Intracellular NASPM allows an unambiguous functional measure of GluA2-lacking calcium-permeable AMPA receptor prevalence

Calcium-permeable AMPA-type glutamate receptors (CP-AMPARs) contribute to many forms of synaptic plasticity and pathology. They can be distinguished from GluA2-containing calcium-impermeable AMPARs by the inward rectification of their currents, which reflects voltage-dependent block by intracellular spermine. However, the efficacy of this weakly permeant blocker is differentially altered by the presence of AMPAR auxiliary subunits – including transmembrane AMPAR regulatory proteins, cornichons and GSG1L – that are widely expressed in neurons and glia. This complicates the interpretation of rectification as a measure of CP-AMPAR expression. Here we show that inclusion of the spider toxin analogue 1-naphthylacetyl spermine (NASPM) in the intracellular recording solution results in complete block of GluA1-mediated outward currents irrespective of the type of associated auxiliary subunit. In neurons from GluA2-knockout mice expressing only CP-AMPARs, intracellular NASPM, unlike spermine, blocks all outward synaptic current. Thus, our results identify an unambiguous functional measure, sensitive solely to changes in CP-AMPAR prevalence.


Intracellular polyamine toxins block CP-AMPARs 102
To determine whether intracellularly applied NASPM or polyamine toxins might offer 103 advantages over spermine for the identification of native CP-AMPARs we examined 104 three compounds: NASPM, which contains the same polyamine tail as spermine, 105 PhTx-433, which has a different distribution of amines, and PhTx-74, which lacks one 106 amine group (Fig. 1a). Initially, we recorded currents in outside-out patches from HEK 107 cells transiently transfected with GluA1 alone or with GluA1 and GluA2, to produce 108 homomeric CP-and heteromeric CI-AMPARs, respectively. The receptors were 109 activated by glutamate (300 μM) in the presence of cyclothiazide (50 μM) to minimize 110 AMPAR desensitization and we applied voltage ramps (100 mV/s) to generate current- presence of 100 μM spermine, they displayed doubly rectifying responses (Fig. 1b). 115 By contrast, when the intracellular solution contained 100 μM NASPM, GluA1 116 receptors displayed inwardly rectifying responses with negligible current passed at 117 potentials more positive than +20 mV (Fig. 1b). Unlike responses from GluA1 alone, 118 currents from GluA1/2 receptors in the presence of NASPM were non-rectifying (Fig.  119   1b). I-V plots showed that, when added to the intracellular solution at 100 μM, each 120 blocker conferred marked inward rectification on the currents mediated by GluA1 121 (rectification index, RI+60/−60 0.02-0.26), but not on those mediated by GluA1/2 (RI+60/−60 122 0.84-1.30) (Fig. 1c). Thus, although differing in structure, they all produced selective 123 cyclothiazide from outside-out patches excised from HEK293 cells expressing either 127 GluA1 or GluA1/2. The voltage was ramped linearly from −80 to +60 mV (100 mV/s). 128 GluA1 displayed outward rectification with a polyamine-free pipette solution. In the 129 presence of 100 μM spermine GluA1 displayed a doubly rectifying relationship, and 130 with 100 μM NASPM GluA1 displayed full inward rectification. GluA1/2 did not rectify 131 in the presence of NASPM. c) Normalized and pooled I-V relationships for GluA1 and 132 GluA1/2 in the presence of 100 μM intracellular polyamines (n = 3-8 voltage-dependent block of the GluA2-lacking CP-AMPARs. Of note, the block by 144 intracellular PhTx-74 was restricted to CP-AMPARs, despite the fact that it produces 145 low affinity block of CI-AMPARs when applied extracellularly (Jackson et al., 2011;146 Nilsen and England, 2007). 147 We next investigated the concentration-and auxiliary subunit-dependence of the block. 148 Specifically, we generated conductance-voltage (G-V) relationships and fit those from 149 inwardly rectifying responses with a single Boltzmann function and those from doubly 150 rectifying responses with a double Boltzmann function (Panchenko et al., 1999). This 151 revealed that as the concentration of added blocker was increased (from 0.1 or 1 μM 152 to 500 μM) there was a progressive negative shift in Vb (the potential at which 50% 153 block occurs) (Fig 2a, b). Plotting Vb against polyamine concentration (Fig. 2b) allowed 154 us to determine the IC50, 0 mV (the concentration expected to result in half maximal block 155 at 0 mV) and thus estimate the potency of each polyamine. This showed that, for 156 steady-state conditions, the order of potency for GluA1 block was spermine > NASPM 157 > PhTx-433 > PhTx-74 (Fig. 2b, c). The same analysis demonstrated that the potency 158 of each blocker was reduced when GluA1 was co-expressed with TARP γ2 (between 159 7-and 18-fold reduction; Fig. 2b, c). 160 to +40 mV, with a large outward conductance seen in the presence of spermine but 198 not NASPM (Fig. 3b). NASPM, with its naphthyl headgroup, might also be expected to display limited 206 permeability of CP-AMPAR channels. This could account for the shape of the ramp I-207 V (Fig. 1c) and G-V plots with NASPM (Fig. 2a). Indeed, the pronounced effect of 10 208 μM intracellular NASPM on steady-state GluA1/γ2 currents at positive potentials ( Fig.  209 3a) is also consistent with limited permeation, leading to the accumulation of channel 210

block. 211
In line with this, we found that the decay of GluA1/γ2 currents recorded in the presence 212 NASPM was strongly voltage-dependent (Fig. 3a, c). At negative potentials τdecay 213 values were similar in the presence and absence of NASPM. However, at positive 214 potentials (from +10 to +80 mV) the kinetics in the two conditions differed markedly; in 215 the absence of NASPM current decay was slowed at positive potentials, while in the 216 presence of NASPM the decay was progressively accelerated (Fig. 3d). In the 217 absence of NASPM, τdecay was slower at +60 mV than at −60 mV (7.3 ± 0.6 ms versus 218 5.7 ± 0.3 ms, n = 8; paired mean difference 1.55 ms [0.86, 2.62], p = 0.019 paired 219 Welch t-test). By contrast, in the presence of NASPM τdecay was markedly faster at 220 +60 mV than at −60 mV (1.7 ± 0.2 ms versus 5.6 ± 0.6 ms, n = 10; paired mean 221 difference −4.14 ms [−5.44, −3.07], p = 0.00012). Of note, at +60 mV, along with the 222 accelerated decay in the presence of NASPM, we also observed a dramatic slowing 223 of the recovery of peak responses following removal of glutamate (Fig. 3e). Currents 224 were elicited by pairs of 100 ms glutamate applications at frequencies from 0.125 to 4 225 Hz. The peak amplitude of successive responses is normally shaped solely by the 226 kinetics of recovery from desensitization. In the absence of polyamines, a small degree 227 of residual desensitization was apparent at 4 Hz (150 ms interval), but full recovery 228 was seen at all other intervals, at both +60 and −60 mV (Fig. 3f). However, in the 229 presence of NASPM, although responses at −60 mV were indistinguishable from those 230 in the absence of polyamines, at +60 mV an additional slow component of recovery 231 (τrec slow 4.9 s; Fig. 3f) was present. Taken together, our data suggest that 10 μM 232 intracellular NASPM produces a pronounced, rapid and long-lasting inhibition of CP-233 AMPARs at positive potentials. 234

NASPM can induce complete rectification that is unaffected by auxiliary 235 subunits 236
We found that with 10 µM NASPM the block of CP-AMPARs was incomplete ( Fig.  237 3a,b) and depended on the recent history of the channel (Fig. 3f). However, as  to those in the polyamine free condition. With NASPM at −60 mV and in the absence 263 of polyamine at both voltages, the second currents broadly recovered to the initial 264 levels. With NASPM at +60 mV however, peak currents recovered much more slowly.  Normalised G-V data for peak and steady-state currents evoked by 10 mM glutamate 273 from GluA1/γ2 receptors with intracellular spermine (10 μM) or NASPM (10 μM). 274 Fig_3b_source_data.csv 275 Thus, any experimentally observed change in spermine-induced rectification cannot 294 be unambiguously attributed to a change in CP-AMPAR prevalence alone. 295 Accordingly, we next sought to determine whether NASPM was able to produce 296 complete rectification and whether its action was affected by the presence of auxiliary 297 subunits. To this end, we compared the effects of 100 μM spermine and 100 μM 298 NASPM on I-V relationships for GluA1 expressed alone or with γ2, γ7, CNIH2 or 299 GSG1L (Fig. 4a-c). With GSG1L, full rectification was seen with both spermine and    With 100 µM spermine added to the intracellular solution mEPSCs were detected in 346 stellate cells from GluA2 −/− mice at both negative and positive voltages (Fig. 5a, c). At 347 −60 mV the mean absolute amplitude of the averaged mEPSCs was 39 ± 5 pA, the 348 20-80% rise time was 0.21 ± 0.02 ms and the τw,decay was 1.14 ± 0.09 ms (n = 6). At 349 +60 mV the corresponding measures were 22 ± 2 pA, 0.23 ± 0.03 ms and 1.58 ± 0.19 350 ms. In each cell, fewer events were detected at +60 mV that at −60 mV (the mean 351 frequency was 2.3 ± 1.6 Hz at +60 mV and 7.7 ± 5.0 Hz at −60 mV). This is consistent recordings with 100 µM intracellular NASPM, mEPSCs were detectable at negative but 359 not positive voltages (Fig. 5b, c). At −60 mV the mean absolute amplitude of the 360 averaged mEPSCs was 36 ± 6 pA, the 20-80% rise time was 0.26 ± 0.02 ms and the 361 τw,decay was 1.45 ± 0.17 ms (n = 6). At −60 mV the average mEPSC frequency was 14.2 362 ± 8.2 Hz (n = 6), but at +60 mV no events were seen (RI+60/−60 = 0; Fig. 5d). Thus, even 363 when synapses contained exclusively GluA2-lacking AMPARs, spermine did not fully 364 block outward currents. By contrast, NASPM produced full inward rectification, 365 providing an unambiguous read-out of a pure population of CP-AMPARs. 366  are unaffected when this molecule is applied from the inside (Fig. 1). This route-specific 427 differential effect of PhTx-74 on CI-AMPARs may arise because its polyamine tail 428 contains only three of the four amine groups found in the other toxins, which leads to 429 an asymmetric distribution. When PhTx-74 is applied from the outside, the electrostatic 430 repulsion mediated by the edited Q/R site may be limited due to the missing amine. 431 When present inside the cell, however, the proposed 'inverted' toxin orientation would 432 maintain Q/R site repulsion through the intact terminal amine groups. Alternatively, when entering from the intracellular side, the NASPM head group may 447 be sterically hindered by the Q/R +4 aspartate residues, so decreasing the affinity in 448 this orientation. 449 We have shown previously that with GluA4, γ2 reduces spermine potency by around 450 20-fold (Soto et al., 2007). This is thought to result from increased spermine 451 permeation (Brown et al., 2018). Here, we found that co-expression of γ2 with GluA1 452 reduced the potency of all intracellular blockers (7 to 18-fold). Unlike spermine,

Advantages of intracellular NASPM for assessing CP-AMPAR contribution 460
Although intracellular spermine is widely used in voltage-clamp recordings to assess 461 the contribution of CP-AMPARs to the recorded currents, intracellular NASPM offers 462 two key advantages. First, it can produce a selective and complete block of CP-463 AMPAR-mediated outward current. Second, this block is unaffected by the presence 464 of auxiliary proteins. Together, these characteristics eliminate the difficulties of 465 interpretation that can arise when using spermine. 466 Rectification of AMPAR-mediated currents is commonly quantified using a rectification 467 index (RI). This is frequently the ratio of conductance or current amplitudes at +40 (or 468 Gria2-deficient (GluA2 −/− ) mice were bred from heterozygous parents (Gria2 tm1Rod , 544 MGI: 1857436) (Jia et al., 1996). Mice were group housed in standard cages and 545 maintained under controlled conditions (temperature 20 ± 2°C; 12 h light-dark cycle). 546 Food and water were provided ad libitum. Both male and female mice were used for 547 generating primary neuronal cultures. To increase likelihood of homozygotes surviving, 548 at P1 the litter numbers were typically reduced to six, by culling the largest pups (those 549 presumed to be wild-type or heterozygous). The genotypes of the remaining pups were 550 subsequently confirmed using PCR analysis using the following primers: oIMR6780 551 GGTTGGTCACTCACCTGCTT (wild type); oIMR6781 TCGCCCATTTTCCCATATAC 552 (common) and oIMR8444 GCCTGAAGAACGAGATCAGC (mutant). All procedures for 553 the care and treatment of mice were in accordance with the Animals (Scientific 554 Procedures) Act 1986. 555

Data analysis 634
Analysis of HEK cell recordings. Records were analyzed using Igor Pro with rectification index (RI+60/−60) was calculated as the ratio of the current at +60 mV and 637 −60 mV (average of 15 data points spanning each voltage). To generate conductance-638 voltage (G-V) curves the reversal potential for each leak-subtracted I-V curve was 639 calculated to ascertain the driving force. The resultant data were normalized, 640 averaged, then converted to conductances. To account for the polyamine-independent 641 outward rectification of AMPARs, the conductance values were divided by those 642 obtained in the polyamine-free condition. For currents that displayed inward 643 rectification only, G-V curves were fitted with the Boltzmann equation: 644

645
where Gmax is the conductance at a sufficiently hyperpolarized potential to 646 produce full relief of polyamine block, Vm is the membrane potential, Vb is the potential 647 at which 50% of block occurs, and kb is a slope factor describing the voltage 648 dependence of block (the membrane potential shift necessary to cause an e-fold 649 change in conductance). For currents that displayed double rectification, G-V curves 650 were fitted with a double Boltzmann equation which contains equivalent terms for 651 voltage-dependent permeation (p) (Panchenko et al., 1999): 652 Plots of Vb against the log of the polyamine concentration were fitted with a linear 654 function, the x-axis intercepts giving the voltage-independent affinity (IC50, 0 mV). 655 Current decays following fast applications of glutamate (10 mM, 100 ms) at positive 656 and negative potentials were described by single or double exponential fits. In the latter 657 case, the weighted time constant of decay ( ',()*"+ ) was calculated according to: condition, values of ()*"+ (from single exponential fits) and ',()*"+ (from double 662 exponential fits) were pooled. 663 Double pulse experiments (100 ms glutamate applications at intervals of 150 ms to 7.9 664 s (4 to 0.125 Hz) were used to assess the recovery of peak responses following 665 removal of glutamate. As expected, some residual desensitization was apparent at the 666 shortest of the intervals in all conditions (Coombs et al., 2017). Thus, the magnitude of 667 the second pulse reflected recovery from desensitization and, at +60 mV with 10 μM 668 NASPM, the relief of NASPM block. In the latter case, the recovery of the second peak 669 was fitted with a biexponential function, as above. 670 mEPSC analysis. mEPSCs were detected using an amplitude threshold crossing 671 method based on the algorithm of (Kudoh and Taguchi, 2002) (NeuroMatic). The 672 standard deviation of the background noise at +60 mV (range 2.4 to 6.4 pA; 3.5 ± 0.4 673 pA, n = 12) was determined by fitting a single-sided gaussian to an all-point histogram 674 from a 500 ms stretch of record. For each cell the same detection threshold (2-3x the 675 standard deviation at +60 mV) was used at both −60 and +60 mV. At each voltage, 676 the mEPSC frequency was determined as the total number of mEPSCs 677 detected/record length, and a mean mEPSC waveform was constructed from those 678 events that displayed a monotonic rise and an uncontaminated decay. One cell 679 recorded with intracellular spermine was excluded from the analysis as <10 events 680 were detected during 40 s recording at −60 mV. For each of the 12 other cells, detected 681 events were aligned on their rising phase before averaging. The rectification index was 682 then calculated as 683

684
where ̅ is the amplitude of the mean mEPSC and is the frequency of 685 mEPSCs at the indicated voltage. For each averaged mEPSC we determined the 20-686 80% rise time and fitted the decay with a double exponential to obtain τw,decay (as 687 described for HEK cell agonist-evoked currents). 688

Data presentation and statistics 689
Summary data are presented in the text as mean ± s.e.m. from n measures, where n 690 represents the number of biological replicates (number of cells or patches in each data 691 set, from 1-5 separate cultures/transfections). Estimates of paired or unpaired mean 692 differences and bias-corrected and accelerated 95% confidence intervals from 693 bootstrap resampling (Weiss, 2016) Fig. 4c and Fig. 5c) indicate the median (black line), the 25-75th percentiles 696 (box), and the 10-90th percentiles (whiskers); filled circles are data from individual 697

patches/cells and open circles indicate means. Comparisons involving two data sets 698
were performed using a two-sided Welch two-sample t test that does not assume equal 699 variance (normality was not tested statistically but gauged from density histograms 700 and/or quantile-quantile plots). Analyses involving data from three or more groups were 701 performed using two-way analysis of variance (Welch heteroscedastic F test) followed 702 by pairwise comparisons using two-sided Welch two-sample t tests. Exact p values are 703 presented to two significant figures, except when p < 0.0001. Statistical tests were 704 performed using R (version 3.6.0, the R Foundation for Statistical Computing, 705 http://www.r-project.org/) and R Studio (version 1.3.1056, RStudio). No statistical test 706 was used to predetermine sample sizes. No randomization was used. 707