Detection and Discrimination of Natural and Synthetic Polysaccharides in a Solid-State Nanopore

In 2007-08 over 100 people died as a result of a contaminated batch of the polysaccharide heparin, an otherwise life-saving anticoagulant drug. After the contaminant was discovered, the development of assays that detect the contaminant, a structurally similar molecule, oversulfated chondroitin sulfate, became a necessity. Solid-state nanopores, which can, with appropriate experimental design, readily detect single molecules of analyte, may be able to help distinguish the two with greater ease than conventional assays, and with greater throughput even at concentrations well below that of USP assays. Polysaccharides, especially naturally occurring polysaccharides, have a vast range of structures characterized by widely varying molecular weights and charge distributions, and variability in linkage type. These polymers are challenging to analyze, and so studies using synthetic glycopolymers with known sizes and charge distributions, should be able to help one establish conditions to probe differences in molecular structure more easily. Under the right experimental conditions, solid-state nanopores were readily able to detect and distinguish between oversulfated chondroitin sulfate and heparin, and also synthetic glycopolymers of varying charge and length. This work may provide the necessary context to use nanopores for drug purity assays, to aid in understanding glycopolymer interactions, and also as a tool for characterizing polysaccharide structure and properties.


PREFACE
This thesis has been completed in manuscript format, and completed with two separate manuscripts.
Naturally occurring glycans are inarguably one of the most important classes of biologically active molecules, yet due to their complex chemical properties they are weakly characterized. Their roles extend across a multitude of biological processes 1-10 , but most recently the safety of therapeutic glycans has come into question 11 . Heparin, a linear unbranched glycan of the glycosaminoglycan family extracted from porcine intestinal mucosa, is a highly sulfated glycan with the largest negative charge density of any biologically active molecule and is a life-saving anticoagulant drug. However, in 2007-08 the deliberate adulteration of heparin resulted in mortality and morbidity across the world, resulting in widespread panic about the safety of this drug. The contaminant responsible for the observed adverse reactions was identified as oversulfated chondroitin sulfate 12 , a more sulfated version of the osteoarthritis supplement chondroitin sulfate, which bears an extremely similar structure to heparin.
As a result of the contamination crisis, methods for distinguishing between the two molecules became a necessity. Possible synthetic alternatives to heparin, glycopolymers, have also been considered 13 .
In the first manuscript of this thesis, nanopores, which have the ability to detect single molecules at a time, were used to distinguish heparin from oversulfated chondroitin sulfate. Preliminary results showed that resistive pulse nanopore measurements are able to distinguish between the two molecules, and after careful v optimization nanopores may serve as useful mediums in drug purity screening devices in the future. Different samples of the glycan sodium alginate were also analyzed with resistive pulse nanopore measurements and were able to highlight the differences in molecular structure of the two samples.
In the second manuscript, nanopores were used to detect synthetic glycopolymers generated by ring opening metathesis polymerization, from the group of Amit Basu at Brown University 14 . Glycopolymers have been shown to serve as natural glycan analogs and unlike natural glycans, have well characterized properties. This has allowed systematic studies of glycopolymer structure versus biological activity to be established elsewhere. Here we show that under the right experimental conditions, glycopolymers of differing length and charge can be selectively detected and differentiated in resistive pulse nanopore measurements, and they can also probe interactions of glycopolymers with other species. This work has highlighted the potential to study glycopolymer interactions at the single molecule level.
vii   unit absorption at ~1100 cm -1 allows calculation of the M/G ratio that varies with particular alginate source. 13 Using this approach, alginate A1 was determined to be ~63%G/37%M, and alginate A2 was ~57%G/43%M. These relative proportions were supported by additional analysis: in Supplementary Figure 3b, the particular alginate lyase was a mannuronic lyase, so that the greater absorption from the digestion of A2 than A1 was consistent with a greater proportion of M in A2………………………56  Oligo-and polysaccharides are ubiquitous in nature, with a broad spectrum of roles that includes energy-storage and provision (including as a foodstuff), structural building block (e.g. cellulose), therapeutic function (e.g. the anticoagulant heparin), and a vital part in biological recognition processes. [1][2][3][4][5][6][7][8][9][10][11] Conventional chemical analysis tools are frequently challenged by the daunting complexity of polysaccharide analysis: 12, 13 identification of monomer composition (~120 naturally occurring monomers!) and sequence, monomer linkage types, stereochemistry, polymer length, and degree of polymer branching. 13 These challenges were tragically driven home in 2008 when undetected contamination of the common anticoagulant heparin by a structurally similar adulterant, oversulfated chondroitin sulfate (OSCS), resulted in profoundly adverse clinical consequences in the United States, including ~100 deaths. [14][15][16][17][18][19] .

LIST OF TABLES
Glycan samples can be challenged by heterogeneity and low abundance in addition to chemical and structural diversity, so while new analysis tools have been broadly called for, 12, 13, 20 single-molecule-sensitive methods are a particularly compelling goal for glycomicsmore so given the absence of sample amplification techniques analogous to PCR for DNA sequencing 21 . Nanopore single-molecule methods have emerged as a powerful tool for characterizing DNA and proteins including aspects of sequence, structure, and interactions. [22][23][24][25][26][27][28] Monomer-resolved length determinations of more prosaic polyethylene glycol samples further buttress the potential of suitably configured nanopore assays for the analysis of polymers with biological utility. 29 The simplest implementation for nanopore measurements places the nanopore-a <100 nm-long nanofluidic channel through an insulating membrane-between two electrolyte solutions ( Figure 1). Ion passage through the nanopore in response to a voltage applied across the pore gives the baseline "open pore" current, i0; passage of a molecule into, across, or through the nanopore disrupts this ion flow to give a blocked-pore current, ib. A discernible current perturbation reveals the presence of an analyte, and the sign, magnitude, and temporal structure of ib depend strongly on size and shape of the analyte-and of the nanopore-and on the applied voltage and bulk and interfacial charge distributions. It thus provides insight into analyte presence, identity, and properties, including interactions between the analyte and pore interior or surface. [29][30][31][32] Analysis of the resistive-pulse characteristics of a sample offers the potential to glean molecular-level insights, but the ib characteristics can also be used more simply as benchmarks in quality assurance assays where atypical ib signal sample impurities.
Much groundwork must be laid, including proof-of-principle experiments, if nanopore methods are to emerge as a tool for glycan profiling-and by extension as a tool for -omics writ-large (spanning genomics, proteomics, and glycomics). Protein nanopores, polymer, and glass-supported nanopores have been used to detect sugar-pore binding, polysaccharides, and enzyme-digested oligosaccharides. [33][34][35][36][37][38][39][40][41][42] While solid-state nanopores in thin (~10 nm) membranes have been often portrayed as the preeminent nanopore platform, their use to profile classes of molecules beyond DNA and proteins is in its infancy. These nanopores can be size-tuned 43 to match analyte dimensions (especially relevant for branched polysaccharides), and when fabricated from conventional nanofabrication materials such as silicon nitride (SiNx), 44, 45 offer resistance to chemical and mechanical insult alongside low barriers to large-scale manufacturing and device integration. The potential for integration of additional instrumentation components, such as control and readout electrodes, around the thin-film nanopore core, is especially compelling. 28,44,45 Recent (nanopore-free) work on recognition electron tunneling measurements on polysaccharides, for example, has reaffirmed the importance of a nanopore development path that values augmented nanopore sensing capabilities. 46 A key question concerning the use of SiNx nanopores for polysaccharide sensing is whether this fabrication material is compatible with sensing glycans. The often challenging surface chemistry of SiNx (giving rise to a complex surface charge distribution) 44, 45, 47 may lead to analyte-pore interactions that hinder or prevent its use. Variability in polysaccharide electrokinetic mobility arising from differences in molecular structures may exacerbate the effect of these interactions. These issues become particularly important when analyte translocation through a constricted pore is required, such as in transverse electron tunneling measurements. 28,46 The aims of the present work were threefold: (1) [14][15][16][17] to make the analysis of heparin (~16 kDa) and OSCS by nanopore a compelling experimental test with clinical relevance.   Figure   2) that physically separated electrodes and nanopore, events were only detected when A1 was injected into the well proximal to the nanopore, thus supporting a signal generation mechanism involving interaction with the nanopore and not with the electrodes. This result did not, however, distinguish between passage-free collision with the nanopore opening ("bumping" or "blocking") or translocation through the pore. 32 Either mechanism (including extending the idea of "bumping" or "blocking" to allow for transient interactions of the analyte with the pore mouth), though, has the potential to deliver analytically useful sensing performance. Low analyte concentrations challenge the direct investigation of polysaccharide translocation through small, single nanopores. In one experiment to investigate this, a solution of A1 was added to the headstage side of a ~22 nm-diameter nanopore and was left overnight with a +200 mV applied voltage. The initially analyte-free contents of the ground-stage side were then transferred to the headstage side of a fresh ~17 nm-diameter pore, and an appreciable number of A1-characteristic events (182 in 1 h) were detected again at +200 mV.
Acid digestion was used as a signal generation and amplification technique (complete details in the Supplementary Information) to convert A1 polymers to many smaller fragment-derived species absorbing at ~270 nm. 51, 52 This spectrophotometric assay (Supplementary Figure 3) was used to confirm translocation of polysaccharide through a ~9 nm SiNx nanopore. residues exceeding that of A2, with values from IR spectroscopy of ~63%G/37%M and ~57%G/43%M, respectively. 48 Nanopore profiling of A2 showed differences compared to A1.
Using the same electrolyte for A2 as for A1, measurements generated a ~7-fold lower event frequency with longer durations for A2 compared to A1, in spite of at the 75-fold higher A2 concentrations required for reasonable measurement times. Enzymatic digestion of A2 produced events at a higher frequency than for undigested A2, but still at lower frequency than for A1. The events for the digested sample of A2 were ten-fold shorter-lived than for the A2 polymer, but not appreciably different in terms of blockage depth ( Figure 3).  SiNx pores at pH ~7. Alginate A2 is more negatively charged and so one would anticipate a stronger electrophoretic driving force; the direction of signal generation is still consistent with electroosmosis. The lower event frequency compared to A1 can be understood as arising from opposing electrophoretic and electroosmotic driving forces, but with the electrophoretic force on A2 being greater than on A1. More detailed exploration of the differences between A1 and A2 must also contend with their different molecular weights and their different chain flexibilities arising from their different M/G ratios. In the case of heparin, the charge density is sufficiently high so that events are detected using a voltage polarity that would drive the anionic polymer towards the nanopore. Experimental investigations including and beyond the ones presented here, exploring the underpinnings of the nanopore-generated signal using (polysaccharide) biopolymers with greater chemical and structural complexity than the canonical nanopore test molecule, DNA, or than homopolymers such as polyethylene glycol, should also provide fertile ground for high-level simulations. Interfacial effects will require additional study in the context of polysaccharides, but hold possibilities for tuning sensing selectivity and sensitivity. Indeed, explicit consideration of sensing conditions-including nanopore size, electrolyte composition, and voltage polarity-already augments the ability to compare nanopore molecular fingerprints as shown in Figure 3.
The failure in 2008 to detect an OSCS contaminant in clinical heparin samples had previously led to patient morbidity and mortality, [14][15][16][17][18] so that our ability to use a simple nanopore-based assay to quantify heparin levels and detect OSCS at clinically meaningful contamination levels, is itself significant. In a broader sense, we expect that these initial results exploring polysaccharide structure can, by analogy with earlier nanopore DNA and protein sensing supporting genomics and proteomics, spotlight the potential of using nanopores as a

Author Contributions
All authors have contributed to, and approve, the manuscript.  Here we present the idea that glycopolymers, glycan analogs that can mimic the activity of a range of molecules, can also serve as useful analogs in nanopore measurements, which allows the ability to understand glycan activity at the single molecule level. We aim to use glycopolymers of different chain length and charge density, and be able to differentiate and selectively detect them using resistive-pulse measurements inside a nanopore. We also aim to test a nanopore's ability to probe the complexation interaction between an anionic glycopolymer and the cationic amino acid, Poly-L-lysine.

RESULTS
The synthetic glycopolymers used in this study carried a pendant galactose ring.     KCl, pH=3, and a voltage of -50 mV were able to detect both Sgal-30 and Sgal-90, but didn't afford the ability to clearly differentiate between the two glycopolymers. Although the glycopolymers are different chain lengths, the mechanism of passage through the nanopore may not be shown in the relative current drops during an event, especially if they pass through the pore in a linear fashion due to the inability to distinguish analyte size by fb values, as Sgal- increase the signal-to-noise ratio in nanopore measurements by increasing the blockage depth during an event by enhancing the electric funneling field near the nanopore entrance 37 , a phenomenon that has been under extensive theoretical study [38][39] . Upon using the salt gradient, the event frequency increased by a factor of >10, and increases in the current blockage magnitude were recorded which allowed discrimination between Sgal-30 and Sgal-90.
Overall, much larger event depths were recorded for Sgal-90 when compared to the most frequent blockage depths of Sgal-30, and were present in a 50/50 Sgal-30+Sgal-90 mixture.  We hope that these initial nanopore measurements with glycopolymers may lead to more detailed studies of their possible interactions with a variety of molecules.

METHODS
A full listing of the experimental details is available in the Supplementary Information. Nanopores were formed via dielectric breakdown 40 in nominally 10 nmthick silicon nitride (SiNx) membranes. Nanopore sizes were inferred from their conductance, G, determined from Ohmic current-voltage data assuming a cylindrical nanopore shape and bulk and surface conductances. Nanopores used for measurements produced stable open-pore (analyte-free) currents in the electrolyte solutions used. Glycopolymers were made by Blais Leeber from Brown University from the group of Amit Basu. For routine measurements, sample aliquots were added to the headstage side ( Figure 1), leaving the ground side free of initially added analyte. Current blockages were extracted using a current-threshold analysis. All applied voltages are stated with the polarity of the electrode on the headstage side relative to ground on the ground side of the sample cell (As shown in Supplementary Figure 1).

Author Contributions
All authors have contributed to, and approve, the manuscript. aliquots were added to the headstage side (Figure 1), leaving the ground side free of initially added analyte. Calibration curves for each nanopore were constructed by repeated cycles of measurement followed by the addition of another analyte aliquot. Current blockages were extracted using a current-threshold analysis. Any current blockages exceeding 100 s (≲ 0.1%)

Supplementary Information
were not included in analyses.  1 3 ⁄ to be ~19 nm for A1 and ~8 nm for A2 (on an n -basis). The corresponding rootmean-squared end-to-end distance, 〈 2 ̅̅̅ 〉 1 2 ⁄ for each sample is equal to 3.1 h .

Acid and Enzymatic Digestion Procedures.
A ~9 nm nanopore was mounted in the PTFE sample holder. A 200 μL amount of 0.2% (w/v) A1 was added to the head stage side in 5 µL aliquots per hour throughout the work day during 4 days of application of a +200 mV cross-membrane voltage. For overnight voltage applications, the electrode polarity was maintained, but the electrodes were placed in the opposite wells. The head-stage and initially analyte-free ground side solutions were extracted, individually mixed with 1 mL of 75% sulphuric acid and heated overnight (16 h) at 80°C.
Samples were diluted with 3 mL of water before spectral acquisition.
A 2250 µL aliquot of 0.2% (w/v) A1 was added to a 150 µL aliquot of 1 unit/mL alginate lyase and heated in a water bath at 37˚C for 30 minutes. Samples of 3% (w/v) A2 were mixed with alginate lyase (1:1 (v/v) mixture with 1 unit/mL enzyme) for 10 minutes at 37°C. 20 μL of this mixture was added to the headstage side and events were detected with the application of +200 mV on the head stage side. Measurements in the presence of 20 μL of 1 unit/mL of alginate lyase, alone, in the headstage side support that the detected events in the presence of analyte originated from enzymatic digestion products.

Preparation of Heat Maps by Histogramming Individual Events.
Heat maps were prepared in Origin (Originlab Corporation, MA) from event data sorted into bins by paired and . The bin width along the axis was set equal to 3.49 ( ) − 1 3 , where ( ) is the standard deviation across all events, and N is the total number of events. 9 Bin size along the axis was set to √10. Heat maps are plotted using log10 of the number of events in each bin.

Recognition Flag Generation
Recognition flag generation was done using custom codes written in Mathematica 11.0.1.0 (Wolfram, Champaign, IL). (1) All individual events were histogrammed with respect to fb using a bin width of 0.0025 (using nanopores with diameters from ~8-14 nm, and determined using the USP heparin data). (2) Any bin with counts below 0.5% of the maximum bin count were removed, and all counts were then normalized. (3) The OSCS identification threshold was taken to be at the nearest bin at the distance of three standard deviations (after the 0.5% filter) from the bin with the maximum number of counts. (4) When events had been detected at fb below this threshold, the recognition flag was set to red to signal the presence of OSCS; it was otherwise left white.
(5) All individual events were then histogrammed with respect to the logarithm (log10) of the event duration (τ) using a bin width of 0.25 (here, determined using the USP OSCS data). (6) The same 0.5% filter was applied to these histograms, which then had their counts normalized. (7) The event duration threshold was taken to be the nearest bin at the distance of three standard deviations (after the 0.5% filter) from the bin with the maximum number of counts. (8) When events had been detected at log10τ above this threshold, the recognition flag was set to red to signal the presence of heparin; it was otherwise left white.
Supplementary Figure 1: Calibration curve of sodium alginate event frequency versus volume of 0.2% (w/v) A1. Two trials were performed, with each data point including at least 1000 events extracted from at least 1 h long measurements at 200 mV applied voltage after consecutive additions of 4 µL aliquots to the headstage side of the same nanopore. Error bars represent the standard deviation across the trials.
Supplementary Figure 2: A special nanopore configuration in which the electrolyte wells proximal to the electrodes and to the nanopore were physically separated. The purpose of this configuration was to determine if the current blockages arose from analyte interaction with the electrodes, or with the nanopore, itself. The electrolyte wells in the lower PTFE cell held the electrodes and were separated by an intact SiNx membrane that did not allow ionic flow. These wells were connected through electrolyte-filled silicone tubing and an electrolytefilled beaker, to a second electrolyte-filled PTFE cell in which the wells were separated by a SiNx nanopore. With analyte injected into the bottom cell, the only possible mechanism of current blockage was either by direct interaction with the electrodes, or by the passage of analyte through the tubing and beaker of solution until it could interact with the nanopore. When a 4 µL aliquot of the alginate was added to the head stage side of the lower cell, only 18 appreciable current transients were detected in a 1 hour measuring period, contrasted with 561 events in 1 hour when the alginate was directly injected adjacent to the head stage side of the nanopore. The additional electrolyte between electrodes and nanopore reduces the cross-pore applied potential compared to the usual singlecell sensing configuration.
Supplementary Figure 3. UV/Vis spectra of acid and enzymatic digestion products. a) Stock A1 subjected to 16 h of sulphuric acid digestion generated a UV/Vis spectrum characteristic of the digested polysaccharide 10, 11 that was replicated in the samples taken from the headstage and from the groundstage sample wells after 4 days of a translocation experiment (200 µL aliquot). The dashed lines denote the UV/Vis spectra of the sample before digestion, and the solid lines denote the spectra after digestion. b) Alginate lyase digestion of alginate is expected to introduce chromophores with a peak absorption at ~232 nm, consistent with observations here. 12

Preparation of Heat Maps by Histogramming Individual Events.
Heat maps were prepared in Origin (Originlab Corporation, MA) from event data sorted into bins by paired fb and τ. The bin width along the fb axis was set equal to Wbin = 3.49σ(fb)N -1/3 , where σ(fb) is the standard deviation across all events, and N is the total number of events. 13 Bin size along the τ axis was set to √10. Heat maps are plotted using log10 of the number of events in each bin.
The distributions of event counts by fb in Supplementary Figure 4 were fit using the function where the parameters of the unmodified Gaussian function are as conventional -Ai, µi, and σi are the magnitude scaling, expected value, and standard deviation. The step function, (1+θ) was set to 1 for fb < fb cutoff + Wbin and 0 otherwise, so that the fit function covers only the accessible experimental data (fb cutoff was the threshold for event extraction The distributions of the log of event counts by duration were fit to a log-normal distribution = −(ln − ) 2 (2 2 ) ⁄ where the parameters had conventional meanings, and the event duration was expressed in µs. The event duration corresponding to the peak of the event count distribution, τp , was found by taking the first derivative of the curve.  Figure 4: Histograms of (top row) <ib>/<i0> (bottom row) duration in log10 of A1 alginate in (a) ~5 nm and (b) ~19 nm pore, A2 in (c) ~22 nm, (d) 10-min enzyme digested A2 in ~23 nm pore, (e) heparin and (f) OSCS in the same ~14 nm pore with the bin size set automatically by the measurement statistics as described above.
Supplementary Figure 5: Plots of log10 of event duration (τ) versus area under each event for alginate A1 in a) ~5 nm and b) ~19 nm diameter pores and c) for alginate A2 in a ~22 nm diameter pore recorded for 1 hour in 1 M KCl at pH ~7. Two distinct event distribution tails are visible corresponding to short-lived spike-like pulses and longer-lived rectangular blockages. The longer-lived tail for A2 is more prominent as a percentage of total events than for A1, consistent with the appearance of the combined heat and scatter plots in Figure 3. The shorter events could be attributed to either "bumps" or fast translocations, and longer-lived events could be attributed to slower translocations or longer-lived interactions with the pore (in both cases, complementary measurements independently confirmed alginate translocation). The low molecular weight and high M/G ratio (more G is attributed to stiffness) of A2 meant, it has a greater probability of Supplementary Figure 7. Infrared spectra of alginate samples. The intensity of the peaks near 1400 and 1600 cm -1 , relative to the remainder of the spectrum, are consistent with a lesser proportion of carboxylic acid salt residues in (a) A1 than in (b) A2. Comparison of the intensity of the guluronic (G) unit absorption at ~1025 cm -1 to the mannuronic (M) unit absorption at ~1100 cm -1 allows calculation of the M/G ratio that varies with particular alginate source. 13 Using this approach, alginate A1 was determined to be ~63%G/37%M, and alginate A2 was ~57%G/43%M. These relative proportions were supported by additional analysis: in Supplementary Figure 3b, the particular alginate lyase was a mannuronic lyase, so that the greater absorption from the digestion of A2 than A1 was consistent with a greater proportion of M in A2.
Supplementary Figure 8. Heparin and OSCS events. A representative a) i) segment of a heparin induced-current trace using a ~10 nm-diameter pore with a magnified current event from the same trace, and from ii) OSCS through the same pore in response to a -200 mV applied voltage in 4 M KCl at pH ~7. b) Contour+scatter plots of i) heparin, ii) OSCS and iii) heparin contaminated with OSCS through a ~14 nm diameter pore.

Recognition Flag Generation
Recognition flag generation was done using custom codes written in Mathematica 11.0.1.0 (Wolfram, Champaign, IL). (1) All individual events were histogrammed with respect to fb using a bin width of 0.0025 (using nanopores with diameters from ~8-14 nm, and determined using the USP heparin data). (2) Any bin with counts below 0.5% of the maximum bin count were removed, and all counts were then normalized. (3) The OSCS identification threshold was taken to be at the nearest bin at the distance of three standard deviations (after the 0.5% filter) from the bin with the maximum number of counts. (4) When events had been detected at fb below this threshold, the recognition flag was set to red to signal the presence of OSCS; it was otherwise left white.
(5) All individual events were then histogrammed with respect to the logarithm (log10) of the event duration (τ) using a bin width of 0.25 (here, determined using the USP OSCS data). (6) The same 0.5% filter was applied to these histograms, which then had their counts normalized. (7) The event duration threshold was taken to be the nearest bin at the distance of three standard deviations (after the 0.5% filter) from the bin with the maximum number of counts. (8) When events had been detected at log10τ above this threshold, the recognition flag was set to red to signal the presence of heparin; it was otherwise left white.
Supplementary Figure 9. Hue plots of show the outcomes of recognition flag generation (and measurement statistics-see procedure detailed above) after steps 3 (top) and 7 (bottom), based on fb = <ib>/<i0> and log10 τ of the individual events. The identification threshold, determined by the measurement statistics of each run, is given by the blue line. The corresponding final recognition flags, showing successful detection of the toxic OSCS impurity across four independent trials in ~8. 6, 9.8, 9.9, and 13.6 nm (left to right), are shown in Figure 5.    Table 1: Number of recorded events in a ~17 nm nanopore after the addition of 3 µL of 0.02% (w/v) Sgal-30 with an applied voltage of -200 mV for 20 minutes. Given the much higher event frequency using KCl, it was used in all further experiments. Figure 4: Current trace of a ~17 nm nanopore at -50 mV when no analyte was present in the headstage side electrolyte well of the nanopore (top) compared to that when 5 µL of 0.02% (w/v) Sgal-30 and 5 µL of 0.02% (w/v) poly-L-lysine was run at +50 mV to try and detect an Sgal-30-poly-L-lysine complex (middle). A zoomed in portion of the middle section (bottom) shows current drops but failed to provide enough signal-to-noise to distinguish the current drops as events.