Review

 [Link]

The Promise of Nanopore Technology: Advances

in the Discrimination of Protein Sequences and Chemical
Modifications
Benjamin Cressiot, Laurent Bacri, and Juan Pelta*

ever, only 21 000 distinct protein-coding

Only a small percent of human genomic DNA encodes for proteins. Addi- genes were identified.[2] Meaning just a tiny
tionally, protein isoforms variants and chemical modifications are not coded percentage of DNA has the information
in the genome read by the cell machinery. The resulting protein diversity is necessary for translation of all the pro-
deeply involved in regular and diseased cellular processes. One challenge teins present in human cells. Among these
genes, only 150 are systematic targets of
for the field of biotechnology, after human genome sequencing, will be to somatic cancer mutations 2850 are drivers
decipher the proteome at a single molecule scale to analyze single-cell protein of rare diseases when mutated and ≈1100
variability. In fact, cellular proteic information, often used as a source of are involved in diseases caused by multiple
biomarkers, is of great importance for early disease detection. This review genes.[3] We understand now that muta-
discusses the proteome’s complexity from its genetic source to fully modi- tions are not the only factor responsible
for human disease. Epigenetics, another
fied proteins. It focuses on the principle of nanopore data analysis and how
factor of interest responsible for disease, is
to obtain information from an electrical current trace. Specifically, the most influenced by several factors such as envi-
recent developments in detection, sequencing and post-translational discrimi- ronment, aging, and lifestyle. One of the
nation of amino acids, peptides and proteins, are described. The main results major sources of epigenetic changes is pro-
obtained in this field are discussed and the nanopore techniques to other tein chemical modification.[4,5] Post-transla-
classical or single-molecule approaches are compared. tional modifications consist of proteolytic
cleavage or covalent addition of a chemical
group to one or more amino acids, altering
protein properties.[6]
1. Introduction Eukaryote chemical modifications regulate many protein
biological functions[7,8] such as apoptosis (cell death), epi-
The international Human Genome Project started over 40 years dermal growth factor receptor signaling, endocytosis, DNA-
ago. The objective was to identify all genes and to understand damage responses, and immunity. These modifications are also
their function to accelerate biomedical research. The human involved in a lot of human diseases:[9–11] cancer apparition and
genome was initially sequenced and analyzed by an interna- dissemination, autoimmune pathologies, neurodegenerative
tional consortium that was published in 2001.[1] The location of diseases or even type 2 diabetes. These chemical modifications
then novel and known genes was described along with a genetic are essential for therapeutic protein manufacturing.[12] More
map to identify genes implied in human diseases.[1] generally, the proteome has a key regulatory role in physiolog-
Using genetic prediction approaches, scientists expected the ical processes that dictate disease development.[5]
human genome to contain about 105 protein-coding genes. How- An immense number of distinct proteins, at least 106 dif-
ferent molecules, resulting from open reading frame (ORFs)
Dr. B. Cressiot variants, splice variants, and functional post-translational modi-
CY Cergy Paris Université
fications generate the basis of the proteome’s complexity. More-
Biology Department
CNRS over, the fluctuations in protein complexes throughout cell and
Laboratoire Analyse, Modélisation, Matériaux pour la Biologie et life cycles and the subcellular localization of distinct proteins
l’Environnement UMR 8587, Cergy 95000, France add further, often overlooked, complexity to the proteome.[5]
Dr. B. Cressiot, Dr. L. Bacri, Prof. J. Pelta We briefly describe the main factors explaining the proteome
Université Paris-Saclay complexity in Figure 1. Because two-thirds of human genes
Université d’Evry
Biology and Physics department are estimated to contain one or more alternative spliced exons
CNRS and a large number of noncoding DNA sequences (introns),
Laboratoire Analyse, Modélisation, Matériaux pour la Biologie et the alternative splicing mechanism leads to around 105 pro-
l’Environnement UMR 8587, Evry 91025, France tein isoforms in eukaryotic cells.[13] This mechanism removes
E-mail: [Link]@[Link]
introns and associates combinations of exons together with
The ORCID identification number(s) for the author(s) of this article multiple potential mature transcripts. Another factor increasing
can be found under [Link]
the complexity of a cell’s protein pool is the variety of chem-
DOI: 10.1002/smtd.202000090 ical modifications they are subject to and their rate. The main

Small Methods 2020, 2000090 2000090 (1 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

post-translational modifications are phosphorylation, ubiq-

uitination, acetylation, methylation, sulfonation, amidation, Benjamin Cressiot is an
hydroxylation, and glycosylation.[4,6–8] Furthermore, proteins assistant professor at CY
can also form complexes from the same subunit or with dif- Cergy Paris University.
ferent subunits. In the case of multi-subunits complexes, the In 2012, he received his
stoichiometry is controlled during translation.[5] Ph.D. in biophysics from
The discovery of additional translated open reading frames Evry University on protein
upstream and downstream of annotated ORFs revealed that transport dynamics and
protein variants can be expressed from alternative start codons unfolding through protein
both in-frame and out-of-frame.[5] Many previously ignored and solid-state nanopores.
ORFs are translated and function in regulatory mechanisms or He then joined the University
as bioactive proteins. The findings showed that genetic infor- of Notre Dame as a post-
mation is more densely coded and that the proteome is more doctoral fellow to work on
complicated than previously anticipated.[14] the secretion mechanism of
For the past 15 years, several significant results paved the autotransporter proteins using the nanopore method as
way to nanopore protein sequencing.[15] Several parameters a model. In 2015, he joined Northeastern University as a
must be controlled to achieve this goal: i) Proteins must be postdoctoral fellow, where he engineered a new protein
properly unfolded by chaotropic agents,[16–20] by tempera- nanopore from a virus capsid for biomolecule detection.
ture,[21,22] by molecular motors[23,24] or by mechanical force;[25,26] He is now working on battery parasitic species detection,
ii) Protein transport dynamics must be controlled; [27–32] iii) A biomarker detection, and design of new nanopores for
specific electrical signature must be associated to each amino sensing applications.
acid sequence;[33–41] iv) Post-translational and chemical modifi-
cations must be identified on proteins[42] or peptides.[35,43–47]
An interesting review recently discussed protein sequencing Laurent Bacri received his
challenges by comparing nanopore single molecular approach Ph.D. degree in wetting of
with fluorescence-fingerprinting, tunneling current/ionic current porous media after studying
blockades and classical bulk experiments.[15] In another recent at the Curie Institute.
review W. Timp and G. Timp described mass spectrometry to Following postdoctoral
proteomics with new alternative single-molecule approaches with research in biophysics at
a nanopore by using fluorescence native folded fingerprint and Munich Technique University,
ionic current blockade of denatured protein in highly confined he was recruited as an
solid-state nanopores.[48] Finally, the driving forces for control- associate professor in the
ling peptide capture and transport through alpha-hemolysin pro- Loïc Auvray’s team at Evry
tein channels were explored in another recent publication. The University in 2001. He has
authors also describe atomic simulation approaches to control developed an expertise in the
the transport dynamics and to distinguish all 20 amino acids.[49] understanding of the physical processes describing the
In this review, we first introduce why DNA decoding is not suffi- confinement of single biomolecules or nanoparticles in
cient for understanding human diseases and cellular functions. We nanopores.
highlight that a single gene can encode a variety of proteins with the
proteome complexity via epigenetics and other factors. Decoding Juan Pelta is professor of
proteome complexity lies not only in deciphering the alphabet of 20 biochemistry and biophysics
amino acids but in being able to evaluate the impact of epigenetic at Evry/Paris-Saclay University.
factors, isoforms, alternative ORF translation, and protein chemical He has been leading a team
modifications (Figure 1). Another hurdle for the clinical transla- of biologists since 2005 (Cergy
tion of this technology is increasing the sensitivity to very small University) and the nanopore
amounts of target protein biomarkers. The manuscript focusses on team at the LAMBE (Evry and
peptide/protein sequencing; the control of protein unfolding and CY universities) since 2010.
translocation; detection of single amino acid chemical and post- During his graduate studies
translational modification identification on peptides and proteins. in biophysics (LPS, Orsay)
and postdoc (Rhône Poulenc
Rorer), he made a significant
2. Data Analysis contribution to the understanding of DNA aggregation by
polyamines, proteins, and the resolubilisation of aggre-
We introduce here the general background and parameters (cur- gates. He then worked (Lecturer at CY, 1997) on multid-
rent blockade, dwell time, event frequency) for obtaining infor- omain protein folding. Since 2006, his activity has focused
mation from a nanopore current recording. These principles will on nanopores, protein translocation, folding, peptide size
provide the base with which to understand the results presented and sequence discrimination, biomarker detection. He is
further in this review. For those who wish to further understand co-founder of the DreamPore start-up.
the analytical aspect of nanopore recoding, we recommend these

Small Methods 2020, 2000090 2000090 (2 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 1. Overview of proteome complexity. Numerous factors contribute to the generation of complex proteomes: alternative splicing; the assembly
of protein complexes; the subcellular location of proteins; the attachment of chemical modifications to proteins; the use of alternative upstream open
reading frames in mRNA translation. Adapted with permission.[5] Copyright 2016, Springer Nature.

reviews.[50–53] To illustrate the principle of data analysis for pep- In a part of the current trace (Figure 2b), current blockades are
tide sequence analysis, we give an example with one protein observed when an equimolar mixture of phenylalanine-arginine
channel: the aerolysin.[54,55] This method is not specific to this (FR7) and tyrosine-arginine (YR7) is added. In this particular
channel and can be used with other nanopore types, including example, the blockades are defined by blockade current Ib larger
solid state nanopores,[56–60] hybrid nanopores[61] or track-etched than a first threshold Th1 = Ip + 3σ ≈ −92 pA, which excludes
nanopores.[62] The Aerolysin channel is characterized by a 99% of the noise. Among these blockades, the ones larger than a
strong confinement (≈1.2 nm) and a strong internal charge.[54] second threshold Th2 = Ip + 5σ = −87 pA are taken into account.
The data presented in this part are extracted from the recent These blockades are defined by their blockade current Ib, their
work performed by Ouldali et al.[35] The peptides are detected duration Tt between the starting point S1, the ending one E1,
by a decrease of the ionic current in the presence of a single and the inter-event time Ti between two consecutive starting
aerolysin channel inserted into a lipidic bilayer (Figure 2a).[27,63] points S1 and S2 (Figure 2b). The distribution of the inter-event
Electrodes are put in each of the compartments separated by the times follows an exponential law ∝ exp(−fc × Ti) because there
lipidic bilayer with the cis compartment usually connected to the is no correlation between two current blockades, where fc =
ground (Figure 2a). The bilayer is painted with a micropipette 4.17 ± 0.04 blockades/s is the characteristic frequency of block-
of determined aperture, the size of which is about 150 µm. This ades (Figure 2c). In the blockade duration scatter plot according
aperture allows the connection between the two compartments. to its amplitude (Figure 2d), two areas are observed. The peak of
The membrane must be thinner than 4–5 nm to allow for the the first one, centered at IFR7 = 34.98 ± 0.02 pA, is attributed to
insertion of a single channel, which is added in the cis compart- the blockades due to phenylalanine-arginine peptides (FR7).[35]
ment.[64] The thickness is usually controlled by its capacitance The peak of the second one, at IYR7 = 35.77 ± 0.02 pA, is attrib-
Cm according to the equivalent circuit model described in[53,54,65] uted to the tyrosine-arginine peptides (YR7) (Figure 3a). The
position of each peak is calculated from the fitting of each distri-
C m = ε 0 ε r A/ m (1) bution with a Gaussian function. From the mean and standard
deviation of the different peak position, the mean value calcula-
where ε0 is the vacuum permittivity and εr the relative dielectric tion allows to reach a 20–30 fA precision, much smaller than
constant of the buffer. A and ℓm are the bilayer area and thick- the distribution standard deviation (≈180 fA).
ness, respectively. Ouldali et al.[35] data illustrates the possibility of discrimi-
The nanopore insertion is controlled by the ionic current nating between two peptides that differ only in the presence of
drop until the open pore current Ip ≈ 99 pA (Figure 2b) is a hydroxyl group on the aromatic ring via their blockade cur-
reached. The signal is filtered with a 4-pole Bessel filter at cutoff rent signature. The blockade level difference for FR7 and YR7
frequency fcutoff = 5 kHz to reduce the standard deviation σ = were just 0.79 ± 0.06 pA, which is below the standard deviation
2.4 pA of the Gaussian current noise. The cutoff frequency amplitude of the open pore current. This discrimination was
must be chosen accordingly to the characteristic duration of the made possible by the long duration of each blockade, between
blockades in the experiments. If the blockade duration is shorter 1 ms and 0.1 s. Moreover, the presence of additional short
than 0.66/fcutoff, i.e., 130 µs here, the signal should be distorted blockades within the long blockades are observed (Figure 2b).
by the filter.[51] This limitation can be overcome.[49,51,64,65] The These short blockades could be due to the exploration and
acquisition is performed at 250 kHz sampling rate. the conformational change of one single peptide and/or their

Small Methods 2020, 2000090 2000090 (3 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 2. Peptide detection. a) Insertion of an aerolysin nanopore into a lipid bilayer. The ionic current through the channel is monitored in pres-
ence of peptides under an applied voltage ΔV. b) Ionic current trace measured in 4 m KCl, 25 × 10−3 m HEPES pH 7.5 containing a micromolar equi-
molar mixture of phenylalanine-arginine (FR7) and tyrosine-arginine (YR7), R being arginines, T = 20.0 ± 0.5 °C. The applied voltage is ΔV = −50 mV.
Ip = 99.2 ± 2.4 pA is the average open pore current. Th1 = 92.2 pA and Th2 = 87.4 pA correspond to detection thresholds defined in the manuscript. S1
and S2 are the starting point of the first and second blockade respectively. E1 is the ending point of the first blockade.[53] Ib is the blockade current. Tt is
the duration (S1E1) of the blockade. Ti = S1S2 is the inter-event time between two following blockades. c) Distribution of the inter-event time Ti. The fitting
curve is an exponential function ≈exp(−fc × Ti) where fc is the characteristic blockade frequency. d) Scatter plot of each blockade duration according to
its blockade current. Two areas are observed according to the nature of the peptides: FR7 (in red) and YR7 (in blue). Data extracted from the data set.[35]

interaction with the inner pore. The interaction mechanism specific interactions between the channel’s internal side and
between peptides and the inner channel is not well understood. the hydroxyl group located in the aromatic ring. Nevertheless,
If the dielectric constant inside the channel (εr = 2) is much the peptides can be distinguished via the blockade current
lower than the one in bulk (εr = 80), the electrostatic interac- observed.
tions are assumed to be strongly enhanced into the nanopore This approach can also be applied to discriminate between
despite the high salt concentration (4 m KCl), which screens the leucine and isoleucine[35] (Figure 3c). This amino acid pair is
electrostatic interactions. discerned by the presence or absence of a methyl group, rather
Because this detection is sensitive, the median, rather than than a hydroxyl group, on the aromatic ring. The structural dif-
the average, value between S1 and E1 is recorded to avoid local ference between these two conformers lead to distinct blockade
blockades and to increase precision. Consequently, the current current distributions of 35.59 ± 0.03 pA and 36.07 ± 0.03 pA for
distribution bin width is 0.06 pA in Figure 3a,c. The current leucine (LR7) and isoleucine (IR7) respectively. The difference
blockade distribution could be understood as the probability of between each distribution center is about 0.476 pA, meaning
a single blockade belonging to FR7 or YR7. Thus, the nature of that the nanopore is sensitive to 0.5% ionic current variations.
each blockade can be statistically defined. This data treatment The blockade duration distributions belonging to leucine and
allows plotting of the Ib distribution of each peptide (Figure 3a) isoleucine peptides can be calculated following the same data
and the distribution of the corresponding duration (Figure 3b). treatment described on the paragraph above. Surprisingly,
The bin width is plotted on a logarithmic scale due to the large in this case, the blockade durations τ = 16.96 ± 0.2 ms and
range of blockade durations, from 1 ms to 0.2 s. Then, this dis- τ = 13.4 ± 1 ms for LR7 and IR7 respectively could discrimi-
tribution is fitted by the function[66] Tt*exp(−Tt/τ) where τ is nate amongst the peptides. This difference means the trans-
the characteristic duration of each peptide (Figure 3b). Since port through the channel is sensitive to the methyl function.
both durations (τ = 21.3 ± 0.3 ms and τ = 21.7 ± 0.4 ms for FR7 Figure 3e,f illustrates the blockade current and duration signa-
and YR7 respectively) could not be discriminated, there are no tures characteristic of each peptide. Nevertheless, the blockade

Small Methods 2020, 2000090 2000090 (4 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 3. Amino acid discrimination. a) (left axis) Blockade current distribution of an equimolar mixture of phenylalanine-arginine (FR7) and
tyrosine-arginine (YR7). Both populations are discriminated: FR7 in blue, YR7 in red. Each distribution is fitted by a Gaussian distribution centered
at 34.98 ± 0.02 and 35.77 ± 0.02 pA for FR7 and YR7 respectively. (right axis) Scatter plot of each blockade duration according to its blockade cur-
rent. Each current blockade could be attributed to the FR7 or YR7 based on the blockade current distribution. b) Blockade duration distribution by
peptide. Each distribution is fitted with the function[66] Tt*exp(−Tt/τ) where τ = 21.3 ± 0.3 ms and τ = 21.7 ± 0.4 ms for FR7 and YR7 respectively.
c) (left axis) Blockade current distribution of an equimolar mixture of leucine-arginine (LR7) and isoleucine-arginine (IR7). Both populations are
discriminated: LR7 in green, IR7 in orange. Each distribution is fitted by a Gaussian distribution centered at 35.59 ± 0.03 and 36.07 ± 0.03 pA for
LR7 and IR7 respectively. (right axis) Scatter plot of each blockade duration according to its blockade current. Each current blockade could be
attributed to the LR7 or IR7 from the blockade current distribution. d) Blockade duration distribution for LR7 or IR7 peptides. Each distribution is
fitted with the function Tt*exp(−Tt/τ) where τ = 16.96 ± 0.2 ms and τ = 13.4 ± 1 ms for LR7 and IR7 respectively. e) Blockade current according to the
peptide nature phenylalanine, leucine, tyrosine, isoleucine. f ) Blockade duration according to the peptide nature phenylalanine, leucine, tyrosine,
isoleucine. Solutions are prepared with micromolar peptide solution in 4 m KCl, 25 × 10−3 m HEPES buffer, at 7.5 pH Data, T = 20.0 °C ± 0.5 °C,
ΔV = −50 mV. Data extracted from the data set.[35]

Small Methods 2020, 2000090 2000090 (5 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

durations between phenylalanine and tyrosine could not be biological nanopores.[67] The same approach has been pro-
discriminated. posed to be applicable to proteins. In 2013, Nivala et al.[24]
described controlled unfolding and translocation of a protein
through a nanopore for the first time using an alpha-hemo-
3. Efforts at Reading Protein Sequences and lysin nanopore and an AAA+ unfoldase ClpX (Figure 4a).
Briefly, a protein containing a ssrA tag is captured into the
Detection of Single Amino Acids
nanopore. The ssrA tag is recognized by the ClpX unfol-
3.1. Controlled Protein Unfolding and Translocation through dase on the other side of the nanopore, and in the presence
a Protein Nanopore of ATP, the molecular motor progressively pulls on the pro-
tein allowing for controlled translocation and unfolding.
Nucleic acids can presently be sequenced using enzyme- The authors were able to discriminate sequence-dependent
based motors that actively control DNA/RNA transport into features attributed to the protein folded domains. However,

Figure 4. Attempt to read the sequence of proteins and detection of single amino acids. a) Ionic current trace during ClpX-mediated protein transloca-
tion. Reproduced with permission.[24] Copyright 2013, Springer Nature. i) Capture of the S1 substrate. ii) Upon protein capture, the ionic current drops
to ≈14 pA (≈0.7 pA RMS noise). iii) ClpX-mediated ramping state. The ionic current decreases to ≈10 pA and is characterized by one or more gradual
amplitude transitions. This pattern is only observed in the presence of ClpX and ATP (trans compartment). iv) Smt3 domain unfolding and translocation
through the nanopore (≈3.8 pA, 1.7 pA RMS noise). i’) Return to open channel current upon completion of substrate translocation to the trans compart-
ment. b) Stochastic detection of the enantiomers of an amino acid by using a Cu(II)–phenanthroline protein nanopore. Reproduced with permission.[33]
Copyright 2012, John Wiley and Sons. The alpha-hemolysin nanopore is modified covalently at Cys-117 on one subunit with an iodoacetamide-containing
phenanthroline. Copper(II) is added from the trans compartment, and binds strongly to the phenanthroline-modified pore. D- and L-amino acids are
added from the cis compartment and reversibly bind to the copper ion within the pore. The presence of the amino acid leads to a transient decrease
in the conductance of the pore, the extent of which depends on which enantiomer is bound. c) Illustration of a single Cys molecule traversing through
an aerolysin nanopore which is embedded in a lipid bilayer. Reproduced with permission.[34] Copyright 2020, Royal Society of Chemistry. The two com-
partments of the bilayer chamber are named as cis and trans. The hydrogen bond acceptors of Lys, Arg, Ser, Thr, Asn, Gln are colored by brown, red,
orange, pink, firebrick, yellow, respectively. The recorded raw current traces in the presence of Cys.

Small Methods 2020, 2000090 2000090 (6 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

they were unable to sequence each amino acid translocating on the specific detection of larger entities such as peptides.
through the nanopore like it had been done with DNA. The Mereuta et al.[69] demonstrated the possibility to control and
same group used the same approach a few years later with a reduce peptide (cecropin A–magainin antimicrobial chimera
ClpXP motor to discriminate among distinct protein domains, peptide) translocation speed through an alpha-hemolysin pore
and detect sequence-dependent variations within those by varying the pH. The authors were able to observe interme-
domains.[23] Protein unfolding was also achieved with a chi- diate single-peptide sub-states associated with distinct struc-
meric molecule, a thiorodoxin protein covalently attached to tural milestones along the pore. On a following publication,
an oligonucleotide, using the electrical driving force.[26] The they used 30 amino acid long engineered peptides, R12-X-E12
unfolding process through an alpha-hemolysin channel is peptides with X being either 6 alanine, or 6 tryptophan, or 3
coupled to the translocation and is dependent on the N/C- alanine and 3 tryptophan (Figure 5a).[37] The opposing charges
terminal entry into the nanopore.[25] Another approach is of the peptide extremities allowed for the peptide to be pulled
to denature the protein with an unfolding agent[16–18,27,30,68] into the nanopore reversibly by inverting the applied voltage.
before translocation. In all the cases with protein channels or They were able to specifically discriminate the neutral amino
solid state nanopores, the experimental sensitivity is incom- acids contained in the center of the primary sequence by volu-
patible with protein sequence reading but is able to control metric measurements (Ala ≈100 Å compared to Trp ≈239 Å).
transport. Singh et al.[70] analyzed the interaction of peptides with a
Recently, a new approach was developed using a sub-nano- Nocardia farcinica channel and measured the binding of tri-,
metric solid-state nanopore to indirectly read a polypeptide penta-, and hepta-arginine by time-resolved ion current block-
chain.[22] The proteins were completely unfolded in the pres- ages. They were able to discriminate the three peptides by
ence of an anionic detergent (SDS), temperature and beta- adjusting the peptide side entry and applied transmembrane
mercaptoethanol, the driving force being the electrical force. voltage.
The fluctuation of the current amplitudes detected during To detect peptides at a single amino acid resolution,
unfolded protein transport were associated to amino acid researchers though to use nanopores as mass spectrometers.
volume. The best indirect resolution obtained to read the poly- The detection of different peptides differing by their mole-
peptide chain is by blocks of 4 amino-acids due to excluded cular weight can be discriminated as a function of the cur-
volume approach. rent blockade amplitude. Chavis et al.[38] used this strategy,
combined with a thiolate-capped gold cluster, to increase cati-
onic peptide residence time into an alpha-hemolysin pore and
3.2. Detection of Single Amino Acids hence, increase the detection sensitivity (Figure 5b). Piguet
et al.[39] performed size discrimination of short uniformly
Up to now, unlike with DNA, nanopore technology has not charged homopeptides (5 to 10 amino acids), mixed in solu-
been sensitive enough to sequence each amino acid of a pro- tion or independently, at single amino acid resolution. In this
tein. Researchers have focused their efforts on discriminating study, the aerolysin nanopore is completely unmodified to
single amino acids. The first reported study on the detection discriminate properly peptides of different lengths. Further-
of single amino acids was performed using a modified alpha- more, they were able to discriminate a hetero-peptide (R5K5)
hemolysin containing a CuII-phenanthroline into the pore from a homo-peptide (R10 or K10) of same molecular weight
lumen[33] (Figure 4b). This single modification allowed the (Figure 5c).
discrimination of enantiomers of underivatized amino acids
(L/D-Cys, L/D-Phe, L/D-Asp and L/D-Trp). Furthermore, the
authors showed that several amino acids can even be discrimi- 4.2. Identification of Peptide Amino Acid Composition
nated simultaneously in a mixture according to the current
pore blockade. More recently, Yuan et al.[34] described the detec- The next challenge facing the field was to identify peptides dif-
tion of single cysteines using an unmodified aerolysin nano- fering by a single amino acid. Huang et al.[36] used a wild-type
pore (Figure 4c). Hydrogen bonds between the cysteines and Fragaceatoxin C (FraC) nanopore to discriminate endothelin
the amino acids composing the nanopore lumen explain this 1 and endothelin 2: near-isomeric polypeptides differing
sensitivity. in 1 out of 21 amino acids and otherwise structural isomers
(Figure 6a). The authors were able to detect different current
blockades arising from an additional bulky tryptophan residue
4. Peptide Amino Acid Detection and Composition in endothelin 2. They also showed that in a mixture of the two
peptides, they were able to discriminate each peptide specifi-
Determination
cally. Later on, the same authors used a mutated FraC with a
4.1. Detection of Peptides by Current Blockade Approach controllable diameter to discriminate Angiotensin II and angi-
otensin A, which differ only by a substitution of an alanine
Although the detection of single amino acids by nanopore with a glutamate[40] (Figure 6b). With this experiment, the
seems promising, several challenges need to be addressed for authors also confirmed a correlation between the depth of cur-
sequencing applications. In fact, single amino acid detection rent blockade and the peptide mass irrespective of chemical
sensitivity is limited by the translocation speed of the mole- composition. By using an aerolysin nanopore, Ouldali et al.[35]
cules through the pore in contrast with the temporal resolution demonstrated the detection of natural amino acids (Figure 6c).
of the nanopore technique. Several studies followed focusing To do so, the authors chemically linked single amino acids to

Small Methods 2020, 2000090 2000090 (7 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 5. Detection of peptides by current blockade approach. a) Detecting amino acids in a peptide from ion current fluctuations across the nanopore.
Reproduced with permission.[37] Copyright 2017, American Chemical Society. The system is composed of the α-HL nanopore embedded in a lipid bilayer,
the peptide, water, and ions (2 m KCl). The α-HL’s constriction (Glu 111, Met 113, and Lys 147) is highlighted in green, while the peptide residues are
colored according to their charge, i.e., the negative Glu tail is in red, the positive Arg tail is in blue, while the middle neutral domain (Ala of Trp) is in
white. The peptide is imported in the pore from the barrel (trans) side with the positive Arg tail directed toward the vestibule (cis) entrance of the pore.
Representative current recordings through a single α-HL pore showing the transient pore blockades by a trapped Pe1 peptide (Ac-(R)12−(A)6−(E)12−
NH2), Pe3 peptide (Ac-(R)12−(W)6−(E)12−NH2), or Pe5 peptide (Ac- (R)12−(A)3−(W)3−(E)12−NH2), added in the trans side. b) Schematic illustration of
the cluster-enhanced nanopore detection. Reproduced with permission.[38] Copyright 2017, American Chemical Society with corresponding representa-
tive current blockade distributions for five different peptides studied ((pink) NT, (gold) QBP1, (blue) A1, (green) A2, and (orange) LE) give rise to
current blockade distributions dominated by single peaks with (bottom) and without (top) a gold cluster in the pore. Representative current blockade
distributions show narrower peaks for (A) angiotensin I, (B) angiotensin II by increasing the mean residence time with a gold cluster in the pore. c)
Discrimination of 10 amino acid long peptides of different sequences. Reproduced with permission.[39] Copyright 2018, Springer Nature. Scatter plot
of blockade duration versus relative blockade current Ib/I0 (top of each subfigure) and histogram of the relative blockade current Ib/I0 values (bottom
of each subfigure) in the case of the interaction of aerolysin nanopore with a solution of 10 amino acid long arginine homopeptides (RRRRRRRRRR,
named RR-10), a solution of 10 amino acid long lysine/arginine heteropeptides (KKKKKRRRRR, named KR-10) and a solution of 10 amino acid long
lysine homopeptides (KKKKKKKKKK, named KK-10). Each histogram corresponding to a given peptide sequence exhibits a main Ib/I0 population clearly
distinct from the other peptide sequences, allowing to discriminate between peptides of same length but of different sequences.

a short polycationic carrier arginine heptapeptides. Each of 5. Chemical and Post-Translational Modifications
the twenty amino acids linked to the polycationic carrier pro- Detection on Peptides and Proteins
duced a current blockade distinct from that of the polycationic
carrier alone. Among the twenty amino acids, peptides con- Detection and determination of chemical and post-translational
taining R, W, F, K, L, N, T, P, D, A, C, S, and G amino acids patterns within individual protein molecules remain chal-
produced blockade populations distinct from one another. In lenging. Rosen et al.[42] published the first and only study to
contrast, the blockade currents of the remaining seven pep- detect chemical modifications on proteins. The authors used
tides fell into two groups, (Q, M, H, I, Y) and (E, V). The thierodoxin as a protein model. The thierodoxin used was
current blockades within each group were too similar to be unphosphorylated, mono-phosphorylated or di-phosphorylated
differentiated but the current blockades of each group could near the C-terminus (Figure 7a). They were able to distinguish
be clearly distinguished from that of the rest of the peptides. these three states through the current blockade ratio. Further-
By using equimolar peptide mixtures, the authors showed more, they were even able to quantify protein population states
they could discriminate between individual peptide species in (compared to isoelectric focusing) and detected which locations
a mixture. were phosphorylated on the protein.

Small Methods 2020, 2000090 2000090 (8 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 6. Identification of peptide amino acid composition. a) Discrimination of endothelin 1 and 2 with wild type FraC nanopore. Reproduced with
permission.[36] Copyright 2017, Springer Nature. (WtFraC, PDB: 4TSY) at pH 4.5. Molecular surface representation of endothelin 1 (PDB: 1EDN) and
endothelin 2 (homology model from endothelin 1, PyMOL) using electrostatic coloring (PyMOL). Amino acid sequences of endothelin 1 and 2. Blue
lines indicate the disulfide bridges in each polypeptide. Ires% and dwell time for endothelin 1 and endothelin 2 blockades at −50 mV in pH 4.5 buffer
(1 m KCl, 0.1 m citric acid, 180 × 10−3 m Tris base). Representative endothelin 1 and endothelin 2 blockades to the same FraC nanopore under −50 mV
applied potential. Histogram of residual currents provoked by 2 × 10−6 m endothelin 1 and corresponding heatplot depicting the standard deviation of
the current amplitude versus Ires%. Addition of 8 × 10−6 m endothelin 2 to the same pore reveals a second population. b) Discrimination of peptides
differing by a single amino acid using type II W116S-fragaceatoxin C (FraC). Reproduced with permission.[40] Copyright 2019, Springer Nature at pH
4.5. Cut through of a surface representation of wild- type FraC (Wt-FraC) oligomer (PDB: 4TSY) colored according to the vacuum electrostatic potential
as calculated by PyMOL. One protomer is shown as a cartoon representation with tryptophans 112 and 116 displayed as spheres. Peptide sequences
of angiotensin II, and A with typical blockades provoked by the two angiotensin peptides measured at −30 mV applied bias. Color density plot of the
Iex% versus the standard deviation of the current amplitude for angiotensin II, and A, respectively. Separation of angiotensin II and A in a mixture.
c) Electrical detection of the twenty proteinogenic amino acids. Reproduced with permission.[35] Copyright 2019, Springer Nature. Schematic of the
peptide constructs used to investigate the current blockade of the twenty proteinogenic amino acids. A cationic carrier of seven arginine amino acids
(R7) is chemically linked at the carboxyl terminus to the eighth amino acid, X, to form twenty XR7 peptides. Fragment of a typical current recording
from a nanopore experiment where equimolar amounts of RR7, KR7, HR7, ER7, and DR7 were introduced into the cis-compartment solution. The bottom
graph shows a zoomed-in view of the top graph. Histogram of Ib/I0 values (black line, left axis) and the scatter plot of the blockade duration (red dots,
right axis) for an equimolar mixture of RR7, KR7, HR7, ER7, and DR7. Histogram of Ib/I0 values (black line, left axis) and scatter plot of blockade duration
(red dots, right axis) recorded from an equimolar mixture of YR7 and FR7 or LR7 and IR7. Colored rectangles indicate the mean (centerline) and the s.d.
(widths) of the Ib/I0 values recorded individually for each XR7 peptide. Experimentally determined mean Ib/I0 value and its s.d. for all twenty XR7 peptides.

Other researches focused their efforts on the detection of respectively (Figure 7c). They showed phosphorylation acceler-
chemical modifications on peptides. Restrepo-Pérez et al.[45] ates a hundred-fold the traversing speed of a negatively charged
characterized six potential tags for a fingerprinting approach substrate and significantly enhances the translocation fre-
using a FraC nanopore (Figure 7b). The different tags ranging quency of a positively charged substrate.
from 427 and 1275 Da are covalently linked to a cysteine and Restrepo-Pérez et al.[44] recently used FraC nanopore for the
can be clearly identified. The authors also changed position of discrimination of phosphorylation and glycosylation. These
the tags and were able to identify each position. Moreover, they modifications often are mutually exclusive and compete for
found that each tag gave specific current blockades correlated protein modification sites in different diseases (Figure 8a).
with their geometry and molecular weight. Ying et al.[46] used The authors tracked dwell times and current blockade depths
the same approach to detect phosphorylated/dephosphoryl- to identify unmodified, phosphorylated or glycosylated serines
ated peptides after exposure to kinase/phosphatase enzymes on each peptide. Furthermore, they were able to discriminate

Small Methods 2020, 2000090 2000090 (9 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 7. Detection of chemical and post-translational modifications of peptides and proteins. a) Single-molecule nanopore detection of phosphoryla-
tion of a model substrate. Reproduced with permission.[42] Copyright 2014, Springer Nature. Current signature of the unfolding and translocation of
TrxS112−P-oligo(dC)30 through an alpha-hemolysin pore showing the four characteristic levels: 1) open pore; 2) oligonucleotide leader threaded into the
pore; 3) C terminus of the protein substrate in the pore; 4) unfolding of the remainder of the protein and diffusion through the pore. Sequences of the
C termini of Trx mutants used. Phosphorylatable serine residues, red; nonphosphorylatable alanine residues, black. Models of Trx mutants, current
signatures and representative scatter plots of IRES% and In. and associated histograms. b) Detection of chemical modifications on peptides. Reproduced
with permission.[45] Copyright 2019, American Chemical Society. Peptide labeled at different positions through FraC. From left to right: Schematic, cur-
rent traces, and relative blockade histograms for unlabeled peptide and for fluorescein-labeled peptide at positions C11, C15, and C20. c) An aerolysin
nanopore sensor for the phosphorylation detection of peptides. Reproduced with permission.[46] Copyright 2019 Yi-Lun Ying et al. Exclusive Licensee
Science and Technology Review Publishing House. The aerolysin nanopore sensing of phosphorylated and unphosphorylated peptides. The phosphoryl-
ated substrate with additional negative charges suffers the strong electrophoretic force (red arrow) through the aerolysin nanopore. The native substrate
bears a weaker electrophoresis force (red arrow) to transverse through the aerolysin nanopore compared to the phosphorylated substrate. The scatter
plots for the aerolysin nanopore sensing of EYQEYQEYQ peptide and LRRASLG peptide.

and identify the three peptides within a mixture from the cur- Such a technology could be incorporated into pre-existing auto-
rent level blockades. Ouldali et al.[35] also explored the specific mated processing platforms for single cell analysis. The cur-
identification of phosphotyrosine, nitrotyrosine, sulfated rent standard method to perform protein sequencing is mass
tyrosine and sulfated methionine (Figure 8b). Each peptide spectroscopy.[71–73] At present, the mass spectrometry’s protein
could be differentiated from their unmodified form according detection sensitivity is insufficient for single-cell resolution
to the relative residual current. medical diagnostics. The number of molecules needed to obtain
a resolved signal is between 106 to 109 molecules (Orbitrap
mass spectrometer). At least 50 000 cells and around 0.1 mg of
6. Other Attempts at Pushing the Limits of protein are required to obtain reproducibility robustness and
reproducibility.[74] If we consider a cell with a low protein copy
Protein Sequencing and Potential Health
number (<1000 molecules), 12 millions of cells (between 0.1 to
Applications 10 femtomole) would be required to reach the system’s limit of
detection.[75] Recent approaches have aimed to circumvent these
There is an unmet need for single molecule analytical methods shortcomings. In order to decrease the number of mammalian
capable of small quantities of protein/peptide biomarkers. cells needed for mass spectrometry (MS) analysis, a nanodroplet

Small Methods 2020, 2000090 2000090 (10 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

Figure 8. Detection of post-translational modifications of peptides. a) Schematic representation of the measurement approach using a FraC nanopore
with an example current trace obtained for a measurement on a mixture of three peptides: an unmodified control peptide, a phosphorylated peptide
and a glycosylated peptide. Reproduced with permission.[44] Copyright 2019, American Chemical Society. Scatter plot of relative blockade versus dwell
time and a relative blockade histogram of the mixture of the three peptides. Three different current blockade levels are observed. b) Blockade currents
of chemically modified YR7 peptides. Reproduced with permission.[35] Copyright 2019, Springer Nature. Histogram of Ib/I0 (top plot of each panel) and
scatter plot of blockade duration versus Ib/I0 (bottom plot of each panel) obtained from nanopore experiments during analysis of an equimolar mixture
of YR7 and (NO2)-YR7 peptides, an equimolar mixture of YR7 and (SO3H2)-YR7 peptides and a mixture of YR7 (1 × 10−6 m final concentration) and (P)-YR7
(10 × 10−6 m final concentration) peptides. Experimentally determined mean Ib/I0 value and its s.d. for all twenty XR7 peptides and phosphotyrosine,
nitrotyrosine, sulfated tyrosine and sulfated methionine peptides.

platform has been designed and manufactured: NoPOTS acid at the N-terminal position. The fluorescence progressively
(nano­droplet Processing in One pot for Trace Sample).[76] This decreases through repeating cycles of Edman degradation.
platform reduced the sample volume needed from 100 µL to This approach is performed in parallel with up to one million
200 nL. When coupled to LC-MS, NoPOTS was capable of ana- molecules in zeptomolar concentrations. Finally, the sequence
lyzing 2000 to 3000 proteins by using a small population of obtained from the fluorescent signature is aligned with a ref-
cells (between 10 to 140 cells). Recent advances in individual ion erence database. This method is restricted by the size of the
mass spectrometry (I2MS) have also managed to increase the peptide of interest as the chemical degradation is only com-
resolving power (by a factor 20) and an increase of the dynamic patible with short peptides. However, there is promise for this
range of MS.[77] Specifically, This I2MS allows mass determina- ingenious and powerful analytical approach when methods are
tion in a mixture of denatured proteins or in native complexes developed to reduce costs and expand the applications to more
with masses ranging from 5 kDa to 3.5 MDa.[78] However, MS complex proteins.
based approaches suffer from significant cost elevations for A recent, exciting review discussed 3 single-molecule level
increasing sensitivity. approaches for protein sequencing.[15] Aside from the nanopore
Another historic biochemistry method to sequence proteins technique, two other interesting single-molecule methods also
is Edman degradation.[79] This technique is limited to short pave the way to protein information decoding: a tunneling cur-
purified peptides (<50 amino acids). The N- and C-terminus rents based technique[81,82] and a fingerprinting approach with
must be free for the degradation to occur, and the reaction fluorescence FRET mode.[83–85] The principle of tunneling cur-
is time-consuming (around 45 min for a degradation cycle). rent is to position an amino acid between two nanogap gold-
Recently, single-molecule fluoro-sequencing techniques have deposited electrodes. The width of the nanogap is comparable to
been paired with Edman degradation to develop a protein/pep- amino acid sizes and compatible with tunneling effect (between
tide detection and sequence platform.[80] Proteins are cleaved 0.55 to 0.7 nm). Changes in tunneling current by the amino acid,
and fluorescent dies are covalently attached on two specific in the order of nS amplitude, are associated with the sequence
amino acids (cysteine and lysine) on the peptides. The pep- identification. Up to 12 different amino acids have been identi-
tides are then immobilized on a glass surface in a TIRF (total fied using 2 nanogap widths, and up to 9 amino acids for a fixed
internal reflection fluorescence) single-molecule-microscope. nanogap.[82] The post-translational modification phosphoty-
Each peptide is imaged, and the fluorescent tag position is rosine has also been detected. The sensitivity of this technique
observed. Edman degradation chemically removes one amino allowed the identification of D/L-asparagine enantiomers and

Small Methods 2020, 2000090 2000090 (11 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

of methylated glycine.[81] On the other hand, this elegant funda- [1] E. S. Lander, L. M. Linton, B. Birren, C. Nusbaum, M. C. Zody,
mental approach would be difficult to apply to proteome analysis. J. Baldwin, K. Devon, K. Dewar, M. Doyle, W. FitzHugh, R. Funke,
The concept of single-molecule fingerprinting[83] is to deter- D. Gage, K. Harris, A. Heaford, J. Howland, L. Kann, J. Lehoczky,
mine the protein sequence by labeling only two amino acids R. LeVine, P. McEwan, K. McKernan, J. Meldrim, J. P. Mesirov,
C. Miranda, W. Morris, J. Naylor, M. Rosetti, R. Santos, A. Sheridan,
on a protein with two fluorescent acceptor dyes. To obtain a
C. Sougnez, Y. Stange-Thomann, N. Stojanovic, et al., Nature 2001,
FRET (Förster resonance energy transfer) signal, a molecular 409, 860.
motor which is labeled with a donor dye, unfolds, transports [2] M. Clamp, B. Fry, M. Kamal, X. Xie, J. Cuff, M. F. Lin, M. Kellis,
and degrades the protein. When one of the acceptor dyes on the K. Lindblad-Toh, E. S. Lander, Proc. Natl. Acad. Sci. USA 2007, 104,
protein of interest comes close enough to the molecular motor 19428.
to accept a photon from its donor dye, a fluorescent signal is [3] E. S. Lander, Nature 2011, 470, 187.
generated. Sequencing is achieved by analyzing the resulting [4] D. Virág, B. Dalmadi-Kiss, K. Vékey, L. Drahos, I. Klebovich, I. Antal,
fluorescence signal. A computational approach predicts the K. Ludányi, Chromatographia 2020, 83, 1.
identity of some protein’s sequences with only two fluores- [5] J. W. Harper, E. J. Bennett, Nature 2016, 537, 328.
cent labels. The experimental validation of this concept has [6] F. Wold, Annu. Rev. Biochem. 1981, 50, 783.
[7] Y. L. Deribe, T. Pawson, I. Dikic, Nat. Struct. Mol. Biol. 2010, 17, 666.
been published by Van Ginkel et al.[85] A donor-labeled ClpXP
[8] D. Vucic, V. M. Dixit, I. E. Wertz, Nat. Rev. Mol. Cell Biol. 2011, 12, 439.
is immobilized on a quartz surface, and the motor recognizes [9] A. M. Bode, Z. Dong, Nat. Rev. Cancer 2004, 4, 793.
an acceptor substrate. The translocation is then associated with [10] L. Martin, X. Latypova, F. Terro, Neurochem. Int. 2011, 58, 458.
fluorescent signals. A wild type and mutant titin protein were [11] H. Xu, Y. Wang, S. Lin, W. Deng, D. Peng, Q. Cui, Y. Xue, Genomics,
used to validate this concept. Proteomics Bioinf. 2018, 16, 244.
[12] G. Walsh, R. Jefferis, Nat. Biotechnol. 2006, 24, 1241.
[13] Q. Pan, O. Shai, L. J. Lee, B. J. Frey, B. J. Blencowe, Nat. Genet.
2008, 40, 1413.
7. Conclusion [14] M. W. Orr, Y. Mao, G. Storz, S.-B. Qian, Nucleic Acids Res. 2019, 74,
Regarding protein channels, ionic current measurements, more 481.
[15] L. Restrepo-Pérez, C. Joo, C. Dekker, Nat. Nanotechnol. 2018, 13,
precisely, the relative current blockade or residual current, are
786.
sensible enough to identify amino acid composition and post- [16] G. Oukhaled, J. Mathe, A. L. Biance, L. Bacri, J.-M. Betton, D. Lairez,
translational modifications. However, reading and differenti- J. Pelta, L. Auvray, Phys. Rev. Lett. 2007, 98, 158101.
ating the 20 different amino acids along an unfolded protein [17] D. S. Talaga, J. Li, J. Am. Chem. Soc. 2009, 131, 9287.
has not yet been achieved and will require further advance- [18] C. Merstorf, B. Cressiot, M. Pastoriza-Gallego, A. Oukhaled,
ments in the field. In contrast, the detection and identification J.-M. Betton, L. Auvray, J. Pelta, ACS Chem. Biol. 2012, 7, 652.
of each of the 20 amino acids inside an engineered and pow- [19] A. Oukhaled, B. Cressiot, L. Bacri, M. Pastoriza-Gallego,
erful nanopore like the aerolysin seems possible. The next step J.-M. Betton, E. Bourhis, R. Jede, J. Gierak, L. Auvray, J. Pelta, ACS
will be to control the capture and entry of the monomers after Nano 2011, 5, 3628.
a protein cleavage inside the nanopore. Protein coding would [20] K. J. Freedman, S. R. Haq, J. B. Edel, P. Jemth, M. J. Kim, Sci. Rep.
2013, 3, 1638.
represent a new frontier in the high-throughput understanding
[21] L. Payet, M. Martinho, M. Pastoriza-Gallego, J.-M. Betton, L. Auvray,
of disease development and would prove to be an invaluable J. Pelta, J. Mathé, Anal. Chem. 2012, 84, 4071.
tool for researchers. [22] E. Kennedy, Z. Dong, C. Tennant, G. Timp, Nat. Nanotechnol. 2016,
11, 968.
[23] J. Nivala, L. Mulroney, G. Li, J. Schreiber, M. Akeson, ACS Nano
2014, 8, 12365.
Acknowledgements [24] J. Nivala, D. B. Marks, M. Akeson, Nat. Biotechnol. 2013, 31, 247.
This research was supported by ANR-17-CE09-0044-02. The authors [25] D. Rodriguez-Larrea, H. Bayley, Nat. Commun. 2014, 5, 4841.
thank John T. Gonzalez for careful reading and for improving the English [26] D. Rodriguez-Larrea, H. Bayley, Nat. Nanotechnol. 2013, 8, 288.
in the manuscript. The authors acknowledge support from Genopole for [27] M. Pastoriza-Gallego, L. Rabah, G. Gibrat, B. Thiebot,
their financing. F. G. van der Goot, L. Auvray, J.-M. Betton, J. Pelta, J. Am. Chem.
Soc. 2011, 133, 2923.
[28] M. Pastoriza-Gallego, M.-F. Breton, F. Discala, L. Auvray,
J.-M. Betton, J. Pelta, ACS Nano 2014, 8, 11350.
Conflict of Interest [29] B. Cressiot, A. Oukhaled, G. Patriarche, M. Pastoriza-Gallego,
J.-M. Betton, L. Auvray, M. Muthukumar, L. Bacri, J. Pelta, ACS
The authors declare no conflict of interest. Nano 2012, 6, 6236.
[30] L. Restrepo-Pérez, S. John, A. Aksimentiev, C. Joo, C. Dekker,
Nanoscale 2017, 9, 11685.
[31] M. M. Mohammad, S. Prakash, A. Matouschek, L. Movileanu,
Keywords J. Am. Chem. Soc. 2008, 130, 4081.
electrical detection, nanopores, post-translational modifications, protein [32] M. M. Mohammad, L. Movileanu, Methods Mol. Biol. 2012, 870, 21.
sequencing, proteome complexity [33] A. J. A. Boersma, H. H. Bayley, Angew. Chem., Int. Ed. 2012, 51, 9606.
[34] B. Yuan, S. Li, Y.-L. Ying, Y.-T. Long, Analyst 2020, 12, 619.
Received: February 9, 2020 [35] H. Ouldali, K. Sarthak, T. Ensslen, F. Piguet, P. Manivet, J. Pelta,
Revised: April 3, 2020 J. C. Behrends, A. Aksimentiev, A. Oukhaled, Nat. Biotechnol. 2019,
Published online: 31, 247.

Small Methods 2020, 2000090 2000090 (12 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[Link] [Link]

[36] G. Huang, K. Willems, M. Soskine, C. Wloka, G. Maglia, Nat. [63] T. Gutsmann, T. Heimburg, U. Keyser, K. R. Mahendran,
Commun. 2017, 8, 935. M. Winterhalter, Nat. Protoc. 2015, 10, 188.
[37] A. Asandei, A. E. Rossini, M. Chinappi, Y. Park, T. Luchian, Lang- [64] G. Baaken, I. Halimeh, L. Bacri, J. Pelta, A. Oukhaled,
muir 2017, 33, 14451. J. C. Behrends, ACS Nano 2015, 9, 6443.
[38] A. E. Chavis, K. T. Brady, G. A. Hatmaker, C. E. Angevine, [65] A. Balijepalli, J. Ettedgui, A. T. Cornio, J. W. F. Robertson,
N. Kothalawala, A. Dass, J. W. F. Robertson, J. E. Reiner, ACS Sens. K. P. Cheung, J. J. Kasianowicz, C. Vaz, ACS Nano 2014, 8,
2017, 2, 1319. 1547.
[39] F. Piguet, H. Ouldali, M. Pastoriza-Gallego, P. Manivet, J. Pelta, [66] O. V. Krasilnikov, C. G. Rodrigues, S. M. Bezrukov, Phys. Rev. Lett.
A. Oukhaled, Nat. Commun. 2018, 9, 966. 2006, 97, 018301.
[40] G. Huang, A. Voet, G. Maglia, Nat. Commun. 2019, 10, 835. [67] M. Jain, H. E. Olsen, B. Paten, M. Akeson, Genome Biol. 2016, 17,
[41] S. Li, C. Cao, J. Yang, Y.-T. Long, ChemElectroChem 2018, 13, 595. 239.
[42] C. B. Rosen, D. Rodriguez-Larrea, H. Bayley, Nat. Biotechnol. 2014, [68] B. Cressiot, E. Braselmann, A. Oukhaled, A. H. Elcock, J. Pelta,
32, 179. P. L. Clark, ACS Nano 2015, 9, 9050.
[43] L. Harrington, L. T. Alexander, S. Knapp, H. Bayley, ACS Nano 2019, [69] L. Mereuta, M. Roy, A. Asandei, J. K. Lee, Y. Park, I. Andricioaei,
13, 633. T. Luchian, Sci. Rep. 2015, 4, 3885.
[44] L. Restrepo-Pérez, C. H. Wong, G. Maglia, C. Dekker, C. Joo, Nano [70] P. R. P. Singh, I. I. Bárcena-Uribarri, N. N. Modi, U. U. Kleinekathöfer,
Lett. 2019, 19, 7957. R. R. Benz, M. M. Winterhalter, K. R. K. Mahendran, ACS Nano
[45] L. Restrepo-Pérez, G. Huang, P. R. Bohländer, N. Worp, R. Eelkema, 2012, 6, 10699.
G. Maglia, C. Joo, C. Dekker, ACS Nano 2019, 13, 13668. [71] H. Steen, M. Mann, Nat. Rev. Mol. Cell Biol. 2004, 5, 699.
[46] Y.-L. Ying, J. Yang, F.-N. Meng, S. Li, M.-Y. Li, Y.-T. Long, Research [72] J. R. Yates, S. K. R. Park, C. M. Delahunty, T. Xu, J. N. Savas,
(Washington, DC, US) 2019, 2019, 1050735. D. Cociorva, P. C. Carvalho, Nat. Methods 2012, 9, 455.
[47] F.-N. Meng, Y.-L. Ying, J. Yang, Y.-T. Long, Anal. Chem. 2019, 91, [73] R. Aebersold, M. Mann, Nature 2016, 537, 347.
9910. [74] S. Shao, T. Guo, C. C. Koh, S. Gillessen, M. Joerger, W. Jochum,
[48] W. Timp, G. Timp, Sci. Adv. 2020, 6, eaax8978. R. Aebersold, Proteomics 2015, 15, 3711.
[49] A. Asandei, G. Di Muccio, I. Schiopu, L. Mereuta, I. S. Dragomir, [75] B. Huang, H. Wu, D. Bhaya, A. Grossman, S. Granier, B. K. Kobilka,
M. Chinappi, T. Luchian, Small Methods 2020, 1900595. R. N. Zare, Science 2007, 315, 81.
[50] D. Pedone, M. Firnkes, Anal. Chem. 2009, 81, 9689. [76] Y. Zhu, P. D. Piehowski, R. Zhao, J. Chen, Y. Shen, R. J. Moore,
[51] Z. Gu, Y.-L. Ying, C. Cao, P. He, Y.-T. Long, Anal. Chem. 2015, 87, A. K. Shukla, V. A. Petyuk, M. Campbell-Thompson, C. E. Mathews,
907. R. D. Smith, W.-J. Qian, R. T. Kelly, Nat. Commun. 2018, 9,
[52] A. Fragasso, S. Schmid, C. Dekker, ACS Nano 2020, 14, 1338. 882.
[53] J. Roman, B. Le Pioufle, L. Auvray, J. Pelta, L. Bacri, Eur. Phys. J. E: [77] J. O. Kafader, R. D. Melani, M. W. Senko, A. A. Makarov,
Soft Matter Biol. Phys. 2018, 41, 99. N. L. Kelleher, P. D. Compton, Anal. Chem. 2019, 91, 2776.
[54] B. Cressiot, H. Ouldali, M. Pastoriza-Gallego, L. Bacri, [78] J. O. Kafader, R. D. Melani, K. R. Durbin, B. Ikwuagwu, B. P. Early,
F. G. van der Goot, J. Pelta, ACS Sens. 2019, 4, 530. R. T. Fellers, S. C. Beu, V. Zabrouskov, A. A. Makarov, J. T. Maze,
[55] C. Cao, D.-F. Liao, J. Yu, H. Tian, Y.-T. Long, Nat. Protoc. 2017, 12, D. L. Shinholt, P. F. Yip, D. Tullman-Ercek, M. W. Senko,
1901. P. D. Compton, N. L. Kelleher, Nat. Methods 2020, 422, 198.
[56] E. C. Yusko, J. M. Johnson, S. Majd, P. Prangkio, R. C. Rollings, J. Li, [79] P. Edman, Arch. Biochem. 1949, 22, 475.
J. Yang, M. Mayer, Nat. Nanotechnol. 2011, 6, 253. [80] J. Swaminathan, A. A. Boulgakov, E. T. Hernandez, A. M. Bardo,
[57] J. Roman, N. Jarroux, G. Patriarche, O. Francais, J. Pelta, J. L. Bachman, J. Marotta, A. M. Johnson, E. V. Anslyn,
B. Le Pioufle, L. Bacri, ACS Appl. Mater. Interfaces 2017, 9, 41634. E. M. Marcotte, Nat. Biotechnol. 2018, 36, 1076.
[58] N. Giamblanco, D. Coglitore, J.-M. Janot, P. E. Coulon, B. Charlot, [81] Y. Zhao, B. Ashcroft, P. Zhang, H. Liu, S. Sen, W. Song, J. Im,
S. Balme, Sens. Actuators, B 2018, 260, 736. B. Gyarfas, S. Manna, S. Biswas, C. Borges, S. Lindsay, Nat. Nano-
[59] J. Sha, W. Si, B. Xu, S. Zhang, K. Li, K. Lin, H. Shi, Y. Chen, Anal. technol. 2014, 9, 466.
Chem. 2018, 90, 13826. [82] T. Ohshiro, M. Tsutsui, K. Yokota, M. Furuhashi, M. Taniguchi,
[60] R. Hu, J. V. Rodrigues, P. Waduge, H. Yamazaki, B. Cressiot, T. Kawai, Nat. Nanotechnol. 2014, 9, 835.
Y. Chishti, L. Makowski, D. Yu, E. Shakhnovich, Q. Zhao, [83] Y. Yao, M. Docter, J. van Ginkel, D. de Ridder, C. Joo, Phys. Biol.
M. Wanunu, ACS Nano 2018, 12, 4494. 2015, 12, 055003.
[61] B. Cressiot, S. J. Greive, M. Mojtabavi, A. A. Antson, M. Wanunu, [84] J. Swaminathan, A. A. Boulgakov, E. M. Marcotte, PLoS Comput.
Nat. Commun. 2018, 9, 4652. Biol. 2015, 11, 1004080.
[62] N. Giamblanco, D. Coglitore, A. Gubbiotti, T. Ma, E. Balanzat, [85] J. van Ginkel, M. Filius, M. Szczepaniak, P. Tulinski, A. S. Meyer,
J.-M. Janot, M. Chinappi, S. Balme, Anal. Chem. 2018, 90, 12900. C. Joo, Proc. Natl. Acad. Sci. USA 2018, 115, 3338.

Small Methods 2020, 2000090 2000090 (13 of 13) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim