Quantification of Fundamental Weak Base and Ion Permeabilities in the Acidic pH Region Utilizing Conjugated Oregon Green in Liposome‐Based Assays

Ammonia and ammonium play an essential role in the metabolism of almost all living organisms. Despite their importance for human health as well as plant growth, knowledge about passive NH3/NH4+ membrane permeabilities is scarce. Large unilamellar vesicles (LUVs), a popular model membrane system, have the potential to be a game changer. However, despite a variety of environmental sensitive dyes awaiting encapsulation into membrane vesicles, a pH sensor for the acidic pH region is currently missing. This gap is filled by introducing conjugated Oregon Green (OG) which can be encapsulated into lipid‐ or polymer‐based vesicles. This allows the quantification of proton or weak base permeabilities across the vesicular membrane and functionally characterizes pH sensitive membrane proteins at an acidic pH as low as pH 4.0. Furthermore, the expanded pH range enables simultaneous estimation of ion permeabilities without the use of membrane potential uncouplers due to an increased ion sensitivity. The utility of the sensor is demonstrated by quantifying passive NH3, NH4+, and Cl− permeabilities through vastly different lipid and polymer membranes. Moreover, its performance is benchmarked against carboxyfluorescein, an established pH‐sensor in the neutral pH‐range.


Introduction
Ammonia and ammonium are fundamental to nitrogen metabolism in all forms of life.Several physiological disorders, like cirrhosis, result in hyperammonemia, which in turn contributes to hepatic encephalopathy, neurologic disease and organic acidurias.For most microorganism's ammonium is the preferred nitrogen source, which is a crucial building block of amino acids.Hence, most plants rely on ammonium and other DOI: 10.1002/adsr.202200097nitrogenous wastes incorporated into the soil or into fertilizer.On the other hand, as high concentrations of ammonium are toxic, ammonium has to be excreted as metabolic waste. [1,2]Thereby, the neutral molecule NH 3 can passively permeate lipid bilayers according to Overton's Rule [3] where K p is the partition coefficient of NH 3 between water and hexadecane (chemical structure comparable to hydrophobic part of lipid), D NH 3 the diffusion constant of NH 3 in water, and h the lipid bilayer thickness.][10][11][12] However, despite P NH 3 being orders of magnitude higher as compared to passive membrane permeabilites of H 2 O, glycerol or urea, [13] NH 3 permeability across lipid bilayers is facilitated by the highly conserved Amt/Mep/Rh superfamily of transmembrane proteins. [14]Trimeric ammonium transport proteins (Amt) are found in bacteria, plants, yeast, and animals. [15]In yeast they are called methylamine permease (Mep) and in animals Rhesus (Rh) proteins.They are supplemented by ammonia-permeable aquaporins [16] which, e.g., colocalize with Rh proteins in erythrocytes and cells associated with either homeostasis or high rates of anabolism. [14]A prominent example is AQP8 which was shown to be involved in fast and selective NH 3 transport. [17]In plants, tonoplast intrinsic proteins are involved in remobilization of ammonia from the vacuoles during senescence or nitrogen starvation conditions. [18]In the pathogenic bacterium Helicobacter pylori, colonizing the gastric mucus in half of the world population, a small pH gated inner membrane channel UreI has been suggested to transport not only urea but also its metabolite NH 3 . [19]espite its biological significance, to the best of our knowledge a surprisingly low number of quantitative passive NH 3 permeability estimations through membranes and incorporated transmembrane channels can be found in the literature. [4,17,20,21]As a consequence, compared to, e.g., the permeation of H 2 O, the mechanistic knowledge of weak base permeation lacks behind.
Previous measurement concepts comprise free standing planar lipid bilayers with or without reconstituted membrane proteins. [13,22]A shift in the pH within the unstirred layers adjacent to the lipid bilayers was measured either via a shift in transmembrane potential after addition of a protonophore [4,21] or the use of ion sensitive microelectrodes which were moved perpendicular to the membrane within the unstirred layer. [17,20]However, despite of challenges in free-standing bilayer and microelectrode preparation, protein reconstitution into these membranes suffers from low yields which furthermore depends on the protein of interest.Large unilameallar vesicles (LUVs) with encapsulated environmental fluorophores may serve as an alternative. [22]t encompasses easy sample preparation with sufficient reconstitution efficiencies.Combined with an elaborate data analysis it allows to quantify passive membrane permeabilities and unitary permeabilities of transmembrane channels as shown for water, [23,24] weak acids, [25][26][27][28][29][30] and ions. [31,32]However, this widely used model membrane system has never been adapted to measure weak base or proton permeabilites in the acidic pH region.The latter being a prerequisite to characterize protein function in the acidic pH region or to increase the sensitivity for ions, e.g., NH 4 + .As ion permeation creates a membrane potential in LUVs, ion permeabilities can only be assessed by either measuring the initial slope of the measurement signal at t = 0 [33] or using membrane potential decouplers.However, efficient decoupling via the latter crucially depends on the lipid to decoupler ratio. [34]Furthermore, several decouplers suffer from unideal selectivity's complicating data analysis.
Herein, we introduce the concept of LUVs loaded with conjugated Oregon Green (OG) as membrane impermeable pH sensor to measure passive membrane and/or membrane channel permeabilites of NH 3 /NH 4 + .In contrast to the widely used pH sensor carboxyfluorescein (Cf), with a pKa value of ≈6.5, OG is highly membrane permeable preventing its application so far.OG labeled lipids on the other hand suffer from the fact that changes in the surface pH are sensed in contrast to changes of bulk pH.The former readout is problematic as the pH close to an interface depends on its surface charge and solutes in the aqueous environment. [35,36]Hence, our novel sensor for the first time enables pH sensitive measurements using LUVs in the acidic pH region.This is among other things inevitable for the characterization of the open and closed states of pH gated membrane channels, e.g., HpUreI which exhibits a pKa of opening of ≈6.0. [37]Moreover, our conceptual advancement opens a measurement window for NH 4 + and Cl − permeabilities without the use of membrane potential decouplers.We demonstrate the use of the sensor quantifying passive NH 3 , NH 4 + , Cl − permeabilities through vastly different membranes including lipid bilayers in fluid and gel state as well as a polymer-based membrane.Furthermore, in control experiments we quantify passive H + permeabilities as well as selected activation energies of permeation.
Thus, our sensor concept will not only help to advance the knowledge of passive solute permeabilities of biological membranes but also help to spur the characterization of wild-type and mutated transmembrane channels as well as artificial channels.0]

Conjugation of OregonGreen
5-Oregon Green 488 succinimidyl ester (AAT Bioquest) was conjugated with Kanamycin sulfate (Sigma Aldrich) containing a mixture of Kanamycin A, B, and C (predominantly Kanamycin A) or m-dPEG 8 -amine (Dextran, Quanta Biodesign, PN 10278).Kanamycin (Kan) molecules have four to five amine groups that could be linked covalently to the NHS group of Oregon Green 488 (OG488).Dextran has one amine group available for linkage.
In brief, 1 mm OG488 NHS was dissolved in PBS buffer at a pH between 8.3 and 8.5.Kanamycin or Dextran was added to the solution with a molar ratio of amine groups to OG488 NHS of roughly 1 to 1.The sample solution was protected from light to prevent photo bleaching and incubated at RT for 2 to 3 days.OG488_Kan and OG488_Dextran were stored at 277 K till further usage.

pKa Determination
The pKa of Cf, OG488_Kan, and OG488_Dextran was estimated with a fluorescence spectrophotometer (F-2700, Hitachi).The samples were prepared in a cuvette with 100 mm NaCl, 5 mm -alanine, 5 mm MES, and 5 mm MOPS at different pH values.The pH was adjusted with HCl or NaOH during constant stirring.The dye concentration was ≈1 μm to avoid inner filter effects.For each pH a wavelength scan from 515 to 650 nm was performed using an excitation wavelength of 480 nm.The area I Area [in p.d.u.] under these curves was calculated and plotted against the corresponding pH.The resulting sigmoidal curve was fitted with [41] : where b g (background), c 0 (amplitude), and pKa are fitting parameters.Since the pKa depends on temperature the pH adjustment and fluorimeter measurement were performed at 277 K and RT, each.

Water Permeability Estimation
LUVs were subjected to a hyperosmotic solution (working buffer + 200 mm sucrose) in a stopped-flow apparatus (SFM-300 or m-SFM, Bio-Logic, Claix, France) at 277 or 295 K.The intensity of scattered light was monitored at 90°at a wavelength of 546 nm. [12,24,43]Water permeabilities (P f ) through vesicles with a hyper osmotic gradient c s were calculated utilizing the analytical solution [23,44] with inner vesicle osmolarity c 0 , Lambert function L, the surface area of lipid vesicles S, the molar volume of water V w , and V 0 the water volume at time t = 0.For estimating the volume V(t) the measured scattered light intensity (I(t)) was Taylor approximated in the form [44] with and constants b and d.I(∞) is the final intensity value at the plateau of the scattered signal.
Intravesicular pH changes were monitored via pH sensitive fluorophores utilizing stopped-flow methodology.Thereby, the estimation of solute permeabilities P m relied on intravesicular pH changes pH in (t) using Cf for the neutral pH range (pH 5.5-7.5) or conjugated OG 488 in the acidic pH range (pH 4.0-6.0).Thereby, pH in depended on the concentration of NH 3 /NH 4 + , buffers and the self-dissociation of water (Figure 1).For fluorescence experiments, monochromatic light of 480 nm wavelength was used to illuminate the LUVs in a stopped-flow apparatus (SFM-300 or m-SFM, Bio-Logic, Claix, France) at 277 or 295 K. Fluorescence emission passed a 515 nm long pass filter before it was detected at were osmolarity optimized to minimize vesicle swelling during the measurement ensuring membrane integrity of the LUVs. [24]o relate the normalized fluorescence intensity I in (t) from the fluorophore inside the vesicle and pH in (t) the following theoretical model was used [25] : where pK dye is the pKa value of the fluorescent agent, pH out the time independent pH outside the vesicle and I b is an outratio factor considering background noise caused by dye leakage, vesicle size distributions and possible aggregated lipids.For a numerical solution the most suitable P m was chosen by estimating pH in (t,P m ).

Vesicle Modeling
If a weak base like NH 3 , or its corresponding protonated form NH + 4 , leaves the vesicle through the membrane pH in changes according to the following reaction scheme: Water dissociation rate k + w 2.5 × 10 −5 s −1 [47]   pKa ammonia at RT pK NH 3 9.25 [48]   pKa ammonia at 277 K pK NH 3 9.90 [48]   pKa beta alanine at RT pK BA 3.55 a) pKa beta alanine at 277 K pK BA 3.63 [49]   pKa MES at RT pK M 6.19 [50, 51]   pKa MES at 277 K pK M 6.35 [50, 51]   pKa  a) Taken from the Chemical Book database. [66]e corresponding protonation rates k p and deprotonation rates k d can be estimated from a linear regression of Figure 4 in Gutman et al. [45] : where pK and K a are pKa values (listed in Table 1) and the corresponding temperature dependent equilibrium constants.The change of NH + 4 ( dt ) due to de/protonation can be expressed as The Goldman-Hodgekin-Katz flux equation j GHK was used to describe passive diffusion of NH + 4 and other ions through membranes [52,53] : where P q is the membrane permeability of species q, z q its valence, S the vesicle surface, F the Faraday constant, R the Gas constant, and [q] in and [q] out the concentration of charged species q inside and outside the vesicle, respectively.In this model q refers to the charged ions NH + 4 , Na + , H + , and Cl − .It was assumed that at time t = 0 the membrane potential V m = 0. Ion permeabilities P q for protons and sodium ions were taken from Table 2.
stance change in the vesicle over time.P OH − could be neglected due to its low contribution in the acidic/neutral pH region. [59]hus, estimating the change of NH + 4 , Equation (10) needed to be subtracted from Equation (9): where the first term describes the diffusion through the membrane and the second term the effect of protonation and deprotonation with rates k N p and k N d , respectively.The passive flux through membranes for neutral charged molecules like NH 3 was described with Fick's first diffusion Law [60] : where [NH 3 ] out is the constant NH 3 concentration outside the vesicle, since the occupied volume of the vesicles was two orders of magnitude smaller than the volume outside of the LUVs.Thus, the total change of amount of substance n NH 3 over time is the sum of Equations ( 12) and ( 9): The efflux of NH 3 and NH + 4 was generated by a concentration gradient inducing an osmotic shift  Osm and vesicle volume change [23] With water permeability P f , molar volume of water v w , (Table 1) and the osmotic gradient  Osm [61] where k permeable [X(t)] inside and n impermeable [Y(t = 0)] inside substrate concentrations were inside the vesicle.[Z(t)] outside was the sum of all m substrate concentrations outside the vesicle.Since charged particles like Cl − can enter the vesicle according to Equation (10), the membrane was modeled as a capacitor with charge difference Q and specific capacity μ [25,46] : Q can be expressed with the amount of charge n q [mol] using the relation Q = n q • z q • F. Thus, V m can be estimated as The net charge of impermeable solutes was initially zero and did not contribute to V m .
Due to pH sensitive measurements the proton concentration played a crucial role, hence it was necessary to consider also the self-dissociation of water into H + and OH − according to [25] : where k + w is the dissociation and the association rate of water with [H 2 O] = v −1 w and K w the water equilibrium constant.Furthermore, the proton concentration [H + ] depended on the protonation k p and deprotonation k d rates of any available buffer M. Thereby, the concentration of the de/protonated buffer [M − ]/[MH] inside the vesicle could be estimated using the following relations between protonated and deprotonated species of buffer M: ) Moreover, in Equations ( 19) and (20) the relation between concentration c(t), vesicle volume V(t), and amount of substance n(t) = c(t) • V(t) were used.
Therefore, the change of the amount of protons dn H + dt inside the vesicle could be calculated as the sum of proton influx (Equation (10)) and any de/protonation effects with ammonia, buffers, water and the fluorescent dye: where index i represents buffer substrate Beta Alanine, MES and the fluorescent dye with corresponding de/protonation rates.pKa values for estimating de/protonation rates according to Equation (8) can be found in Table 1.Mathematica's NDSolve IDA package (V.12) (RRID:SCR_014448) was used to numerically solve the system of differential equations for estimating solute flux rates for NH 3 , NH + 4 and Cl − .

Activation Energy of Passive NH 3 Permeation
The activation energy E a for NH 3 and NH + 4 through the membranes was estimated with the Arrhenius Equation with solute permeability P m [25] : where R, T, and P 0 is the Gas constant, temperature, and a constant, respectively.The linearized form of Equation ( 22) was fitted to the Arrhenius Plot

Proton Permeability Estimation
Cf loaded DOPE/DOPG, PLE or PBD-b-PEO-based LUVs at a concentration of ≈ 0.5 mg mL −1 in 100 mm KCl, 2 mm MES at pH 7.0 were incubated with 1 μm Valinomycin for ≈ 5 min before the measurement (Fluorescence Spectrophotometer, F-2700, Hitachi).Sample excitation and fluorescence emission at 90°were chosen to be at 480 nm and above 520 nm, respectively.Fluorescence intensity changes due to proton flux through the membranes were recorded for ≈900 s at 277 K.At t = 0 HCl was added to the sample cuvette during constant stirring to create an outside oriented pH gradient (Figure S2, Supporting Information).For each of the three pH gradients (Figure S2, Supporting Information) two to four single measurements were averaged.The averages were normalized for a global fit using the same theory as for NH 3 permeability (see previous section).

Statistical Analysis
Stopped-flow data are presented neglecting the deadtime of the system of ≈1 ms.This is the time it takes the solution to flow from the mixer, where the reaction starts, to the measurement cuvette.Three to nine raw data traces with a sampling time between 100 μs and 2 ms, depending on the time duration of the experiment, were averaged.To enable the correlation between fluorescent intensity and pH in according to Equation ( 6) averaged data was normalized to one.Normalization was done with I max at t = 0, which was found fitting the first 10 data points exponentially.pKa values and activation energies are depicted as mean ± SD.SD values were created using the command

Acidic pH Range Enables the Measurement of Ion Properties
LUVs are small nano-compartments enabling the quantification of solute permeabilities across its membrane utilizing encapsulated environmental sensitive dyes.The adaptation of our recently published mathematical model [61] to the herein presented passive permeability sensors enabled us to predict solute concentrations over time in dependence of the experimental conditions.
Comparing the sensitivity of such a passive permeability sensor in the neutral regime, at a pH of 7.0, to the acidic region, at a pH of 4.5, revealed an increased sensitivity to ions in the acidic region (Figure 2).Furthermore, the dynamic time range for resolving passive NH 3 permeabilities is greatly enhanced.Emanating from the passive membrane permeabilities for PLE calculated herein and literature values listed in Table 1, we show that at a pH of 7.0 the sensor is insensitive to passive NH 4 + and Cl − permeabilities (Figure 2).On the contrary, at a pH of 4.5 the corresponding time dependences expanded over several orders of magnitude enabling its experimental quantification.

Sensor Implementation
As depicted in Figure 2 it is critical to measure in the acidic pH region to ensure accurate NH 3 permeability values and be able to resolve NH 4 + and Cl − permeabilities within one and the same experiment.Due to a lack of membrane impermeable pH sensitive probes we attached OG to the membrane impermeable molecule kanamycin (OG_Kan).The latter being known as antibiotic against bacterial infections and tuberculosis and being used in cell culture at concentrations of up to several 100 μm. [62]o ensure that kanamycin is not interfering with our permeability estimations via membrane specific interactions we measured the water permeability of PLE in dependence of increasing kanamycin concentrations (Figure S3, Supporting Information).From this control experiment it is evident that kanamycin concentrations between 0 and 100 μm do not change the measurement results and 50 μm can be used without hesitation.Similar to our Cf based measurements, the use of 50 μm kanamycin results in an overall dye concentration of ≈0.25 mm due to its multiple amine groups which are available for labeling.0.25 mm of encapsulated dye depict a good tradeoff between minimizing the dye concentration and maximizing the signal to noise ratio during pH dependent measurements.
Moreover, we compared P NH 3 measured with OG_Kan with OG_Dextran.The latter being also membrane impermeable but suffering from lower signal amplitudes and a higher outratio in our hands.These experiments revealed a reasonable difference in P NH 3 of ≈20% for PLE at 277 K, most probably caused by the experimental challenges with OG_Dextran (Figure S4, Supporting Information).
Next, we set out to benchmark the performance of our new sensor against LUVs filled with Cf, a widely used pH sensitive dye with a pKa of ≈6.5 (Figure S5, Supporting Information).A direct comparison at pH 5.5 highlights the potential of OG_Kan with consistent NH 3 permeabilities (Figure S6, Supporting In- formation).Moreover, we compared P NH 3 ,Cf and P NH 3 ,OG_Kan from global fits to the experimental data of three lipid mixtures (PLE, Epithelial lipid composition, and DPPC).Thereby, P NH 3 ,Cf is calculated from a global fit to two to three intensity traces at different pH conditions between pH 6.0 and 7.0.Similarly, P NH 3 ,OG_Kan is calculated from global fits to three intensity traces at acidic buffer conditions with a pH between pH 4.0 and 5.5.These experiments show in two out of three cases a slightly lower P NH 3 ,Cf of ≈30% as compared to P NH 3 ,OG_Kan (Figure S7, Supporting Information).In the case of the epithelial lipid composition, measured at 277 K, the two values perfectly matched with P NH 3 = 3.2 m s −1 .For PLE and DPPC P NH 3 ,OG_Kan perfectly matches P NH 3 calculated from a global fit of the entire pH region from 4.0 to 7.0 (Figure S8, Supporting Information).This emphasizes the crucial importance of the acidic pH region for an accurate P NH 3 estimation and discloses small uncertainties in fitting Cf data due to a higher I b value and a smaller dynamic measurement range.
Finally, we benchmarked our new sensor against the established Cf based sensor measuring the activation energy E a of NH 3 permeation through PLE-based lipid membranes.Again, we subjected LUVs with encapsulated OG_Kan and Cf, respectively, to an inside-out gradient of NH 4 Cl.This time at pH 6.0 for Cf and pH 4.5 for OG_Kan.These experiments were repeated with the same batch of LUVs for different temperatures between 277 and 305 K or 310 K, respectively.Fitting Equation (23) to the Arrhenius Plot (Figure 3) with the membrane permeabilities plotted over the corresponding tempera- tures we could extract E a, NH 3 ,pH6.0= 10.3 ± 1.4 kcal mol −1 and E a, NH 3 ,pH4.5 = 10.1 ± 1.4 kcal mol −1 for the Cf-based as well as the OG_Kan-based sensor.Both values are in perfect agreement being slightly smaller as compared to the activation energy of water permeation through PLE of 11.05 ± 0.5 kcal mol −1 . [6]This is in line with the increased permeability of PLE to NH 3 as compared to H 2 O. Furthermore, as explicated in the previous chapter, OG_Kan enabled us to calculate NH 4 + permeabilities for PLE.In a similar fashion as for NH 3 we calculated the activation energy of NH 4 + permeation through PLE to E a, NH + 4 ,pH4.5 = 16.9 ± 1.0 kcal mol −1 .

Passive ion Permeabilities
With our novel OG_Kan based sensor we quantified weak base and ion permeabilities for a diverse set of lipid membranes and one representative polymer membrane.Utilizing stopped-flow methodology we subjected LUVs to an outside oriented NH 4 Cl gradient and measured the time dependent changes in fluorescence as depicted for PLE lipids in Figure 4. Similar measurements were performed for a mixture of DOPE/DOPG, mimicking the two major components of PLE but lacking the negatively charged cardiolipin.Moreover, we assayed DPPC in the gel state and a mimic of the sterol containing epithelial cell membrane (3:2:1 molar ratio of cholesterol:PLE:sphingomyelin).Finally, as a representative of synthetic membranes, we used PBD-b-PEO with a hydrophobic core thickness of ≈6.5 nm [63] (Figure S8, Supporting Information).
To ensure accuracy of the corresponding global fits to the data presented in Figure 4 and Figure S8 (Supporting Information) we measured P H + for PLE, DOPE/DOPG, and the PBD-b-PEO-based membrane in a separate set of experiments (Figure S2, Supporting Information).As can be seen from Figure S10 (Supporting Information) the P H + values of the other lipid-based membranes as well as all P f and P Na + values (listed in Table 2) are not so crucial for the accuracy of the respective global fits and could therefore be taken from the literature without hesitation.The combined results of our global fits to the vastly different membranes are presented in Table 3.
In line with previous work on passive water and weak acid permeability through membranes, [13,30] P NH 3 , as an example of weak base permeability, depends on membrane thickness, the bilayer phase state and the sterol content of the membrane.Moreover, our experiments show that the presence of 10 mol% cardiolipin reduce P NH 3 by a factor of 2.5 (Table 3).Interestingly, cardidid not have an effect on P NH + 4 which is nearly six orders of magnitude lower compared to P NH 3 .However, in line with a decreased P NH 3 through the polymeric membrane P NH + 4 is reduced 2 times in contrast to PLE and DOPE/DOPG containing membranes.P Cl − in the sterol rich epithelial like lipid composition is ≈9 orders of magnitude smaller as P NH 3 , with an estimated activation energy of 21.6 and 12.8 kcal mol −1 for Cl − and NH 3 permeation from the data in Table 3, respectively.Furthermore, the upper bounds of P Cl − < 1 × 10 −14 m s −1 for PLE, DOPE/DOPG and PBD-b-PEO compared to P Cl − = 0.4 × 10 −14 m s −1 for the epithelial mix at 277 K suggests a very different permeability barrier as for the neutral molecule NH 3 with more than an order of magnitude difference of P NH 3 for the same experimental conditions.Moreover, the effect of cholesterol and sphingomyelin on P NH + 4 is an order of magnitude larger as on P NH 3 .As a general outline our data suggests that P Na + < P Cl − < P NH + 4 < P H + < P H 2 O < P NH 3 irrespective of the membrane barrier.
Compared to literature values of DphpC membranes with P NH 3 = (480 ± 50) m s [20] our results of P NH 3 = 119 m s −1 for DOPE/DOPG and P NH 3 = 48 m s −1 for PLE are a factor of 4 to 10 smaller.However, P NH 3 = 13.1 m s −1 of epithelial cell membrane mimicking lipids at RT are in perfect agreement with P NH 3 = 16 m s −1 . [20]Besides these literature values which were estimated via studies on planar lipid bilayers with ion sensitive microelectrodes, measurements of a shift in the transmembrane potential after addition of a protonophore to the membranes revealed P NH 3 = 1300 m s −1 for egg PC-decane bilayers [4] and P NH 3 = 3.7 × 10 −2 cm s −1 for egg lecithin:cholesterol (1:1) bilayers. [21]Both values being more than an order of magnitude higher as compared to our results on comparable lipids.

Implications for Membrane Protein Characterization
Our knowledge of passive weak base membrane permeabilities are also essential for the characterization of membrane proteins.Insertion of membrane proteins into the membrane replaces a small patch of lipids thereby reducing the background permeability through the lipid matrix but enabling the substrate with a new, additional transport pathway.From the biological point of view, it seems clear that solute flux through transmembrane proteins is only then of significance if it enhances the overall permeability as compared to the background permeability through the lipid matrix.However, from a membrane biophysical point of view it is evident that the membrane protein of interest can only be functionally characterized if its substrate permeability is orders of magnitude higher as compared to the permeability through a similar sized membrane batch.One possibility to overcome this issue is to increase the amount of protein reconstituted into the model membrane system.However, membrane protein densities of more than several dozens of proteins per LUV (≈120 nm diameter) are seldomly reported.In addition, resolvable experimental kinetics have an upper limit too.These considerations define a range of membrane protein permeabilities which can be assessed by in vitro assays as explicated herein.As a rule of thumb, the higher the background permeability through the membrane matrix, the narrower the range of experimentally accessible unitary solute permeability (p m ) values.Let us consider an example of p NH 3 estimation through membrane proteins utilizing a LUV based sensor.With p NH 3 = 2.7 × 10 −14 cm 3 s −1 for RnAQP8 at RT [17] (recalculated to p NH 3 = 1.7 × 10 −14 cm 3 s −1 for 277 K with 4 kcal mol −1 [64,65] ) this would translate into 76, 126, and 311 AQP8 monomers per LUV to enhance P NH 3 = P NH 3 ,lipid + P NH 3 ,protein of PBD-b-PEO, PLE, and DOPE/DOPG at 277 K by a factor of 2, respectively.Only in the case of the epithelial lipid composition at 277 K or the epithelial lipid composition and DPPC at 295 K, the number of RnAQP8 monomers per LUV translates into a realistic number of 8, 22, and 1, respectively.Not surprisingly, an epithelial lipid composition was used to calculate the mentioned p NH 3 of RnAQP8 utilizing planar lipid bilayers and ion sensitive microelectrodes. [17]ence, calculation of p NH 3 values seem unrealistic in highly fluidic membranes usually used in in vitro assays.Increasing the sterol content of the membrane or changing the phase state of the lipid bilayer on the other hand are practicable options for the characterization of ammonia channeling proteins.

Conclusion
The herein reported dye OG_Kan expands the list of environment sensitive dyes which can be used in vesicle assays for membrane permeability quantification.Encapsulated into lipid-or polymerbased membranes with and without transmembrane proteins, its pH sensitivity enables passive weak base and ion permeability estimation and transmembrane protein characterization in the acidic pH region.Compared to OG488 labeled lipids, experimental results are independent of surface charge effects and solely reflect membrane permeability processes or differences in membrane protein function (Figure S12, Supporting Information).Moreover, the concentration dependent self-quenching properties of OG488 make it a potential sensor for the quantification of water and other neutral solute permeabilities like urea or glycerol in the acidic pH region.

Figure 1 .
Figure 1.Sensor principle.Schematic representation of lipid/polymerbased vesicles with encapsulated fluorescent reporter molecules.Relevant protonation and deprotonation reactions of the buffer M, NH 3 /NH 4+ and the pH sensitive dye as well as the self-dissociation of water are indicated.Application of an outward oriented NH 4 Cl gradient results in an acidification of the vesicle interior, which can be measured as a drop in fluorescence.Solute permeabilities P m can be extracted from a system of differential equations as outlined herein.

Table 3 . 4 [ 10 −13 m s − 1 ] 3 .
Summarized passive permeabilities through lipid-and polymerbased membranes.Corresponding datasets are shown in Figure 4 and Figure S8 (Supporting Information).LUVs are prepared from an epithelial lipid mixture (3:2:1 molar ratio of cholesterol:PLE:sphingomyelin), E. coli PLE, a DOPE/DOPG mixture (molar ratio of 2:1), PBD-b-PEO, and DPPC.Fit results which are at the resolution limit of the respective fits are highlighted in gray and are replaced by upper bound limits (Figure S11, Supporting Information).perr = np.sqrt(np.diag(pcov)) in Python (version 3.5) with n the number of data points varying between 8 and 16.The uncertainties of the global fit values were extracted utilizing Mathematica's NDSolve IDA package (V.12) (RRID:SCR_014448) and the Non-linearModelFit routine with a confidence interval of at least 95%.Values in the low mill range to neglect them in Tables 2 and Sample sizes are indicated for each measurement in the corresponding figure legend.

Figure 2 .
Figure 2. Influence of NH 3 , NH 4 + , and Cl − permeabilities on simulated time dependent fluorescence emission signals utilizing passive membrane permeability sensors at a pH of 4.5 (left column) compared to a pH of 7.0 (right column).Based on the parametrization of a PLE LUV at 277 K the passive membrane permeabilities of NH 3 , NH 4 + , and Cl − were varied to visualize their specific contribution on the measurement signal.Standard buffer conditions with an inside-out NH 4 Cl gradient of 7.5 mm were used.

Figure 3 .
Figure 3. Activation energy of NH 3 and NH 4 + permeability through E. coli PLE membranes.Arrhenius plot of P NH 3 through PLE membranes.A linear fit to the semilogarithmic plot enables the calculation of E a values.Experimental conditions were 300 mm NaCl, 10 mm of NH 4 Cl, 10 mm MES, 10 mm -Alanin, and 0.25 mm dye inside the LUVs and 300 mm NaCl, 3.3 mm of NH 4 Cl, 10 mm MES, and 10 mm -Alanin outside the LUVs.Experiments were performed with Cf-based sensors at pH 6.0 and OF_Kan-based sensors at pH 4.5.As exemplified in Figure 2 the acidic pH region enables the measurement of passive ion permeabilities.Raw data can be found in Figure S9 (Supporting Information).

Figure 4 .
Figure 4. NH 3 /NH 4 + permeability through E. coli PLE membranes at 277 K.The results of the global fit (red dashed line) to the data can be found in Table 3.Similar datasets for the epithelial lipid mixture at 277 K and RT, DOPE/DOPG at 277 K, PBD-b-PEO at 277 K, and DPPC at RT are depicted in Figure S8 (Supporting Information).The corresponding size distributions are shown in Figure S11 (Supporting Information).At time zero LUVs in working buffer are subjected to an outside oriented gradient of NH 4 Cl of 6.66 mm at varying pH conditions.Sample size is specified in brackets next to the corresponding pH value.

Table 1 .
of used parameters in the model with corresponding references.

Table 2 .
Used permeabilities in the model with corresponding references.