Location of Phosphorylation Sites within Long Polypeptide Chains by Binder-Assisted Nanopore Detection

The detection and mapping of protein phosphorylation sites are essential for understanding the mechanisms of various cellular processes and for identifying targets for drug development. The study of biopolymers at the single-molecule level has been revolutionized by nanopore technology. In this study, we detect protein phosphorylation within long polypeptides (>700 amino acids), after the attachment of binders that interact with phosphate monoesters; electro-osmosis is used to drive the tagged chains through engineered protein nanopores. By monitoring the ionic current carried by a nanopore, phosphorylation sites are located within individual polypeptide chains, providing a valuable step toward nanopore proteomics.


■ INTRODUCTION
Post-translational modifications (PTMs) of proteins are pivotal in cell regulation and typically involve the enzymatic addition of chemical groups to amino acid side chains. 1 Phosphorylation, the process of adding a phosphate group to predominantly serine, threonine, and tyrosine residues, is the most prevalent PTM, with over ∼10 6 phosphorylation sites that account for >60% of all reported PTMs. 1 Dysregulation of phosphorylation is commonly associated with diseases such as cancer, Parkinson's, and Alzheimer's. 2 For example, tau proteins in pathological lesions of Alzheimer's are heterogeneously and highly phosphorylated, with more than 50 identified phosphorylation sites. 3Bottom-up mass spectrometry is routinely applied to detect PTMs on peptide fragments derived from disease-related proteins but faces challenges to determine if widely separated modifications, whether identical or distinct, are present on the same polypeptide chain.For example, cross-talk between phosphorylation and O-GlcNAcylation was reported to regulate subcellular localization of proteins, such as tau. 4 However, there lacks a straightforward technique to correlate the presence of PTMs at distant sites within individual polypeptide chains. 5Nanopore nucleic acid sequencing has emerged as a powerful technology to provide ultralong DNA or RNA reads for long-range correlation of genomic or transcriptomic features. 6,7−12 Electro-osmosis has been demonstrated to propel unfolded polypeptides through nanopores 13−15 and PTMs deep within long polypeptide chains have been located during translocation. 14This work is a first step toward the label-free analysis of modified proteins extracted from biological samples. 14In parallel, the identification of PTMs on short peptides (up to ∼30 amino acids) has been achieved, 16−19 either when the peptides are sensed as a whole or when a peptide is transported through the pore as a conjugate to an oligonucleotide. 18Although PTMs containing branched structures (e.g., glycans) or entire proteins (e.g., ubiquitin) might be challenging to detect on polypeptides translocating through nanopores of ∼1−2 nm in internal diameter, >80% of the ∼400 PTM types are small (<∼300 Da) or narrow in shape. 1 In addition, nanopores with wider internal geometries (e.g., ClyA) might be applicable for sensing bulkier PTMs.To demonstrate the broad applicability of the approach, we previously detected three PTMs (phosphorylation, glutathionylation, and glycosylation) on full-length proteins when segments of singly modified individual thioredoxin (Trx)-linker concatemers were stalled during translocation through a nanopore. 14To our surprise, glutathionylation and phosphorylation, placed at a particular site in the polypeptide, produced similar current blockades and noise patterns. 14This prompts the question of how many PTMs can be discriminated among the 400 different natural PTMs identified so far by their perturbation to the ionic current driven through a protein nanopore. 1To distinguish PTMs with similar electrical signatures or to allow targeted detection of specific PTMs, we sought to use PTM-specific binders to generate distinct current characteristics.To this end, we have explored a phosphorylation-specific reversible chemical binder, Phos-tag,  -P) 274 amino acids apart on a Trx-linker pentamer were detected.Level A1 for linker 3 showed a slightly lower I res% compared to that of linker 1 (I res% of linker 1 is shown in orange dash).This difference was attributed to the additional amino acid sequence in linker 3 (Table S1).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events, ΔI res% = I res% (A1, linker 1) − I res% (A1-P), where I res% (A1, linker 1) is the I res% value of the A1 level for linker 1 within an individual translocation event.If there were two Ser-P molecules detected in different segments within a single translocation event, they were analyzed individually.(c) Left: Phos-tag dizinc complexes bound to phosphoserine generated alternating current levels (A1-P-PZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light purple are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PZn 2 -H), while data points in dark purple are the I r.m.s. and ΔI res% values for the lower level of the two-level A1 state (A1-P-PZn 2 -L).(d) Left: Phos-tag-acrylamide dizinc complexes bound to serine phosphate produced alternating current levels (A1-P-PAZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light green are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PAZn

Journal of the American Chemical Society
which binds selectively and strongly to phosphate monoesters when complexed with zinc ions (e.g., for phosphoserine or phosphothreonine residues within model peptides, K d = ∼0.7 μM; for phosphotyrosine residues within model peptides, 20,21 Phos-tag produced distinctive modulation of the associated ionic current as phosphorylated polypeptide chains were translocated through an engineered nanopore, thereby mediating enhanced localization of phosphorylation sites within long polypeptide chains.

■ RESULTS AND DISCUSSION
In our previous research, we employed an anion-selective αhemolysin (αHL) mutant (NN-113R) 7 (permeability ratio P Na + /P Cl − = 0.33) 22 to generate electro-osmotic flow, thereby driving the capture, linearization, and translocation of polypeptide chains.We identified and located PTMs on long polypeptide chains of up to nine thioredoxin units (Trx, 108 amino acids (aa)) connected by linkers (29 aa). 14Each Trx unit within the Trx-linker concatemers had the two catalytic cysteines removed (Trx: C32S/C35S). 8Chaotropic reagents (e.g., guanidinium chloride (GdnHCl) or urea) at nondenaturing concentrations were used to promote cotranslocational unfolding. 14During the electro-osmotic translocation of the Trx-linker concatemers, features comprising three levels were seen (A1, A2, and A3) (Figure 1a,b).We provisionally assigned level A1 to be produced by the nanopore containing a threaded linker ahead of a folded Trx unit, level A2 to be produced when a partially unfolded C-terminus of a Trx unit extends into the nanopore, and level A3 to be produced by the spontaneous unfolding and passage of the remaining Trx polypeptide chain through the nanopore.In the presence of a PTM in the linker, a phosphate group (P), for instance, level A1 exhibited a reduced percentage residual current (I res% ) value and higher root-mean-square noise (I r.m.s.) 14 (Figure 1b).Here, we examined the detection of phosphorylation in association with phosphate-specific binders: Phos-tag dizinc complex (PZn 2 ) and Phos-tag-acrylamide dizinc complex (PAZn 2 ).Phos-tag is commonly immobilized in SDS-PAGE gels to detect phosphoproteins 23,24 and applied to generate mass shifts in MALDI-TOF mass spectrometry. 25We constructed a Trx-linker pentamer ((Trx-linker) 5 ) containing two phosphorylation sites (RRAS) in linker 2 and linker 4 (Figures 1a, S1, and Table S1), which were phosphorylated on serine by the catalytic subunit of protein kinase A (Figure S2).Phosphorylated polypeptides were captured, unfolded, and translocated by electro-osmosis through the (NN-113R) 7 αHL pore.GdnHCl (750 mM) was employed to accelerate cotranslocational unfolding.Consistent with prior findings, translocation of the pentamer, C-terminus first, generated current patterns with a maximum of 4 A1−A3 repeats following an initial spike (Figure 1b).The spike to around 0 pA at the beginning of nearly all the translocation events was attributed to rapid unfolding and translocation of the first Cterminal Trx-linker unit.While only ∼6% of the doubly phosphorylated Trx-linker pentamers produced 4 repeating A1−A3 features following an initial spike, >72% of the recorded translocation events contained at least one A1 level with a reduced I res% value and a higher I r.m.s., compared to A1 levels for unmodified segments (Table S2).These characteristics were consistent with the electrical profiles previously identified for a phosphorylated linker and were therefore assigned as level A1-P.In events where 4 repeats of A1−A3 features were observed following an initial spike, the level A1-P was recorded for both the second and fourth units, consistent with the presence of two phosphorylated serine residues (Ser-P) within linker 2 and linker 4, 274 amino acids apart within the polypeptide chain.
Next, we sought to determine if PAZn 2 would enable us to distinguish phosphorylation from a PTM that exhibits a similar ionic blockade. 14We constructed a Trx-linker pentamer with distinct modification sites in linker 2 (RRASAA) and linker 4 (RRAAAC).We carried out phosphorylation and glutathionylation reactions sequentially to obtain a Trx-linker pentamer with Ser-P in linker 2 and glutathionylated cysteine (Cys-GS) in linker 4 (Figure 2a).In line with the characteristic current patterns recorded separately with Trx-linker nonamers containing a single Ser-P or Cys-GS residue within the same linker sequence, 14 the signals from Ser-P and Cys-GS within the same Trx-linker pentamer exhibited indistinguishable residual currents and noise when linker 2 and linker 4 were located within the pore (Figure 2a).Pleasingly, the introduction of PAZn 2 altered the signal derived from linker 2 to give a pattern similar to that of level A1-P-PAZn 2 , while the signal from linker 4 was unchanged, allowing clear differentiation between phosphorylation and glutathionylation (Figure 2b).

■ CONCLUSIONS
Here, we demonstrate the nanopore detection of widely separated phosphorylation sites (e.g., >250 aa apart) within a polypeptide chain by using the Phos-tag dizinc complex (PZn 2 ) and the Phos-tag-acrylamide dizinc complex (PAZn 2 ).The binder created a distinct two-level current feature when phosphorylated polypeptide segments were inside the nanopore, which resembled current patterns observed during divalent cation chelation within an engineered αHL pore 26 or with amino acids interacting with immobilized Ni 2+ in an engineered nanopore. 28We were able to saturate Ser-P with PAZn 2 (10 mM HEPES, pH 7.2, 750 mM GdnHCl, 2.37 μM Trx-linker pentamer, 2.37 mM Phos-tag-acrylamide, 4.74 mM ZnCl 2 ((Trx-linker) 5 :Phos-tag-acrylamide:ZnCl 2 = 1:1000:2000).The phosphorylation-specific current feature enabled the discrimination of phosphorylation from PTMs that produced similar current blockades such as glutathionylation.We envision that combinations of PTM-specific binders will allow the simultaneous detection of multiple PTMs.Suitable binders should recognize PTMs irrespective of adjacent amino acid sequences, exhibit fast association and slow dissociation kinetics, and ideally generate characteristic current signatures, such as the subconductance states seen in this work, which facilitate the discrimination of PTMs.Given that most PTMs are enzymatically installed and regulated, they tend to be located within the flexible or exposed regions of proteins. 29For the small fraction of PTMs that are nonenzymatically installed within the buried regions of proteins (e.g., disulfide bonds), partial unfolding might occur in the presence of chaotropic reagents to allow binder association.So far, we have identified PTMs in polypeptide segments while they are transiently arrested within a nanopore.In our ongoing efforts to study biologically relevant protein targets, the use of bulky binders (e.g., antibodies) holds promise for temporarily halting protein translocation at the pore entrance, thereby mediating PTM identification in any region of a protein.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.4c03912.Amino acid sequences of the thioredoxin-linker pentamers; and I r.m.s.characteristics of A1-P, A1-P-PZn 2 , and A1-P-PAZn 2 ; mean dwell times (<τ>) for two-level A1-P-PAZn 2 ; an SDS-polyacrylamide gel of the Trx-linker pentamer; ESI LC-MS characterization of Trx-linker pentamers; fractions of phosphorylated linkers detected in the PAZn 2 -bound state in the absence and presence of competing phosphoserine; an ionic current trace showing phosphorylated linkers in excess Zn 2+ generated similar A1-P current signals; a current trace showing transition between level A1-P-PAZn 2 and level A1-P when a phosphorylated segment was inside the (NN-113R)

Figure 1 .
Figure 1.Detection of serine phosphate bound to Phos-tag in a polypeptide chain.(a) Monitoring the Trx-linker pentamer traversing the αhemolysin nanopore (NN-113R)7 .The Trx-linker pentamer contained two RRAS sequences within linker 2 and linker 4, which were phosphorylated enzymatically on serine.(b) Left: phosphorylated serine residues (Ser-P) 274 amino acids apart on a Trx-linker pentamer were detected.Level A1 for linker 3 showed a slightly lower I res% compared to that of linker 1 (I res% of linker 1 is shown in orange dash).This difference was attributed to the additional amino acid sequence in linker 3 (TableS1).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events, ΔI res% = I res% (A1, linker 1) − I res% (A1-P), where I res% (A1, linker 1) is the I res% value of the A1 level for linker 1 within an individual translocation event.If there were two Ser-P molecules detected in different segments within a single translocation event, they were analyzed individually.(c) Left: Phos-tag dizinc complexes bound to phosphoserine generated alternating current levels (A1-P-PZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light purple are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PZn 2 -H), while data points in dark purple are the I r.m.s. and ΔI res% values for the lower level of the two-level A1 state (A1-P-PZn 2 -L).(d) Left: Phos-tag-acrylamide dizinc complexes bound to serine phosphate produced alternating current levels (A1-P-PAZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light green are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PAZn 2 -H), while data points in dark green are the I r.m.s. and ΔI res% values for the lower level of the two-level A1 state (A1-P-PAZn 2 -L).If there were two A1-P-PZn 2 or A1-P-PAZn 2 detected in different segments in a single translocation event, they were analyzed individually.Conditions in (b): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, and 2.37 μM Trx-linker pentamer (cis).Conditions in (c): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, 2.37 μM Trx-linker pentamer (cis), 118.5 μM Phos-tag (cis), and 237 μM ZnCl 2 (cis).Condition in (d): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, 2.37 μM Trx-linker pentamer (cis), 118.5 μM Phos-tag-acrylamide (cis), and 237 μM ZnCl 2 (cis).All of the measurements were conducted at +140 mV (trans) and 23 ± 1 °C.
Figure 1.Detection of serine phosphate bound to Phos-tag in a polypeptide chain.(a) Monitoring the Trx-linker pentamer traversing the αhemolysin nanopore (NN-113R)7 .The Trx-linker pentamer contained two RRAS sequences within linker 2 and linker 4, which were phosphorylated enzymatically on serine.(b) Left: phosphorylated serine residues (Ser-P) 274 amino acids apart on a Trx-linker pentamer were detected.Level A1 for linker 3 showed a slightly lower I res% compared to that of linker 1 (I res% of linker 1 is shown in orange dash).This difference was attributed to the additional amino acid sequence in linker 3 (TableS1).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events, ΔI res% = I res% (A1, linker 1) − I res% (A1-P), where I res% (A1, linker 1) is the I res% value of the A1 level for linker 1 within an individual translocation event.If there were two Ser-P molecules detected in different segments within a single translocation event, they were analyzed individually.(c) Left: Phos-tag dizinc complexes bound to phosphoserine generated alternating current levels (A1-P-PZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light purple are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PZn 2 -H), while data points in dark purple are the I r.m.s. and ΔI res% values for the lower level of the two-level A1 state (A1-P-PZn 2 -L).(d) Left: Phos-tag-acrylamide dizinc complexes bound to serine phosphate produced alternating current levels (A1-P-PAZn 2 ).Right: scatter plot of I r.m.s. and ΔI res% for individual translocation events.Data points in light green are the I r.m.s. and ΔI res% values for the higher level of the two-level A1 state (A1-P-PAZn 2 -H), while data points in dark green are the I r.m.s. and ΔI res% values for the lower level of the two-level A1 state (A1-P-PAZn 2 -L).If there were two A1-P-PZn 2 or A1-P-PAZn 2 detected in different segments in a single translocation event, they were analyzed individually.Conditions in (b): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, and 2.37 μM Trx-linker pentamer (cis).Conditions in (c): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, 2.37 μM Trx-linker pentamer (cis), 118.5 μM Phos-tag (cis), and 237 μM ZnCl 2 (cis).Condition in (d): 10 mM HEPES, pH 7.2, 750 mM GdnHCl, 2.37 μM Trx-linker pentamer (cis), 118.5 μM Phos-tag-acrylamide (cis), and 237 μM ZnCl 2 (cis).All of the measurements were conducted at +140 mV (trans) and 23 ± 1 °C.