Imaging the electrical activity of organelles in living cells

Eukaryotic cells are complex systems compartmentalized in membrane-bound organelles. Visualization of organellar electrical activity in living cells requires both a suitable reporter and non-invasive imaging at high spatiotemporal resolution. Here we present hVoSorg, an optical method to monitor changes in the membrane potential of subcellular membranes. This method takes advantage of a FRET pair consisting of a membrane-bound voltage-insensitive fluorescent donor and a non-fluorescent voltage-dependent acceptor that rapidly moves across the membrane in response to changes in polarity. Compared to the currently available techniques, hVoSorg has advantages including simple and precise subcellular targeting, the ability to record from individual organelles, and the potential for optical multiplexing of organellar activity.

plasma membrane 7-11 . 48 Membrane potential imaging using voltage-sensitive dyes has been extensively used for mitochondria and 49 ER; and more recently, for phagosomes and lysosomes [12][13][14][15] . Still, a precise and standardized method 50 allowing for the recording of electrical signals generated at individual organelles is not yet available. To this 51 end, we have developed a general methodology for the recording of electrical signals from individual 52 organelles. The method relies on the use of a Hybrid Voltage Sensor (hVoS) 16 and here we show its 53 effectiveness for the fast imaging of variations in Ψ org (∆Ψ org ) in intact living cells. 54 The hVoS approach is extremely sensitive, capable of measuring rapid changes in the membrane potential 55 of both excitable and non-excitable cells [16][17][18][19] . The method takes advantage of a FRET pair consisting of a 56 membrane-anchored fluorescent protein acting as donor and the colorless hydrophobic anion 57 dipicrylamine (DPA) acting as acceptor. Due to its small size, negatively charged DPA has the ability to 58 rapidly transit across the membrane in response to changes in the membrane potential, acting as voltage 59 sensor 16 . Conveniently, the imaging read-out of membrane potentials at the plasma membrane is linear 60 within a broad dynamic voltage range (from -130 to +40 mV) 18 . 61 In principle, when combined with DPA, it is possible for any membrane-bound fluorescent marker to 62 targeted construct expressing the pH sensor pHluorin at the luminal side and the red-emitting fluorescent 112 protein mKate facing the cytoplasm 26 . Under the same conditions used for Lamp1-EGFP, pHluorin signal 113 remains stable while mKate fluorescence rapidly quenches upon the addition trypsin (Fig. 2a, bottom). 114 Wide field images revealed a characteristic ring shape on several of the Lamp1-EGFP positive structures 115 ( Supplementary Fig. 1a). The ring-shaped objects were usually between 1.2 and 0.9 µm enclosing a hollow 116 space of varying size, consistent with the dimensions of lysosome organelles ( Supplementary Fig. 1 b and  117 c). Analyzing the intensity of moving objects usually introduce error on the imaging data and requires 118 correction. As lysosomes are moving objects inside the living cell, we first evaluated the relative mobility of 119 Lamp1-EGFP labeled membranes 27 . By computing mobility maps we determine that at 20˚C Lamp1-120 EGFP positive structures are relatively immobile during the two-minute window required for our 121 recordings, eliminating the need for further correction of motion ( Supplementary Fig. 1d). 122 An important contributor to the lysosomal membrane potential is the pH gradient (∆µH + ), maintained by 123 the vesicular proton pump (v-ATPase) 28 . Thus, perturbation of the ∆µH + would provide a simple 124 approach to test our ability to measure Ψly using hVoS org in living cells. We first asked whether 125 alkalinization of the lysosomal lumen can be induced directly by ammonium 15,28,29 . As expected, 126 ammonium incubation (10 mM) causes a rapid and strong change in luminal pH, observed as an increase 127 of fluorescence signal measured from the pH sensor pHluorin, which we localized to the lysosomal lumen 128 ( Fig. 2 b and d). Next, we repeated the experiment using Lamp1-EGFP in the presence and absence of 129 DPA (Fig. 2 c and d). It has been estimated that 20 mM ammonium in the extracellular solution will 130 depolarize the lysosomal membrane in about 40mV 15 . Accordingly, in the presence of DPA, ammonium 131 quenches about 20% of GFP's fluorescence indicating a voltage-dependent DPA transit within the 132 lysosomal membrane (Fig. 2d). Such quenching is absent when the voltage sensor is not present. On the 133 contrary, a modest increase in EGFP's fluorescence can be detected after the ammonium treatment in the 134 absence of DPA (Fig. 2d). This could be explained because NH4 + will alkalinize not only cellular 135 compartments but also the cytoplasm. It has been reported that 20 mM ammonium in the external solution 136 will cause a change in cytoplasmic pH of about 0.3 pH units 30 . Given the high buffer capacity of the 137 cellular cytoplasm, we reasoned that normal fluctuations in cytoplasmic pH would not contaminate our 138 membrane potential measurements. Under our imaging conditions (i.e. GFP facing the cytosol), a negative 139 deflection of the fluorescence signal is caused by the redistribution of DPA molecules to the less negative 140 outer leaflet of the lysosomal membrane (Fig. 2d), and we interpret this as depolarization of the lysosomal 141 membrane. For the case of single membrane organelles (i.e. ER, lysosomes, endosomes, and golgi) ΔΨ is 142 calculated by Vm org = V cytosol -V lumen 31 , therefore our results indicate that alkalinization of the lysosome lumen 143 causes a rapid depolarization of the organelle's membrane as reported before 15 . 144 To further examine whether hVoS org is capable of following rapid changes in Ψ ly , we used the optogenetic 145 tool Lyso-pHoenix. Lyso-pHoenix is a large protein sensor composed by mKate at the cytoplasmic N-  Fig. 2 b and c). The observed increase in mKate fluorescence suggests a 156 repolarization of the lysosomal membrane. Taken together, our approach not only demonstrates the ability 157 to follow the amplitude and kinetics of changes in Ψ ly but also confirms the importance of luminal pH in 158 setting the resting potential of lysosomes. 159

Calibration of the hVoS org signal 161
The transport of ions into the lumen of the lysosome depends on the electrochemical gradient. At rest, the 162 electrochemical gradient of lysosomes can be roughly separated in two components -the chemical gradient 163 of protons (H + ) (∆pH ly ), and the lysosomal membrane potential (Ψ ly ) 32 . Ammonium treatments provides 164 us with a rough estimate of the voltage versus fluorescence response of our probe. To better control over 165 the voltage across the lysosomal membrane, we performed an In-cell calibration of hVoS org by using 166 potassium (K + ) as the only permeating ion. Gently digitonin-permeabilized cells were incubated with 167 nigericin, an antiporter of H + and K + (to dissipate ∆pH ly ), and the K + -selective ionophore valinomycin, to 168 have control of Ψ ly simply by changing the K + concentration in the extracellular solution now in contact 169 with the external membrane of the lysosome (Fig. 3 a and b). By doing this, we observed a linear of the 170 response up to 120 mV (positive inside). Considering the signal-to-noise ratio we estimated the limit of 171 detection to be 0.9±0.4 % of ΔF/F, corresponding to about 8mV (Fig. 3b). supplementary movie 1). When observed in more detail, we observe that the Lamp1-positive structures on 176 the periphery appear to have a smaller Ψ ly at rest ( Supplementary Fig. 3). We estimated that Ψ ly ranges from 177 60 to 110 mV (positive inside) and would not be unreasonable to propose that these differences can be 178 explained by the pH gradient observed in lysosomes during maturation 33 . 179 We used the same experimental approach to examine the resting transmembrane potential of Golgi (Ψ go ) 180 and ER (Ψ ER ) endomembranes. The Golgi marker manosidase II fused to EGFP (EGFP-ManII) reported 181 79±6 mV (positive inside; n=4) and the ER marker Sec61b fused to a HaloTag 34 (ht-Sec61b + Janelia 182 Fluor 556) reported -25±9 mV (negative inside; n=3) ( Figure 3d). Measuring the transmembrane potential of trans Golgi network has been elusive and previously estimated close to zero mV 35 . By localizing hVoS org 184 to the Golgi membrane we present the first direct measurement of resting membrane potential of this 185 compartment in living cells. Moreover, the confirmation that HaloTag can be used in combination with 186 DPA opens many possibilities for multiplexing intracellular voltage signals and to explore more subtle 187 aspects of lysosomal and cellular physiology 22,23,36,37 . 188 189

Modulation of lysosomal membrane potential by TRPML1 and TPC channels 190
To explore the contribution of known ion channels that are native in lysosomal membranes, we used 191 hVoS org to estimate Ψ ly at rest with overexpression of TPC1 and TRPML1. In both cases a smaller 192 quenching of the GFP signal was obtained, suggesting that the lysosomal membrane is less polarized upon 193 overexpression (Fig. 3d). This observation suggests that the overexpressed channels are active and that 194 their activity cannot be compensated efficiently by the v-ATPase or other lysosomal control mechanisms. 195 TPC1 seems to be particularly effective on collapsing the resting potential of the lysosome. This suggests 196 that the intrinsic voltage-sensitivity of the channel might create a positive-feedback loop when is not well 197

compensated. 198
The mammalian target of rapamycin (mTOR) is a kinase that integrates intracellular level of nutrients, the 199 energetic state, and growth factor signaling in higher eukaryotes 38 . Starvation and/or rapamycin treatment 200 (a general mTOR inhibitor) induce a robust electrical response of the endosome/lysosome (EL) vesicular 201 compartment, linking mTOR signaling and TPC sodium channels 39 . Moreover, the activity of mTORC1 202 has been associated to the activity of the SLC sodium-coupled amino acid transporter and also to the 203 activity of the lysosomal v-ATPase 40,41 . Consistent with the notion that the mTOR-signaling network is 204 associated to lysosomal electrical activity 42 , we observed that incubations with rapamycin (5 µM) evoked a 205 strong and transient depolarization in intact lysosomes of living cells (Fig. 4; supplementary movie 2). The 206 depolarization of the lysosomal membrane is in agreement with previous reports showing a Na + efflux 207 from the lumen of enlarged endolysosomes into the cytosol 6,39 . It is worth noticing that even in the 208 presence of rapamycin a late repolarization component is clearly visible, which may correspond to a voltage 209 dependent component that could be fulfilled by BK channels present at the lysosomal membrane (Fig. 4c) 210 43 . Additionally, we attempt to multiplex the signals by co-expressing Lamp1-EGFP and ht-Sec61b. 211 Simultaneous imaging of ΔΨ ly and ΔΨ ER was performed in response to rapamycin and followed by 212 digitonin permeabilization, showing a clear temporal separation of responses and the ability to space-213 resolve the signals from both compartments (supplementary movie 3). 214 We then compared the averaged response of lysosomes to rapamycin in cells expressing Lamp1-EGFP 215 alone or co-expressed with either hTPC1 or hTRPML1 ion channels (Fig. 4a). We calculated that the 216 rapamycin-dependent depolarization dissipates the lysosomal resting potential by 85 ± 4 mV (n=4) (Fig. 4  217 c and d), which corresponds to nearly 75% reduction of Ψ ly at rest. In contrast, the overexpression of both 218 TPC1 and TRPML1 channels showed a lower effect on Ψ ly (30 ± 6 and 15 ± 4 mV respectively; n=3) ( Fig.  219   4 c and d). The time course of the response can be described by a single exponential decay with a 220 characteristic time constant that is significantly different between normal lysosomes (12.3±3.3 s; n=4) and 221 those with lysosomal channels overexpressed (p<0.01; Fig. 4c). The kinetics of the lysosomal response is 222 shifted towards smaller values for both TPC1 (4.8±1.6 s; n=3) and TRPML1 transfected cells (5.7±1.2 s; 223 n=3) (Fig. 4c). The acceleration observed in the response when TPC1 channels are overexpressed suggests 224 that they are not close to the maximal open probability under overexpression conditions 6 . 225 Although the smaller dissipation of the voltage gradient correlates well with the ability to set a more 226 depolarized resting potential, the acceleration on depolarization kinetics together with the appearance of a 227 late seemingly voltage-dependent component, suggests to us that we cannot work under the assumption 228 that the lysosome membrane operates as a simple resistor-capacitor circuit. Thus, our results suggest that 229 the overexpression of ion channels that are residents of the lysosome affect the resting potential and in 230 doing so the dynamic response of the lysosome. 231

Discussion 233
The membrane potential is a major regulator of electrogenic transport across membranes. Therefore, By targeting hVoS org we effectively measure changes in membrane potential resolved in time and space 251 within living cells (Fig. 5), an experimental feat that has not been previously possible. The present hVoS org 252 approach not only allows recording of membrane potentials in intact cells but demonstrate to be robust 253 enough measure the elusive resting potential of intact individual sub-cellular structures that include 254 lysosomes, Golgi, and ER. 255 We also showed the ability of performing simultaneous measurements from different intracellular 256 membranes, demonstrating the capacity of hVoS org to provide new insights for cell biologists into whether 257 organelle-localized signals are modulated as a result of changes in organellar membrane potentials. We 258 foresee that combinations of sensors having different spectral properties, targeted to distinct sub-cellular 259 compartments, will allow for detailed space-time correlations of organelle's activity in living cells 37,44 .

551
ATPase leads to acidification, causing a more hyperpolarized membrane potential in the mature lysosome (orange 552 arrow).