Millisecond\nLabel-Free Single Peptide Detection and\nIdentification Using Nanoscale Electrochromatography and Resistive\nPulse Sensing

We are developing a unique protein identification method\nthat consists\nof generating peptides proteolytically from a single protein molecule\n(i.e., peptide fingerprints) with peptide detection and identification\ncarried out using nanoscale electrochromatography and label-free resistive\npulse sensing (RPS). As a step in realizing this technology, we report\nherein the nanoscale electrochromatography of model peptides using\nthermoplastic columns with surfaces engineered to identify peptides\nvia their molecularly dependent mobility (i.e., time-of-flight, ToF).\nToFs were elucidated using a dual in-plane nanopore sensor, which\nconsisted of two in-plane nanopores placed on either end of the nanoelectrochromatography\ncolumn. The surface of the nanocolumn, which consisted of poly(methyl\nmethacrylate) (PMMA), was activated with an O<sub>2</sub> plasma,\ncreating surface carboxylic acid groups (−COOH) inducing a\nsurface charge on the column wall as well as affecting its hydrophilicity.\nTo understand scaling effects, we carried out microchip and nanochannel\nelectrochromatography of the peptides labeled with an ATTO 532 reporter\nto allow for single-molecule tracking. Our results indicated that\nthe apparent mobilities of the model peptides did not allow for their\nseparation in a microchannel, but when performed in a nanocolumn,\nclear differences in their apparent mobilities could be observed especially\nwhen operated at high electric field strengths. We next performed\nlabel-free detection of peptides using the dual in-plane nanopore\nsensor with the two pores separated by a 5 μm (length) column\nwith a 50 nm width and depth. When a single peptide molecule passed\nthrough an in-plane nanopore, the sensor read a pair of resistive\npulses with a time difference equivalent to ToF. We identified the\npeptides by evaluating their ToF, normalized RPS current transient\namplitude (Δ<i>I</i>/<i>I</i><sub>0</sub>), and RPS peak dwell time (<i>t</i><sub>d</sub>). We could\nidentify the model peptides with nearly 100% classification accuracy\nat the single-molecule level using machine learning with a single\nmolecule measurement requiring <10 ms.