Standardized workflow for multiplexed charge detection mass spectrometry on orbitrap analyzers

Abstract

Individual ion mass spectrometry (I²MS) is the Orbitrap-based extension of the niche mass spectrometry technique known as charge detection mass spectrometry (CDMS). While traditional CDMS analysis is performed on in-house-built instruments such as the electrostatic linear ion trap, I²MS extends CDMS analysis to Orbitrap analyzers, allowing charge detection analysis to be available to the scientific community at large. I²MS simultaneously measures the mass-to-charge ratios (m/z) and charges (z) of hundreds to thousands of individual ions within one acquisition event, creating a spectral output directly into the mass domain without the need for further spectral deconvolution. A mass distribution or ‘profile’ can be created for any desired sample regardless of composition or heterogeneity. To assist in reducing I²MS analysis to practice, we developed this workflow for data acquisition and subsequent data analysis, which includes (i) protein sample preparation, (ii) attenuation of ion signals to obtain individual ions, (iii) the creation of a charge-calibration curve from standard proteins with known charge states and finally (iv) producing a meaningful mass spectral output from a complex or unknown sample by using the STORIboard software. This protocol is suitable for users with prior experience in mass spectrometry and bioanalytical chemistry. First, the analysis of protein standards in native and denaturing mode is presented, setting the foundation for the analysis of complex mixtures that are intractable via traditional mass spectrometry techniques. Examples of complex mixtures included here demonstrate the relevant analysis of an intact human monoclonal antibody and its intricate glycosylation patterns.

Introduction

Mass spectrometry (MS) is uniquely positioned for the mass measurement and molecular identification of biological molecules and their assemblies¹. MS has gained increasing attention over the past few decades with advancements in instrumentation techniques^2–5. Conventional MS measures the mass-to-charge ratio (m/z) of ions, where the absolute mass and the number of charges of the ions are convoluted. Charge detection MS (CDMS) is an MS technique that simultaneously measures charges and m/z values of single particles, from which mass values can be directly determined⁶. CDMS opens up a new analytical avenue for megadalton-sized biological assemblies and extremely heterogeneous mixtures of highly modified proteins and their complexes^7–9. CDMS has traditionally been built on custom instrumentation platforms such as electrostatic linear ion traps^10,11 and time-of-flight mass spectrometers^12–14. These platforms rely on highly specialized components such as energy filters and cryogenic detectors, which raises challenges for them to be accessed by the broad scientific community. In addition, the measurement of one or a few ions at a time imposes a throughput barrier for the dissemination of CDMS to large-scale applications such as reaction screening and proteome profiling¹¹.

Orbitrap-based individual ion MS (I²MS) is the first CDMS application on a broadly accessible MS platform⁹. It provides mass measurement with high resolution, high mass accuracy, improved throughput (tens of ions per measurement at a minimum) and top-down tandem MS capabilities^9,15. Moreover, I²MS can be readily accessed from existing commercial Orbitrap ultra-high mass range (UHMR) mass spectrometers that require no hardware modifications. Demonstrated I²MS applications encompass both academia and industry, ranging across therapeutic diagnostics including virus-like particle composition and cargo determination⁹, vaccine development^16,17, characterizations of highly modified immunoprecipitated proteins^16,18,19, single-cell proteomics^20,21 and biological tissue imaging^22,23.

Despite the advances in I²MS, the scientific community lacks a standardized protocol and a user-friendly software interface to generate meaningful spectra in the mass domain. Herein, we present a detailed protocol for I²MS analysis of native and denatured protein standards and complex mixtures, including critical steps and challenges in the experimental setup and data acquisition, along with the capabilities of software dedicated to I²MS data processing and visualization. We also discuss the fundamentals, application scenarios, advantages and limitations of Orbitrap-based I²MS.

Development of the protocol

I²MS is a CDMS technique that uses Orbitrap MS platforms to determine the molecular mass of multiply charged individual ions^9,24. The Orbitrap mass analyzer detects ions by using image current induced on the outer electrodes. In an I²MS data-acquisition event, the Orbitrap mass analyzer records the induced image current of many individual ions, and the m/z values of the ions are determined from their axial oscillation frequencies^2,25–27. Simultaneously during the detection event, the induced image current of each ion as a function of time on the Orbitrap outer electrode is used to generate a Selective Temporal Overview of Resonant Ions (STORI) plot, from which the quantized charge value of each individual ion is determined^9,28. The neutral masses of the protein ions are calculated by

$$\text{Neutral}\text{Mass}=\left(\frac{m}{z}\times z\right)-\left(z\times M_{\text{proton}}\right)$$

where $m$ and $z$ are the mass and charge of the ion, and $M_{\text{proton}}$ is the mass of a proton ($M_{\text{proton}}=1.007825\mathrm{D}\mathrm{a}$). This equation applies to positive protein ions with protonation. The general concept can also be applied to negative ion mode, in which neutral mass is obtained by adding the number of charges multiplied by $M_{\text{proton}}$ ²⁹. I²MS directly benefits from the high mass resolution of Orbitrap mass analyzers to assign an individual ion signal to a specific accurate mass^24,26. While using I²MS to determine the mass distribution of a protein, the molecular masses of many copies of the same protein with different isotopic compositions are individually determined⁹. A mass distribution of the protein is generated by compiling all the molecular masses in the true mass domain⁹.

Our group has published a standard operation procedure for native MS analysis using Orbitrap platforms^30,31. In that work, we described a sample-preparation procedure for the analysis of standard native protein complexes using nanoelectrospray ionization (nESI), along with instrument optimization for obtaining high-quality intact MS and tandem MS spectra of the native complexes³⁰. In contrast to these traditional ‘ensemble-mode’ Orbitrap MS measurements in which the m/z of proteins is obtained by using Fourier transform of a time-domain signal, in this protocol, we are making a major transition by using the I²MS technology, which enables the direct readout of mass distribution of these native protein complexes⁹. A few major improvements over the previously published protocol are higher mass resolution, higher mass accuracy and higher intraspectral dynamic range that can be obtained from highly heterogeneous samples⁹. In particular, we can obtain isotopic resolution for protein complexes up to 232 kDa (tetrameric pyruvate kinase)³² within 15 min of data acquisition (~500 MS spectra). Furthermore, using this protocol, we can routinely obtain isotopically resolved mass measurement of intact monoclonal IgG antibody with the ability to map associated glycoforms³³.

In the initial I²MS publications, we performed the data processing by using in-house-built software^9,32. In this protocol, we introduce STORIboard, a commercially available software that allows for I²MS data processing and visualization. Initially, an instrument-specific charge-calibration curve needs to be constructed from standard protein complexes by using STORIboard; once the charge calibration is complete, mass determination of the unknown samples can be performed within the software by user-defined parameters.

The protocol is composed of four sections including (i) protein sample preparation and ion source optimizations, (ii) the optimization of the experimental setup and instrumental parameters to obtain individual ion signals, (iii) the use of the STORIboard software to create an instrument-specific charge-calibration curve by using standard proteins for charge assignment and (iv) the production of a meaningful mass-domain spectral output for a complex or unknown sample. All the steps in this protocol are demonstrated by using the Q Exactive UHMR hybrid quadrupole-Orbitrap mass spectrometer as the instrument platform.

Applications of the method

In this protocol, we focus on the applications of I²MS on a specific instrument platform (Q Exactive UHMR) commercially available as direct mass technology mode (DMT mode) for the analysis of large proteins and their complexes. The same instrument is also suitable for the analysis of other biomolecular complexes such as protein-DNA complexes and virus particles³⁴. We have previously demonstrated the I²MS mass analysis of intact nucleosomes, which are structural units of the eukaryote chromatin composed of a DNA segment and eight histone proteins³⁵.

The QE Orbitrap UHMR system uses a central electrode voltage of −5 kV for positive ion mode, which is suitable for the analysis of native protein complexes without any hardware modification³⁶. Instrument modification is required when extending I²MS to denatured proteins and proteoforms because of the more rapid decay of denatured proteins of larger cross sections in the Orbitrap mass analyzer^28,36–38. In particular, the electronic circuit board that provides the potential on the central electrode of the Orbitrap mass analyzer can be modified to provide −1 kV instead of −5 kV for positive ion mode on a Q Exactive Plus system to reduce the kinetic energy of the ions in the Orbitrap and decrease collision-induced ion decay⁹. This instrument modification allowed for the identification of ~500 proteoforms from a 0–30-kDa fraction in a human cell lysate⁹. In addition, the same system has been used to image and identify proteoforms up to 70 kDa directly from tissue^22,23.

Comparison with other methods

I²MS uses the Orbitrap mass analyzer to determine the charge and the m/z of ions⁹. The core of CDMS instrumentation is the integration of two analytical modalities that simultaneously measures charge and m/z of single ions in the same acquisition event. Aside from Orbitrap systems^34,39, the electrostatic linear ion trap (ELIT)^10,40–43 is a mainstream CDMS instrumentation platform that simultaneously determines the m/z of an ion from the frequency of oscillation between a pair of electrostatic ion mirrors, and ion charge from the induced image charge on a cylindrical metal tube as the ion passes through it^44,45. The ELIT is particularly advantageous in measuring large particle masses in the megadalton range with hundreds of charges while circumventing the ambiguity in mass determination in highly heterogeneous samples⁶. A range of large biological and polymeric particles, including but not limited to native protein complexes^46,47, DNAs⁴⁵, adeno-associated virus particles^48,49, virus-like particles⁵⁰, bacteriophages⁵¹, vaccines⁵², protein fibers⁵³, exosomes^54,55 and lipoproteins⁵⁶, have been characterized by using ELITs. Despite significant advances in the determination of mass and charge accuracy^11,57–60, most ELIT-based CDMS systems have limitations in throughput, measuring up to ~20 ions at a time⁶¹. Advanced approaches in signal processing and instrumentation, such as high-order frequency harmonics^47,61,62 or arrays of ELITs⁶³, have been developed to improve the throughput of measurement. In addition, ELIT-based CDMS relies on custom-built instrument platforms, which is challenging for the dissemination of CDMS to the scientific community⁶. Aside from Orbitrap and ELIT, other CDMS instrument platforms use time-of-flight mass spectrometers coupled to cryogenic detectors, which requires sophisticated instrumentation and specialized components for broad and large-scale applications^12,13.

ESI is the ionization method used in this protocol, which generates multiply charged protein ions via desolvation of charged microdroplets^64,65. ESI is a soft ambient ionization technique that has been used for the ionization of analytes from lipids and metabolites to large non-covalent complexes and biological particles^66,67. Other ionization methods that generate multiply charged protein ions from solid or liquid samples including desorption electrospray ionization (DESI)⁶⁸, nanospray DESI⁶⁹, paper spray ionization⁷⁰ and laser-ablation techniques such as matrix-assisted laser DESI⁷¹ and desorption by impulsive excitation⁷².

In this protocol, protein samples are directly infused into a mass spectrometer through nESI from an MS-friendly buffer solution. In cases in which protein samples are stored in MS-incompatible buffer solutions, MS spectra will contain highly heterogeneous adducted protein ions challenging for interpretation⁷³. To address this problem, a few front-end separation approaches have been coupled to ESI for protein sample introduction, which can be readily adapted to I²MS analysis. Capillary electrophoresis⁷⁴ and liquid chromatography (LC)⁷⁵ have been widely used for denatured protein analysis. I²MS data acquisition has been demonstrated on shorter time scale implementation for capillary electrophoresis⁷⁶ and LC⁷⁷. For online protein purification in native mode, a few techniques such as capillary zone electrophoresis⁷⁸, flow field-flow fractionation⁷⁹ and size-exclusion chromatography⁸⁰ may be coupled to I²MS.

More broadly speaking, Orbitrap-based I²MS/CDMS is a technique that enables the mass measurement and molecular characterization of single particles. Aside from MS, frequency-addressed nanomechanical resonators determine the mass of a neutral particle by using the frequency shift of a nanoscale resonator when the particle is adsorbed⁸¹. Single-particle mass measurements in solution have been achieved by using mass photometry, which determines the mass of a particle from light scattering upon adsorption of the particle onto a glass slide at the solid-liquid interface^82–84. Despite the accessibility to broad mass ranges and the ability to measure the mass of biological particles that are difficult to purify and isolate, the mass resolution and mass accuracy in these techniques are limited.

Limitations

Although I²MS offers a new opportunity for accurate mass measurement of complex protein samples via measuring charge and m/z individually, the precision of charge assignment in I²MS is still limited²⁴. In theory, one STORI slope value corresponds to a specific charge state. However, because of the uncertainty in the data-acquisition process, the measured STORI slope value of an ion may be misassigned to neighbor charge states. In the data-analysis process embedded in STORIboard, we have applied statistical analysis coupled to either Central Limit Theorem (‘>500 kDa’ option) or an iterative voting algorithm (‘<500 kDa’ option) to address this issue. However, these algorithms require many ions to build the statistics for confident charge assignment, and overfitting the data is a concern, particularly for heterogeneous samples. An improved signal processing algorithm (misSTORI) is available for mass shifting ions in the megadalton range⁸⁵. Moreover, accuracy in STORI analysis may be improved by acquiring a longer integrated induced signal⁸⁶. Although this can be achieved by extending the transient length, longer trapping time may result in more ion decay and longer data acquisition time to acquire the same number of ions. Other groups have worked on precision in charge assignment on ELIT platforms^57,58 and developed computational tools⁸⁷ for charge assignment of individual ions collected in Orbitrap, which may inspire future improvement in I²MS charge assignment.

Despite the ability to achieve isotopic resolution for large proteins and their complexes not obtainable by using ensemble MS measurements, the experimental throughput of I²MS is limited. In the I²MS workflow, we obtain individual ions by attenuating the total ion signals in the Orbitrap MS acquisition events⁹. As a result, a long data acquisition time is necessary to obtain a mass distribution with acceptable ion statistics. This limitation is more pronounced for small- or medium-sized proteins (<30 kDa), in which ensemble mode data acquisition using Orbitrap or other high-resolution MS systems is feasible and more efficient. For example, Orbitrap MS in ensemble mode generates a spectrum for native carbonic anhydrase (29 kDa) in <5 min at isotopic resolution (Extended Data Fig. 1). However, 30 min of data acquisition was needed to collect spectra by using the I²MS workflow.

I²MS is advantageous for accurate mass measurements of larger proteins and their complexes (>50 kDa), which is increasingly challenging for traditional high-resolution MS as the protein size becomes larger⁸⁸. The increased challenge is due to (i) the limited resolving power to differentiate densely packed isotopes of highly charged protein peaks and (ii) complex ion dynamics including collision-induced ion decay at increased measurement times. However, the experimental throughput is more important than accurate mass determination in many cases for >200-kDa protein complexes because of their molecular heterogeneity in native MS. In particular, many molecular forms of the native protein spaced by subtle mass differences may overlap in the mass-domain spectrum obtained from I²MS. Although isotopic resolution may be obtained for these overlapping features, identifying the molecular forms and their relative abundances is overwhelmingly challenging. In these cases, ensemble mode measurement provides enough chemical and biological context for the sample at higher throughput even though the mass definition is not accurate.

Another limitation of I²MS is the challenge of obtaining meaningful data for samples with little or no background information. To get a spectrum from an unknown sample, it may take a few iterations to optimize the instrument settings and data-acquisition parameters. Future research will focus on advancing the automated data-acquisition algorithm to properly maximize the number of single ions in a single MS spectrum. For an easy entry to challenging samples, we recommend using automatic ion control (AIC) for injection time control⁷⁶ and considering the instrument settings and optimization procedures discussed in this protocol (e.g., Instrument parameter optimization; Supplementary Table 1).

Experimental design

Overview of the protocol

This protocol is composed of four sections (Fig. 1) including (i) protein sample preparation and ion source optimizations for nESI (Steps 1–13), (ii) the optimization of the experimental setup and instrumental parameters to obtain individual ion signals (Steps 14–18), (iii) the creation of an instrument-specific charge-calibration curve by using standard proteins for charge assignment (Steps 19–21) and (iv) the production of meaningful mass-domain spectral output for a complex sample (Steps 22–32). In the following sections, we highlight a few key factors that we recommend considering while following this protocol.

Fig. 1 |. Workflow overview of the four main parts of the protocol.

(i) Sample preparation and introduction into the mass spectrometer (Steps 1–13), (ii) MS optimization for individual ion data acquisition (Steps 14–18), (iii) charge calibration generation (Steps 19–21) and (iv) application to samples (Steps 22–32). HV, high voltage.

Sample preparation

Many common buffer solutions that store protein samples are not MS compatible and generate heavily adducted signals in MS⁷³. To address this issue, protein samples need to be desalted and exchanged to MS-friendly buffer solutions for nESI analysis (see Reagents). In our previous publications, we have described the procedures for desalting and buffer exchange by using centrifugal filters^30,31. To obtain high-purity samples with good recovery rate, a few key factors in the filtration process need to be considered. In particular, except for carbonic anhydrase, which is purified by using a 10-kDa molecular-weight-cutoff (MWCO) filter, we use 100-kDa MWCO filters for all the native standards used in the protocol, which helps to eliminate high-molecular-weight impurities below the analyte of interest. In addition, we acknowledge the variations across different centrifugation devices. Nevertheless, we recommend using a lower centrifugal speed (10,000–12,000g) than the 14,000g recommended in the product instruction manual, to avoid mechanical breakdown of the filter. For each sample, 7–10 consecutive washes with AmAc solutions (100 mM) is recommended to achieve high signal quality for nESI.

The concentration of the samples for nESI is a critical factor in the collection of individual ions. At high concentration, individual ions may be obtained by attenuating total ion signal (Step 16) via a decrease in the MS injection time. However, this results in an overall low duty cycle for data collection because of the short injection time used. We recommend protein concentrations of 0.1–1 μM for both native and denatured samples used for I²MS. If the sample is heterogeneous, higher total protein concentrations (1–10 μM) might be considered.

Ion source optimization

Before and during sample introduction into the mass spectrometer, a few optimization steps for the nESI emitter should be considered (Steps 9–13; Fig. 2b, 1–3). The emitter tip should be carefully inspected under a microscope before installation onto the nanoFlex source (Step 9). Depending on the product, the tips may be received as closed, in which case the sharp end should be gently clipped by using precision tweezers under the microscope. The tip should be clipped as little as possible to keep a small diameter on the sharp end. Tips with large diameters will generate large ESI droplets that are difficult to desolvate and generate clean analyte ion distributions⁸⁹. In this protocol, we use commercial nESI emitters (ES387, Thermo Fisher Scientific), with a typical appearance under the microscope shown in Fig. 2a. While infusing denaturing buffer solutions with a ~30–40% organic solution content, the flow rate is typically higher because of lower solution viscosity. Thus, a sufficient amount of sample needs to be loaded into the emitter (>5 μl) to maintain the spray for a long acquisition of >1 h.

Fig. 2 |. Workflow overview for setting up the nESI source.

a, A picture of an nESI emitter tip under the microscope as received from the supplier (scale bar: 250 μm). b, Steps to load a nESI emitter into the fitting on the nanoFlex source head. c, Overview of the nanoFlex source with microscope camera and nESI emitter installed. d, Field of view under the microscope camera showing the mass spectrometer inlet capillary nozzle (left) and the nESI emitter tip (right). The dashed line indicates the alignment between the emitter tip and the center of the inlet capillary.

After installing the nESI emitter and applying voltage to initiate the electrospray process, the tip location should be optimized by using the XYZ manipulator (Step 10). The goal of this optimization is to achieve a stable spray and obtain desired protein signals⁹⁰. The two critical parameters are the nESI voltage and the distance between the emitter tip and the MS capillary inlet nozzle. A low nESI voltage (1.0–1.4 kV) is suitable for nESI tips prepared according to the above recommended protocol. High voltage beyond 2 kV may result in material buildup at the tip region and reduced spray stability because of corona discharge⁹¹. An optimal distance between the emitter tip and the inlet may be found for each nESI tip to generate properly desolvated protein ions in a continuous and stable manner. To examine and optimize ion source conditions, we recommend using ensemble MS mode at moderate MS resolution for a brief survey⁹². In Extended Data Fig. 2, we show a typical spectra obtained from averaging 200 MS spectra corresponding to one charge state of the standards at MS resolution of 12,500 (DMT mode off). We recommend comparing the peak m/z value and full width at half maximum of the charge state cluster signal with the optimized spectra shown in Extended Data Fig. 2. Moreover, for intact monoclonal antibody (SILu)⁹², baseline glycoform separation within the same charge state cluster in native mode is an indicator of good ion source quality (Extended Data Fig. 2). This is critical because individual ion collection under the same conditions in the next stage will be proportional to the quality of their ensemble counterparts in the survey. We note that the same nanoESI tips should not be used for the subsequent measurement, so the next sample may be subjected to signal variability because of potential variations in the new tips.

When the protein signals are heavily adducted, activation voltages in a few ion-transfer stages may be applied to improve the ion detection, as has been described in detail elsewhere⁹³. In particular, in-source collision-induced dissociation (CID) voltage generates a DC gradient between the S-lens and the injection flatapole to desolvate the ions (Tune → Instrument Control → Scan Parameters → Fragmentation → In-source CID); in-source trapping further desolvates the ions by lowering the DC bias of the injection flatapole (Tune → Instrument Status → Settings → Ion transfer → In-Source Trapping Mode); extended trapping expands the collisional cooling in C-trap to the higher-energy collisional dissociation (HCD) cell space with a DC activation gradient applied between the two chambers, which allows for additional desolvation in the instrument backend (Tune → Instrument Status → Control → Settings → Define Scan → Misc.). We note that in-source trapping is a pulsed event that is not synchronized with the ion injection into the C-trap⁹⁴. When in-source trapping is used, a continuous ion beam from the source will be collected and released every 4.5 ms^95,96. When injection times <4.5 ms are used to attenuate ion signals for relatively concentrated samples, in-source trapping may be mismatched with C-trap injection, leading to complete loss of ion signals in some of the acquisitions. The sample should be diluted to avoid the use of extremely low injection time (<0.1 ms) when in-source trapping is used in I²MS data collection.

Charge calibration

It is critical to obtain a calibration curve from samples with well-resolved charge states for the assignment of ion charges before mass-domain spectrum generation. Charge calibration in the software is typically conducted by using standard protein ions with resolved charge states¹⁵. The charge-calibration curve may be created by using different data collected from the same instrument and detector mode. This section of the procedure can be applied to any protein or complex that has visible or defined charge state peaks in the m/z domain. Alternatively, the software has an ‘autocalibration’ function that generates a self-consistent charge calibration directly from the individual ions in the sample by using a regression error minimization model. In this protocol, we use a combination of four native standards corresponding to a charge range of 9–52. We recommend using standards that have charges similar in m/z and charge of the samples of interest. A detailed description of the software functions mentioned in this section can be found in Supplementary Note 1.

Instrument parameter optimization

The next stage after ion source optimization is to adjust instrument parameters to obtain individual ions. First, a survey acquisition in the ensemble MS mode under optimized conditions should be performed to record the m/z and charge state distribution of the protein (Steps 14 and 15). Next, DMT mode should be switched on for the transition to I²MS data collection (Step 16). In this section, we explain how the parameters in the DMT mode window affect the I²MS data collection (Fig. 3a).

Fig. 3 |. Instrument parameter optimization for obtaining an individual ion signal.

a, The ‘Direct Mass’ window in ‘Tune’ showing the adjustable parameters. b–e, MS spectrum of tetrameric pyruvate kinase (100 nM) at MS injection time of 100 ms (b), 10 ms (c), 1 ms (d) and 0.2 ms (e). The red dashed lines in c and d indicate the single ‘ion ledge’ with descending peak intensities at higher m/z.

In Supplementary Table 1, we show a series of recommended MS instrument settings for different standards. We note that the recommended settings are instrument specific and may be subject to change. The following discussion provides guidance on how to rationally adjust a few critical parameters. In particular, we recommend selecting ‘High m/z’ for the ‘Detector m/z Optimization’ and the ‘Ion Transfer Target m/z’ for most of the native protein samples (m/z > 2,000). We note that ‘Low m/z’ detector mode is recommended when small proteins with m/z < 2,000 need to be considered in an experiment, while each detector mode needs its own charge-calibration curve. The ‘Trapping Gas Pressure Setting’ is a normalized manufacturer-specified parameter that controls the pressure in the HCD cell. In the QE Orbitrap UHMR system, the HCD gas pressure causes a change in the pressure of the C-trap and the Orbitrap chamber as gas leaks over. As a result, changing the ‘Trapping Gas Pressure Setting’ directly affects ion lifetimes within the Orbitrap. In the Orbitrap chamber, individual ions may undergo energetic collisions with gas molecules and fragment during the 2-s transient recording time, which leads to ion losses³⁷. To maximize the ion survival rate during the 2-s transients, the pressure in the Orbitrap should be minimized. However, a low-pressure setting may result in inefficient collisional cooling and ion trapping in the C-trap, leading to a lower total number of ions collected in the Orbitrap. We recommend using a ‘Trapping Gas Pressure Setting’ of 0.1–0.4 (arbitrary units) for native protein analysis, which applies well to the native protein standards that we used in this protocol. In specific cases in which the number of ions per spectrum is low, a setting between 1 and 2 can be used to increase the number of ions per spectrum and enhance the experimental throughput. However, for relatively large proteins in denatured mode, the pressure should be minimized if possible because of their larger collision cross sections and more severe ion decay⁹⁷. In Anticipated results, we further discuss the pressure effects on denatured SILu antibody.

We also provide here an explanation of the additional parameters mentioned in Supplementary Table 1. ‘Source DC offset’ is the DC bias applied to the S-lens, where ions from the transfer tube are focused and directed to the injection flatapole and downstream optics. It is kept at 0.1 V by default when ‘in-source CID’ voltage is not applied and should be adjusted to [in-source CID voltage + 0.1] V when it is applied. The effect of in-source CID has been described earlier; ‘eFT’ refers to enhanced Fourier Transform⁹⁸.

Ion flux control (Fig. 3a; Step 16) is a critical consideration in obtaining individual ion signals. A high number of ions per acquisition period will result in multiple ions at the same m/z value. Ion population as the time integral of ion flux is primarily controlled by adjusting the MS injection time^9,76. When conducted manually, the process starts from a survey spectrum in which ensembles of ions for all the charge states of a protein are collected in one acquisition event (Fig. 3b). When dropping the MS injection time to attenuate the number of ions in the Orbitrap, the unresolved charge state clusters transition to groups of discrete bars⁹ (Fig. 3c). For less-abundant charge states (e.g., 35+ and 31+), all the peaks in the corresponding m/z range for a given charge state are of similar intensities, which indicates single ions. Peaks of twice as high and multiple times higher intensities are surrounded by single ions at m/z of the more abundant charge states (e.g., 33+). These ‘multiple ion events’ are good reporters for practitioners to calibrate the level of the individual ions while adjusting the MS injection time manually. Importantly, a trend can be observed for the intensities of the individual ions of different charge states of a protein. Individual ions of higher charge state at lower m/z show higher intensities, whereas ions of lower charge state at higher m/z show lower intensities. This gives rise to a descending intensity level called the ‘ion ledge’ for the individual ions from low to high m/z (Fig. 3d, red dashed line)⁹. This phenomenon is a manifestation of the image charge detection in Orbitrap-based MS systems, in which non-decaying ions with more charges induce higher image current on the outer electrode corresponding to higher intensities. Observing this phenomenon acts as a simple way to validate the injection time being used, and denatured samples often exhibit a clearer ‘ion ledge’ because charge states are far closer together in m/z space than those in native mode.

Once the ‘ion ledge’ is confirmed for the protein, further optimization of MS injection time is focused on minimizing the contribution of double and multiple ion events while maintaining as many individual ion signals as possible. From Fig. 3c,d, multiples ion events are eliminated at lower injection time, and the spectrum in Fig. 3d is desirable for I²MS data collection. Figure 3e shows an example spectrum in which too few ions are collected.

Manual injection time control requires the practitioner to monitor the spectrum appearance and manually adjust the injection time accordingly. To address this issue, an automated ion flux control option is developed and made available in DMT mode⁷⁶. In particular, it uses an AIC algorithm that allows for real-time adjustment of the injection time according to the ion flux changes in the source. The ‘AIC Target Density (%)’ option (Fig. 3a) describes the occupancy of the m/z space when the individual ion signal is maximized with minimal interference from multiple ions. A value of 100% has been validated theoretically and experimentally and is thus recommended for optimal I²MS data acquisition. Increasing the AIC density value further increases the total ion number collected per acquisition event, whereas decreasing this value lowers the total ion number collected.

Data analysis

In this protocol, we introduce an I²MS data analysis platform named ‘STORIboard’. Charge calibration, mass spectrum generation and data visualization are all integrated in this software.

The first stage in I²MS processing is to analyze the STORI plot of each individual ion (Steps 22–25), the process of which has been extensively described elsewhere²⁸. Briefly, when the instrument performs data acquisition in DMT mode, the instrument records time-domain data for each MS spectrum within the .RAW file. The transient segments were recorded around frequency values that are identified as individual ions²⁸. For each individual ion, a STORI plot can be created by integrating the signal around the frequency over the detection time²⁸. This allows for the extraction of the slope of the STORI plot, which is used to obtain the charge of the individual ion by comparing it to the charge-calibration curve⁹. Moreover, in the analysis of STORI data, ions that are short lived and go through transitions such as fragmentation in the detection period are filtered out to minimize mass and charge misassignment. In particular, the software default processing template (Fig. 4a, top) sets a filtering threshold for the STORI plot at R² = 0.996 and an ion lifetime at ≥0.84 for a 2-s acquisition (‘DurationThreshold’ at 0.42 corresponding to 42% of the transient length). The default settings work for most applications (Fig. 4a). In specific cases when a more stringent ion lifetime filtering needs to be applied, increase ‘RSquareThreshold’, ‘DurationThreshold’, ‘MinimumTimeOfDeath’ and ‘MaximumTimeOfBirth’ (maximum = 1 for all above parameters, in which the range of 0–1 acts as a percentage of 0–100% of the transient time). All parameters can be lowered to allow more individual ions into the charge assignment and mass spectrum generation, but there is a risk of introducing charge misassignments that would lead to inaccurate mass spectra.

Fig. 4 |. Screenshots of STORIboard software.

a, Default parameters in ‘STORI Processor’ and ‘Charge Assignment’ (‘Voting v3’ and ‘Central Limit’). b, ‘Calibration File Manager’ window. A representative calibration curve generated from the four standards used in this protocol is shown in the bottom half of the panel.

The second stage in I²MS processing is to perform charge assignment (Steps 26 and 27). Two algorithms are available for charge assignment (Fig. 4a, bottom): ‘Voting v3’ (‘<500 kDa’ option) and ‘Central Limit’ (‘>500 kDa’ option)⁹. Briefly, ‘Voting v3’ is an iterative voting algorithm that identifies isotopic clusters of ions in the slope versus m/z space that correspond to different charge states of the same mass distribution⁹⁹. Alternatively, ‘Central Limit Theorem’ determines the charge of a number of ions by grouping ions with similar m/z values and excluding ions with >30% outlying slope values²⁴. A suitable algorithm should be applied to perform charge assignment of the I²MS data depending on the range and space of the m/z and charge distributions of the sample. In particular, voting heavily relies on resolved isotopes for its iterative charge-assignment algorithm. Although <500 kDa is listed, accurate mass distributions can be produced only if isotopes are resolvable. ‘Central Limit Theorem’ does not rely on high resolution of the input data and as a result is more suitable for processing data at higher molecular masses for which isotopic resolution is lost.

Materials

Biological materials

Pre-purified proteins or protein complexes (see Reagent setup).

In this protocol, for our standards, we use carbonic anhydrase isozyme II from bovine erythrocytes (Sigma-Aldrich, cat. no. C2522), SILuLite SigmaMAb universal antibody standard human (Sigma-Aldrich, cat. no. MSQC4), pyruvate kinase (PK) from rabbit muscle (Sigma-Aldrich, cat. no. 10109045001) and β-galactosidase from Escherichia coli (Sigma-Aldrich, cat. no. G3153).

Reagents

• Water, Optima LC/MS grade (Fisher Scientific, cat. no. W64)

• Acetonitrile, Optima LC/MS grade (Fisher Scientific, cat. no. A955–4)

  CAUTION Acetonitrile is a flammability hazard category 2, toxic hazard category 4 and eye irritant category 2. Wear proper personal protective equipment (PPE) when handling and avoid contact with your eyes. Keep away from heat, sparks and open flame. Use per safety data sheet (SDS) recommendations.

• Methanol, Optima LC/MS grade (Fisher Scientific, cat. no. A4564)

  CAUTION Methanol is a health hazard category 1, toxic hazard category 3 and flammability hazard category 2. Wear proper PPE when handling and avoid contact with your skin and eyes. Keep away from heat, sparks and open flame. Use per SDS recommendations.

• Ammonium acetate (AmAc), crystalline/certified American Chemical Society (ACS) reagent (Fisher Scientific, cat. no. A637–500) or AmAc solution (7.5 M) (Sigma-Aldrich, cat. no. A2706)

• Acetic acid, Optima LC/MS grade (Fisher Scientific, cat. no. A11310X1AMP)

  CAUTION Acetic acid is a flammability hazard category 3 and corrosive category 1. Wear proper PPE when handling. Keep away from heat, sparks, open flames and hot surfaces. Wash your skin thoroughly after handling. Use per SDS recommendations.

Equipment

• 1.5-ml protein LoBind microcentrifuge tubes (Fisher Scientific, cat. no. 13–698-794)

• 0.5-ml protein LoBind microcentrifuge tubes (Fisher Scientific, cat. no. 13–698-793)

• Amicon Ultra 0.5-ml centrifugal filter unit, 10-kDa MWCO (Millipore Sigma, cat. no. UFC5010)

• Amicon Ultra 0.5-ml centrifugal filter unit, 100-kDa MWCO (Millipore Sigma, cat. no. UFC5100)

• Centrifugation device. In this protocol, we use Thermo Scientific Sorvall Legend Micro 21 microcentrifuge and rotor packages (Fisher Scientific, cat. no. 75–772-436)

• Syringe (10 μl; Model 1701 RN SYR, small removable needle, 32 gauge, 2 inches, point style 3; Hamilton, cat. no. 80014)

  CRITICAL A small needle size (0.24 mm) is preferred. Large needles may not be suitable for the inner diameter of the borosilicate emitter for sample loading.

• Q Exactive UHMR hybrid quadrupole-Orbitrap mass spectrometer with DMT mode license

  CRITICAL The Q Exactive UHMR is the only instrument model that currently supports the DMT mode license.

• Nanospray flex ion source (Thermo Fisher Scientific, cat. no. ES071)

• Offline nanospray (Nano ES) source head kit (Thermo Fisher Scientific, cat. no. ES259), including borosilicate emitters (0.8-mm tip, 52-mm long, Au/Pd double coating, 100 pieces, cat. no. ES387), offline nanospray ion source head (cat. no. ES260), precision tweezers and ceramic glass cutter (cat. no. ES250) and static air pressure device (cat. no. ES242)

• Dino-Lite universal serial bus cameras and software disc (Thermo Fisher Scientific, cat. no. ES218) and universal serial bus extension cords (Thermo Fisher Scientific, cat. no. 00302–99-00095)

• Analytical balance

• LevGo smartSpatula disposable polypropylene spatula (Fisher Scientific, cat. no. 18–001-018)

• Low-nitrogen weighing paper (Fisher Scientific, cat. no. 09–898-12A)

• Pipette tips (ShopRAININ, Green-Pak, Universal, 1,000 μl 768A/8, cat. no. 30389279; GPS UNV 250 μl 960A/10, cat. no. 30389287)

• Nunc 15- and 50-ml conical sterile polypropylene centrifuge tubes (Fisher Scientific, cat. nos. 12–565-269 and 12–565-271, respectively)

Software

• Xcalibur version 3.0 or newer (Thermo Fisher Scientific): used to analyze data acquired on the Q Exactive UHMR instrument (2.13 or newer)

• DinoCapture 2.0: used to optimize the position of the nESI tip

• STORIboard (Proteinaceous, Inc.): used to perform charge calibration, I²MS data processing and visualization

Reagent setup

AmAc solution

To prepare 50 ml of a 100 mM AmAc solution for native protein analysis, dissolve 0.385 g of AmAc in 50 ml of water in a 50-ml centrifuge tube. Alternatively, if using a premade 7.5 M AmAc stock solution, add 0.67 ml of stock solution into 49.33 ml of water in a 50-ml centrifuge tube. Mix well and store in ice before further use. AmAc should be made fresh.

Denaturing buffer

For denatured protein analysis, make 1 ml of denaturing buffer composed of 60%:39%:1% (vol/vol/vol) water/acetonitrile/acetic acid. Denaturing buffer solutions should be made fresh.

Equipment setup

Instrument setup

1. Set up the nanoFlex source on the mass spectrometer. Connect the high voltage probe to the bottom socket of the source.

   CAUTION A high spray voltage of 1–2 kV is applied to the probe. Put the mass spectrometer into standby mode or change the voltage to 0 V, and wear proper PPE while handling the high-voltage cable.

2. Set up the two DinoLite cameras on the top and the side of the source. The cameras should be calibrated according to the instructions in the manual. Open the DinoCapture 2.0 software to activate the cameras. Adjust the position and focus of the cameras to bring the inlet into view. Leave the inlet off from the center of the view so that the nESI tip can be seen when it is brought in proximity to the inlet.

3. Open the ‘Tune’ window of the instrument control personal computer and make sure that no warnings are reported from the ‘Tune’ window. Switch the instrument to ‘Standby’ state and load a generic template .mstune file for direct infusion nESI or use the example .mstune files that we provided in Supplementary Code.

4. Modify the .mstune file in the ‘Tune’ window for I²MS acquisitions as needed by using the parameters in Supplementary Table 1 and refer to the recommendations in Experimental design, Instrument parameter optimization. These .mstune files can be used in many experiments.

   CRITICAL For best results, mass calibrations (spectral m/z calibrations) need to be completed in time intervals according to MS instrument guidelines. m/z drift changes mass peak placement as expected.

5. (Optional) Create a method file by using Thermo XCalibur for collecting streamlined survey and I²MS spectra. This method file can be used in many experiments. For survey scans, method files may not be necessary. Data acquisition can be conducted in the ‘continuous’ mode and manually stopped when the quality check of the spectra is finished.

Procedure

Buffer exchange

TIMING 1 h

CRITICAL We use SILu as an example for the buffer exchange section, but other protein standards should follow a similar procedure as described for SILu. All proteins used in this protocol need to be buffer-exchanged for optimal MS data quality by using Amicon centrifugal filters (https://www.emdmillipore.com/US/en/life-science-research/protein-sample-preparation/rotor-guides/Amicon-Ultra-0.5/5kub.qB.SLUAAAFE3XEMfo2_,nav)^30,31.

• 1Equilibrate the filter unit with 500 μl of water and spin for 3 min at 11,000g. Open the container of protein standards and weigh 1 mg of SILu powder.
• 2Dissolve 1 mg of SILu (stored at −80 °C for ≤12 months) in 100 mM AmAc, load the solution into the 100-kDa MWCO filter unit and spin at 4 °C for 3 min at 11,000g 7–10 times. Each time, spin the solution down to <100 μl and fill the filter unit up to ~500 μl with AmAc. In the final centrifugation step, replace the outer tube of the filter unit with a new one for the collection of the concentrated solution. Flip the filter and push it against the outer tube. Spin at 500g for 2 min to collect the solution in the outer tube.
• 3Estimate the volume of the final solution. We recommend using a 100-μl pipette loaded with a clean and empty pipette tip (assuming the total volume is <100 μl). Draw all the solution into the tip with a pipette set at 100 μl. An air column will form at the tip of the pipette tip after all the liquid is drawn (Extended Data Fig. 3, top). Gently adjust the level of the pipette to push the air column out from the tip by holding the lock button of the pipette (Extended Data Fig. 3, bottom). The final volume reading on the pipette corresponds to the estimated volume of the solution. The concentration of the solution can be estimated by the weight of the starting material divided by volume. Various online calculators for protein concentration are available (e.g., https://www.bioline.com/media/calculator/01_04.html).
• 4Distribute aliquots of the final solution into multiple tubes and keep them in a −80 °C freezer for storage before further analysis.

  CRITICAL STEP We strongly recommend dividing the stock solution into multiple vials for storage and thawing one aliquot for analysis each time. Multiple freeze-thaw cycles may result in protein deactivation and decomposition. Flash-freezing is not required.

Sample preparation

TIMING 10 min

CRITICAL Steps 5–18 apply to sample preparation both for standards and for unknown samples.

CRITICAL When performing calibration, Steps 5–18 should be repeated for each standard. Here, we have used carbonic anhydrase, SILu, tetrameric pyruvate kinase and tetrameric beta-galactosidase all analyzed in native condition.

• 5Prepare the nESI tip. Use a microscope to check the tip and make sure that the tip is intact and clean. Clip a tiny section of the sharp end of the tip to open the aperture.
• 6
  Take the buffer-exchanged protein standard stock solutions out from −80 °C and let them thaw in an ice bucket. Leave the 100 mM AmAc or denaturing buffer in ice until the temperature is equilibrated. Once the stock solutions and the dilution buffer are thawed and equilibrated to 0 °C (5–10 min), make the protein solutions for nESI. For calibration, we used the following concentrations for our standards:

  • carbonic anhydrase: 100 nM
  • SILu: 500 nM
  • tetrameric pyruvate kinase: 100 nM
  • tetrameric β-galactosidase: 300 nM

  Make a final volume of 20 μl for each standard solution in a 0.5-ml Eppendorf tube.

  CRITICAL STEP For multiple injections on the same day for the same sample, make a volume of ≥50 μl for the diluted solution, to avoid multiple rounds of freezing and thawing of the stock solution.

  CRITICAL STEP For complex unknown samples, we suggest starting from relatively high concentrations below the millimolar range and diluting the sample as needed in subsequent MS optimizations.

  

• 7Load the sample into the nESI tip. For each sample, draw ~5 μl of solution from the Eppendorf tube into a 10-μl syringe and remove any air columns or bubbles. Load the sample into the nESI tip by inserting the needle all the way into the tip. The syringe needle may be hard to observe inside the nESI tip with the coating. In this case, check the position of the syringe needle by shining a white light on the tip. Slowly dispense the solution into the nESI tip to avoid the formation of air columns or bubbles. Check the nESI tip through a white light to check for any air gaps within the solution column. A continuous liquid column is necessary for continuous spray for long experiments. When needed, rapidly flick the nESI tip back and forth by hand to force the liquid toward the spray side of the tip.

  CAUTION Perform this action carefully to avoid hurting your wrist.

• 8Use a ceramic cutter to trim the back end of the nESI tip by 1 cm. This is to reduce the air gap in the nESI tip and make sure that the tip fits the housing of the nanoFlex source.

Setting up the nESI tip on the source

TIMING 10 min

• 9Load the clipped tip onto the nanoFlex source (Fig. 2). First, pull the stage away from the inlet. Next, loosen the union and insert the tip by using precision tweezers to avoid breaking the tip region. Lastly, tighten the union so that the nESI tip is sealed by the rubber O-ring inside the union.
• 10Adjust the XYZ manipulator to position the stage properly. Set the Z direction (perpendicular to the instrument front end wall) all the way back before sliding the stage back to the position slowly. Adjust the XYZ manipulator if needed to avoid the tip crashing. Watch the cameras and slowly bring the tip into the field of view by adjusting the Z direction of the manipulator.
• 11Use an air syringe to fast forward the solution through the tip. When a drop is formed in front of the tip, it indicates that the tip is not clogged and ready for infusion.

  

• 12Use the finer Z axis manipulator to adjust the tip to 3–5 mm from the inlet as a starting position (Fig. 2d).
• 13Set the spray voltage to 1 kV before switching the instrument to the ‘On’ state. Observe the real-time MS spectra. If no signal is observed, slowly increase the spray voltage with 0.1-kV steps.

  

Obtaining spectra for protein

TIMING 30–60 min

CRITICAL Charge calibration using the protein standards is not required every time to measure unknown samples. A charge calibration curve is typically stable for 6–12 months for the instrument. Venting the instrument does not strongly affect the charge calibration. Accordingly, charge calibration is not required after regular front-end maintenance of the instrument. However, when the Orbitrap voltage recalibration/optimization procedure or deep cleaning of the instrument in the C-trap and HCD cell region is performed, a new charge-calibration curve should be obtained, because this process may cause perturbation of the ion energy and trajectory of individual ions entering the Orbitrap analyzer.

CRITICAL Each UHMR instrument has two detector modes (high and low m/z). Charge calibration for each detector mode needs to be performed separately.

• 14Load the I²MS ‘Tune’ file created in step 4 of Equipment setup. Record a survey ensemble MS spectrum at MS resolution 12,500 for the sample. Uncheck ‘Direct Mass’ mode, set ‘AGC target’ to ‘fixed’, ‘Enhanced Fourier Transform’ to ‘On’ and ‘MS resolution’ to 12,500. Check the ensemble spectra in Fig. 5 to set the proper ‘Scan range’ and adjust ‘Maximum inject time’ until all charge state clusters in Fig. 6 are clearly visible. When >100 MS spectra are acquired, check the averaged MS spectrum by using QualBrowser in the XCalibur software. For each protein standard, check the charge state distributions of the standards in averaged survey spectra and compare them with the spectra shown in Extended Data Fig. 2 and Fig. 5. Adjust the tip position while watching the real-time signals to obtain a stable temporal total ion signal profile from spectrum to spectrum. Check the most recent averaged spectrum (10–50 MS spectra) and the total ion chronogram to obtain this information. The goal is to make sure that the ion source generates stable, decent protein signals (check Extended Data Fig. 2).

  

• 15Optimize the instrumental conditions on the basis of the ‘Tune’ file (from step 4 of Equipment setup) and recommendations in Experimental design, Ion source optimization to obtain protein signals with a minimal level of solvent adduction. The tune file should in principle be optimized ahead of time. In specific instances when the sample is heavily adducted, consider optimizing the source settings (e.g., in-source CID voltage, in-source trapping and extended trapping voltage).

  

• 16Check the ‘Direct Mass’ mode under ‘Instrument Control’ and start the MS acquisition in the individual ion mode. Set ‘High m/z’ for the ‘Detector m/z Optimization’ and the ‘Ion Transfer Target m/z’ and minimize ‘Trapping Gas Pressure Setting’ when possible (Check Experimental design for further details). To obtain individual ions in each MS spectrum, set ‘AIC Target Density (%)’ at 100. From the ‘Scan Parameters’ tab, select ‘240000’ for ‘Resolution’ and ‘1’ for ‘Microscans’.

  

• 17Collect 1,000–2,000 MS spectra for the sample. The number of MS spectra collected for different protein standards may vary. Alternatively, use the ‘Real Time Processing’ option in STORIboard to track valid signals and stop the acquisition when the desired ion count is reached.
• 18Once the desired .RAW file is collected, stop data acquisition and put the instrument in the ‘Standby’ mode.

  

Fig. 5 |. Mass spectra of native protein standards.

a–d, m/z- and mass-domain spectra of native standards carbonic anhydrase (a), SILu monoclonal antibody (b), tetrameric pyruvate kinase (c) and tetrameric β-galactosidase (d). The left panels show ensemble-mode spectra (top, 12,500 resolution) and single spectra in the I²MS mode with individual ‘ion ledges’ (bottom, intensity trend highlighted by using red dashes). A zoomed-in region of the m/z spectrum of the 25+ charge state of the SILu antibody is shown in the bottom left of panel b. The panels on the right show the corresponding mass-domain spectra processed by using I²MS. Zoomed-in regions of isotopically resolved mass spectra of SILu and PK are shown in b and c. The right side of panel c shows the mass-domain spectra processed by using 3,000 (black), 1,000 (blue) and 500 (red) MS spectra. Theoretical mass distributions of canonical forms of the standards are overlayed with the experimental mass spectra.

Fig. 6 |. Mass spectra of intact antibody in native and denatured modes.

a,b, Ensemble MS measurement of denatured SILu antibody showing the full-scale charge-state distribution (a) and the major glycoforms (b). c, Denatured (left) and native (right) single spectrum of SILu in black traces. The combs in orange and green indicate the positions of the charge states in each spectrum. d, Mass-domain spectrum of denatured SILu antibody obtained by using the workflow presented in the protocol. Calculated mass distributions of the glycoforms (colored) are overlayed with the experimental data.

Obtaining charge calibration by using STORIboard

TIMING 15 min

CRITICAL As discussed above, charge calibration using the protein standards is required only when the back end of the instrument (i.e., C-trap, Orbitrap and HCD cell) is opened and maintained (see Overview of the protocol).

• 19Copy the .RAW files for charge calibration to a single directory accessible by the STORIboard software.

  CRITICAL STEP When using data from multiple files to generate one calibration file, place all the data files in the same directory.

• 20Open STORIboard. To obtain a charge-calibration curve for the instrument, open ‘Manage Calibration Files’ (Extended Data Fig. 4, stage 1), click ‘Create’ (Extended Data Fig. 4, stage 2) and select ‘From .raw (Auto Cal)’ (Extended Data Fig. 4, stage 3). The software will open a pop-up window named ‘Generate Calibration File’ (Extended Data Fig. 4, stage 4). Click ‘Browse’ and load the data files (Extended Data Fig. 4, stage 4). Available calibration file formats include .RAW, .csv and .dmt. .RAW files should be used if following the previous sections of the procedure. Multiple analyte input files can be incorporated into the same charge calibration under the same directory. .csv files can be obtained by data export from a database engine such as SQLite Studio and can be customized by the user and incorporated into charge calibration. Alternatively, under the ‘Create’ button, the user can choose to use the charge-assigned ions (‘From .dmt file (Calibrated Ions)’) or all the ions (‘From .dmt file (All Ions)’) from the .dmt file for calibration. When using data from multiple files to generate one calibration file, select all the data files. Click ‘Open’, and the software will start generating the calibration file with a progress bar shown (Extended Data Fig. 4, stage 5).
• 21When the process is finished, click ‘Save’ (Extended Data Fig. 4, stage 6). By default, the resulting calibration file is in .csv format and is saved under ‘C:\Users\Public\Documents\Proteinaceous\STORIboard\CalibrationFiles’. A new row will show up in ‘Calibration File Manager’ (Extended Data Fig. 4, stage 7, dark gray). The user can manage the calibration file by clicking ‘Open Folder’ (Extended Data Fig. 4, stage 7). If the new file does not show up in the ‘Calibration File Manager’ window, click ‘Refresh’ (Extended Data Fig. 4, stage 7).

  

Analysis of an unknown sample

TIMING 30–60 min

CRITICAL The ‘ion ledge’ may not be cleanly visible because of many signals from overlapping analyte charge states in close m/z proximity. In this case, especially for unknown samples, manually drop the injection time in the ensemble mode survey experiment or use the ‘Target Ion Density (%)’ option for automated control using AIC.

• 22Open STORIboard and create a processing template (Extended Data Fig. 5). Click ‘Manage Processing Templates’ (Extended Data Fig. 5, stage 1) to open a pop-up window called ‘Processing Template Manager’ (Extended Data Fig. 5, stage 2). To create a new template, click ‘Create’ (Extended Data Fig. 5, stage 2). Select ‘<500 kDa’ for the Voting v3 algorithm or ‘>500 kDa’ for the Central Limit algorithm (refer to Experimental design for the selection between the algorithms; Fig. 4 and Extended Data Fig. 5, stage 3). If the user wishes to apply the same parameters used in a previously processed file obtained elsewhere, select ‘From .dmt’.
• 23Define processing parameters (Extended Data Fig. 5, stages 4 and 5). For both algorithms, the software will open a pop-up window named ‘Processing Template Parameters’. Define the ‘File name’. Click the ‘STORI Processor’ tab to expand the options. The ‘STORI Processor’ tab is identical in both algorithms. The default parameters have been optimized for most of the samples. If the user wishes to adjust the parameters under ‘Charge Assignment’, refer to Experimental design for further instructions. Click ‘Save’ to exit the window.

  CRITICAL STEP Adjusting the processor parameters may allow decayed ions into the process, leading to more charge misassignments and negatively affecting the spectrum generation.

• 24When a new processing template is generated, a new row will appear in ‘Processing Template Manager’ (Extended Data Fig. 5, stage 6, dark gray). The template is saved as a .JSON file. The user can manage the template files by clicking on the ‘Open Folder’ button. If the new file does not show up in the ‘Processing Template Manager’ window, click ‘Refresh’. The user can edit the template by clicking ‘Edit’ (Extended Data Fig. 5, stage 6, dashed box).
• 25Load data for processing (Extended Data Fig. 6). Click ‘Processing Workflow’ (Extended Data Fig. 6, stage 1). Select ‘Add Input Files’ to import individual data files or ‘Add Input directories’ to import all files in the folder (or drag and drop; Extended Data Fig. 6, stage 2). The software will open a pop-up window where loaded files can be managed (Extended Data Fig. 6, stage 3). Additional files can be loaded by clicking ‘+ Add Input’ (Extended Data Fig. 6, stage 4). When multiple files are loaded, the ‘Merge Output’ option will show up in the window (Extended Data Fig. 6, stage 3, black bar). Checking ‘Merge Output’ will merge all the loaded files for charge assignment to generate a single mass spectrum; unchecking ‘Merge Output’ allows for batch processing of multiple data files by using the same parameters at improved throughput.

  CRITICAL STEP Multiple files may be processed together for charge assignment. Previously processed data in ‘.dmt’ or ‘.i2ms’ format obtained before applying this protocol can be loaded to save processing time.

• 26Load the processing template (Steps 22–24) and calibration file (Step 21) for processing (Extended Data Fig. 6, stage 3, arrows). Select the desired template and calibration file from the corresponding dropdown list. ‘Pre-Charge Assignment Filtering’ may be used for preselection of m/z or range of spectrum for processing (Extended Data Fig. 6, stage 5). Click ‘Run’ to start the data processing.
• 27Once data processing is completed, the software will automatically open the resulting .dmt file. The file can be loaded later from ‘Load a Direct Mass Technology file’ in the main window of STORIboard.

Data visualization using STORIboard

TIMING 20–120 min

CRITICAL In this section, we introduce the generic steps in I²MS data visualization using the STORIboard software. In Steps 28–32, we describe the generic steps to visualize mass spectra in a .dmt file generated in the analysis of a standard (Step 21) or unknown sample (Step 27). A detailed software walkthrough is presented in Supplemental Information and Extended Data Figs. 7–10.

• 28By default, the mass-domain spectrum is presented on the bottom section of the ‘Spectrum’ window when the .dmt file is opened (Extended Data Fig. 7). Switch the spectrum ‘Type’ to visualize the spectrum in either mass- or m/z-domain. Other custom options for spectrum visualization are described in detail in Supplemental Information.
• 29Chronological ion statistics in absolute ion count (labeled as ‘Frequency’) are presented on the top section in the same window (‘Chromatogram’ tab, Extended Data Fig. 7). The purple, blue and green scatter plots correspond to the raw ion count, the processed ion count (after STORI processing) and the charge assigned ion count, respectively. Switch the chromatogram ‘Type’ to ‘Stacked’ to visualize the processed/charge assigned ion statistics in fractional scale (Extended Data Fig. 7, 1–3).
• 30Switch to the ‘Heatmap’ tab to visualize the data in 2D heatmap view (Extended Data Fig. 8). By default, the ions are plotted by their charge (Y-axis) and mass (X-axis). Switch the axis definition of the heatmap to visualize the data in different formats. Available options for the axes are ‘slope’ (STORI slope), ‘Charge’ (charge state), ‘m/z (Th)’, ‘Mass (kDa)’, ‘Lifetime’ (the duration of ions that pass the STORI processing) and ‘Scan Number’.
• 31Switch to the ‘Calibration’ tab for visualization of the charge-calibration curve used for processing the data (Extended Data Fig. 9). The data used this curve for charge calibration that was created in Steps 19–21.
• 32Switch to the ‘Summary’ tab to display the ‘STORI Processor Parameters’ and ‘Charge Assignment Parameters’ (Extended Data Fig. 10). The right panel in the ‘Summary’ tab shows the number of remaining ions in sequential steps of processing labeled on the ‘Threshold Filter’ axis. Switch the tab (‘Ions Filtered Counts’ and ‘% Ions Filtered’) on the top of the graph to view the ion statistics in absolute count or relative percentage scale.

Troubleshooting

For Troubleshooting advice, see Table 1.

Table 1 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
6	Low individual ion count with maxed-out injection time	The sample is too heterogeneous	Make a 10× more concentrated sample solution
11	No large drop can be formed at the front of the tip	The tip is clogged	Slightly clip the tip of the emitter with the clamping tweezers and carefully flick the emitter capillary to push the sample forward
13	No signal observed even with elevated nESI voltage	An air bubble is formed, or the tip loses the metal coating	Apply a slight back gas pressure or replace the tip
14	The signal in the survey spectrum is too low	The pressure in the C-Trap is too low	Bring the value up in 'HCD Pressure Setting'
15	Fragmentation of the standards is observed	The source/ion transfer voltage is too high	Lower the in-source CID voltage/extended trapping voltage
	Significant fluctuations in the total ion chronogram are observed	Mismatch of in-source trapping and C-trap ion injection	Avoid using an injection time <10 ms. The sample may need to be diluted to acquire a single ion signal at higher injection time
16	The output spectrum does not contain meaningful protein peaks	Automated injection time control does not find the proper injection time	Use the 'Manual Inject Control' option. Look for the descending 'ion ledge' of individual ions and drop the injection time until very few of the ion signals go above the 'ion ledge' level (Fig. 5). Maximize the individual ion population in each acquisition period whenever possible while minimizing the multiple ion events
18	No signal is observed after a few samples have been analyzed	Ion optics charging	Initiate a charging test. If the diagnostic fails, clean necessary optics or contact Technical Support
21	Uncertainty about the validity of the output spectrum	Uncertainty about the accuracy of charge calibration	Run an experiment with a known standard (e.g., carbonic anhydrase), process the data by using the newly obtained charge-calibration curve and compare the output mass spectrum with the examples shown in Fig. 5

Timing

Steps 1–4, buffer exchange: 1 h (multiple protein standards can be processed at the same time)

Steps 5–8, sample preparation and loading into the nESI tip: 10 min

Steps 9–13, setting up the nESI tip on the source: 10 min

Steps 14–18, obtaining spectra for protein standards for charge calibration: 30–60 min per standard, up to 300 min total

Steps 19–21, obtaining charge calibration by using STORIboard: 15 min

Steps 22–27, analysis of an unknown sample: 30–60 min

Steps 28–32, analysis of an unknown sample and data visualization using STORIboard: 20–120 min, depending on the data-acquisition time for the unknown sample and the specifications of the computer used to process the data

Anticipated results

The present protocol enables the direct mass readout of complex protein samples. We first demonstrate that by applying the charge calibration obtained from the standards, we can generate mass-domain spectra of these standards containing direct mass information³⁰. Figure 5 shows the comparison of spectra of the native standards in low-resolution ensemble MS mode, with examples of a single spectrum containing individual ions and mass-domain spectra obtained by using the I²MS workflow. Native carbonic anhydrase is observed with a charge state distribution from 9+ to 11+. The I²MS spectrum of carbonic anhydrase (Fig. 5a, right) shows the canonical and a minor sodiated form both in the Zn-bound native state. It is not surprising that isotopic resolution can be reached for native carbonic anhydrase, which can be achieved by using ensemble Orbitrap-MS data collection at high resolution. However, isotopically resolved mass distributions at >150 kDa using a commercially available instrument are currently possible only by using the QE Orbitrap UHMR system with the I²MS workflow (DMT mode). Figure 5b (left) shows the ensemble-mode spectrum for native human monoclonal IgG antibody (SILu). With optimized source conditions, adjacent glycoforms can be well separated at baseline level. Increasing the MS resolution did not give rise to isotopically resolved spectra in the m/z domain²⁴. When using the I²MS workflow, we obtained an isotopically resolved mass spectrum of native SILu antibody shown in Fig. 5b (right). In this spectrum, the five major glycoforms can be clearly observed, and the mass distributions of the glycoforms match well with the calculated distributions (overlaid in color). The broader mass distributions obtained experimentally are due to salt and solvent adduction in the native mode.

In addition, we report that isotopic resolution of molecular mass of up to ~230 kDa can be routinely measured. An ensemble-mode spectrum of tetrameric PK with a charge state distribution from 32+ to 36+ with a minor distribution of its truncated form is shown in Fig. 5c (left). The right panel of Fig. 5c shows the mass-domain spectrum summed from ~52,000 individual ions in 3,000 MS spectra during an ~90-min data-acquisition period. In this spectrum, the intact PK homotetramer and its truncated form (three intact monomers + one truncated monomer) can be clearly observed. The observed deviation of the experimental spectrum (black trace) from the theoretical mass distribution is attributed to water losses induced by the in-source CID. We have previously demonstrated isotopic resolution of tetrameric PK (232 kDa) obtained by summing ~90,000 individual ions over the course of ~3 h of data acquisition³². In this experiment, we obtained a spectrum of comparable quality by using half of the data-acquisition time. Furthermore, we examined a smaller data size that gives isotopically resolved mass distribution at ~230 kDa. In the right panel of Fig. 5c, we show the mass spectra obtained from 1,000 MS spectra (~8,200 ions, ~30-min data acquisition) and 500 MS spectra (~2,900 ions, ~15-min data acquisition). Although the truncated form is not visible, isotopic resolution is still maintained at the major distribution corresponding to the intact tetrameric form.

We have previously reported the record of isotopically resolved mass measurement of an ~466-kDa β-galactosidase homotetramer with specially tuned instrument conditions (6-s Orbitrap transient) that are not currently available commercially³². However, in Fig. 5d, we demonstrate that mass distribution of a protein complex of ~0.5-MDa molecular weight can still be measured and approximated to theoretical calculations (red trace). The observed positive mass shift of the experimental spectrum from the calculation may be attributed to solvent adduction and metal binding³².

The present protocol also enables the characterization of more complex samples. In this example, we apply the protocol to unravel the glycoform landscape of denatured SILu antibody¹⁰⁰. In contrast with the native SILu analysis discussed above, the denatured mode significantly reduces salt and solvent adduction to the antibody, which provides higher sensitivity and more spectral clarity than the native condition²⁴. However, intact SILu antibodies in denatured buffer generate higher charge states and thus more closely spaced isotopic peaks in the m/z domain, which is challenging to resolve. In addition, denatured antibodies adopt an unfolded structure in the gas phase, which gives rise to large collision cross sections. As a result, we anticipate more ion losses because of collision-induced ion decay during the 2-s transients.

Figure 6a shows an ensemble MS spectrum of denatured SILu antibody that has a charge state distribution spanning from 40+ to 70+ centered at 54+. At a mass resolution of 12,500, the four most-abundant glycoforms are well resolved, with the fifth one (labeled with a purple dot) visible above the noise level (Fig. 6b). Peaks at higher m/z in each of the charge state clusters may correspond to additional low-abundance glycoforms. I²MS data acquisition of the same sample was performed after the low-resolution survey spectra. Figure 6c shows a comparison of an example single spectrum of denatured and native SILu containing individual ions. In contrast to the well-defined ‘ion ledge’ in the individual ion native mode (Fig. 6c, right), the denaturing-mode single spectrum shows a complex ion intensity distribution. The peaks with intensities below the ‘ion ledge’ are produced from ions that were lost during the detection period. This may be attributed to the signal losses of ions of higher charge states with larger cross sections, which suffer more severe collisional ion decay during the measurement⁹⁷.

Figure 6d shows the mass-domain spectrum obtained from ~3,600 spectra containing ~58,000 charge-assigned ions. The inversed spectrum in Fig. 6d is the native SILu spectrum shown for comparison. The five major SILu glycoforms are all detected in the denatured mode. The experimental mass distribution matches well with the calculation, with nearly isotopic resolution (zoomed-in spectrum for G0F+G1F). We note that the width of the isotopic envelopes in the denatured mode are comparable to the calculated distributions and are much narrower than the native mode. This is attributed to less adduction and better desolvation of the ions in the denatured mode. We also note that the total number of ions collected for denatured SILu is ~4.4 million, out of which only 1.3% (~58,000 ions) survived the I²MS processing and are charge assigned. Most of the ions have short lifetimes below the STORI processor threshold and are filtered out in the I²MS processing step. This is attributed to large cross sections of denatured SILu ions, leading to ion decay upon collisions with the gas molecules in the Orbitrap chamber. Shorter ion lifetimes also negatively affect mass resolution. This can be observed from the comparison with native SILu (Figs. 5b and 6d) in which adjacent isotopes can be resolved to baseline level.

Extended Data

Extended Data Fig. 1 |. MS spectra of carbonic anhydrase.

m/z spectrum of native carbonic anhydrase in the ensemble acquisition mode with the zoomed-in regions of 10+ and 9+ charge states showing isotopic resolution.

Extended Data Fig. 2 |. Low-resolution MS spectra of standards.

m/z spectrum of native carbonic anhydrase, 10⁺ (a); SILu antibody, 25⁺ (b); pyruvate kinase, 34⁺ (c); and β-galactosidase, 50⁺ (d) at MS resolution of 12,500. b, The 25⁺ charge state of native SILu antibody in an optimized electrospray condition with baseline-resolved glycoform separation (dark trace) and in a poor electrospray condition (red trace).

Extended Data Fig. 3 |. Volume measurement using the pipette.

An ~50-μl solution drawn by the pipette tip when the pipette is set at 100 μl (top) and the same solution after pushing the air column out from the tip by holding the lock button of the pipette (bottom).

Extended Data Fig. 4 |. Charge calibration using STORIboard.

Stepwise instructions for charge calibration using STORIboard.

Extended Data Fig. 5 |. Generating a processing template using STORIboard.

Stepwise instructions for generating a processing template using STORIboard.

Extended Data Fig. 6 |. Data processing using STORIboard.

Stepwise instructions for data processing using STORIboard.

Extended Data Fig. 7 |. Spectrum’ tab and ‘Compare-Mirror Plot’ window under the ‘Spectrum’ tab in STORIboard.

‘ Example screenshot of the ‘Spectrum’ tab in STORIboard (a) and the ‘Compare Spectra’ window (b).

Extended Data Fig. 8 |. ‘Heatmap’ tab in STORIboard.

Example screenshot of the ‘Heatmap’ tab in STORIboard.

Extended Data Fig. 9 |. Calibration’ tab in STORIboard.

‘ Example screenshot of the ‘Calibration’ tab in STORIboard.

Extended Data Fig. 10 |. Summary’ tab in STORIboard.

‘ Example screenshot of the ‘Summary’ tab in STORIboard.

Key points.

• This protocol describes a standardized workflow for multiplexed charge detection mass spectrometry (CDMS) on an Orbitrap analyzer. This allows individual ion mass spectrometry to be performed on a broadly accessible MS platform.

• Compared to custom-built CDMS instrument platforms, this approach does not require sophisticated instrumentation or specialized components, enabling broad and large-scale applications.

Key references.

Kafader, J. O. et al. Nat. Methods 17, 391–394 (2020): https://doi.org/10.1038/s41592-020-0764-5

Su, P. et al. Sci. Adv. 8, eabp9929 (2022): https://doi.org/10.1126/sciadv.abp9929

Su, P. et al. J. Proteome Res. 23, 1883–1893 (2024): https://doi.org/10.1021/acs.jproteome.4c00075

Melani, R. D. et al. J. Proteome Res. 21, 274–288 (2022): https://doi.org/10.1021/acs.jproteome.1c00882

Drown, B. S. et al. Anal. Chem. 96, 4455–4462 (2024): https://doi.org/10.1021/acs.analchem.3c04

Data availability

.RAW data and the processed .dmt data files used to demonstrate I²MS can be found in the MassiVE repository under the identifier MSV000094321. Additional data in the Supplemental Information are available from the corresponding authors upon request.