A microtranslatome coordinately regulates sodium and potassium currents in the heart

Catastrophic arrhythmias and sudden cardiac death can occur with even a small imbalance between inward sodium currents and outward potassium currents, but mechanisms establishing this critical balance are not understood. Here, we show that mRNA transcripts encoding INa and IKr channels (SCN5A and hERG, respectively) are associated in defined complexes during protein translation. Using biochemical, electrophysiological and single-molecule fluorescence localization approaches, we find that roughly half the hERG translational complexes contain SCN5A transcripts. Moreover, the transcripts are regulated in a way that alters functional expression of both channels at the membrane. Association and coordinate regulation of transcripts in discrete “microtranslatomes” represents a new paradigm controlling electrical activity in heart and other excitable tissues.


77
Cardiac I Kr is critical for normal repolarization 2 and is a major target of acquired and 78 congenital long QT syndrome 3,4 . I Kr channels minimally comprise hERG1a and hERG1b 79 subunits 5,6 , which associate cotranslationally 7 and preferentially form heteromultimers 8 .

80
Underlying heteromultimerization is the cotranslational association of hERG1a and 1b mRNA 81 transcripts 9 . Because current magnitude is greater in heteromeric hERG1a/1b vs. homomeric 82 hERG1a channels, and loss of hERG1b is pro-arrhythmic 5,10 , the mechanism of 83 cotranslational assembly of hERG subunits is important in cardiac repolarization 9 .

85
In this study we found that association of transcripts could occur not only between alternate 86 hERG transcripts encoded by a single gene locus, but also between transcripts encoding 87 entirely different ion channel types whose balance is critical to cardiac excitability. Indeed, we 88 show that SCN5A, encoding the cardiac Na v 1.5 sodium channel, associates with hERG    Table S3). Plotted against each other, hERG1a and 161 SCN5A mRNA numbers exhibited a coefficient of determination (R 2 ) of 0.57 (P=0.00001; 41 162 cells; Fig. 3a

171
To determine potential hERG1a and SCN5A transcript association using smFISH, we 172 measured proximity between the two signals using the centroid position, scored from 173 touching to 67% (1 pixel) overlap ( Fig. 4a and

179
To test specificity of interaction between hERG1a and SCN5A transcripts, smFISH and 180 pairwise comparisons were also performed with RyR2 and GAPDH transcripts, which 181 revealed no significant association ( Fig. 4d and   To further explore whether colocalized mRNAs were part of a translational complex, we 190 combined smFISH with immunofluorescence using hERG1a antibodies. We observed close 191 association between hERG1a and SCN5A mRNAs and hERG1a protein significantly greater 192 than that expected by chance ( Fig. 5a and b and Supplementary Fig. S6a and b).

199
We monitored association of hERG1a protein and transcript in the presence of puromycin, 200 which releases translating ribosomes from mRNAs 13 (Fig. 6a). We observed no change due 201 to puromycin in the total number of respective mRNAs detected per cell (Fig. 6b). As 202 expected, puromycin reduced association between hERG1a mRNA and hERG1a protein 203 (antibody) and the S6 ribosomal protein (Fig. 6c). In addition, triple colocalization of hERG1a 204 and SCN5A transcripts and either hERG1a protein or the ribosomal subunit S6 was robustly 205 reduced (Fig 6d). These findings further support the conclusion that hERG1a and SCN5A 206 associate cotranslationally.

209
We previously demonstrated that targeted knockdown of either hERG1a or 1b transcripts by 210 specific short hairpin RNA (shRNA) caused a reduction of both transcripts not attributable to 211 off-target effects in iPSC-CMs or in HEK293 cells 9 . To determine whether hERG and SCN5A 212 transcripts are similarly subject to this co-knockdown effect, we evaluated expression levels 213 by performing RT-qPCR experiments in iPSC-CM. We found that hERG1a, hERG1b and 214 SCN5A expression levels were all reduced by about 50% upon hERG1a silencing compared 215 to the effects of a scrambled shRNA (Fig. 7a, orange bars). RYR2 transcript levels were 216 unaffected. We observed similar results using the specific hERG1b shRNA (Fig. 7a, blue 217 bars). Expressed independently in HEK293 cells, only hERG1a mRNA was affected by the 218 1a shRNA, and only hERG1b was affected by the 1b shRNA (Fig. 7b). SCN5A was 219 unaffected by either shRNA, indicating that the knockdown in iPSC-CMs was not due to off-220 target effects and levels of associated hERG1a and SCN5A are quantitatively coregulated.

221
Similar results of approximately 40% co-knockdown of discrete hERG1a and SCN5A mRNA 222 particles were obtained using smFISH ( Supplementary Fig. S7). Even more than the total 223 population of mRNA, the number of colocalized particles is decreased by approximately 224 55%, indicating that physically associated transcripts are subjected to co-knockdown (   Table S4). These results are in accordance to our previous studies reporting 236 a reduction in I Kr density upon hERG1b-specific silencing, and indicate that transcripts 237 targeted by shRNA are those undergoing translation 9,10 . To determine whether hERG1b 238 silencing also affects translationally active SCN5A, we measured peak I Na density in iPSC-

239
CMs and detected significant reduction of about 60% when hERG1b was silenced, compared 240 to control cells (Fig. 7f, g and h). Peak G max was decreased but no alterations in voltage 241 dependence of activation or inactivation were detected ( Fig. 7h and Supplementary Tables 242 S4 and S5). Late I Na , measured as the current integral from 50 to 800 ms from the beginning 243 of the pulse 14 , was similarly reduced in magnitude (Fig. 7i, j and k). This analysis indicates 244 that coregulation via co-knockdown results in quantitatively similar alteration of I Na , late and I Kr , 245 which operate together to regulate repolarization 15 . I to , which does not regulate action 246 potential duration in larger mammals 16 , is unaffected by hERG1b silencing (Fig. 8a

253
We have demonstrated using diverse and independent approaches the association and 254 coregulation of transcripts encoding ion channels that regulate excitability in cardiomyocytes.

255
By co-immunoprecipitating mRNA transcripts along with their nascent proteins, we have 256 shown that hERG and SCN5A transcripts associate natively in human ventricular 257 myocardium and iPSC-CMs as well as when heterologously expressed in HEK293 cells.

258
Using smFISH together with immunofluorescence in iPSC-CMs, we demonstrate that the     action. Perhaps by proximity to RISC, translation of the nontargeted mRNA is also disrupted, 324 but to our knowledge no current evidence is available to support this idea. A transcriptional 325 feedback mechanism seems unlikely given that co-knockdown can occur with plasmids 326 transiently expressed from engineered promoters and not integrated into the genome of 327 HEK293 cells. It is also important to note that it is unknown whether SCN5A is the only 328 sodium channel transcript coregulated by hERG knockdown. In principle, transcripts 329 encoding other sodium channels implicated in late I Na , such as Nav1.8 36,37 , could also be 330 affected, as could transcripts encoding auxiliary subunits associated with Nav1.5 38 .

332
Whether disrupting the integrity of these complexes gives rise to some of the many

427
FISHQUANT was used as a second method for spot detection and gave similar values.

428
Briefly, background was substracted using a Laplacian of Gaussian (LoG) and spots were fit 429 to a three-dimensional (3D) Gaussian to determine the coordinates of the mRNA molecules.

430
Intensity and width of the 3D Gaussian were thresholded to exclude non-specific signal 11,12 .

431
To evaluate the number of mRNA molecules, the total fluorescence intensity of smFISH

498
Cells were recorded using a holding potential of -50 mV, followed by a pulse at -40 mV to 499 inactivate sodium channels, then 3-second depolarizing steps (from -50 to +30 mV in 10 mV 500 increments) to activate hERG channels and finally to -40 mV for 6 seconds. Steady-state I Kr 501 was measured as the 5 ms average current at the end of the depolarizing steps. Tail currents 502 were measured following the return to -40 mV.

503
To record I Na , cells were perfused with an external solution containing (in mM): NaCl 50,

508
I Na activation was investigated by applying pulses between -140 and +20 mV in 10 mV 509 increments from a holding potential of -120 mV. To measure inactivation of sodium channels, 510 conditioning pulses from -140 to +20 mV in 10 mV increments were applied from a holding 511 potential of -120 mV following by a test pulse to -20 mV.

514
Pipette were filled with an internal solution containing (in mM): NaCl 5, CsCl 133, Mg-ATP 515 2, TEA 20, EGTA 10, HEPES 5, and pH was adjusted to 7.33 with CsOH. I Na,late was 516 measured by applying an 800 ms single pulse to -30 mV from a holding potential of -120 mV.

517
The mean current was measured at the last 200 ms of the pulse. An external solution 518 containing 30 μM TTX was perfused after the first pulse to determine if the current was due 519 to the activity of sodium channels.

520
To record I to , cells were continuously perfused with an external solution containing (in mM):

549
The source data corresponding to Figures 1b, 2b, 2c      Figure 7: Co-knockdown of I Kr and I Na by hERG transcript-specific shRNA. a, Effects of hERG1a or hERG1b silencing on channel mRNA expression levels detected by RT-qPCR (mean ± 95% CI) in IPSC-CMs. A non-targeting shRNA (scrambled shRNA) is used as a control. b, Effects of specific hERG1a or hERG1b silencing on ion channel mRNAs expressed alone in HEK293 cells. c, Representative family of traces show I Kr in presence of control (upper) or hERG1b shRNA (lower). d, Summary of steady-state current density vs. test potential shows effect of hERG1b shRNA (mean ± SE). e, Effects of 1b shRNA on peak tail current vs. pre-pulse potential (mean ± SE). f, Representative family of traces recorded from iPSC-CMs showing effects of hERG1b-specific shRNA compared to control shRNA on peak I Na . g, Summary current-voltage plot of peak I Na vs. test potential (mean ± SE). h, Summary conductance (G)-voltage plot based on data from g (mean ± SE). i, Late sodium current representative trace in control and 1b shRNA-transfected cells, measured by applying a single pulse protocol of 800 ms. j, Summary statistics of peak I Na showed a decrease upon transfection with hERG1b shRNA (mean ± SE). k, Late I Na measured as the integral from 50 to 800 ms from the beginning of the pulse showed a decrease upon transfection with hERG1b shRNA (mean ± SE).    Table S1: List of probes used in smFISH experiments. The probes were designed using StellarisⓇ probe Designer software with the following parameters: 18 to 20 nucleotides oligo length, a masking level of 5, a minimum spacing length of 2 nucleotides and a maximum number of probes of 48. Due to the length of the N-terminal specific sequence for hERG1a mRNA, the number of probes used to detect hERG1a is limited to 35.

Table S2
Table S2: Summary of colocalization analysis perfomed in iPSC-CMs for different association criteria. Comparison of the average number of mRNAs particles observed to be associated and the expected number based on chance alone using centroid positions and different association criteria (from touching to 67% overlap). The significance is tested with a paired t-test Bonferroni's correction. The number of hERG1a and SCN5A mRNAs observed to be associated is significantly above that expected by chance alone for all association criteria tested while no significant differences are observed for hERG1a/RyR2, hERG1a/GAPDH and SCN5A/GAPDH associations.  -33.2 ± 9.7 -88.6 ± 1.9 6.8 ± 0.5 9 Condition I max peak-tail (pA/pF) V 1/2 (mV) k (mV) n Control 0.50 ± 0.01 -26.0 ± 0.5 5.6 ± 0.5 5 1b shRNA 0.21 ± 0.03 -23.1 ± 4.5 7.1 ± 5.3 4 Table S4: Voltage dependence of activation and inactivation parameters for the sodium channels in cells transfected with a control shRNA or a hERG1b specific shRNA. Parameters were obtained after fitting to a Boltzmann equation activation and inactivation data.   8-14 shows the corresponding RNA-IP's using an anti-hERG1a antibody, followed by a bead-only control and H2O control. The next group shows the corresponding RNA-IP's using the anti-Nav1.5 antibody, followed by a group of IgG controls. H2O and beads lanes show absence of template contamination; control (+) represents signal amplified from purified plasmid template.      GAPDH (28 cells, a), and hERG1a and GAPDH (13 cells, b). The Pearson's correlation coefficient (R 2 ) were calculated for each pairs of mRNAs.