., 2004), regulation by positive allosteric modulators, such as CTZ, aniracetam, CX614 
and PEPA (4-[2-(Phenylsulfonylamino)ethylthio]-2,6-Difluoro-Phenoxyacetamide) (Jin et al., 
2005; Partin et al., 1995; Partin et al., 1994; Sekiguchi et al., 1997; Sun et al., 2002) as well as 
controlling AMPAR secretion from the endoplasmic reticulum (ER) (Coleman et al., 2006; Penn 
et al., 2008) (Fig. 6E). 
 
Markov models describing the effect of alternative splicing on AMPAR channel gating and its 
regulation by allosteric modulators have assumed that the flip/flop cassette impacts AMPARs 
only after agonist binding (Koike et al., 2000; Partin et al., 1996). Our data provides an 
alternative explanation whereby the key events that shape channel gating and allosteric 
modulation occur before agonist binding pre-determined by the intrinsic mobility of the apo 
state. In keeping with this, kinetic differences between flip/flop isoforms are controlled by 
three closely-positioned residues in the flip/flop cassette (Fig. 3B and 6E), including Ser775Asn 
(Quirk et al., 2004). As this same residue regulates the apo state, it establishes a causal link 
between the mobility of the apo state with channel gating and allosteric modulation. 
 
 23 
The effect of alternative splicing on AMPAR secretion from the ER has also been associated with 
motions triggered by agonist-bound AMPARs, primarily through the Val/Leu779 residue in 
coordination with Ser/Asn775 (Coleman et al., 2006; Penn et al., 2012) (Fig. 6E). Interestingly, 
mutating the Thr765 and Pro766 residues of the flip cassette to Asn and Ala in the flop isoform 
(Fig. 3A and B) has no effect on AMPAR secretion from the ER (Penn et al., 2008) but yet with 
Ser775Asn, they fully account for the kinetic differences between flip and flop isoforms (Quirk 
et al., 2004). This distinction reveals that some of the amino acid residues of the flip/flop 
cassette that govern differences in channel gating and ER exit of AMPARs are separable. 
Importantly, the Ser/Asn775 residue is implicated in all of these known regulatory effects of 
alternative splicing, which is in keeping with its central structural position at the kink between 
helices J and K of the flip/flop cassette (Fig. 6E) separating the 765/766 residues, that control 
channel gating, from the 779 residue, that controls ER exit. This structural arrangement has 
important ramifications for the developing and adult CNS since the expression of the flip/flop 
cassette in neurons is developmentally-regulated (Monyer et al., 1991), activity-dependent 
(Penn et al., 2012) and cell-type specific (Sommer et al., 1990). Consequently, our work 
establishes for the first time that fine-tuning nanoscale movements of apo or resting AMPARs 
may be a critical factor governing glutamatergic signaling in the mammalian brain. Finally, 
whether the nearby arginine of the R/G site, which fits directly into the LBD dimer interface 
(Greger et al., 2006), could influence LBD stability as well as anion and TARP sensitivity awaits 
to be studied. 
 
 
 24 
TARP stoichiometry dictates the functional behavior of alternatively-spliced AMPA receptors 
Our study establishes that TARP stoichiometry modifies the actions of the flip/flop cassette on 
recombinant AMPAR heteromers, generating two functionally-distinct classes that correspond 
to partially- and fully-TARPed AMPARs (Fig. 8F and G). Biochemical analysis of AMPARs native to 
the cerebellum has suggested that TARP stoichiometry is fixed (Kim et al., 2010), although it 
was not possible to determine whether the number of TARPs per AMPAR tetramer 
corresponded to partial (i.e. 1, 2 or 3 TARPs) or full occupancy (i.e. 4 TARPs). Cryo-EM structures 
and single-molecule imaging studies of AMPARs have shown that the make-up of AMPAR-TARP 
complexes can be quite variable, with assemblies containing 1 or 2 (Hastie et al., 2013; Twomey 
et al., 2016) or 4 (Hastie et al., 2013; Twomey et al., 2017a; Zhao et al., 2016) TARP auxiliary 
subunits per AMPAR tetramer complex (Chen and Gouaux, 2019). Our data from AMPARs 
expressed by cerebellar Purkinje and stellate cells suggest that TARP stoichiometry is variable 
and, most likely, fixed in each neuronal class, in agreement with a previous study comparing 
TARP stoichiometry between hippocampal granule and CA1 pyramidal cells (Shi et al., 2009). 
These conclusions are reliant on the similarity between the data from cerebellar neurons and 
recombinant GluA1/A2 heteromers. Whether a similar relationship can be extended to 
GluA2/A3 heteromers, another abundant AMPAR composition in the CNS (Bowie, 2012; Henley 
and Wilkinson, 2016; Jacobi and von Engelhardt, 2017), remains to be determined. Since data 
from Purkinje and stellate cells are consistent with AMPARs being fully-and partially-occupied 
by TARPs, respectively, it is possible that CNS neurons may regulate AMPAR responsiveness by 
varying the number of auxiliary subunits per tetramer. Since differences in the duration and 
amplitude of AMPAR-mediated excitatory postsynaptic potentials determines whether 
 25 
postsynaptic neurons operate as integrators of synaptic activity or coincidence detectors (Konig 
et al., 1996; Shadlen and Newsome, 1994), it will be interesting in future studies to examine the 
role of TARP stoichiometry in shaping the complex behavior of neuronal circuits.   
 
Long-range allosteric control of the amino-terminal domain by the flip/flop cassette 
Several studies have reported dynamic motions in the NTD of both AMPA- and kainate-type 
iGluRs (Dürr et al., 2014; Dutta et al., 2015; Matsuda et al., 2016; Meyerson et al., 2014; 
Nakagawa et al., 2005) which have been assumed to be a consequence of agonist binding and 
receptor desensitization (Dürr et al., 2014; Meyerson et al., 2014; Nakagawa et al., 2005). Our 
data reveal unexpectedly that motions in the NTD occur in the apo state through long-range 
control exerted by the flip/flop cassette which primes the receptor prior to agonist binding (Fig. 
4) and establishes its intrinsic mobility (Fig. 5). This mechanism is distinct from the allosteric 
coupling described for NMDA-type iGluRs where it is the NTD that dictates the behavior of the 
LBD of different GluN2-containing isoforms to determine agonist potency and channel kinetics 
(Gielen et al., 2009; Yuan et al., 2009). Flip variants promote moderate NTD movement and give 
rise to slower channel desensitization and robust regulation by anions and auxiliary subunits. 
The greater mobility imparted by the flop cassette overrides this allosteric regulation and acts 
as a master-switch presumably by rendering the LBD dimer interface less stable (Dawe et al., 
2015). Removal of the NTD has only a modest slowing effect on AMPAR gating and allosteric 
regulation by anions (Table S1); consequently, it still unclear what role, if any, movements in 
the NTD may fulfill. An attractive possibility is that nanoscale mobility of the NTD controls trans-
synaptic contact formation at glutamatergic synapses (Elegheert et al., 2016; Garcia-Nafria et 
 26 
al., 2016), permitting AMPAR trafficking during synapse strengthening (Diaz-Alonso et al., 2017; 
Watson et al., 2017) as recently proposed for GluA1 and GluA2 subunits. Given that most native 
AMPARs are either GluA1/A2 or GluA2/A3 heteromers (Bowie, 2012; Henley and Wilkinson, 
2016; Jacobi and von Engelhardt, 2017), it will be interesting in future studies to determine how 
the flip/flop cassette contributes to receptor trafficking and synapse strengthening. 
 
On a broader perspective, the unappreciated role of the apo or resting state may have far 
reaching implications for our understanding of the inner workings of many types 
of signaling proteins, such as other ion channel families, G-protein coupled-receptors (GPCRs), 
transporters and kinases. For example, alternative splicing also dramatically impacts the 
signaling properties of other iGluRs (Regan et al., 2018), as well as many other ligand- and 
voltage-gated ion channels (Catterall et al., 2005; Kadowaki, 2015; Latorre et al., 2017; 
Lipscombe and Andrade, 2015; Soreq, 2015). Whether it can also explain the multiplicity of 
drug action on signaling proteins, such as GPCR biased agonism and modulation (Lane et al., 
2017) also awaits future investigation. 
 
 
 
 
 
 
 
  
 27 
Acknowledgments 
We would like to thank technician Heidi Peterson for contributing to expression and 
purification of protein for crystallization studies, and MAX-lab, Lund, Sweden for providing 
beamtime and help during data collections. We would also like to thank members of the Bowie 
lab for the critical reading and comments on the manuscript. 
 
Funding: This work was supported by operating grants from the CIHR (D. Bowie), Lundbeck 
Foundation, Danscatt, and Biostruct-X (J. Kastrup), the Biotechnology and Biological Sciences 
Research Council (J.M. Edwardson), and Newton-Picarte grant DPI-20140080 (N. Barrera and 
J.M. Edwardson). A. Perozzo and Y. Yan were supported by NSERC and FRQS Masters 
fellowships, respectively and G. Dawe and R. Alexander were supported by NSERC Doctoral 
fellowships. M.F. Kadir was supported by a scholarship from the Islamic Development Bank. E.A. 
Santander was supported by a CONICYT-Chile Cambridge Scholarship and Raminta 
Venskutonytė was supported by the Lundbeck Foundation. 
 
Author contributions: In the Bowie lab, GBD, AMP, YY, RPDA and MA carried out the 
electrophysiology experiments, made figures and analyzed the data. GBD, AMP, YY and RPDA 
also contributed to the writing of the manuscript. MRPA generated the constructs for 
electrophysiology and AFM, made figures and contributed to the discussions of the manuscript. 
DB conceived of the study, coordinated the experiments between the different labs and wrote 
the manuscript. In the Edwardson and Barrera labs, MFK, CN, CF and EAS carried out the AFM 
experiments and analyzed data. NPB designed the AFM experiments and analyzed data. JME 
designed the AFM experiments, analyzed data and contributed to writing the manuscript. In the 
Kastrup lab, RV carried out crystallization of GluA2i and GluA2o, collected diffraction data, 
refined structures, analyzed the structures, made figures, and contributed to writing of the 
manuscript. KF was involved in initial crystallization experiments, data collection and 
refinements of GluA2o. JSK hypothesized the existence of an anion binding site in AMPARs, 
participated in data collections, performed the final refinements of the structures, analyzed the 
structures, made figures, participated in writing of the manuscript, and generally supervised the 
work.  
 
Declaration of Interests: The authors declare no competing interests.  
 28 
MAIN FIGURE TITLES AND LEGENDS 
Figure 1. External anions selectively modulate AMPA receptor desensitization. (A) Cryo-EM 
structure of the GluA2-TARP γ2 (pale pink) receptor complex (PDB: 5KBU) in an antagonist-
bound form. (B) Side view of the GluK2 (left, PDB: 3G3F) and GluA2 (right, PDB: 4IGT) LBD 
dimer, depicting the binding pockets for two Na+ ions (purple) and one Cl- ion (green) in GluK2 
and two Li+ ions (magenta) in GluA2. (C-D) Typical current responses of GluA2i receptors to a 
250 ms (C, patch 140228p6) or 1 ms (D, patch 150825p10) application of 10 mM L-Glu in 
external NaCl (black), NaBr (grey), and NaF (light grey). Inset: responses scaled to compare 
decay kinetics. The uppermost trace (black) shows the junction current recorded after the 
experiment to monitor the solution exchange rate. (E) Recovery from desensitization for GluA2i 
receptors (patch 151201p9) in external NaCl (black) and NaI (orange). (F-G) Mean time 
constants of current decay after 250 ms (F, τdesensitization) or 1 ms (G, τdeactivation) L-Glu applications 
plotted against ionic radius. Data are mean ± SEM, from 7-15 (F) or 5-12 (G) independent patch 
experiments. (H) Recovery from desensitization experiments in different external anion 
solutions. Data are mean ± SEM, from 6 (NaI), 7 (NaBr), or 13 (NaCl) independent patch 
experiments. See also Table S1. 
 
Figure 2. Detection of bromide ions in the GluA2-LBD dimer interface. (A-B) Side (A) and top 
(B) views of the GluA2o-LBD N775S dimer. Bromide ions are shown in the dimer interface as 
brown spheres and glutamate with carbon atoms as green spheres. Nitrogen atoms are blue 
and oxygen atoms are red. (C-D) Anomalous difference electron density map (black; contoured 
at 7σ; before introduction of bromide ions in the structure) and Fo-Fc difference map (green; 
contoured at 3σ) from omitting Ser775/Asn775, bromide ions and water molecules within 4 Å 
of bromide from GluA2o-LBD N775S (C) and GluA2o-LBD (chain A) (D). (E-F) Magnified side (E) 
and top (F) views of the bromide binding sites in the GluA2o-LBD N775S dimer interface. Water 
molecules are shown as grey spheres and amino acid residues surrounding the binding sites are 
orange or cyan sticks based on their subunit of origin. See also Figure S1 and Table S2.  
 
Figure 3. Ser/Asn 775 residue of the flip/flop cassette governs anion modulation of AMPARs. 
(A) Sequence alignment of the GluA1 and GluA2 AMPAR flip/flop cassette, located toward the 
C-terminal end of the LBD. Positions that differ in both subunits are shaded grey, while residues 
are colored by chemical property (blue, positive charge; red, negative charge; pink, small polar; 
purple, large polar; green, hydrophobic). (B) GluA2o-LBD dimer with residues Asn765, Ala766 
and Asn775 shown as yellow spheres and the two bromide ions as brown spheres. The flip/flop 
cassette is orange in one subunit and cyan in the other subunit. (C-E) Typical current responses 
(250 ms, 10 mM L-Glu) of GluA2i (C, patch 151123p15), GluA2o (D, patch 160218p14), and 
GluA2i S775N (E, patch 160119p3) receptors in external NaCl (black) and NaI (colored trace). 
Inset: scaled responses in external NaI (colored trace), NaBr (grey), NaCl (black), and NaF (light 
grey). The uppermost trace (black) shows the junction current recorded after the experiment to 
monitor solution exchange rate. (F) Desensitization time constants for data shown in panels C-
E, plotted against different halide ions. Data are mean ± SEM from 6-15 independent patch 
experiments. Prop refers to propionate. (G) Recovery from desensitization of GluA2i S775N 
 29 
receptors (patch 160401p15) in external NaCl (black) and NaI (green). (H) Mean time constants 
of recovery from desensitization for GluA2i (orange), GluA2o (blue), and GluA2i S775N (green) in 
different external anions. Data are mean ± SEM from 5-32 independent patch experiments. See 
also Figures S2 and S3 and Table S1.   
 
Figure 4. Anions control the resting and active quaternary structure of the AMPA receptor 
through the LBD. (A) Top: Representative AFM images of a bilayer containing GluA2i receptors 
before (left) and after (right) application of L-Glu (10 mM). Scale bar, 100 nm; color-height 
scale, 0-10 nm. Bottom: Sections through the receptor at the position indicated by the white 
line above. (B) Top: Representative AFM images of a bilayer containing GluA2i receptors before 
(left) and after (right) a switch from NaCl to NaI. Scale bar, 100 nm; color-height scale, 0-10 nm. 
Bottom: Sections through the receptor at the position indicated by the white line above. (C) 
Average height changes of GluA2i, GluA2o and GluA2i S775N in response to L-Glu. Data are 
mean ± SEM for 10-14 receptors. (D) Average height changes of GluA2i, GluA2o and GluA2i 
S775N in response to anion switches. Data are mean ± SEM for 12-14 receptors. (E-G) Mean 
height changes of GluA2i ΔNTD (E), A2i ΔNTD-eGFP (F), and A2o ΔNTD-eGFP (G) receptors in 
response to anion substitution as well as 10 mM L-Glu application in different external anions. 
Data are mean ± SEM for 10-22 receptors. Cartoon uses GFP (PDB: 1GFL) and AMPAR LBD (PDB: 
1FTJ) structures for illustrative purpose only. See also Figures S4, S5 and S6 and Table S3.  
 
Figure 5. Mobility of the resting state of GluA2 AMPARs is variable between flip and flop 
isoforms. (A) Galleries of zoomed (120 x 120 nm) AFM images of individual flip (left) and flop 
(right) AMPARs. Scale bar, 20 nm; color-height scale 0-8 nm. (B) Representative cumulative 
second-by-second (up to 28 s) mobility data for individual flip, flop and flip S775N AMPARs in 
the resting state. (C) Representative cumulative second-by-second (up to 28 s) mobility data for 
individual flip AMPARs in the resting state and in the presence of L-Glu (10 mM). (D) Combined 
mobility data (at 28 s) for flip and flop AMPARs in the resting state, in the presence of L-Glu and 
in the presence of L-Glu plus CNQX (0.5 mM). Asterisks indicate significant differences (p < 0.05) 
between groups (one-way ANOVA, Fisher test). NS, not significant. (E) Combined mobility data 
(at 28 s) for flip and flop AMPARs in the resting state and in the presence of different 
concentrations of L-Glu. Asterisks indicate significant differences (p < 0.05) between flip and 
flop receptors under equivalent conditions (Mann-Whitney U-test). CSD refers to cumulative 
squared displacement. See also Table S4 and Movies S1-S4. 
 
Figure 6. The flop cassette acts as a master-switch to override regulation of AMPARs by TARP 
auxiliary subunits. (A) Typical current responses of GluA2i (patch 171120p1), GluA2o (patch 
160324p7) and GluA2i S775N (patch 180813p9) receptors to a 250 ms application of 10 mM L-
Glu when co-expressed with 2 (colored trace). The AMPAR responses in the absence of 2 
were taken from patch numbers 180301p3 (GluA2i), 160218p14 (GluA2o) and 160119p3 (GluA2i 
S775N). The uppermost trace (black) shows the junction current recorded after the experiment 
to monitor solution exchange rate. (B) Typical current responses in different external anions of 
wildtype and mutant GluA2i AMPARs co-assembled with 2. (GluA2i, patch 151214p3; GluA2o, 
patch 160324p7; GluA2i S775N, patch 180813p9). (C) Mean equilibrium current amplitude as a 
 30 
percentage of the peak response. (D) Mean time constants of current decay in the continued 
presence of L-Glu. (E) GluA2 LBD dimer highlighting the flip/flop cassette in orange (front) and 
cyan (back). Although different residues are responsible for the functional differences between 
flip and flop GluA2 isoforms in terms of channel gating (residues 765, 766 and 775), allosteric 
regulation by anions and CTZ (residue 775) and ER exit (residues 775 and 779), the 775 residue 
is implicated in all of these regulatory functions. See also Table S1. 
 
Figure 7.  Native AMPARs are modulated by external anions.  (A) Typical current responses of 
an excised Purkinje cell membrane patch (left, patch 190129p5) or stellate cell nucleated patch 
(right, patch 190205p4) to a 250 ms (black) or 1 ms (grey) application of 10 mM L-Glu. Insets 
show non-desensitizing responses in the presence of 100 µM CTZ (blue) (Purkinje, patch 
190214p8; stellate, patch 190129p3). (B) Mean equilibrium current amplitude as a percentage 
of peak response (Iequilibrium). (***p < 0.0001, unpaired t-test) (C) Mean weighted time constants 
of current decay following 250 ms L-Glu application (τdes). (***p < 0.0001, Mann-Whitney U 
test). (D) Normalized current responses to 250 ms application of L-Glu in NaCl (black), NaBr 
(dark grey) or NaF (light grey) in Purkinje (left, patch 190213p4) and stellate cell (right, patch 
190205p4) patches. (E) Mean weighted τdes across anion conditions in Purkinje (black squares) 
and stellate cell (black circles) patches. Data are presented as mean ± SEM. See also Table S5. 
 
Figure 8. The functional behavior of AMPAR heteromers is shaped by TARP stoichiometry. (A) 
Typical current responses of heteromeric AMPARs to a 250 ms application of 5 mM L-Glu in 
tandem with TARP 2 when indicated. Example traces are in the following sequence from left to 
right: GluA1i/2+GluA2i/2 (patch 190215p2), GluA1o/2+GluA2i/2 (patch 190215p13), 
GluA1i/2+GluA2o/2 (patch 190228p2), GluA1o/2+GluA2o/2 (patch 190301p2), 
GluA1o/2+GluA2i (patch 190213p3), and GluA1o+GluA2i/2 (patch 190204p11). All GluA2i/o 
plasmids are Q/R edited. (B) Mean equilibrium current amplitude as a percentage of the peak 
response. (C) Mean time constants of current decay in the continued presence of L-Glu. Data 
are mean  SEM with values of individual patches plotted as white circles. (D) Overlay of 
current responses of heteromeric AMPARs co-expressed with 4 TARPs (A, left) on a shorter time 
scale. (E) Left: typical current responses of GluA1i/2+GluA2i/2 (patch 190215p2) and 
GluA1o/2+GluA2o/2 (patch 190301p1) in external NaCl (black), NaBr (dark gray) or NaF (light 
grey). Right: summary of des of GluA1i/2+GluA2i/2 (orange circles), GluA1o/2+GluA2i/2 
(white circles), GluA1i/2+GluA2o/2 (white squares) and GluA1o/2+GluA2o/2 (blue squares) in 
external NaCl, NaBr and NaF. (F-G) Mean equilibrium current percentage plotted against mean 
time constants of desensitization. In (F), GluA1+GluA2 combinations in tandem with 4 TARPs 
(black circles) are included with pulled patches from cerebellar Purkinje cells (red circles). In (G), 
GluA1+GluA2 combinations in tandem with 2 TARPs (black circles) are included with pulled 
patches from cerebellar stellate cells (cyan circles). Data were fit by linear regression: (F) black, 
y = 1.76 - 0.30, r = 0.991; red, y = 1.70 - 2.23, r = 0.948; p = 0.10, one-way ANOVA; (G) black, y = 
0.36 - 0.18, r = 0.738; cyan, y = 0.36 + 1.15, r = 0.614; p = 0.20, one-way ANOVA. See also Table 
S5. 
  
 31 
STAR METHODS 
 
CONTACT FOR REAGENT AND RESOURCE SHARING 
Further information and requests for resources and reagents should be directed to and will be 
fulfilled by the Lead Contact, Derek Bowie (derek.bowie@mcgill.ca). 
 
EXPERIMENTAL MODEL AND SUBJECT DETAILS 
Cell Culture 
Electrophysiology experiments: human embryonic kidney cells (HEK293T/17) were purchased 
from ATCC (CRL-11268). These cells constitutively express the simian virus 40 (SV40) large T-
antigen and 17 refers to the clone number selected for its high transfectability. Cells were 
grown at 37°C under 5% CO2 in Minimum Essential Medium with GlutaMAX (i.e. MEM 
GlutaMAX) supplemented with 10% fetal bovine serum. The sex of the cell line is not 
determined.   
AFM experiments: tsA201 cells (a subclone of HEK293 cells stably expressing the SV40 large T-
antigen, see (Suzuki et al., 2013)) were grown at 37°C under 5% CO2 in Dulbecco’s Modified 
Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum, 100 μg/mL streptomycin 
and 100 units/mL penicillin. The sex of the cell line is not determined. 
Mice 
All experiments have been approved by the local authorities, were performed in accordance 
with the guidelines of the Canadian Council on Animal Care and were approved by the Animal 
Care Committee of McGill University. Wild-type mice with a C57BL/6J background were 
obtained from Jackson Laboratories (Bar Harbor, USA) and maintained as a breeding colony at 
McGill University. Both male and female wild-type mice were used for experiments and ranged 
from postnatal days 18 to 25.  
 
 
  
 32 
METHOD DETAILS 
 
Recombinant Electrophysiology 
 
HEK293 cells were used to express recombinant GluA1 and/or GluA2 AMPAR subunits for 
outside-out patch recordings. For GluA2 homomers, the Q/R unedited flip and flop isoforms 
(GluA2Qi and GluA2Qo) were used. For GluA1/A2 heteromers, the Q/R edited flip and flop 
isoforms of GluA2 (GluA2Ri and GluA2Ro) were used. Residue numbering includes the signal 
peptide. Mutant receptors were generated using site-directed mutagenesis. External and 
internal recording solutions typically contained (in mM): 150 NaX (X = halide ion), 5 HEPES, 0.1 
CaCl2, 0.1 MgCl2, and 2% phenol red at pH 7.4, and 115 NaCl, 10 NaF, 5 HEPES, 5 Na4BAPTA, 0.5 
CaCl2, 1 MgCl2, and 10 Na2ATP at pH 7.4, respectively. For GluA1/A2 heteromer recordings, 30 
µM spermine was included in the internal solution; data points were only included when the I/V 
plot was linear, typical of GluA2(R)-containing AMPARs (Partin et al., 1995) (see Fig. S7). Sucrose 
was supplemented to maintain the osmotic pressure at 300 mOsm. L-Glu was typically applied 
at 10 mM and CTZ at 100 µM, unless otherwise indicated.  
 
Recording pipettes were composed of borosilicate glass (3-5 MΩ, King Precision Glass, Inc.) 
coated with dental wax. The reference electrode was connected to the bath via an agar bridge 
of 3 M KCl. Agonist solutions were applied using a piezo-stack driven perfusion system, and 
measured solution exchange time was under 400 µs. Series resistances (3-15 MΩ) were 
routinely compensated by 95%. All recordings were performed using an Axopatch 200B 
amplifier (Molecular Devices, LLC). Current records were low-pass filtered by an 8-pole Bessel 
filter at 10 kHz and sampled at 25-50 kHz. Data were acquired using pClamp9 software 
(Molecular Devices, LLC) and illustrated using Origin 7 (OriginLab Corp.). 
 
Slice Electrophysiology 
 
Slice preparation 
Mice were anesthetized with isoflurane and immediately decapitated. A block of cerebellar 
vermis was rapidly dissected from the mouse head and submerged in ice-cold cutting solution 
perfused with carbogen gas (95% O2, 5% CO2). Cutting solution contains (in mM): 235 sucrose, 
2.5 KCl, 1.25 NaH2PO4, 28 NaHCO3, 0.5 CaCl2, 7 MgCl2, 28 D-glucose, 1 ascorbic acid, 3 sodium 
pyruvate (pH 7.4; 305–315 mOsmol/L). The block of vermis is then fastened to a platform, 
transferred to the slicing chamber and again submerged in ice-cold cutting solution, bubbled 
with carbogen throughout the remainder of the procedure. Thin slices of cerebellar vermis (300 
µm) were obtained with a vibrating tissue sectioner (Leica VT1200; Leica Instruments, Nussloch, 
Germany). The slices were transferred to oxygenated artificial cerebrospinal fluid (aCSF) and 
held at room temperature (21°C-23°C) for at least 1 h before recordings were performed. aCSF 
contained the following (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 2 CaCl2, 1 MgCl2, 
25 D-glucose (pH of 7.4; 305–315 mOsmol/L). 
 
 
 33 
Acute slice electrophysiology 
Slice experiments were performed on an Olympus BX51 upright microscope (Olympus, Southall, 
UK) equipped with differential interference contrast/infrared optics. Recordings were made 
from either visually-identified stellate or Purkinje cells in acute sagittal slices of cerebellar 
vermis. Patch pipettes were prepared from thick-walled borosilicate glass (GC150F-10, OD 1.5 
mm, ID 0.86 mm; Harvard Apparatus Ltd, Kent, UK) and had open tip resistances of 3–6 MΩ 
when filled with an intracellular recording solution. Internal solution contained (in mM): 140 
CsCl, 10 HEPES, 10 EGTA, 2 MgCl2 and 60 µM spermine-HCl to examine rectification due to 
polyamine channel block (pH of 7.4; 295-305 mOsmol/L). Local agonist/antagonist applications 
were performed using a homemade flowpipe from theta tubing with a tip diameter of 300–400 
μm. External solution was the same as described above with the addition of 10 μM D-APV to 
block NMDA receptors. Nucleated (stellate) or excised membrane (Purkinje) patches were 
placed near the mouth of a double-barreled flowpipe, which was rapidly jumped between 
control and solution containing 10 mM L-Glu (1 - 250 ms duration). Recordings were performed 
using a Multiclamp 700A amplifier (Molecular Devices, Sunnyvale, CA, USA). The bath was 
continuously perfused at room temperature (21–23 °C) with aCSF at a rate of 1–2 mL/min. 
Currents were filtered at 5 kHz with an eight-pole low-pass Bessel filter (Frequency Devices, 
Haverhill, MA, USA) and digitized at 25 kHz with a Digidata 1322A data acquisition board and 
Clampex 10.1 (pClamp) software. 
 
Crystallization 
 
Wildtype GluA2o and N775S (flip-like) mutant ligand binding domains (LBDs) were expressed 
and purified as described previously (Krintel et al., 2012). Crystallization was performed using 
the vapor diffusion hanging drop method at 6oC. The crystallization drop consisted of 1 μl 
GluA2o-LBD solution (8 mg/ml) or GluA2o-LBD N775S mutant (4 mg/ml) in a buffer containing 10 
mM HEPES, 20 mM NaCl, 1 mM EDTA (pH 7.0), and 1 μl of reservoir solution. Before setting up 
the crystallization drops, the protein solution was mixed with L-Glu to a final concentration of 2 
mM L-Glu and 300 mM NaBr (GluA2o-LBD) or 4 mM L-Glu and 250 mM RbBr (GluA2o-LBD 
N775S). Crystals used for diffraction data collections were obtained at conditions consisting of 
reservoir solution: 25% PEG4000, 0.2 M Na2SO4, and 0.1 M CH3COONa, pH 5.5 (GluA2o) or 22-
24% PEG4000, 0.2-0.3 M Li2SO4 and 0.1 M cacodylate, pH 6.5 (GluA2o-LBD N775S). Before data 
collection, the crystals were cryo-protected in reservoir solution containing 20% glycerol. 
 
Data collection and structure determination 
X-ray diffraction data on GluA2-LBD crystals were collected at the Max-Lab beamline I911-3 
(Lund, Sweden) (Ursby et al., 2013) at 100 K. Diffraction images were processed in XDS (Kabsch, 
2010). Data were scaled and merged using SCALA (Evans, 2006) within CCP4 (Winn et al., 2011) 
and the structures were solved by molecular replacement in Phaser (McCoy et al., 2007) using 
GluA2-LBD structures as search models (PDB: 3TDJ, molA for GluA2o, (Krintel et al., 2012) and 
PDB: 4O3A, molA for GluA2o-LBD N775S, (Krintel et al., 2014)). Initially, the structures were 
rebuilt in AutoBuild (Terwilliger et al., 2008) within Phenix (Adams et al., 2010). Structures were 
further improved using Coot (Emsley et al., 2010) and refinement in Phenix. Both structures 
displayed good quality indicators as calculated by MolProbity (Chen et al., 2010) within Phenix. 
 34 
Figures were prepared with the PyMOL Molecular Graphics System (Version 1.7.4, Schrödinger, 
LLC). For statistics on data collection and refinements, see Table S2. 
 
AMPA receptor purification and AFM imaging 
 
The following constructs were used, all with an HA tag at the N-terminus and in the vector 
pRK5: rat GluA2i and GluA2o (Q/R site unedited), GluA2i with the point mutation S775N, NTD-
GluA2i (minus NTD), NTD GluA2i-GFP and NTD GluA2o-GFP. The HA tag contained the 
residues YPYDVPDYA, located after the first amino acid following the signal peptide (i.e. 
between residues 22 and 32). 
 
Isolation of AMPARs 
DNA (250 μg) was used to transfect 5 x 162 cm2 flasks of tsA201 cells using polyethylenimine. 
After transfection, cells were incubated for 24-48 h at 37°C to allow expression of receptors. 
Proteins were isolated from transfected cells by immunoaffinity chromatography; all steps were 
carried out at 4°C. Cell pellets were resuspended in solubilization buffer [25 mM Tris-HCl, pH 
7.5, 150 mM NaCl, 10 mM EDTA, 1% Triton X-100, 1 mM PMSF, and protease inhibitor cocktail 
(Roche), prepared in Biotechnology Performance Certified (BPC) water (Sigma)]. Precipitated 
DNA was removed by low-speed centrifugation, and the sample was then centrifuged in a 70 Ti 
rotor at 35,000 rpm for 1 h. Solubilized extracts were incubated with anti-HA-agarose beads 
(Sigma) for 3 h. The immunobeads bearing captured protein were then washed extensively with 
solubilization buffer containing 0.1% CHAPS (Sigma) instead of 1% (v/v) Triton X-100, and the 
bound protein was eluted with HA peptide (200 μg/mL). Protein purity was evaluated by silver 
staining and immunoblotting. 
 
Integration of receptors into liposomes 
Chloroform solutions of L-α-phosphatidylcholine (PC) and 1,2-dioleoyl-sn-glycero-3-phospho-L-
serine (DOPS; Avanti Polar Lipids) were mixed at a molar ratio of 3:1. The chloroform was then 
evaporated under a stream of nitrogen gas, and the lipids were resuspended in HEPES-buffered 
saline (HBS; 150 mM NaCl, 20 mM HEPES, pH 7.6) containing CHAPS and mixed with purified 
receptor to give a final lipid concentration of 2 mg/mL and a final CHAPS concentration of 1% 
(w/v). The mixture was dialysed at 4°C against detergent-free buffer for 2 days, with several 
changes of buffer. To detect receptors, where appropriate, the dialysed sample was incubated 
for 12 h at 4°C with anti-HA antibody. 
 
AFM imaging 
A droplet (20 μL) of proteoliposome suspension was deposited onto the surface of a freshly 
cleaved mica disc (diameter, 1 cm), followed by incubation for 5 min at room temperature 
(20°C), during which time the proteoliposomes collapsed to form supported lipid bilayers 
containing integrated receptors. The mica surface was gently washed several times with HBS to 
remove unadsorbed proteoliposomes. AFM imaging under fluid was carried out at room 
temperature using a Bruker-AXS FastScan Dimension AFM instrument. The instrument was used 
in micro-volume fluid mode to facilitate the application of agonist, antagonist or ions directly 
while imaging. All images were collected in ‘tapping’ mode, using FastScan-D silicon probes 
 35 
(Bruker). The cantilevers (with a typical spring constant of 0.25 N/m) were tuned to a resonance 
frequency of between 90 and 140 kHz. The microscope was engaged with a 2 μm x 2 m scan 
area and a 5 nm target amplitude to allow for tuning. The amplitude setpoint was adjusted to 
the highest setting that allowed imaging with little noise, to minimize the force applied to the 
sample. To measure receptor heights, images were captured at a scan rate of 20 Hz (25 
seconds/frame) with 512 scan lines per area. Individual particles were identified, and particles 
with heights between 5 and 10 nm were taken to represent AMPARs. 
 
To follow the dynamics of the NTDs, sequential high-magnification (120 nm x 120 nm) images of 
receptor-containing bilayers were captured at a frequency of 1 frame/second with a fixed 
integral gain of 2.5 and a target amplitude of 1 nm. The target amplitude was kept at 1 nm to 
exert minimum force on the receptor. Individual particles were identified, and particles with 
heights between 7 and 10 nm were taken to represent AMPARs. N.B. The particles were slightly 
taller when using target amplitude 1 nm instead of the conventional target amplitude of 5 nm. 
Movement of the two globular structures relative to each other was followed under various 
conditions. The centers of the structures were identified by using Gaussian fitting (ImageJ plug-
in, Adrian’s FWHM), and NTD mobility was expressed as the cumulative squared displacement 
(CSD): 
𝐶𝑆𝐷(𝑡) =  ∑(𝑥𝑖+1 − 𝑥𝑖)
2
𝑡
𝑖=1
 
 
where x is the distance between the centers of the globular structures. 
 
Data analysis 
Image analysis was performed using the Nanoscope analysis 1.5 software and ImageJ 
(Schneider et al., 2012). Data analysis was carried out using Microsoft Excel, OriginPro 8.5 or 
SigmaPlot 12.5. Histograms were drawn with bin widths chosen according to Scott’s equation:  
 
ℎ = 3.5𝜎 / 𝑛1/3 
 
where h is the bin width, σ is an estimate of the standard deviation and n is the sample size. 
 
  
 36 
QUANTIFICATION AND STATISTICAL ANALYSIS 
Additional details of data analysis and statistical analysis can be found in the Method Details, 
main and supplemental figures, supplemental tables, and corresponding legends. 
Electrophysiological data 
Electrophysiological data were analyzed using Clampfit 10.5 (Molecular Devices, LLC). To 
measure deactivation and entry into desensitization, current decay rates were fitted using 1st 
or 2nd order exponential functions of the form y = Ai*exp(-x/τi). Where two exponential 
components were used, time constants are expressed as a weighted mean. To measure 
recovery from desensitization, a two-pulse protocol was delivered using variable interpulse 
intervals, and the peak amplitude of the second pulse was expressed as a fraction of the first 
peak. Recovery data were fitted with the Hodgkin-Huxley equation y = N0+(1-N0)*(1-exp(-
krec*x))^n, where N0 is the equilibrium response at the end of the first pulse, krec is the recovery 
rate, and n is an exponent that reflects the number of kinetic transitions contributing to the 
recovery time course. The value of n was set to 2 (see (Robert et al., 2005)). Distributions were 
first evaluated using Kolmogorov-Smirnova test. Non-normally distributed data were analyzed 
using non-parametric tests like the Mann-Whitney U test where indicated. Normally distributed 
data were compared using one-way ANOVA and unpaired two-tailed Student’s t-test. 
Significance is defined as p < 0.05. Data are presented as mean ± SEM, with n referring to 
individual patches. 
Atomic force microscopy data 
Data are presented as mean ± SEM, with n referring to individual receptors. Heights of 
individual AMPARs before and after either the addition of L-Glu or an anion switch were 
compared using a Student’s paired, two-tailed t-test. ATD mobility data for flip and flop 
AMPARs in the resting state, in the presence of L-Glu, and in the presence of L-Glu plus CNQX 
were compared using a one-way ANOVA, with a Fisher test. Combined ATD mobility data for flip 
and flop AMPARs in the resting state and in the presence of different concentrations of L-Glu 
were compared using a Mann-Whitney U test. 
 
DATA AND SOFTWARE AVAILABILITY 
The X-ray crystal structures of GluA2o-LBD (PDB code, 6GL4) and the flip-like mutant GluA2o-
LBD N775S (PDB code, 6GIV) have been uploaded to the Protein Data Bank. 
 37 
Supplemental Movie Titles 
 
Videos showing NTD movements in individual receptors. Images were captured at 1 frame/s 
and are played back at 4 frames/s. 
Movie S1. Flip receptor (resting state), Related to Figure 5. 
Movie S2. Flop receptor (resting state), Related to Figure 5. 
Movie S3. Flip receptor (10 mM L-Glu), Related to Figure 5. 
Movie S4. Flop receptor (10 mM L-Glu), Related to Figure 5. 
 
  
 38 
References 
 
Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., 
Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010). PHENIX: a comprehensive Python-based system for 
macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66, 213-221. 
Balasuriya, D., Takahashi, H., Srivats, S., and Edwardson, J.M. (2014). Activation-induced structural 
change in the GluN1/GluN3A excitatory glycine receptor. Biochem Biophys Res Commun 450, 1452-
1457. 
Barbour, B., Keller, B.U., Llano, I., and Marty, A. (1994). Prolonged presence of glutamate during 
excitatory synaptic transmission to cerebellar Purkinje cells. Neuron 12, 1331-1343. 
Bats, C., Soto, D., Studniarczyk, D., Farrant, M., and Cull-Candy, S.G. (2012). Channel properties reveal 
differential expression of TARPed and TARPless AMPARs in stargazer neurons. Nat Neurosci 15, 853-861. 
Bowie, D. (2002). External anions and cations distinguish between AMPA and kainate receptor gating 
mechanisms. J Physiol 539, 725-733. 
Bowie, D. (2008). Ionotropic glutamate receptors & CNS disorders. CNS Neurol Disord Drug Targets 7, 
129-143. 
Bowie, D. (2010). Ion-dependent gating of kainate receptors. J Physiol 588, 67-81. 
Bowie, D. (2012). Redefining the classification of AMPA-selective ionotropic glutamate receptors. J 
Physiol 590, 49-61. 
Cais, O., Herguedas, B., Krol, K., Cull-Candy, S.G., Farrant, M., and Greger, I.H. (2014). Mapping the 
interaction sites between AMPA receptors and TARPs reveals a role for the receptor N-terminal domain 
in channel gating. Cell Rep 9, 728-740. 
Catterall, W.A., Goldin, A.L., and Waxman, S.G. (2005). International Union of Pharmacology. XLVII. 
Nomenclature and structure-function relationships of voltage-gated sodium channels. Pharmacol Rev 
57, 397-409. 
Chen, S., and Gouaux, E. (2019). Structure and mechanism of AMPA receptor - auxiliary protein 
complexes. Curr Opin Struct Biol 54, 104-111. 
Chen, V.B., Arendall, W.B., 3rd, Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J., Murray, L.W., 
Richardson, J.S., and Richardson, D.C. (2010). MolProbity: all-atom structure validation for 
macromolecular crystallography. Acta Crystallogr D Biol Crystallogr 66, 12-21. 
Coleman, S.K., Moykkynen, T., Cai, C., von Ossowski, L., Kuismanen, E., Korpi, E.R., and Keinanen, K. 
(2006). Isoform-specific early trafficking of AMPA receptor flip and flop variants. J Neurosci 26, 11220-
11229. 
Dawe, G.B., Aurousseau, M.R., Daniels, B.A., and Bowie, D. (2015). Retour aux sources: defining the 
structural basis of glutamate receptor activation. J Physiol 593, 97-110. 
Dawe, G.B., Musgaard, M., Andrews, E.D., Daniels, B.A., Aurousseau, M.R., Biggin, P.C., and Bowie, D. 
(2013). Defining the structural relationship between kainate-receptor deactivation and desensitization. 
Nat Struct Mol Biol 20, 1054-1061. 
Dawe, G.B., Musgaard, M., Aurousseau, M.R., Nayeem, N., Green, T., Biggin, P.C., and Bowie, D. (2016). 
Distinct Structural Pathways Coordinate the Activation of AMPA Receptor-Auxiliary Subunit Complexes. 
Neuron 89, 1264-1276. 
Diaz-Alonso, J., Sun, Y.J., Granger, A.J., Levy, J.M., Blankenship, S.M., and Nicoll, R.A. (2017). Subunit-
specific role for the amino-terminal domain of AMPA receptors in synaptic targeting. Proc Natl Acad Sci 
U S A 114, 7136-7141. 
Dingledine, R., Borges, K., Bowie, D., and Traynelis, S.F. (1999). The glutamate receptor ion channels. 
Pharmacol Rev 51, 7-61. 
 39 
Dürr, K.L., Chen, L., Stein, R.A., De Zorzi, R., Folea, I.M., Walz, T., McHaourab, H.S., and Gouaux, E. 
(2014). Structure and dynamics of AMPA receptor GluA2 in resting, pre-open, and desensitized states. 
Cell 158, 778-792. 
Dutta, A., Krieger, J., Lee, J.Y., Garcia-Nafria, J., Greger, I.H., and Bahar, I. (2015). Cooperative Dynamics 
of Intact AMPA and NMDA Glutamate Receptors: Similarities and Subfamily-Specific Differences. 
Structure 23, 1692-1704. 
Elegheert, J., Kakegawa, W., Clay, J.E., Shanks, N.F., Behiels, E., Matsuda, K., Kohda, K., Miura, E., 
Rossmann, M., Mitakidis, N., et al. (2016). Structural basis for integration of GluD receptors within 
synaptic organizer complexes. Science 353, 295-299. 
Emsley, P., Lohkamp, B., Scott, W.G., and Cowtan, K. (2010). Features and development of Coot. Acta 
Crystallogr D Biol Crystallogr 66, 486-501. 
Engel, A.G., Shen, X.M., Selcen, D., and Sine, S.M. (2010). What have we learned from the congenital 
myasthenic syndromes. J Mol Neurosci 40, 143-153. 
Evans, P. (2006). Scaling and assessment of data quality. Acta Crystallogr D Biol Crystallogr 62, 72-82. 
Garcia-Nafria, J., Herguedas, B., Watson, J.F., and Greger, I.H. (2016). The dynamic AMPA receptor 
extracellular region: a platform for synaptic protein interactions. J Physiol 594, 5449-5458. 
Geisheker, M.R., Heymann, G., Wang, T., Coe, B.P., Turner, T.N., Stessman, H.A.F., Hoekzema, K., 
Kvarnung, M., Shaw, M., Friend, K., et al. (2017). Hotspots of missense mutation identify 
neurodevelopmental disorder genes and functional domains. Nat Neurosci 20, 1043-1051. 
Gielen, M., Siegler Retchless, B., Mony, L., Johnson, J.W., and Paoletti, P. (2009). Mechanism of 
differential control of NMDA receptor activity by NR2 subunits. Nature 459, 703-707. 
Greger, I.H., Akamine, P., Khatri, L., and Ziff, E.B. (2006). Developmentally regulated, combinatorial RNA 
processing modulates AMPA receptor biogenesis. Neuron 51, 85-97. 
Greger, I.H., Watson, J.F., and Cull-Candy, S.G. (2017). Structural and Functional Architecture of AMPA-
Type Glutamate Receptors and Their Auxiliary Proteins. Neuron 94, 713-730. 
Hansen, K.B., Naur, P., Kurtkaya, N.L., Kristensen, A.S., Gajhede, M., Kastrup, J.S., and Traynelis, S.F. 
(2009). Modulation of the dimer interface at ionotropic glutamate-like receptor delta2 by D-serine and 
extracellular calcium. J Neurosci 29, 907-917. 
Hastie, P., Ulbrich, M.H., Wang, H.L., Arant, R.J., Lau, A.G., Zhang, Z., Isacoff, E.Y., and Chen, L. (2013). 
AMPA receptor/TARP stoichiometry visualized by single-molecule subunit counting. Proc Natl Acad Sci U 
S A 110, 5163-5168. 
Henley, J.M., and Wilkinson, K.A. (2016). Synaptic AMPA receptor composition in development, plasticity 
and disease. Nat Rev Neurosci 17, 337-350. 
Herguedas, B., Garcia-Nafria, J., Cais, O., Fernandez-Leiro, R., Krieger, J., Ho, H., and Greger, I.H. (2016). 
Structure and organization of heteromeric AMPA-type glutamate receptors. Science 352, aad3873. 
Herguedas, B., Watson, J.F., Ho, H., Cais, O., Garcia-Nafria, J., and Greger, I.H. (2019). Architecture of the 
heteromeric GluA1/2 AMPA receptor in complex with the auxiliary subunit TARP gamma8. Science. 
Herring, B.E., and Nicoll, R.A. (2016). Long-Term Potentiation: From CaMKII to AMPA Receptor 
Trafficking. Annu Rev Physiol 78, 351-365. 
Horning, M.S., and Mayer, M.L. (2004). Regulation of AMPA receptor gating by ligand binding core 
dimers. Neuron 41, 379-388. 
Jackson, A.C., and Nicoll, R.A. (2011). The expanding social network of ionotropic glutamate receptors: 
TARPs and other transmembrane auxiliary subunits. Neuron 70, 178-199. 
Jackson, M.B. (1984). Spontaneous openings of the acetylcholine receptor channel. Proc Natl Acad Sci U 
S A 81, 3901-3904. 
Jackson, M.B. (1986). Kinetics of unliganded acetylcholine receptor channel gating. Biophys J 49, 663-
672. 
 40 
Jacobi, E., and von Engelhardt, J. (2017). Diversity in AMPA receptor complexes in the brain. Curr Opin 
Neurobiol 45, 32-38. 
Jadey, S.V., Purohit, P., Bruhova, I., Gregg, T.M., and Auerbach, A. (2011). Design and control of 
acetylcholine receptor conformational change. Proc Natl Acad Sci U S A 108, 4328-4333. 
Jin, R., Clark, S., Weeks, A.M., Dudman, J.T., Gouaux, E., and Partin, K.M. (2005). Mechanism of positive 
allosteric modulators acting on AMPA receptors. J Neurosci 25, 9027-9036. 
Kabsch, W. (2010). Xds. Acta Crystallogr D Biol Crystallogr 66, 125-132. 
Kadowaki, T. (2015). Evolutionary dynamics of metazoan TRP channels. Pflugers Arch 467, 2043-2053. 
Kim, K.S., Yan, D., and Tomita, S. (2010). Assembly and stoichiometry of the AMPA receptor and 
transmembrane AMPA receptor regulatory protein complex. J Neurosci 30, 1064-1072. 
Klein, R.M., and Howe, J.R. (2004). Effects of the lurcher mutation on GluR1 desensitization and 
activation kinetics. J Neurosci 24, 4941-4951. 
Kohda, K., Wang, Y., and Yuzaki, M. (2000). Mutation of a glutamate receptor motif reveals its role in 
gating and delta2 receptor channel properties. Nat Neurosci 3, 315-322. 
Koike, M., Tsukada, S., Tsuzuki, K., Kijima, H., and Ozawa, S. (2000). Regulation of kinetic properties of 
GluR2 AMPA receptor channels by alternative splicing. J Neurosci 20, 2166-2174. 
Konig, P., Engel, A.K., and Singer, W. (1996). Integrator or coincidence detector? The role of the cortical 
neuron revisited. Trends Neurosci 19, 130-137. 
Krintel, C., Frydenvang, K., Ceravalls de Rabassa, A., Kaern, A.M., Gajhede, M., Pickering, D.S., and 
Kastrup, J.S. (2014). L-Asp is a useful tool in the purification of the ionotropic glutamate receptor A2 
ligand-binding domain. FEBS J 281, 2422-2430. 
Krintel, C., Frydenvang, K., Olsen, L., Kristensen, M.T., de Barrios, O., Naur, P., Francotte, P., Pirotte, B., 
Gajhede, M., and Kastrup, J.S. (2012). Thermodynamics and structural analysis of positive allosteric 
modulation of the ionotropic glutamate receptor GluA2. Biochem J 441, 173-178. 
Lane, J.R., May, L.T., Parton, R.G., Sexton, P.M., and Christopoulos, A. (2017). A kinetic view of GPCR 
allostery and biased agonism. Nat Chem Biol 13, 929-937. 
Latorre, R., Castillo, K., Carrasquel-Ursulaez, W., Sepulveda, R.V., Gonzalez-Nilo, F., Gonzalez, C., and 
Alvarez, O. (2017). Molecular Determinants of BK Channel Functional Diversity and Functioning. Physiol 
Rev 97, 39-87. 
Lipscombe, D., and Andrade, A. (2015). Calcium Channel CaValpha(1) Splice Isoforms - Tissue Specificity 
and Drug Action. Curr Mol Pharmacol 8, 22-31. 
Matsuda, K., Budisantoso, T., Mitakidis, N., Sugaya, Y., Miura, E., Kakegawa, W., Yamasaki, M., Konno, K., 
Uchigashima, M., Abe, M., et al. (2016). Transsynaptic Modulation of Kainate Receptor Functions by 
C1q-like Proteins. Neuron 90, 752-767. 
Mayer, M.L., and Armstrong, N. (2004). Structure and function of glutamate receptor ion channels. Annu 
Rev Physiol 66, 161-181. 
McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C., and Read, R.J. (2007). 
Phaser crystallographic software. J Appl Crystallogr 40, 658-674. 
Meyerson, J.R., Kumar, J., Chittori, S., Rao, P., Pierson, J., Bartesaghi, A., Mayer, M.L., and Subramaniam, 
S. (2014). Structural mechanism of glutamate receptor activation and desensitization. Nature 514, 328-
334. 
Monyer, H., Seeburg, P.H., and Wisden, W. (1991). Glutamate-operated channels: developmentally early 
and mature forms arise by alternative splicing. Neuron 6, 799-810. 
Mosbacher, J., Schoepfer, R., Monyer, H., Burnashev, N., Seeburg, P.H., and Ruppersberg, J.P. (1994). A 
molecular determinant for submillisecond desensitization in glutamate receptors. Science 266, 1059-
1062. 
 41 
Moykkynen, T., Coleman, S.K., Semenov, A., and Keinanen, K. (2014). The N-terminal domain modulates 
alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor desensitization. J Biol 
Chem 289, 13197-13205. 
Nakagawa, T., Cheng, Y., Ramm, E., Sheng, M., and Walz, T. (2005). Structure and different 
conformational states of native AMPA receptor complexes. Nature 433, 545-549. 
Partin, K.M., Bowie, D., and Mayer, M.L. (1995). Structural determinants of allosteric regulation in 
alternatively spliced AMPA receptors. Neuron 14, 833-843. 
Partin, K.M., Fleck, M.W., and Mayer, M.L. (1996). AMPA receptor flip/flop mutants affecting 
deactivation, desensitization, and modulation by cyclothiazide, aniracetam, and thiocyanate. J Neurosci 
16, 6634-6647. 
Partin, K.M., Patneau, D.K., and Mayer, M.L. (1994). Cyclothiazide differentially modulates 
desensitization of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor splice variants. 
Mol Pharmacol 46, 129-138. 
Penn, A.C., Balik, A., Wozny, C., Cais, O., and Greger, I.H. (2012). Activity-mediated AMPA receptor 
remodeling, driven by alternative splicing in the ligand-binding domain. Neuron 76, 503-510. 
Penn, A.C., Williams, S.R., and Greger, I.H. (2008). Gating motions underlie AMPA receptor secretion 
from the endoplasmic reticulum. EMBO J 27, 3056-3068. 
Plested, A.J., and Mayer, M.L. (2007). Structure and mechanism of kainate receptor modulation by 
anions. Neuron 53, 829-841. 
Purohit, P., and Auerbach, A. (2009). Unliganded gating of acetylcholine receptor channels. Proc Natl 
Acad Sci U S A 106, 115-120. 
Quirk, J.C., Siuda, E.R., and Nisenbaum, E.S. (2004). Molecular determinants responsible for differences 
in desensitization kinetics of AMPA receptor splice variants. J Neurosci 24, 11416-11420. 
Regan, M.C., Grant, T., McDaniel, M.J., Karakas, E., Zhang, J., Traynelis, S.F., Grigorieff, N., and Furukawa, 
H. (2018). Structural Mechanism of Functional Modulation by Gene Splicing in NMDA Receptors. Neuron 
98, 521-529. 
Robert, A., Armstrong, N., Gouaux, J.E., and Howe, J.R. (2005). AMPA receptor binding cleft mutations 
that alter affinity, efficacy, and recovery from desensitization. J Neurosci 25, 3752-3762. 
Schneider, C.A., Rasband, W.S., and Eliceiri, K.W. (2012). NIH Image to ImageJ: 25 years of image 
analysis. Nat Methods 9, 671-675. 
Schwarz, M.K., Pawlak, V., Osten, P., Mack, V., Seeburg, P.H., and Kohr, G. (2001). Dominance of the 
lurcher mutation in heteromeric kainate and AMPA receptor channels. Eur J Neurosci 14, 861-868. 
Sekiguchi, M., Fleck, M.W., Mayer, M.L., Takeo, J., Chiba, Y., Yamashita, S., and Wada, K. (1997). A novel 
allosteric potentiator of AMPA receptors: 4--2-(phenylsulfonylamino)ethylthio--2,6-difluoro-
phenoxyaceta mide. J Neurosci 17, 5760-5771. 
Shadlen, M.N., and Newsome, W.T. (1994). Noise, neural codes and cortical organization. Curr Opin 
Neurobiol 4, 569-579. 
Shaikh, S.A., Dolino, D.M., Lee, G., Chatterjee, S., MacLean, D.M., Flatebo, C., Landes, C.F., and 
Jayaraman, V. (2016). Stargazin Modulation of AMPA Receptors. Cell Rep 17, 328-335. 
Shepherd, J.D., and Huganir, R.L. (2007). The cell biology of synaptic plasticity: AMPA receptor 
trafficking. Annu Rev Cell Dev Biol 23, 613-643. 
Shi, Y., Lu, W., Milstein, A.D., and Nicoll, R.A. (2009). The stoichiometry of AMPA receptors and TARPs 
varies by neuronal cell type. Neuron 62, 633-640. 
Sobolevsky, A.I., Rosconi, M.P., and Gouaux, E. (2009). X-ray structure, symmetry and mechanism of an 
AMPA-subtype glutamate receptor. Nature 462, 745-756. 
Sommer, B., Keinanen, K., Verdoorn, T.A., Wisden, W., Burnashev, N., Herb, A., Kohler, M., Takagi, T., 
Sakmann, B., and Seeburg, P.H. (1990). Flip and flop: a cell-specific functional switch in glutamate-
operated channels of the CNS. Science 249, 1580-1585. 
 42 
Soreq, H. (2015). Checks and balances on cholinergic signaling in brain and body function. Trends 
Neurosci 38, 448-458. 
Sun, Y., Olson, R., Horning, M., Armstrong, N., Mayer, M., and Gouaux, E. (2002). Mechanism of 
glutamate receptor desensitization. Nature 417, 245-253. 
Suzuki, Y., Goetze, T.A., Stroebel, D., Balasuriya, D., Yoshimura, S.H., Henderson, R.M., Paoletti, P., 
Takeyasu, K., and Edwardson, J.M. (2013). Visualization of structural changes accompanying activation 
of N-methyl-D-aspartate (NMDA) receptors using fast-scan atomic force microscopy imaging. J Biol 
Chem 288, 778-784. 
Taverna, F., Xiong, Z.G., Brandes, L., Roder, J.C., Salter, M.W., and MacDonald, J.F. (2000). The Lurcher 
mutation of an alpha-amino-3-hydroxy-5-methyl- 4-isoxazolepropionic acid receptor subunit enhances 
potency of glutamate and converts an antagonist to an agonist. J Biol Chem 275, 8475-8479. 
Terwilliger, T.C., Grosse-Kunstleve, R.W., Afonine, P.V., Moriarty, N.W., Zwart, P.H., Hung, L.W., Read, 
R.J., and Adams, P.D. (2008). Iterative model building, structure refinement and density modification 
with the PHENIX AutoBuild wizard. Acta Crystallogr D Biol Crystallogr 64, 61-69. 
Traynelis, S.F., Wollmuth, L.P., McBain, C.J., Menniti, F.S., Vance, K.M., Ogden, K.K., Hansen, K.B., Yuan, 
H., Myers, S.J., and Dingledine, R. (2010). Glutamate receptor ion channels: structure, regulation, and 
function. Pharmacol Rev 62, 405-496. 
Turrigiano, G.G. (2017). The dialectic of Hebb and homeostasis. Philos Trans R Soc Lond B Biol Sci 372. 
Twomey, E.C., Yelshanskaya, M.V., Grassucci, R.A., Frank, J., and Sobolevsky, A.I. (2016). Elucidation of 
AMPA receptor-stargazin complexes by cryo-electron microscopy. Science 353, 83-86. 
Twomey, E.C., Yelshanskaya, M.V., Grassucci, R.A., Frank, J., and Sobolevsky, A.I. (2017a). Channel 
opening and gating mechanism in AMPA-subtype glutamate receptors. Nature 549, 60-65. 
Twomey, E.C., Yelshanskaya, M.V., Grassucci, R.A., Frank, J., and Sobolevsky, A.I. (2017b). Structural 
Bases of Desensitization in AMPA Receptor-Auxiliary Subunit Complexes. Neuron 94, 569-580 e565. 
Ursby, T., Unge, J., Appio, R., Logan, D.T., Fredslund, F., Svensson, C., Larsson, K., Labrador, A., and 
Thunnissen, M.M. (2013). The macromolecular crystallography beamline I911-3 at the MAX IV 
laboratory. J Synchrotron Radiat 20, 648-653. 
Watson, J.F., Ho, H., and Greger, I.H. (2017). Synaptic transmission and plasticity require AMPA receptor 
anchoring via its N-terminal domain. Elife 6, e23024. 
Winn, M.D., Ballard, C.C., Cowtan, K.D., Dodson, E.J., Emsley, P., Evans, P.R., Keegan, R.M., Krissinel, E.B., 
Leslie, A.G., McCoy, A., et al. (2011). Overview of the CCP4 suite and current developments. Acta 
Crystallogr D Biol Crystallogr 67, 235-242. 
Wong, A.Y., Fay, A.M., and Bowie, D. (2006). External ions are coactivators of kainate receptors. J 
Neurosci 26, 5750-5755. 
Yamazaki, M., Le Pichon, C.E., Jackson, A.C., Cerpas, M., Sakimura, K., Scearce-Levie, K., and Nicoll, R.A. 
(2015). Relative contribution of TARPs gamma-2 and gamma-7 to cerebellar excitatory synaptic 
transmission and motor behavior. Proc Natl Acad Sci U S A 112, E371-379. 
Yuan, H., Hansen, K.B., Vance, K.M., Ogden, K.K., and Traynelis, S.F. (2009). Control of NMDA receptor 
function by the NR2 subunit amino-terminal domain. J Neurosci 29, 12045-12058. 
Zhao, Y., Chen, S., Yoshioka, C., Baconguis, I., and Gouaux, E. (2016). Architecture of fully occupied GluA2 
AMPA receptor-TARP complex elucidated by cryo-EM. Nature 536, 108-111. 
Zuo, J., De Jager, P.L., Takahashi, K.A., Jiang, W., Linden, D.J., and Heintz, N. (1997). Neurodegeneration 
in Lurcher mice caused by mutation in delta2 glutamate receptor gene. Nature 388, 769-773. 
 
 KEY RESOURCES TABLE 
REAGENT or RESOURCE SOURCE IDENTIFIER 
Antibodies 
Mouse monoclonal anti-hemagglutinin antibody Covance, HA.11 clone 
16B12 
MMS-101P 
Bacterial and Virus Strains  
E. coli Origami B (DE3) Novagen N/A 
Chemicals, Peptides, and Recombinant Proteins 
Spermine tetrahydrochloride Sigma-Aldrich Cat#S2876 
CAS#306-67-2 
Cyclothiazide Tocris Bioscience Cat#0713 
CAS#2259-96-3 
BAPTA, tetrasodium salt Thermo Fisher Cat#B1214 
CAS#126824-24-6 
EGTA tetrasodium salt Sigma-Aldrich Cat#E9145  
CAS#13368-13-3 
D-APV Abcam Cat#ab120003 
CAS# 79055-68-8 
L-glutamic acid monosodium salt hydrate Sigma-Aldrich Cat#G1626 
CAS#142-47-2 
ATP disodium salt hydrate Sigma-Aldrich Cat#A2383 
CAS#34369-07-8 
Phenol red solution Sigma-Aldrich Cat#P0290 
CAS#143-74-8 
Penicillin-Streptomycin Sigma-Aldrich Cat#P4333 
Polyethyleneimine Sigma-Aldrich Cat#764604 
CAS#9002-98-6 
Anti-hemagglutinin agarose Sigma-Aldrich Cat#A2095 
Complete protease inhibitor cocktail Sigma-Aldrich Cat#11697498001 
Biotechnology performance certified water Sigma-Aldrich Cat#W3513 
CHAPS Sigma-Aldrich Cat#C9426 
CAS#331717-45-4 
Hemagglutinin peptide Sigma-Aldrich Cat#I2149 
CAS#92000-76-5 
L--phosphatidylcholine Avanti Polar Lipids Cat#840053C 
1,2-dioleoyl-sn-glycero-3-phospho-L-serine Avanti Polar Lipids Cat#840035C 
Fast Scan silicon AFM probes Bruker FASTSCAN-D 
CNQX Tocris Bioscience Cat#0190 
CAS#115066-14-3 
Triton X-100 
 
Sigma-Aldrich Cat#93443 
CAS#9002-93-1 
Poly(ethylene glycol) 4,000 Sigma-Aldrich Cat#81240 
CAS#25322-68-3 
Deposited Data 
Structure of GluA2o-N775S LBD (S1S2J) in complex 
with glutamate and RbBr, at 1.75 A 
This paper PBD: 6GIV 
Structure of GluA2o LBD (S1S2J) in complex with 
glutamate and NaBr, at 1.95 A 
This paper PBD: 6GL4 
Experimental Models: Cell Lines 
Human: HEK293T/17 cells (for electrophysiology) ATCC CRL-11268 
Key Resource Table
 Human: tsA201 cells (for AFM) Suzuki et al., 2013 ECACC 96121229 
Experimental Models: Organisms/Strains 
Mouse: C57/BL6J  The Jackson 
Laboratory 
RRID:IMSR_JAX:00
0664 
Recombinant DNA 
pRK5-ΔΔGluA1-flip Dr. M. Mayer, NIH, 
Maryland, USA 
Partin et al., 1995 
pRK5-ΔΔGluA1-flop This Paper N/A 
pRK5-GluA2(Q/R)-flip Dr. P. Seeburg, Max 
Planck Institute for 
Medical Research, 
Heidelberg, Germany 
N/A 
pRK5-GluA2(Q/R)-flop Dr. P. Seeburg, Max 
Planck Institute for 
Medical Research, 
Heidelberg, Germany 
N/A 
pRK5-ΔΔGluA1-flip/2 tandem This paper N/A 
pRK5-ΔΔGluA1-flop/2 tandem This paper N/A 
pRK5-GluA2(Q/R)-flip/2 tandem Dawe et al., 2016 N/A 
pRK5-GluA2(Q/R)-flop/2 tandem This paper N/A 
pRK5-GluA2(Q)-flip NTD This paper N/A 
pRK5-GluA2(Q)-flip NTD-eGFP This paper N/A 
pRK5-GluA2(Q)-flop NTD-eGFP This paper N/A 
pRK5-GluA2(Q)-L504A-flip This paper N/A 
pRK5-GluA2(Q)-L504C-flip This paper N/A 
pRK5-GluA2(Q)-T765N/P766A-flip This paper N/A 
pRK5-GluA2(Q)-T765N/P766A/S775N-flip This paper N/A 
pRK5-GluA2(Q)-S775N-flip This paper N/A 
His8-GluA2o-LBD, plasmid PET-22B(+) Dr. M. Mayer, NIH, 
Maryland, USA 
N/A 
His8-GluA2o- N775S-LBD, plasmid PET-22B(+) Krintel et al., 2012 N/A 
Software and Algorithms 
pCLAMP Software Molecular Devices https://www.molecul
ardevices.com 
Origin 7.0 and OriginPro 8.5 OriginLab Corporation https://www.originlab
.com 
NanoScope Analysis v1.5 Bruker https://www.bruker.c
om 
SigmaPlot Systat Software Inc., 
San Jose, CA 
https://systatsoftwar
e.com/products/sigm
aplot/ 
ImageJ ImageJ, Schneider et 
al., 2012 
https://imagej.nih.go
v/ij/index.html 
Adrian’s FWHM Software (ImageJ Plugin) ImageJ https://imagej.nih.go
v/ij/plugins/fwhm/ 
XDS  Kabsch, 2010 http://xds.mpimf-
heidelberg.mpg.de/ 
CCP4i Winn et al., 2011 https://www.ccp4.ac.
uk/ccp4i_main.php 
 Coot Emsley et al., 2010 https://www2.mrc-
lmb.cam.ac.uk/perso
nal/pemsley/coot/ 
Phenix Adams et al., 2010 https://www.phenix-
online.org/ 
PyMOL Molecular Graphics System Version 1.7.4, 
Schrödinger, LLC 
https://pymol.org/2/ 
 
 
 
 
 
 
 
Figure 1 Click here to access/download;Figure;Fig1.tif
Figure 2 Click here to access/download;Figure;Fig2-revised.jpg
Figure 3 Click here to access/download;Figure;Fig3_Amanda_5.tif
Figure 4 Click here to access/download;Figure;Fig4-revised.tif
Figure 5 Click here to access/download;Figure;Fig5.tif
Figure 6 Click here to access/download;Figure;Fig6-revised.tif
Figure 7 Click here to access/download;Figure;Fig7-revised.tif
Figure 8 Click here to access/download;Figure;Fig8-revised_2_final.tif
Figure S1 
 
 
 
Supplemental Text and Figures
Figure S2 
Figure S3 
Figure S4 
Figure S5 
Figure S6 
Figure S7 
 SUPPLEMENTAL FIGURES TITLES AND LEGENDS 
 
Figure S1. Bromide ions in the GluA2-LBD dimer interface and sensitivity to cyclothiazide, 
Related to Figure 2. (A) Alignment of the GluA2o (magenta/blue) and “flip-like” GluA2o-LBD 
N775S (cyan/orange) LBD dimers. Bromide ions at the dimer interface are shown as spheres 
(GluA2o, grey; N775S, red). L-Glu is shown in green sticks representation. Nitrogen atoms are 
shown in blue and oxygen atoms in red. Left: side view; right: top view. Occupancy of bromide 
ions was refined to 0.4. (B) Anomalous difference electron density map (before introduction of 
bromide ions in the structure) from GluA2o-LBD N775S, contoured at 7σ. Two bromide ions (1) 
were located in the vicinity of Pro515 at the dimer interface. (C) Anomalous difference electron 
density map from GluA2o-LBD (chain A), contoured at 7σ. In addition to two bromide ions (1) 
located at the dimer interface, bromide ions lining the side chain of Arg506 in the glutamate 
binding site (2) and bromide ions involved in crystal packing (3) were observed in this structure. 
(D) Fo-Fc difference map (green; contoured at 3σ) from refining with water molecule instead of 
bromide ion in GluA2o-LBD N775S. A 2Fo-Fc electron density map (grey; contoured at 1σ) is 
shown for the bromide site and water molecules within 4 Å from the bromide site. (E) Fo-Fc 
difference map (green; contoured at 3σ) from refining with water molecule instead of bromide 
ion in GluA2o-LBD (chain A). A 2Fo-Fc electron density map (grey; contoured at 1σ) is shown for 
the bromide site and water molecules within 4 Å from the bromide site. (F) Top views of the 
GluA2o-LBD N775S dimer (PDB: 3H6T; cyan/orange), showing two CTZ molecules (yellow sticks) 
bound in the dimer interface, aligned with bromide ions (brown spheres). (G) Response of 
GluA2i receptors to a sustained (250 ms) application of 10 mM L-Glu in different external halide 
ions, prior to (black) and during (blue) CTZ exposure. Responses in NaCl (patch 160516p4), NaBr 
(patch 160512p4), and NaI (patch 160510p4) were obtained from separate patch experiments. 
(H) Mean equilibrium current amplitude (Iequilibrium) of GluA2i receptors, as a percentage of the 
peak response (left axis, grey), in NaCl prior to CTZ application (Con.), or various halide ions in 
the presence of CTZ. The potentiation of the peak current response induced by CTZ in the same 
set of ionic conditions is also shown (right axis, blue). Data are mean ± SEM, from six (NaCl, NaI) 
or seven (NaBr) independent patch experiments.  
 
Figure S2. Mutation of residue Leu504 disrupts anion modulation of AMPAR desensitization, 
Related to Figure 3. (A-B) Side (A) and zoom-in (B) views of the GluA2i LBD dimer interface 
highlighting the Ser775 and Leu504 residues in proximity to bromide ions (brown spheres) at 
the anion binding site. (C-D) Typical current responses of GluA2i L504C (C) and L504A (D) 
receptors to a 250 ms application of 10 mM L-Glu in external NaCl (black), NaBr (dark grey), and 
NaF (light grey). Inset: scaled responses in different external anions to compare decay kinetics. 
The uppermost trace (black) shows the junction current recorded after each experiment to 
monitor solution exchange. (E-F) Mean time constants of current decay (E) and mean 
equilibrium current amplitude as a percentage of the peak response (F) after a 250 ms L-Glu 
application for the experiments described in panels C and D, in the presence of different 
external anions. (G) Average height changes of GluA2i, GluA2o and GluA2i L504A in response to 
L-Glu. Data are mean ± SEM for 10-14 receptors. (H) Average height changes of GluA2i, GluA2o 
and GluA2i L504A in response to anion switches. Data are mean ± SEM for 10-15 receptors. 
 Figure S3. The GluA2i T765N/P766A mutations do not influence anion sensitivity unless S775N 
is present, Related to Figure 3. (A-B) Typical current responses of GluA2i T765N/P766A (A) and 
T765N/P766A/S775N (B) receptors to a 250 ms application of 10 mM L-Glu in external NaCl 
(black), NaBr (cyan), and NaI (blue). Inset: scaled responses to compare decay kinetics including 
a representative response in NaF (grey). The uppermost trace (black) shows the junction 
current recorded after each experiment to monitor solution exchange. (C-D) Representative 
voltage-clamp traces showing recovery from desensitization for GluA2i T765N/P766A (C) and 
T765N/P766A/S775N (D) receptors in external NaCl (black) and NaI (dark blue). Both ionic 
conditions were recorded from the same patch. (E) Mean time constants of current decay after 
a 250 ms L-Glu application (τdes) for the experiments described in panels A and B, in the 
presence of different external anions. (F) Mean time constants of recovery from desensitization 
for GluA2i T765N/P766A and T765N/P766A/S775N receptors in the presence of different 
external anions.  
 
Figure S4. AFM imaging of bilayer-integrated AMPARs, Related to Figure 4. (A) Isolation of HA-
tagged GluA2i, GluA2o and GluA2i S775N, by anti-HA immunoaffinity chromatography. Samples 
were analyzed by SDS-PAGE followed by either silver staining (left panels) or immunoblotting 
using an anti-HA antibody (right panels). (B) Schematic illustration of the engagement of the 
AFM tip with a single AMPAR. (C) AFM image (2 µm x 2 µm) of a protein-free lipid bilayer. Scale 
bar, 400 nm; color-height scale, 0-10 nm. (D) AFM image (2 µm x 2 µm) of bilayer-integrated 
GluA2i receptors. Arrows indicate receptor positions. Scale bar, 400 nm; colour-height scale, 0-
10 nm. (E) Frequency distribution of heights of the GluA2i particles in bilayers. The curve 
indicates the fitted Gaussian functions. The peaks of the distribution (±SEM) are indicated. 
Based on our previous experience with AFM imaging of NMDA receptors (Suzuki et al., 2013), 
we assumed that particles in the taller population (peak 7.2 nm) represent assembled AMPARs. 
Particles in the shorter population (peak 4.7 nm) likely represent a mixture of unfolded or 
incompletely assembled receptors, together with receptors integrated into the bilayer 
cytoplasmic side-up. Only the assembled receptors were subjected to analysis. (F) Zoomed AFM 
image (200 nm x 200 nm) of a GluA2i receptor decorated by two anti-HA antibodies. Scale bar, 
50 nm; color-height scale, 0-13 nm. The paired-globule appearance of the NTD layer is clearly 
visible in this zoomed image. The height of the antibody-decorated receptor falls within the 
taller of the two populations shown in (E). A schematic illustration of the decorated receptor is 
shown at the right of the AFM image. 
 
Figure S5. Effect of L-Glu and anions on AMPAR height, Related to Figure 4. (A-L) Heights of 
individual GluA2 AMPARs before (left-hand point) and after (right-hand point) the indicated 
anion switch or addition/washout of L-Glu. The asterisks indicate statistical 
significance (*p < 0.05; **p < 0.01; ***p < 0.001; Student’s paired, two-tailed t-test). Drug 
concentrations were: L-Glu, 10 mM; CNQX, 0.5 mM; CTZ, 0.1 mM. 
 
Figure S6. Effect of L-Glu and anions on the height of individual truncated AMPARs, Related to 
Figure 4. (A-C) Isolation of HA-tagged GluA2i NTD (A), GluA2i NTD-GFP (B) and GluA2o NTD-
GFP (C) by anti-HA immunoaffinity chromatography. Samples were analyzed by SDS-PAGE 
followed by either silver staining (left panels) or immunoblotting using anti-HA antibody (right 
 panels). (D-F) Heights of individual GluA2i NTD (D), GluA2i NTD-GFP (E) and GluA2o NTD-
GFP (F) receptors before (left-hand point) and after (right-hand point) either addition of L-Glu 
or Cl--to-I- anion switch. Asterisks indicate statistical significance (*p < 0.05; **p < 0.01; 
Student’s paired, two-tailed t-test). 
 
Figure S7. GluA2(R)-containing heteromers are not blocked by polyamines, Related to Figure 
8. (A-C) Example traces from GluA1i+GluA2i (A, patch 190201p7), GluA1i/2+GluA2o (B, patch 
190204p4), and GluA1i/2+GluA2i/2 (C, patch 190215p5) exposed to 5 mM L-Glu (250 ms) at 
different membrane potentials (range, -100 to 100 mV, 10-mV increments) in the presence of 
30 μM internal spermine (Spm). (D-F) Mean I–V plots, normalized to +100 mV in Spm, for the 
receptor complexes shown in A-C. (G-H) Mean G–V plots for the receptor complexes shown in 
A-C. Data are mean ± SEM. 
  
Table S1. Gating behavior of GluA2 receptors in the presence of different external anions, 
Related to Figures 1, 3, and 6. Wildtype and mutant GluA2 flip (i) and flop (o) isoform receptors 
were expressed alone or with γ2.  
 
 I Br Cl F Propionate 
GluA2i           
τdesensitization 1.6 ± 0.1 (7) 4.3 ± 0.2 (7) 7.3 ± 0.3 (15) 9.2 ± 0.4 (8) 10.4 ± 0.8 (7) 
Iequilibrium (%) 0.6 ± 0.08 0.6 ± 0.09 1.1 ± 0.1 2.1 ± 0.3 3.3 ± 0.3 
Inorm 0.64 ± 0.07 0.93 ± 0.04 1 0.54 ± 0.05 0.23 ± 0.03 
τdeactivation 0.5 ± 0.04 (5) 0.5 ± 0.04 (7) 0.5 ± 0.03 (12) 0.5 ± 0.04 (7) - 
τrecovery 62.5 ± 3.0 (6) 37.2 ± 5.4 (7) 24.1 ± 1.5 (13) - - 
GluA2o      
τdesensitization 0.8 ± 0.05 (6) 1.1 ± 0.08 (6) 1.3 ± 0.06 (12) 1.7 ± 0.14 (6) 1.3 ± 0.05 (6) 
Inorm 0.60 ± 0.03 0.84 ± 0.04 1 0.54 ± 0.04 0.20 ± 0.03 
τrecovery 50.8 ± 5.2 (7) 35.0 ± 3.0 (13) 21.3 ± 1.2 (32)  - 
GluA2i S775N      
τdesensitization 3.6 ± 0.5 (6) 3.4 ± 0.2 (6) 5.2 ± 0.4 (12) 5.8 ± 0.4 (6) 5.6 ± 0.6 (6) 
Inorm 1.06 ± 0.10 1.05 ± 0.06 1 0.46 ± 0.05 0.25 ± 0.02 
τrecovery 40.9 ± 3.1 (5) 23.3 ± 1.9 (5) 15.1 ± 0.9 (9) -  
GluA2i 
T765N/P766A      
τdesensitization 0.7 ± 0.05 (9) 2.2 ± 0.2 (13) 2.9 ± 0.2 (25) 4.1 ± 0.3 (8) - 
τrecovery 37.1 ± 2.8 (7) 23.2 ± 2.1 (11) 15.1 ± 0.7 (22) - - 
GluA2i L504A      
τdesensitization 17.2 ± 1.1 (5) 20.1 ± 1.4 (5) 14.5 ± 0.7 (10) 5.7 ± 0.9 (5) 8.6 ± 0.8 (5) 
Inorm 0.69 ± 0.08 0.86 ± 0.05 1 0.39 ± 0.05 0.24 ± 0.05 
τrecovery 96.3 ± 13.1 (6) 32.9 ± 4.2 (6) 22.0 ± 2.8 (12) - - 
GluA2i ΔNTD      
τdesensitization 2.6 ± 0.1 (5) 6.8 ± 0.4 (6) 15.1 ± 0.6 (6) 19.0 ± 1.5 (6) - 
τrecovery 43.3 ± 6.0 (5) 22.2 ± 1.4 (6) 12.7 ± 1.1 (10) - - 
GluA2i/γ2      
τdesensitization - 13.4 ± 1.4 (8) 29.6 ± 2.5 (8) 39.7 ± 2.8 (8) - 
Iequilibrium (%) - 12.2 ± 1.5 22.6 ± 1.7 33.2 ± 1.7 - 
GluA2o/γ2      
τdesensitization - 1.4 ± 0.1 (8) 1.7 ± 0.09 (8) 4.3 ± 0.4 (8) - 
Iequilibrium (%) - 2.2 ± 0.4 2.8 ± 0.4 6.1 ± 1.0 - 
GluA2i S775N + γ2      
τdesensitization - 4.6 ± 0.3 (8) 8.2 ± 0.7 (12) 14.7 ± 0.9 (9) - 
Iequilibrium (%) - 2.9 ± 0.5 5.9 ± 0.6 11.7 ± 1.4 - 
  
  
GluA2 receptors were activated by long application (250 or 500 ms) or short (1 ms) applications 
of 10 mM L-Glu to measure desensitization and deactivation kinetics, respectively. In the 
presence of auxiliary subunits, current decay associated with desensitization was fit using bi-
exponential functions to obtain the components τfast and τslow. Weighted time constants 
(τdesensitization) were calculated based on the relative area fit by the fast and slow components. 
Time constants are listed in ms and Inorm refers to peak current amplitudes relative to the same 
receptor recorded in chloride. The number of patch recordings for each condition (n) is 
indicated, and all values are mean ± SEM. 
 
Table S2. Data collection and refinement statistics, Related to Figure 2. 
 
 GluA2o-LBD (NaBr) GluA2o-LBD N775S (RbBr) 
Data collection   
Wavelength 0.91949 0.91976 
Space group P2 C2 
Cell dimensions   
a, b, c (Å) 46.7, 47.7, 116.7 123.0, 47.5, 49.8 
, ,  (deg.) 90, 93.8, 90 90, 110.2, 90 
No. in asymmetric unit 2 1 
Resolution (Å) 46.6-1.95 (2.06-1.95)a 46.8-1.75 (1.84-1.75) 
No. unique reflections 37,848 (5,450) 27,430 (3,998) 
Rmerge (%) 8.9 (46.8) 7.4 (37.3) 
I/I 7.8 (1.6) 7.5 (1.8) 
Completeness (%) 100 (100) 100 (100) 
Redundancy 
(merged/anomalous) 
7.6 (7.6) / 3.9 (3.9) 5.0 (4.7) / 2.6 (2.6) 
   
Refinement   
Rwork / Rfree
b
 (%) 16.3/20.0 15.6/19.0 
No. atoms   
     Protein  4162 2092 
     Glutamate/bromide/other 20/6/12 10/1/23 
     Water 430 267 
B-factors (Å2)   
     Protein 27.3/27.6 22.1 
     Glutamate/bromide/other 18.0/35.5/43.4 13.6/32.2/48.3 
     Water 31.5 28.7 
R.m.s. deviations   
     Bond lengths (Å) 0.005 0.0033 
     Bond angles (deg.) 0.74 0.71 
Ramachandran plot   
     Favored (%) 99.04 99.23 
     Outliers (%) 0 0 
 
aValues in parantheses are for the highest resolution shell. 
b5% of data were used for calculation of Rfree. 
Table S3. Summary of height change data, Related to Figure 4 
__________________________________________________________________________________ 
Construct  Condition             Height reduction (nm)     SEM       Sample Size 
__________________________________________________________________________________ 
GluA2i   Cl- to Br-   0.74  0.06  12 
   Cl- to I-    0.87  0.11  13 
Br- to Cl-   -0.60  0.19  5 
I- to Cl-    -0.56  0.07  6 
L-Glu in Cl-   0.69  0.11  11 
L-Glu in Cl- washout  -0.58  0.16  6 
L-Glu in Br-   0.25  0.05  14 
L-Glu in I-   0.21  0.06  12 
L-Glu plus CNQX  -0.11  0.07  12 
L-Glu plus CTZ   -0.03  0.07  14 
Cl- to Br- plus CTZ  0.45  0.19  6 
Cl- to I- plus CTZ  0.66  0.05  13 
 
GluA2o   Cl- to Br-   0.03  0.06  14 
Cl- to I-    0.08  0.06  14 
L-Glu in Cl-   0.61  0.07  10 
L-Glu in Br-   0.70  0.14  10 
L-Glu in I-   0.58  0.12  10 
 
GluA2i S775N   Cl- to I-    0.00  0.05  12 
L-Glu in Cl-   0.56  0.10  10 
L-Glu in I-   0.70  0.06  12 
 
GluA2i L504A  Cl- to Br-   0.08  0.06  15 
Cl- to I-    0.04  0.13  10 
L-Glu in Cl-   0.69  0.07  14 
L-Glu in Br-   0.62  0.11  10 
L-Glu in I-   0.74  0.09  10 
 
GluA2i ATD   L-Glu in Cl-   0.03  0.06  13 
L-Glu in I-   0.13  0.10  11 
   Cl- to I-     -0.05  0.08  15 
 
GluA2i ATD-GFP L-Glu in Cl-   0.28  0.04  10 
L-Glu in I-   0.06  0.10  17 
Cl- to I-    0.32  0.09  18 
 
GluA2o ATD-GFP L-Glu in Cl-   0.27  0.07  12 
L-Glu in I-   0.25  0.12  22 
Cl- to I-    0.02  0.06  22 
 
__________________________________________________________________________________ 
 
Table S4. Combined and raw mobility data, Related to Figure 5.  
 
Combined 
 
Construct Condition       CSD at 28 s (nm2)  SEM       Sample Size 
__________________________________________________________________________________ 
GluA2i  Control         5.58   0.92  5 
L-Glu (0.1 mM)        7.67   1.65  5 
L-Glu (1 mM)        7.13   0.95  3 
L-Glu (10 mM)       13.90   2.11  9 
 
L-Glu (10 mM) plus CNQX      5.33   0.82  4 
 
GluA2o  Control       15.53   2.54  7 
L-Glu (0.1 mM)      14.07   0.81  3 
L-Glu (1 mM)      12.04   0.83  3 
L-Glu (10 mM)      17.12   0.59  5 
 
L-Glu (10 mM) plus CNQX      9.50   1.95  3 
 
GluA2i S775N Control       14.00   1.70  9 
 
Raw 
 
Construct Condition                                CSD at 28 s (nm2)   
__________________________________________________________________________________ 
GluA2i       Control                      8.12, 2.36, 5.81, 6.05, 5.57 
              L-Glu (0.1 mM)                     5.09, 2.46, 10.51, 10.06, 10.25 
               L-Glu (1 mM)       8.50, 7.58, 5.30 
              L-Glu (10 mM)       11.06, 15.71, 6.17, 18.42, 18.32, 4.69, 25.00, 12.63, 13.09 
 
               L-Glu (10 mM) plus CNQX   5.53, 4.12, 4.12, 7.57 
 
GluA2o       Control                     9.92, 9.28, 8.36, 20.53, 23.39, 23.24, 13.87 
               L-Glu (0.1 mM)       13.64, 12.93, 15.64 
               L-Glu (1 mM)       10.96, 11.47, 13.68 
              L-Glu (10 mM)       16.41, 15.38, 17.17, 18.80, 17.83 
 
               L-Glu (10 mM) plus CNQX   10.04, 5.89, 12.57 
 
GluA2i S775N Control         7.19, 16.08, 17.31, 20.41, 6.91, 15.74, 8.43, 18.31, 15.66 
__________________________________________________________________________________ 
 
Table S5. Gating behavior of native AMPARs and GluA1/A2 heteromers in the presence of 
different external anions, Related to Figures 7 and 8.  
 
 Br Cl F 
Native Cells 
Purkinje Cell       
τdesensitization 3.9 ± 0.5 (5) 7.5 ± 0.4 (20) 10.6 ± 0.9 (6) 
Iequilibrium (%) 5.4 ± 2.0 (4) 8.3 ± 0.6 (21) 17.1 ± 2.8 (4) 
Iequilibrium in CTZ (%) - 89.1 ± 2.5 (7) - 
τdeactivation - 1.0 ± 0.1 (5) - 
Stellate Cell    
τdesensitization 2.0 ± 0.2 (8) 2.9 ± 0.1 (30) 5.0 ± 0.6 (5) 
Iequilibrium (%) 2.4 ± 0.5 (7) 1.4 ± 0.1 (29) 3.2 ± 0.5 (5) 
Iequilibrium in CTZ (%) - 88.6 ± 3.2 (6) - 
τdeactivation  - 0.8 ± 0.04 (17) - 
A1/A2 Heteromers 0 TARPs 
A1i+A2i    
τdesensitization 3.2  0.2 (6) 4.5  0.3 (7) 7.9  0.7 (6) 
Iequilibrium (%) 0.3  0.1 (3) 0.8  0.2 (7) 1.3  0.3 (5) 
A1o+A2o    
τdesensitization 1.7  0.3 (4) 2.0  0.1 (7) 2.4  0.1 (6) 
Iequilibrium (%) 0.3  0.2 (3) 0.0 (6) 0.3  0.2 (4) 
A1/A2 Heteromers 2 TARPs 
A1i/2+A2o    
τdesensitization 2.6 0 .4 (4) 3.4  0.3 (8) 5.8  0.8 (4) 
Iequilibrium (%) 0.0 (3) 1.5  0.5 (8) 3.0  2.5 (3) 
A1o+A2i/2    
τdesensitization 3.6  0.2 (2) 4.7  0.2 (3) 7.8  0.7 (2) 
Iequilibrium (%) 0.6  0.1 (2) 1.1  0.03 (3) 2.6  0.05 (2) 
A1o/2+A2i    
τdesensitization 4.8  0.4 (6) 7.0  0.4 (6) 12.0  1.2 (6) 
Iequilibrium (%) 1.1  0.3 (5) 1.5  0.3 (6) 3.2  0.5 (6) 
A1i/2+A2i    
τdesensitization - 8.2  0.8 (10) - 
Iequilibrium (%) - 3.7  0.9 (10) - 
A1i+A2i/2    
τdesensitization - 6.9  0.9 (7) - 
Iequilibrium (%) - 4.0  0.9 (7) - 
A1/A2 Heteromers 4 TARPs 
A1i/2+A2i/2    
τdesensitization 6.6  0.3 (5) 10.6  0.3 (6) 19.4  1.4 (6) 
Iequilibrium (%) 12.2  1.2 (5) 18.5  1.4 (6) 34.4  1.3 (6) 
    
A1i/2+A2o/2 
τdesensitization 2.7  0.2 (5) 3.3  0.3 (7) 7.9  0.6 (6) 
Iequilibrium (%) 6.0  0.6 (5) 7.1  0.7 (7) 14.3  1.3 (6) 
A1o/2+A2i/2    
τdesensitization 6.3  0.4 (8) 8.9  0.5 (8) 16.8  0.9 (8) 
Iequilibrium (%) 11.7  1.1 (8) 14.3  1.5 (8) 28.2  2.0 (8) 
A1o/2+A2o/2    
τdesensitization 2.2  0.2 (6) 3.0  0.2 (7) 4.8  0.4 (7) 
Iequilibrium (%) 1.4  0.6 (5) 3.5  0.8 (7) 7.4  1.0 (7) 
 
For native receptors, outside-out and nucleated patches were excised from cerebellar Purkinje 
and stellate cells, respectively, and exposed to rapid application of 10 mM L-Glu (250 ms or 1 
ms duration) to measure desensitization and deactivation kinetics. Recombinant GluA1/A2 
heteromers were activated by application of 5 mM L-Glu (250 ms) to measure receptor 
desensitization. Current decay was fit using bi-exponential functions to obtain the components 
τfast and τslow. Weighted time constants (τdesensitization) were calculated based on the relative area 
fit by the fast and slow components. Time constants are listed in ms. Iequilibrium refers to the 
steady-state current as a percentage of the peak response. The number of patch recordings for 
each condition (n) is indicated, and all values are mean ± SEM.