Large‐Peptide Permeation Through a Membrane Channel: Understanding Protamine Translocation Through CymA from Klebsiella Oxytoca

Abstract Quantifying the passage of the large peptide protamine (Ptm) across CymA, a passive channel for cyclodextrin uptake, is in the focus of this study. Using a reporter‐pair‐based fluorescence membrane assay we detected the entry of Ptm into liposomes containing CymA. The kinetics of the Ptm entry was independent of its concentration suggesting that the permeation through CymA is the rate‐limiting factor. Furthermore, we reconstituted single CymA channels into planar lipid bilayers and recorded the ion current fluctuations in the presence of Ptm. To this end, we were able to resolve the voltage‐dependent entry of single Ptm peptide molecules into the channel. Extrapolation to zero voltage revealed about 1–2 events per second and long dwell times, in agreement with the liposome study. Applied‐field and steered molecular dynamics simulations added an atomistic view of the permeation events. It can be concluded that a concentration gradient of 1 μm Ptm leads to a translocation rate of about one molecule per second and per channel.


Introduction
Rapid and label-free monitoring of drug uptake into Gram-negative bacteria is ab ottleneck in antibiotic drug discovery.T he outer membrane of Gram-negative bacteria acts as as elective barrier for the uptake of small molecules including antibiotics and antimicrobial peptides.T he outer membrane itself is nearly impermeable but rich in water-filled transmembrane proteins called porins that facilitate the passive permeation of small hydrophilic molecules below molecular weights of 600 Da. [1,2] In contrast, large molecules are generally excluded but selected ones find their way via self-promoted uptake. [2][3][4][5] Antimicrobial peptides often act via their lytic activity. [6,7] Ty pical examples are cationic antimicrobial peptides that have attracted wider interest as substitutes for classical antibiotics. [8][9][10][11][12][13][14][15][16] Whether these polycationic peptides permeate using as elf-promoted pathway or through channel proteins remains an open but crucial question in understanding their antimicrobial activity and other putative functions.O ne such molecule is protamine (Ptm) which is a32amino-acid-long polycationic peptide with amolecular mass of 5.1 kDa that contains 21 arginine residues ( Figure 1A). This molecule is ac heap byproduct from fish industry and used as an antibiotic in fish farming. [17] Ptm does not show lytic activity but seems to enter the periplasmic space of Escherichia coli, Salmonella typhimurium,a nd Pseudomonas aeruginosa. [5] So far,t he mode of action, i.e., self-promoted uptake or uptake via ap rotein channel, is unknown.
Recently,w eq ualitatively characterised the permeation of Ptm through OmpF by using af luorescence assay [18,19] which is based on the encapsulation of ahost-guest complex between calixarene (CX4) and lucigenin (LCG) into the aqueous interior of vesicles. [20] Theh ost-guest pair has to be selected for negligible membrane permeability and has to be sufficiently large in order not to permeate through bacterial membrane proteins like OmpF. [19] Analytes that permeate across the membrane and bind to the synthetic CX4 receptor (such as cationic peptides) affect ad isplacement of the dye LCG. [18,21] This technique is known as "tandem assay" and has been extensively applied to assess enzyme activity, [22] chirality recognition, [23] and drug delivery, [24] as well as for chemosensing in cells. [25] Herein, we unravel the molecular details of the passive porin uptake of large substrates employing aj oint experimental and computational approach. As model porin, we selected CymA from Klebsiella oxytoca ( Figure 1B), ac hannel which allows the passage of large cyclic a-cyclodextrin molecules (approx. 1kDa) [26] and which, after deletion of 15 N-terminal residues,c onstitutes ac ation-selective hollow channel with a1 5 diameter termed DCymA. [27,28] This channel has been considered because the large pore size allowing passive permeation of large cyclic molecules also provides ap otential route for the uptake of antimicrobial peptides like Ptm.
First, we used the tandem membrane assay to characterise the Ptm permeation into the interior of the liposome by fluorescence ensemble measurements. [18] To obtain an accurate single-molecule picture of the translocation, we further employed single-channel electrophysiology in order to monitor potential translocations across DCymA. Finally,a nd to gain detailed atomistic insight into the translocation mechanism that is elusive to experiments,applied-field and steered all-atom MD simulations were performed.

Ensemble Measurements of Protamine Translocation through CymAb yt he Tandem Membrane Assay
In af irst series of experiments,w ei nvestigated the translocation of Ptm across the DCymA channel by using the previously introduced tandem membrane assay. [19] Briefly, the procedure includes the encapsulation of the CX4/LCG reporter pair into the interior of liposomes as schematically shown in Figure 2A.S ubsequent insertion of DCymA channels into the liposome membranes followed by an addition of Ptm is expected to lead to afluorescence turn-on response by displacement of the dye by the protamine from the macrocyclic cation receptor CX4 (see Section 1ofthe SI for further details).
Fluorescence measurements in two control experiments performed without using Ptm molecules or DCymA channels resulted in af lat line excluding al eakage of CX4 or LCG through the pore or ap assage of Ptm molecules through the lipid membrane (see Figure S2). Subsequently,d uring the actual translocation experiments,a ddition of Ptm molecules and DCymA pores in any order led to an initially rapid increase in the fluorescence followed by as lower rise.T hese results strongly suggest that Ptm permeation occurs selectively through the DCymA channels.S ubsequently,w e performed as et of experiments by varying the Ptm concen-  tration (0.1-1.0 mm)a nd monitoring the initial rates for the fluorescence increase,w hich were found to be closely comparable ( Figure 2B with az oom in Figure 2C). Accordingly,w es uggest that the rate-limiting step for Ptm passing through the DCymA channel is the exit from the channel, most probably due to strong electrostatic interactions of the arginine-rich Ptm with the acidic residues of the pore ( Figure 1B). From these experiments,t he flux for a1mm Ptm concentration gradient is estimated to be roughly equivalent to 1-3 Ptm molecules per second through one DCymA channel (see Section 1i nt he SI for more details). From ab iological perspective,t he transport through wildtype (WT) CymA channels has al arger significance.T he Nterminal loop is weakly bound inside the CymA channel and modulates the translocation of large peptides through the porin ( Figure 1B). [28] Indeed, the experiments performed using WT CymA as ap ore show also an increase in fluorescence,c onfirming an influx of Ptm molecules (see Figure S3 in the SI), but the kinetics,asjudged from the initial rate,i ss lightly slower than for the mutant DCymA pore lacking the back-folded loop.Nevertheless,itshould be noted that inside abacterial outer membrane the behavior of the Nterminal loop might be different or it may have an unknown biological function.

Determining the Translocation Rates Using Electrophysiology
Single-channel electrophysiological measurements were performed next in order to investigate the translocation of Ptm molecules through CymA channels at as ingle molecule level. In essence,atransmembrane electric field is applied to pull the cationic peptide through as ingle channel reconstituted into af lat lipid bilayer membrane.T he entry of each Ptm peptide,w hich can be resolved at the single-molecule level with this technique,i so bserved by ac omplete ion current blockage.Astatistical analysis of the blockage lengths or dwell times as af unction of external voltage is used to reveal potential Ptm translocations as well as its kinetics through the CymA channel or its mutant. Concentrationdriven translocation can further be investigated by measuring the reversal potential created by aPtm sulfate gradient.
Starting with two compartments filled with 1 m KCl, as ingle WT CymA channel was reconstituted into the DPhPC bilayer from the side that is electrically grounded, also called the cis side,a nd ion current traces were recorded by applying at ransmembrane potential, V m ,a tt he trans side (see Section 2inthe SI for Materials and Methods). As can be seen from Figure S5A in the SI, WT CymA is an electrically noisy channel because of the flexible N-terminal loop that is weakly attached inside the channel from the periplasmic side. [28] Thep utative movement of the N-terminal loop into and back out of the pore creates rapid blockages of the channel. Removing the 15 amino acids long N-terminal peptide silences the channel in an open form and this structural modification allows the experimental quantification of Ptm translocations.
The DCymA channel was inserted from the cis side.After asuccessful single-channel reconstitution was achieved, 1 mm of the substrate Ptm was added to the cis side of the chamber and at ransmembrane potential V m of negative polarity was applied on the trans side.This resulted in discrete ion-current blockages with complete closures of the pore current and resolvable dwell times in the millisecond range ( Figure 3A). Such abehaviour is expected as the Ptm molecule is strongly cationic (+ 21 e)a nd with the application of negative potential, the molecules are pulled through the pore due to the electrophoretic force.Whenthe substrate is administered on the opposite side (trans side), ap ositive potential creates similar ion current blockages ( Figure S6A), suggesting that apermeation from the other side is also feasible,with similar efficiency.I ncreasing the magnitude of V m further increased the number of blockages in both cases ( Figure S7 in SI), meaning that the molecules are pulled at af aster rate across the channel by virtue of astronger electrophoretic force.
To quantify the findings,w eh ave analysed the event rate f e which specifies the number of events per second and the dwell (or residence) time t d of Ptm molecules in the channel that is obtained from an exponential fit of the dwell time distribution. Forb oth quantities at least 1000 events were analysed for each transmembrane voltage applied in the range from AE 20 to AE 100 mV.InFigure 3B,C,the values of f e and t d obtained from the experiments performed after cis-side addition of Ptm and negative applied voltages are shown. Larger negative voltages increase the number of events, f e , and decrease the dwell times, t d .Asimilar trend was observed for insertion of Ptm at the trans side and positive applied voltages ( Figure S6B,C). This behaviour confirms that by increasing the magnitude of the applied voltages,m ore molecules are dragged through the channel. In previous studies,asimilar trend was observed for the permeation of shorter poly-arginine molecules through hetero-oligomeric channels and through the general diffusion channel OmpF. [29] In asecond set of electrophysiological measurements,we characterised the concentration-driven relative permeation by using the so-called reversal potential measurement technique [30,31] (see Section 2inthe SI for details on Material and Methods). Starting with two compartments having 0.1 mm potassium sulfate separated by a DCymA channel embedded in aDPhPC bilayer,aconcentration gradient of 0.25 mm Ptm sulfate was introduced on the cis side.Inprinciple,the largesize Ptm molecules are expected to have al ower electrophoretic mobility compared to that of the sulfate counterions. Application of a0 .25 mm salt gradient caused an ion flux leading to aquasi-equilibrium transmembrane voltage,the socalled reversal potential, of about À20 mV (see Figure S9 in the SI). Thereversal potential approach is relatively simple to measure but less straightforward to analyse and interpret for more complex ions such as highly charged polypeptides.F or the further analysis,weassumed the effective charge of aPtm molecule to be + 21 e due to the 21 arginine residues. Moreover,1 1m obile sulfate ions,e ach having ac harge of À2 e, are assumed to roughly counterbalance the respective charges.A sd etailed in the SI, these charge states lead to apermeability ratio for K + /Ptm 21+ /SO 4 2À of 1:1:48. This ratio implies that Ptm molecules as well as sulfate ions may permeate across the DCymA channel and that for each Ptm molecule 48 counterions are permeating until the reversal potential has balanced this flux. It can be assumed that divalent SO 4 2À ions binding at the inner pore surface will retain the natural cation selectivity of the pore,u nlike Mg 2+ ions which change the pore preference to anions. [27] Fora1mm concentration gradient of protamine sulfate across the membrane,t he approximate flux can be determined to be three molecules per second for protamine and 60 molecules per second for sulfate,w hich is in close agreement with the rates estimated from the fluorescence assay and from the single-channel electrophysiology experiments (see above).

Atomistic Details of the Protamine Translocation Mechanism by MD Simulations
Lastly,w es tudied the translocation of aP tm molecule through the DCymA channel by using molecular dynamics simulations (see Section 3inthe SI for details on the Material and Methods). Tw os tates of Ptm, i.e., the helical and the disordered conformation ( Figure 1A), were considered for the simulations in order to elucidate which one will be the best suitable form for translocation. Moreover,t he intention was to determine the feasibility of aP tm molecule translocation through the pore.
Initially,w ec arried out applied-field MD simulations by placing both Ptm conformations at the extracellular side in separate simulations.Apositive external bias voltage of 1V was applied and the simulations were conducted for 4and 6 ms for the helical and disordered conformations,respectively.As shown in Figure S10 in the SI, the C-terminus of both Ptm structures enters the channel first. This orientation is likely caused by the presence of more densely packed arginines at this end compared to the N-terminus,w hich is electrostatically favoured by acidic residues on the channel wall. TheCtermini of both Ptm conformations reach the middle of the channel within 1 mss imulation time,b ut full translocations have not been achieved in the remaining simulation times.It is clear that the limited but already long simulation times are not sufficient to observe full translocation events.N otably, ad eformation in the b-barrel of the channel was observed towards the extracellular side when the helical Ptm conformation was moving further into the channel (see Figure S11 in the SI). This finding suggests that Ptm is too bulky in its helical state and it has to become fully or partially unfolded into the disordered form before or while permeating through the channel.
To achieve insights into the complete permeation process, we additionally conducted steered MD simulations by pulling the C-terminus of both Ptm conformations along the channel axis.T he resulting force profiles shown in Figure 4A suggest that the forces required to pull the helical form through the channel are slightly higher than those for the disordered form, i.e., the respective maximum forces approach 950 and 890 pN. Interestingly,t he metastable states extracted from the permeation process reveal that helical Ptm unfolds into the disordered form before traversing the channel completely ( Figure 4B). This unfolding process is the reason behind the higher forces observed for the helical Ptm because the permeation of the disordered Ptm conformation occurs in ar ather straightforward manner (see Figure S12 in the SI). This finding also supports our earlier claim that the helical Ptm conformation cannot directly traverse through the channel but has to transform into the less bulky disordered form. Moreover,t he observed forces for the Ptm molecules are not extremely high in comparison to those observed for the "native" large-molecule analytes of CymA, namely a-and b -cyclodextrins,since the highest values for the cyclodextrins are approximately 500 to 700 pN [27] for similar steering velocities.A lthough the obtained simulated data are only average forces and not free-energy profiles,the comparison of the absolute values certainly suggests that it is feasible for Ptm molecules to permeate through the DCymA channel. At the molecular level, the negatively charged acidic residues found along the interior surface of the channel facilitate the gliding of the arginine-rich Ptm through the DCymA channel. Interestingly,P tm molecules have four arginine-rich regions ( Figure 1A)a nd entrance of each of these regions into the pore resulted in the four metastable states It oI V ( Figure 4B). It can be also suggested that states III and IV,during which most of the Ptm structure interacts strongly with the pore,m ust be the rate-limiting states,w hich is supported by the highest necessary pulling forces.

Discussion
Thep resent study demonstrates that the combination of different complementary techniques allows an improved understanding of the translocation of large (5.1 kDa) polypeptides across the DCymA nanopore.T he fluorescence study on liposomes unambiguously established that translocation of Ptm through the wild-type and DCymA channels takes place with atranslocation time in the order of seconds. Concentration-driven fluxes can be estimated by using reversal potential measurements on planar lipid bilayers and revealed ap ermeability ratio of 48:1 for sulfate vs.P tm molecules.F urthermore,s ingle-channel electrophysiology experiments revealed that the polycation can be pulled to and through the pore by electric fields and the comparison of these single-molecule data with the ensemble measurements from the fluorescence assay also establishes that the observed events are actually due to translocation and not blockage. Higher negative voltages accelerate the translocation process and extrapolation of these results to zero voltage revealed rates of about one molecule s À1 for agradient of 1 mm Ptm, in good agreement with the ensemble fluorescence kinetics. Moreover,the available high-resolution structure allowed allatom modelling which revealed molecular details of the translocation process,m ost important, ap referential uptake of Ptm in its disordered conformation.
It is interesting to compare the present results to the translocation of neutral a-cyclodextrin molecules (M W = 973 Da) through DCymA. Fort he latter molecules,a n extrapolation to similar concentration gradients reveals about 10 molecules per second with slightly shorter translocation times in the ms range.N ote that the slightly larger bcyclodextrin (M W = 1135 Da) is unable to permeate the membrane.A nother comparison can be made with the antibiotic molecule kanamycin (M W = 484 Da) for which very strong interactions with OmpF were observed during translocation. [32] Although the molecular size is an order of magnitude smaller than that of Ptm and the pore has half the diameter of that of DCymA, the event rate is about the same whereas the dwell time is in the ms range.Overall, this trend suggests that in the case of bacterial channels,t he molecular weight is only arough exclusion criterion whereas the molecular shape and pore-mediated interactions will finally control the translocation kinetics.M oreover, we conclude that even small concentration gradients are able to enforce the diffusion of large molecules like Ptm through narrow orifices.

Conclusion
Our investigation revealed that a5kDa large peptide can permeate passively through mesoscopic channels.Atconcentration gradients relevant for antimicrobial activity (around 1 mm)a bout one molecule per channel and per second can enter bacteria, which is in the relevant range to cause activity inside the bacteria. Note that acommon strategy in developing novel antibiotics is to create hybrid molecules.C rosslinking peptides,sugars,ornucleic acids as a"carrier" with the antibiotic as acargo is expected to circumvent the resistance barrier. Our finding might encourage the development of hybrid antibiotics covalently attached to peptides.Some years ago,L ee and co-workers pulled short peptides through ab iological channel and correlated the resulting ion current flickering with the peptide sequence. [33] By now this field is much more advanced and under favourable conditions even allows for adiscrimination of single amino acids. [34] Moreover, wide channels such as CymA allowing the passage of peptides are of key interest for peptide sequencing.