High-throughput proteomics of nanogram-scale samples with Zeno SWATH MS
Abstract
The possibility to record proteomes in high throughput and at high quality has opened new avenues for biomedical research, drug discovery, systems biology, and clinical translation. However, high-throughput proteomic experiments often require high sample amounts and can be less sensitive compared to conventional proteomic experiments. Here, we introduce and benchmark Zeno SWATH MS, a data-independent acquisition technique that employs a linear ion trap pulsing (Zeno trap pulsing) to increase the sensitivity in high-throughput proteomic experiments. We demonstrate that when combined with fast micro- or analytical flow-rate chromatography, Zeno SWATH MS increases protein identification with low sample amounts. For instance, using 20 min micro-flow-rate chromatography, Zeno SWATH MS identified more than 5000 proteins consistently, and with a coefficient of variation of 6%, from a 62.5 ng load of human cell line tryptic digest. Using 5 min analytical flow-rate chromatography (800 µl/min), Zeno SWATH MS identified 4907 proteins from a triplicate injection of 2 µg of a human cell lysate, or more than 3000 proteins from a 250 ng tryptic digest. Zeno SWATH MS hence facilitates sensitive high-throughput proteomic experiments with low sample amounts, mitigating the current bottlenecks of high-throughput proteomics.
Editor's evaluation
This paper presents a novel method for high-throughput proteomic data. This is an important improvement allowing a 5-8 fold increase in detection sensitivity over methods of similar throughput. A compelling set of data acquired on human and yeast whole-cell lysate illustrate this gain in performance.
https://doi.org/10.7554/eLife.83947.sa0Introduction
The proteome consists of large numbers of proteins that differ in concentration, distribution, and temporal dynamics (Eriksson and Fenyö, 2007; Hixson et al., 2017). The study of the proteome hence requires the ability to address both its large dynamic range and complexity, as well as the dynamic nature of biological systems. While proteomics typically aims to quantify as many proteins as possible, many of the biological questions that are to be addressed also necessitate the processing of large sample series. For example, a minimal drug treatment experiment of a wild-type and a single mutant cell line, with five drug concentrations, five time points, and triplicate measurements, already produces 150 samples. Epidemiological studies, exemplified by the Fenland study (O’Connor et al., 2015), can reach tens of thousands of samples. Processing such high sample amounts requires fast and robust measurement technologies that are quantification precise and cost effective.
In bottom-up proteomics, data-dependent acquisition (DDA) selects precursors from full-scan MS1 mass spectrum for fragmentation, generating tandem (MS/MS or MS2) mass spectra that can be matched to spectra in a database. While DDA ensures selectivity, it can result in high numbers of missing values that can originate from, for example, the stochastic nature of precursors selected for fragmentation, which limits the application in high-throughput studies. Data-independent acquisition (DIA) techniques such as SWATH acquisition (Gillet et al., 2012) address this problem by generating fragment ion spectra from all precursor ions that fall within a predefined precursor m/z range window. DIA shifted the data-processing strategies from an established spectrum-centric processing model to a peptide-centric model, requiring a shift in algorithm design and assumptions. DIA methods have facilitated applications in large-scale proteomics, including systems biology studies in various model organisms, disease states, and species (Hou et al., 2015; Gao et al., 2017; Basak et al., 2015; Zelezniak et al., 2018; Grobbler et al., 2015; Gillet et al., 2012), with one of the most common clinical applications being the discovery of (compound) biomarkers for disease in body fluids, particularly blood serum or plasma (Anjo et al., 2017; Gajbhiye et al., 2016; Russell et al., 2016; Hou et al., 2015; Pires et al., 2016).
Throughput, the required measurement precision, and sample amount are major factors that need to be balanced when conducting large-scale proteomics measurements. Moreover, particular attention has to be given to batch effects that exist within large proteomic experiments. Proteomics has historically relied on nano-flow chromatography, which is highly sensitive as sample dilution is minimal (Muntel et al., 2019), but the low flow rate requires a longer gradient time, which is more susceptible to technical distortions. Moreover, columns compatible with nano-LC systems typically require exchange after a maximum of a few hundred injections, which together prolongs the acquisition time and leads to complex batch effects in larger studies. Recent advances in instrumentation and software have facilitated the use of faster gradients and higher chromatographic flow rates (Messner et al., 2020; Bian et al., 2020; Vowinckel et al., 2018). These accelerate proteomic experiments and allow highly precise measurements. These come, however, at the cost of a higher sample amount required and can yield lower proteomic depth.
Here, we present a new acquisition technique, Zeno SWATH MS, that exploits a linear ion trap, the Zeno trap (Loboda and Chernushevich, 2009), in combination with SWATH MS acquisition, in order to increase the sensitivity of DIA-MS acquisition, specifically in combination with micro-flow-rate and analytical flow-rate chromatography. We demonstrate that Zeno SWATH MS increases sensitivity and expands the proteomic dynamic range coverage, generating new possibilities in the design of high-throughput proteomic experiments.
Results
Zeno SWATH MS increases proteomic depth and quantification precision
The sensitivity of SWATH MS in time-of-flight (TOF) mass spectrometers is largely determined by the MS/MS duty cycle. The pulse in the TOF accelerator, which results in a small proportion of ions situated at the right position of entering TOF, has a disproportionate impact on lower-m/z ions, resulting in a duty cycle of TOFs of only 5–25%. Here, we describe a setup of the Zeno trap (Loboda and Chernushevich, 2009; Figure 1—figure supplement 1), where all fragment ions are trapped in an axial pseudopotential well and then released by potential energy with timing aligned to the next pulse at the accelerator, allowing a duty cycle increase, thus increasing intensity 4–20 times (Baba et al., 2021). We speculated that the Zeno trap could be employed in combination with SWATH MS to boost the MS/MS sensitivity, facilitating higher identification and quantification performance while requiring lower sample amounts.
To generate platforms for medium- and high-throughput proteomic experiments, we coupled a ZenoTOF 7600 system (SCIEX) equipped with a Zeno trap (Figure 1—figure supplement 1) to both a micro-flow LC system (ACQUITY M-Class, Waters) and an analytical flow-rate LC (1290 Infinity II, Agilent), respectively. For data processing, we applied our deep-neural-network-based software suite, DIA-NN, which has been specifically optimised for handling complex data arising from fast chromatographic methods (Demichev et al., 2020).
First, we examined how consistently precursor (i.e. peptide at a specific charge state) and protein intensities were measured between triplicate injections of 62.5 ng K562 human cell line standard, comparing the performance of Zeno SWATH MS with SWATH MS on the same instrument setup. Peptides were separated using the micro-flow LC running at a flow rate of 5 µl/min, 20 min micro-flow gradient chromatography (Materials and methods). For analysis, we used an experimental spectral library generated using the ZenoTOF 7600 system (Materials and methods). Zeno SWATH MS yielded 5179 consistently quantified proteins among an average of 6226 proteins identified in triplicates. These were quantified with a median coefficient of variance (CV) of 6%, and 89% of identifications or 4602 shared proteins were quantified with a CV better than 20% (Figure 1a). At the precursor level, on average 49,172 precursors were identified, of which 40,209 were repeatedly quantified. Of these, 77% were quantified with a CV of less than 20% (Figure 1b). Zeno SWATH MS showed a considerable improvement in quantification consistency compared to conventional SWATH MS using the same chromatography, mass spectrometry, and sample. SWATH MS consistently quantified 2743 proteins, of which 76% or 2083 proteins were quantified with a CV <20% (Figure 1a). At the precursor level, SWATH identified an average of 21,771 precursors in triplicate injections, of which 16,022 were consistently quantified, with a median CV of 11%.

Comparison between SWATH and Zeno SWATH MS in proteome coverage and quantification precision and using 5 µl/min, 20 min micro-flow-rate chromatography.
(a,b) Reproducibility of protein identification using SWATH MS and Zeno SWATH MS on human cell line standard (K562) separated by micro-flow chromatography with a 62.5 ng load. Average identification numbers of proteins (a) and precursors (b) across three technical replicates (grey background bar) from a human cell line (K562) proteome tryptic digest are given; numbers of consistent identifications in technical replicates are given in dark grey; proteins or precursors quantified with coefficient of variation (CV) better than 20% in grey, and those quantified with a CV better than 10% in light grey. (c) Protein-level LFQbench results for Zeno SWATH MS. Quantification precision was benchmarked using yeast proteome tryptic digests that were spiked in two different proportions (A and B, three repeat injections each) into human cell line standard (K562) (A: 30 ng K562 + 35 ng yeast; B: 30 ng K562 + 17.5 ng yeast). Raw data were processed by library-free mode DIA-NN analysis. Protein ratios between the mixtures were visualised using the LFQbench R package (Navarro et al., 2016). Black dashed lines represent the expected log2(A/B) values for human (=0) and yeast (=1). Left pane, log-transformed ratios (log2(A/B)) of proteins plotted for DIA-NN over the log-transformed intensity of sample B. Right pane, protein quantification performance shown as box plots (boxes, interquartile range; whiskers, 1–99 percentile; n=3414 [human] and n=1816 [yeast]). (d) Average protein quantity differences between mixture A and B. Compared to SWATH MS, Zeno SWATH MS shows an improved separation of the two-species proteomes and more accurately represents the real ratio of the precursors (log2 of human quantity ratio close to 0, while log2 of yeast quantity ratio close to 1). (e) Correlation of protein quantities between both MS schemes in different mixtures. Human and yeast protein quantities in both mixtures show a high correlation between methods. (f) Distribution of protein quantities of each species in each mixture (A and B). In both mixtures and both species, Zeno SWATH MS shows deeper proteomic coverage by quantifying more low-abundant proteins.
The analysis of proteomes of different species mixed in different ratios, as introduced with the LFQbench test (Navarro et al., 2016; Kuharev et al., 2015), has proven an efficient way for evaluating the capabilities of DIA workflows and data-processing software (Demichev et al., 2020; Bruderer et al., 2015). The multiple-species benchmarks evaluate both the identification and quantification precision, which is determined by the ability of the proteomic method to deconvolute the multiplexed proteomes. Herein, we assessed how well the known ratios between two-species lysates – human and yeast lysates mixed in the defined proportions: 1:1 (A/B) for human, 2:1 for yeast (Materials and methods) – were recovered in triplicates using SWATH MS and Zeno SWATH MS, followed by a library-free analysis in DIA-NN. We injected triplicates of 1 µl of both mixtures (65 ng [mixture A] and 47.5 ng [mixture B]) with micro-flow-rate chromatography. Compared with SWATH MS, Zeno SWATH MS revealed more precursors and proteins quantified at better quantification precision. At the precursor level, Zeno SWATH MS quantified 34,872 human and 15,610 yeast precursors, while SWATH quantified 16,167 human and 6847 yeast precursors (SWATH MS, Figure 1—figure supplement 2b; Zeno SWATH MS, Figure 1—figure supplement 2c). At the protein level, 3414 (human) and 1816 (yeast) proteins were quantified with Zeno SWATH MS, while 2064 human and 973 yeast proteins were quantified using SWATH MS. Further, 3282 human and 1692 yeast proteins were quantified with Zeno SWATH MS, showing a ratio close to the known ratio (1:1 (A/B) for human, 2:1 for yeast) in at least two replicates in each of mixture A and B, with a median CV value for human proteins measured in triplicates of mixture A and B of 6.9% and 7%, respectively. The human proteins measured with SWATH MS show a median CV value of 12.7% and 12.6% for triplicate mixture A and B, with 1741 human and 776 yeast proteins quantification close to the known ratio (Zeno SWATH MS, Figure 1c; SWATH, Figure 1—figure supplement 2a). To exclude the bias introduced by data normalisation, we also compared the non-normalised data in both acquisition methods (SWATH MS, Figure 1—figure supplement 2d; Zeno SWATH MS, Figure 1—figure supplement 2e).
In order to account for the protein amount’s effect on performance of the two-species benchmarks, we conducted a similar analysis with a 50% lower injection amount. Compared with SWATH MS, 118% more human and 133% more yeast proteins were quantified in Zeno SWATH MS with 0.5 µl injection; while 65% more human and 86% more yeast proteins quantified in Zeno SWATH MS were quantified with 1 µl injection (SWATH MS, Figure 1—figure supplement 2f; Zeno SWATH MS, Figure 1—figure supplement 2g). Moreover, when comparing the ratio of human and yeast proteome quantified in both mixtures (log2(A/B)), Zeno SWATH MS shows less inter-species interference and is able to represent the real quantification ratio in both species more accurately (log2(A/B)=0 (human), = 1 (yeast)) (Figure 1d).
We speculated that the more accurate quantification ratio provided by Zeno SWATH MS comes from the better identification and more accurate quantification of the lower-abundant proteins. We therefore correlated the human and yeast proteomes shared in both acquisition methods (scatter plot, Figure 1e). For both species, the quantification results from SWATH MS and Zeno SWATH MS are in very good agreement for high-abundant proteins, but diverge for low-abundant ones. Then, we compared the protein quantities of SWATH MS and Zeno SWATH MS in both species (density plot, Figure 1f). In both species, Zeno SWATH MS performs better in quantifying the lower-abundant proteins.
Zeno SWATH MS increases protein identification rate with low sample amounts
Next, we tested the sensitivity of Zeno SWATH MS by analysing a concentration series of the K562 human cell line standardusing both SWATH MS and Zeno SWATH MS. The concentration series was acquired starting with the lowest concentrations injected after several blank measurements with a 20 min, 5 µl/min chromatographic gradient (Materials and methods), covering sample amounts ranging from near 1 to 500 ng (Materials and methods). For a comparison of MS systems, a similar concentration series was acquired using SWATH MS on a TripleTOF 6600 system (SCIEX) with the same 5 µl/min, 20 min chromatography setup (Materials and methods). To avoid the potential bias that system-specific experimental spectral libraries have on data from different MS systems, the benchmark was carried out using library-free data analysis.
Zeno SWATH MS consistently yielded better quantification performance. For instance, from the 62.5 ng K562 human cell line standard, 16,915 precursors or 2886 proteins were identified using the TripleTOF 6600 system at a 1% false discovery rate (Figure 2a). Using SWATH MS on the 7600 instrument, these numbers increased to 29,697 precursors identified from three technical replicates and an average of 3724 proteins (Figure 2a; protein identification in Figure 2—figure supplement 1a, first-pass precursor identification in Figure 2—figure supplement 1b). With Zeno SWATH MS on the same 7600 setup, an average of 62,769 precursors, or 5206 proteins, were identified from triplicate injections of 62.5 ng of the K562 human cell line standard. Interpreting the data in a different way, Zeno SWATH MS identified a similar number of precursors and proteins from 62.5 ng of K562 human cell line standard as SWATH MS did from 500 ng of the same sample on the same instrument (SWATH identified 64,316 precursors or 5242 proteins from 500 ng) (Figure 2a). Normalised per gradient time, this means that up to 4636 precursors were detected per minute of active runtime in Zeno SWATH MS, whereas SWATH from the ZenoTOF 7600 system quantified 2874 precursors/min and SWATH from the TripleTOF 6600 system reached 1153 precursors/min (Figure 2b).

Zeno SWATH MS and its performance on K562 human cell line standard using 5 µl/min, 20 min micro-flow chromatography (SWATH MS on the TripleTOF 6600 system [blue], SWATH MS on the ZenoTOF 7600 system [yellow], and Zeno SWATH MS on the ZenoTOF 7600 system [red]).
(a) Precursor identification performance using SWATH and Zeno SWATH. Illustrated is the average number of precursors identified (n=3, error bars [±1 SD] shown in plot) for a K562 standard dilution series under three acquisition methods with library-free DIA-NN analysis (second-pass data). (b) Visualisation of precursor identification across gradient time with SWATH and Zeno SWATH MS of 62.5 ng K562 standard injection. (c) Histogram (bins = 50) of the protein abundance distribution represented as copy number in log10 scale of a human proteomic reference dataset (Bekker-Jensen et al., 2017). Zeno SWATH MS increases protein identification numbers by quantifying more low-abundant proteins. (d) Correlation of protein quantity between SWATH and Zeno SWATH MS. Illustrated is the correlation of protein quantity between two MS acquisition schemes; 3671 proteins. proteins were identified and quantified in both acquisition schemes, 19 proteins only quantified with SWATH MS, 1535 proteins quantified only by Zeno SWATH MS. Among the 3671 proteins quantified by both acquisition schemes, there is a high correlation of protein quantities between the proteins identified in both acquisition schemes (r2=0.93.).
We speculated that the identification gains could be contributed by the better identification of low-abundant proteins by Zeno SWATH MS. We therefore assessed the proteome concentration dynamic range of the proteins covered by comparing our K562 human cell line standard results to a proteomic reference dataset with copy-number estimates for 14,178 human protein groups (Bekker-Jensen et al., 2017). SWATH MS on the 7600 system (3690 protein groups quantified) showed a larger protein concentration dynamic range coverage compared to the TripleTOF 6600 system (2881 protein groups quantified). Zeno SWATH MS increased the number of proteins identified (5206 protein groups quantified). Comparison with the standard dataset revealed that the gains are due to more low-abundant proteins quantified (Figure 2c). Also here, we report a high agreement of protein quantities between methods overall (r2=0.93), and in particular among high-abundant proteins (Figure 2d).
Zeno SWATH MS facilitates high-throughput analytical flow-rate proteomics with low sample amounts
Because of its high analytical performance and stability in combination with short chromatographic gradients, analytical flow-rate chromatography is a method of choice for large-scale projects and in diagnostic laboratories. Analytical flow-rate chromatography is also attractive for practical reasons, because columns equilibrate quickly, and the problem of dead volumes and sample carryover is substantially reduced. In our recent studies, we reduced the time between injections to 3 min or less (Messner et al., 2021). Specifically, for applications such as neat-plasma proteomics or drug screens, we have previously demonstrated proteomic experiments using flow rates of 800 µl/min, and obtained quantification of precise proteomes with gradient times that ranged from 30 s to 5 min (Messner et al., 2020; Messner et al., 2021). Consequently, even with the slowest of these gradients – 5 min of active separation – the throughput of a proteomics study is increased to 180 samples per day on a single instrument. The downside so far is a reliance on relatively large sample amounts, which restricts the application space and shortens the instrument cleaning cycles. We therefore examined whether Zeno SWATH MS’s performance would allow us to conduct proteomic experiments with analytical flow-rate chromatography and lower sample amounts.
We used the ZenoTOF system coupled to the 1290 Infinity II HPLC system (Agilent), running at 800 µl/min, with a 5 min gradient chromatography. We analysed the K562 human cell line standard in a concentration series, injecting increasing amounts from 4 ng to 2 µg. The obtained raw data were analysed using a spectral library. On average, Zeno SWATH MS identified 29,681 precursors and 4907 proteins with a 2 µg load of sample on the column, whereas SWATH MS identified 17,212 precursors or 3250 proteins with the same amount of sample (protein data shown in Figure 3a, respective precursor data in Figure 3—figure supplement 1a). Importantly, with Zeno SWATH MS, proteomic depth was reached with analytical flow-rate chromatography even with low sample amounts. For instance, with 250 ng of K562 human cell line standard injected, around 3000 proteins were quantified with Zeno SWATH MS while a similar protein quantification number can only be reached by SWATH MS with 2 µg injection on the same instrument and chromatography (Figure 3a). Further, the stability of the analytical flow-rate LC system was reflected in the consistency of the quantification and yielded a median protein quantification CV of 6%. Expressed differently, 92% of the proteins were consistently identified with a CV <20% (Figure 3b, respective precursor data in Figure 3—figure supplement 1b).

Protein identification performance on K562 human cell line standard using 800 µl/min, 5 min gradient chromatography in SWATH and Zeno SWATH MS.
(a) Comparison of sample injection amount and identification performance with SWATH and Zeno SWATH MS. Illustrated is the average number of proteins identified (n=3, error bars [±1 SD] shown in plot) for a K562 human cell line standard dilution series under two acquisition methods with spectral library DIA-NN analysis. (b) Reproducibility of proteins quantified in SWATH and Zeno SWATH MS on 2 µg of K562 human cell line standard separated by analytical flow-rate chromatography. Average protein identification number in three replicates (background), number of proteins identified in all replicates (dark grey), proteins with coefficient of variation below 20% (grey) and below 10% (light grey) of SWATH and Zeno SWATH MS. Raw data were analysed by spectral library-based DIA-NN analysis.
Evaluation of Zeno SWATH MS on plasma, plant, fungal, and bacterial samples
A key application space for high-throughput proteomics is human plasma and serum proteomics, as well as systems and synthetic biology. Plasma, bacterial, plant, and fungal sample matrices differ, however, in dynamic range and proteomic complexity from a human cell line extract such as the K562 human cell line standard. For instance, in human plasma, 99.9% of the proteomic mass is attributable to less than 200 proteins (Anderson and Anderson, 2002), while a yeast proteome coverage saturates at around 4000 proteins even with prefractionation (Thakur et al., 2011). To assess the performance of Zeno SWATH MS on various matrices, we hence assayed a concentration series (1–500 ng) of human plasma as well as cell extracts from the yeast Saccharomyces cerevisiae, the bacterium Escherichia coli, and seedlings of chickpea (Cicer arietinum). The standards and samples generated were separated using 20 min micro-flow-rate chromatography (Materials and methods) and acquired with both Zeno SWATH MS and SWATH MS.
In non-depleted human plasma, Zeno SWATH MS identified 3957 precursors derived from 204 proteins with a 62.5 ng injection. In human neat plasma, Zeno SWATH MS identified 3947 precursors from 245 proteins, while with SWATH MS, 2516 precursors from 156 proteins were identified from the same amount of sample. Notably, 15 ng of sample was sufficient to quantify the dominating fraction of the plasma proteome (160 proteins) (Figure 4). Zeno SWATH MS could hence be applied to high-throughput plasma samples where sample amounts are limited, for instance to make proteomics applicable for analysing finger-prick samples or in combination with nano-sampling devices.

Protein identification number with SWATH MS and Zeno SWATH MS in different sample types with 5 µl/min, 20 min micro-flow-rate chromatography.
We generated tryptic digests from human plasma, and protein extracts from the yeast Saccharomyces cerevisiae S228c, the Escherichia coli lab strain K-12, and chickpea (Cicer arietinum) seedlings germinated in the lab. Data were processed with library-free DIA-NN analysis. Shown data are the identification numbers of and concentration, detailed information of numbers can be found in Supplementary file 1.
S. cerevisiae is one of the dominating workhorses of biotechnology, a key model organism for basic research, closely related to important clinical pathogens such as Candida glabrata (Oliver, 2022), and a pathogen on its own right. In a S. cerevisiae protein extract tryptic digest, micro-flow-rate Zeno SWATH MS quantified 29,339 precursors from 2915 proteins from a 62.5 ng injection, compared to 1660 proteins quantified by SWATH on the same instrument setup. From an extract of E. coli, the dominating bacterial model organism, Zeno SWATH MS identified 24,880 precursors or 2131 proteins at a 62.5 ng load, whereas conventional SWATH acquired 16,139 and 1768 precursors and proteins, respectively. And finally, we identified 33,694 precursors and 3815 proteins in 62.5 ng of chickpea (C. arietinum) seedling extract, chosen as a representative sample for a plant proteome. In comparison, 500 ng of sample was needed to reach similar identification numbers (34,573 precursors, 3807 proteins) using SWATH acquisition on the plant proteome tryptic digest (an overview of achieved protein identification is shown in Figure 4; details for respective precursor identification are shown in Figure 4—figure supplement 1; identification numbers for both proteins and precursors are shown in Supplementary file 1).
Discussion
Proteomic experiments can be limited by sample amount or instrument sensitivity. We and others have previously presented high-throughput proteomic experiments using micro-flow-rate (Vowinckel et al., 2018; Bian et al., 2020; Bruderer et al., 2019; Zelezniak et al., 2018) and analytical flow-rate chromatography (Messner et al., 2020; Messner et al., 2021). Specifically, the latter presents highly desirable properties to increase the throughput of proteomic experiments, for example, ease of use, the ability to run fast gradients, and high reproducibility. The disadvantage of higher flow rates is however the higher dilution, resulting in a requirement for higher sample amounts and, typically, a decrease in the proteomic depth. However, new acquisition techniques can mitigate these problems.
The fast acquisition speed required to conduct proteomic experiments with fast chromatography is often reached using TOF mass spectrometers. Nonetheless, in TOF instruments, all fragment ions are recorded in parallel (quasi-simultaneously), where the ions spread by m/z across the distance between the exit of the collision cell and the TOF accelerator. When the subsequent pulse is observed at the accelerator, only a small proportion of the total fragment ion population is situated at the right location to enter the TOF, leading to a proportion of ions not being able to be detected, thus limiting sensitivity. This disproportionately impacts low-m/z ions, with the duty cycle of TOFs falling in the range of only 5–25%, depending on concentration range and m/z. Because of this duty cycle deficiency, many ions (especially low-m/z ions) are not detected in low-abundance samples, resulting in TOF-MS/MS not fully realising its full sensitivity potential. A linear ion trap introduced after the collision cell (Q2) – the Zeno trap – can compensate for these limitations in TOF-MS/MS (Loboda and Chernushevich, 2009). When the Zeno trap is enabled, the fragment ions in the selected mass range are trapped in an axial pseudopotential well created by an additional RF (‘AC’) voltage applied with the same amplitude and phase to all four rods of the trap in a required focal point. Subsequently, the release of the ions occurs by potential energy with timing aligned to the next pulse at the accelerator, enabling a >90% duty cycle, thus increasing intensity 4–20 times (Baba et al., 2021) without loss of mass accuracy and resolution (Figure 1—figure supplement 1).
To use the Zeno trap technology in high-throughput proteomics, we have established and benchmarked a DIA method termed Zeno SWATH MS. To achieve this, we modified the instrument acquisition software on a Zeno TOF 7600 system that is equipped with the Zeno trap to be able to acquire data in DIA mode, coupled the system to different chromatographic systems, and optimised acquisition parameters. We then tested Zeno SWATH MS on tryptic digests generated of protein extracts derived from a human cell line standard (K562), human plasma, as well as from yeast cultures, E. coli cultures, and chickpea seedling samples. Furthermore, we benchmarked the method on two-species mixtures of yeast and human cell line standard samples. In all cases, we identified more precursors and proteins, and quantified those with higher precision using Zeno SWATH MS in comparison to SWATH MS on the same instrument, as well as its predecessor, the TripleTOF 6600. We show that gains mostly result from the better identification and quantification of low-abundant precursors. Moreover, we report that Zeno SWATH MS can generate deep proteomes from low sample amounts, even when used in conjunction with 800 µl/min analytical flow-rate chromatography. For instance, on most matrices, we quantified a similar number of precursors from 62.5 ng of tryptic digest using Zeno SWATH MS as from 500 ng of the same tryptic digest using SWATH MS. Lowering the sample dependency in high-throughput proteomics has at least three different applications: First, high-throughput proteomics can be extended to studies where only low amounts of material are available. Second, injecting lower amounts of samples decreases the cleaning cycles of mass spectrometers and increases their uptime. Third, injecting lower sample amounts allows one to reduce costs. Reassuringly, Zeno SWATH MS improves both the number of proteins identified and quantification precision in all chromatographic methods tested. To this end, further applications of Zeno SWATH MS remain to be explored. For instance, our benchmark experiments were acquired with a ZenoTOF 7600 system running SCIEX OS 2.1.6, with an ionisation threshold for Zeno SWATH MS to prevent detector damage (‘ion striking’) caused by excessive ions hitting the detector at the same time. This detector protection procedure limits the maximum injection amount in our benchmark to 62.5 ng, preventing us from testing the effect of high amounts of sample on the MS system. The ionization threshold (which limited the load to 62.5 ng) was experimental in nature at the time performing experiments, however, further re-optimizations to this threshold occurred enabling higher loads with Zeno SWATH MS with newer releases of SCIEX OS.
Beyond the here investigated application of Zeno SWATH MS for medium-to-high flow-rate chromatography or high-throughput proteomics, the gain in sensitivity and expansion of proteomics coverage can be useful for a range of other applications. For instance, when coupled with nano-flow chromatography systems, the sensitive Zeno SWATH MS might benefit the field of single-cell proteomics. Moreover, further improvements in sensitivity in high-throughput proteomics are still possible in the future. For instance, our lab has previously introduced a scanning SWATH acquisition technique that significantly increased the identification rate of proteins using the TripleTOF 6600 system (SCIEX), particularly in ultra-fast proteomic experiments (Messner et al., 2021). Scanning SWATH combines the benefits of data-dependent and data-independent techniques by employing a ‘sliding’ quadrupole (Q1) to assign precursor masses to MS/MS traces, allowing for a faster Q1 scan than stepwise SWATH acquisition. A combination of the Zeno trapping technology with scanning SWATH is technically possible and might result in further gains in the sensitive analysis of proteomes at high throughput.
Materials and methods
Materials and reagents
Request a detailed protocolWater was from Merck (LiChrosolv LC-MS grade; Cat# 115333), acetonitrile was from Biosolve (LC-MS grade; Cat# 012078), 1,4-dithiothreitol (DTT; Cat# 6908.2) from Carl Roth, iodoacetamide (IAA; Bioultra; Cat# I1149) and urea (puriss. P.a., reag. Ph. Eur.; Cat# 33247) were from Sigma-Aldrich, ABC (Eluent additive for LC-MS; Cat# 40867), formic acid (LC-MS Grade; Eluent additive for LC-MS; Cat# 85178) was from Thermo Fisher Scientific, trypsin (sequence grade; Cat# V511X), K562 human cell line standard (K562, Mass Spec-Compatible Human Protein Extracts, Cat# V6951), this commercially available cell line is Homo sapiens origin, the identity has been authenticated, and tested by Promega, RRID: CVCL_0004. Digested yeast protein extract (Mass Spec-Compatible Yeast Protein Extracts, Cat# V7461) were from Promega, commercial human plasma samples (Human Source Plasma, LOT# 20CILP1034) were from ZenBio.
Sample preparation
K562 human cell line standard (Promega, #V6951) and yeast tryptic digest (Promega, #V7461) stock samples were prepared via reconstitution of the powder form of the product using 200 µl of reconstitution buffer (1% ACN, 0.1% formic acid in LC grade water) and aliquoted into 4×50 µl stock samples, stored at –80°C and defrosted before sample preparation.
Plasma stock samples were prepared as follows: 5 µl of commercially available plasma sample was added to 55 µl of denaturation buffer (50 µl of 8 M urea, 100 mM ammonium bicarbonate [ABC], 5 µl of 50 mM DTT). The samples were incubated for 1 hr at room temperature before the addition of 5 µl of 100 mM IAA. After a 30 min incubation at room temperature in the dark, the samples were diluted with 340 µl of 100 mM ABC and digested overnight with 22.5 µl of 0.1 µg/µl trypsin at 37°C. The digestion was quenched by adding 50 µl of 10% formic acid. The resulting tryptic peptides were purified on a 96-well C18-based solid-phase extraction (SPE) plate (BioPureSPE Macro 96-well, 100 mg PROTO C18, The Nest Group). The purified samples were resuspended in 120 µl of 0.1% formic acid.
E. coli samples were prepared from 1010 cells, thawed in 200 µl lysis buffer (7 M urea, 0.1 M ABC, 2 mM MgCl2, 0.125 U benzonase). The cell pellets were disrupted using a GenoGrinder in the presence of 100 mg of 0.1 mm ceramic beads. The lysate was incubated at 37°C for 15 min to ensure optimal benzonase activity. The supernatant was clarified by centrifugation at 4°C, 20,000 × g for 15 min.
Following addition of DTT to a final concentration of 5 mM, the samples were incubated for 90 min at room temperature before the addition of IAA to a final concentration of 10 mM. After 30 min incubation at room temperature in the dark, the samples were diluted fivefold with 100 mM ABC and digested overnight with 0.1 µg/µl trypsin at 37°C. The digestion was quenched by adding formic acid to a final concentration of 1%. The resulting tryptic peptides were purified using C18-based (AttractSPE, Affinisep) stage tips as described by Rappsilber et al., 2007, and dried before resuspension in 0.1% formic acid.
Chickpea seedling sample preparation was performed essentially as described by Wang et al., 2006. Briefly, 500 mg of chickpea seedlings, created by the germination of chickpeas in the lab, were pulverised in liquid nitrogen. The powder was washed with trichloroacetic acid/acetone, methanol, and acetone at 4°C. After removal of acetone, the pellet was dried at 50°C for 20 min, the protein was subjected to phenol extraction and precipitated by the addition of ice-cold 0.1 M ammonium acetate in 80% methanol. The resulting pellet was washed once with ice-cold methanol and once with ice-cold 80% acetone, then air-dried. The protein pellet was dissolved in 8 M urea, 100 mM Tris, pH 8.0. Following dilution to 2 M urea, the protein was reduced with 2 mM DTT for 30 min at room temperature and alkylated for 30 min with 15 mM chloroacetamide. The sample was then digested overnight with LysC and trypsin. The resulting peptides were filtered through a 30 kDa MWCO centrifugal filter and further purified by C18 stage tips as described above.
LFQbench hybrid samples
Request a detailed protocolTo generate the hybridisation of different species samples, tryptic peptides were combined as previously published (Navarro et al., 2016; Meier et al., 2020), featuring similar benchmarks justified to the maximum injectable amount of MS system to avoid ion striking, the sample mixtures were in the following ratios: MIX A was composed of 30 ng/µl of K562 human cell line standard and 35 ng/µl yeast tryptic digest; MIX B was composed of 30 ng/µl of K562 human cell line standard and 17.5 ng/µl yeast tryptic digests.
Instrumentation
Request a detailed protocolAll ZenoTOF 7600 system-related methods were developed on research instruments in which an ionisation threshold for Zeno SWATH MS was introduced to prevent detector damage caused by too many ions hitting the detector at the same time (‘ion striking’). SWATH window size and accumulation time were optimised and chosen to avoid ion striking while maintaining adequate precursor and protein identification numbers. The ionisation threshold has re-optimised and allow higher loads with Zeno SWATH MS with a new release of SCIEX OS.
To eliminate the potential for unreliable results due to low sample amount, before all the dilution series is acquired (as described below), we first injected replicates of blanks (50/50 v/v ACN/H2O) to exclude the identification numbers that are explained by carryover, then started with the highest dilution (lowest concentration) and performed triplicate injections to prove similar results among replicates.
ZenoTOF 7600
Micro-flow-rate system
Request a detailed protocolK562 human cell line standard dilution series and all sample species were acquired on an ACQUITY UPLC M-Class system (Waters) coupled to a ZenoTOF 7600 mass spectrometer with an Optiflow source (SCIEX). Prior to MS analysis, samples were chromatographically separated with a 20 min gradient (time, % of mobile phase B: 0 min, 3%; 0.86 min, 7.1%; 2.42 min, 11.2%; 5.53 min 15.3%; 9.38 min, 19.4%; 13.02 min, 23.6%; 15.48 min, 27.7%;17.27 min, 31.8%; 19 min, 40%; 20 min, 80% followed by re-equilibration for 10 min before the next injection; Supplementary file 2) on HSS T3 column (300 µm×150 mm, 1.8 µm, Waters) heated to 35°C, using a flow rate of 5 µl/min (Zelezniak et al., 2018) where mobile phases A and B are 0.1% formic acid in water and 0.1% formic acid in ACN, respectively. To compensate for injection volume variance, 1 µl of each dilution series sample or mixture sample was loaded prior to samples entering the MS. A SWATH acquisition scheme with 60 variable-size windows covering a precursor mass range of 400–900 m/z (Supplementary file 2) and 11 ms accumulation time was used. A Zeno SWATH acquisition scheme with 85 variable-size windows and 11 ms accumulation time was used (Supplementary file 2). For both methods, ion source gas 1 and 2 were set to 12 and 60 psi, respectively; curtain gas to 25, CAD gas to 7, and source temperature to 150°C; spray voltage was set to 4500 V. (Unique precursor and protein identification via different windows in 100 ng K562 human cell line standard injection with various SWATH acquisition schemes are shown in Figure 5, showing very little difference in identification performance with window number of 60 or 85.)

Precursor and protein identification performance with 100 ng K562 human cell line lysate injection using 5 µl/min, 20 min micro-flow chromatography coupled with different SWATH window number acquisition schemes.
For the 40 variable window SWATH acquisition scheme, 26,115 precursors from 3904 proteins were identified; the 60 variable window SWATH acquisition scheme identified 25,884 precursors from 4025 proteins; and the 85 variable window SWATH acquisition scheme identified 25,945 precursors from 3995 proteins.
Analytical flow-rate system
Request a detailed protocolStandards and samples were acquired on an 1290 Infinity II UHPLC system (Agilent) coupled to a ZenoTOF 7600 mass spectrometer with a DuoSpray TurboV source (SCIEX). Prior to MS analysis, samples were chromatographically separated on an Agilent InfinityLab Poroshell 120 EC-C18 1.9 µm, 2.1 mm × 50 mm column heated to 30°C. A 5 min 0.8 ml/min flow rate of linear-gradient ramping from 3% ACN/0.1% formic acid to 40% ACN/0.1% formic acid was applied (Messner et al., 2021). 1 µl of each dilution series sample was loaded prior to samples entering MS. A SWATH acquisition scheme with 60 variable-size windows and 11 ms accumulation time was used (Supplementary file 2). The Zeno SWATH MS has the same 60 windows as the SWATH MS and 13 ms accumulation time. For both acquisition methods, ion source gas 1 (nebuliser gas), ion source gas 2 (heater gas), and curtain gas were set to 60, 65, and 55 psi, respectively; CAD gas was set to 7, source temperature to 700°C, and spray voltage to 3500 V.
TripleTOF 6600
Request a detailed protocolMicro-flow system: For inter-platform comparison, K562 human cell line standard dilution series were also acquired on a nanoAcquity UPLC System (Waters) coupled to a SCIEX TripleTOF 6600 mass spectrometer. The peptides were separated with the same 20 min gradient on a Waters HSS T3 column (300 µm×150 mm, 1.8 µm) using a flow rate of 5 µl/min. The SWATH MS/MS acquisition scheme was used as described previously (40 variable-size windows and 35 ms accumulation time; Zelezniak et al., 2018).
Spectral libraries
Request a detailed protocolThe library for high-flow benchmarks using the K562 human cell line standard was generated from high pH reverse-phase fractionation of commercial K562 human cell line standard and HeLa (Thermo Fisher Scientific, #88329) digests. Individual pH fractions from each cell line were analysed with a ZenoTOF 7600 in IDA/DDA mode in-line with a Waters M-Class HPLC and searched with a ProteinPilot application within the OneOmics suite against a human Swiss-Prot canonical and isoform FASTA file (July 2020). The ProteinPilot results from each fraction were then merged using the Extractor application in the OneOmics suite to build a spectral library for further processing of DIA results.
For the analysis of plasma samples, a public spectral library (Bruderer et al., 2019) was used.
Data processing and analysis
Request a detailed protocolAll raw data from ZenoTOF 7600 system were acquired by SCIEX OS (v2.1.6) (note that SCIEX OS 3.0, which supports Zeno SWATH MS, is the commercial software available on the market), processed with DIA-NN (v1.8 beta 20) (Demichev et al., 2020) using mass accuracy of 20 and 12 ppm at the MS2 and MS1 level, respectively, scan window of 7, protein inference disabled (spectral library as described above) or relaxed (library-free, with additional command --relaxed-prot-inf), quantification strategy of ‘Robust LC (high precision)’, and library generation set as ‘IDs, RT&IM profiling’. All other settings were kept at default.
The high-flow K562 human cell line lysate benchmark data was analysed using the DDA-based library described above, with MBR disabled.
The plasma acquisition analysis was performed as described previously (Messner et al., 2020). Specifically, the ‘Deep learning-based spectra, RTs and IMs prediction’ option was activated in DIA-NN, to replace all spectra and retention times in the public spectral library (Bruderer et al., 2019) with ones predicted in silico. Further, the protein annotation in the library was replaced using the ‘Reannotate’ function in DIA-NN with the annotation from the Human UniProt (The UniProt Consortium, 2016) isoform sequence database (UP000005640, 19 October 2021). The processing was performed using the MBR mode in DIA-NN – a two-step analysis wherein a spectral library (empirical or predicted in silico) is first refined based on the DIA experiment in question, and subsequently used to reanalyse it (Demichev et al., 2020).
For the analysis of micro-flow K562 human cell line standard acquisition as well as samples from other species, the respective FASTA databases from UniProt (human: UP000005640 [10 March 2022] yeast: UP000002311_559292 [28 September 2021] E. coli: UP000000625 [19 October 2021]; chickpea: UP000087171_3827 [22 October 2021]) were used by DIA-NN in library-free mode. Specifically, following a two-step MBR approach described previously (Demichev et al., 2020), an in silico spectral library is first generated by DIA-NN from the FASTA file(s); this library is then refined based on the DIA dataset and subsequently used to reanalyse the dataset, to obtain the final results.
The data were filtered in the following way. First, a 1% run-specific q-value filter was automatically applied at the precursor level by DIA-NN. Further, to compare the protein identification numbers in the dilution series between different injection amounts and LC-MS modes, a 1% global protein group q-value filter (Lib.PG.Q.Value column in the report) was also applied. For the LFQbench-type two-species benchmarks for quantification accuracy, a 1% global protein group q-value filter and 1% run-specific protein group q-value were used instead (Lib.PG.Q.Value and PG.Q.Value column in the report). We note that in any experiment processed using the MBR mode in DIA-NN, 1% global precursor q-value filtering is also applied automatically (Demichev et al., 2020).
CV were calculated for each protein or precursor as its empirical standard deviation divided by its empirical mean and are reported in percentages. CV values were calculated for proteins or precursors identified in all three replicate measurements.
We included the DIA-NN pipeline file in PRIDE (Perez-Riverol et al., 2022) PXD036786, which fully specifies the DIA-NN configuration in processing of each of the datasets analysed.
Data availability
Request a detailed protocolThe mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2022) partner repository with the dataset identifier PXD036786.
Dataset used for data analysis and visualisation are available online, DOI:10.17632/nwjcn4fy87.1 (Wang et al., 2022).
Data availability
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD036786. Dataset used for data analysis and visualisation are available online, https://doi.org/10.17632/nwjcn4fy87.1.
-
Mendeley DataHigh-throughput proteomics of nanogram-scale samples with Zeno SWATH DIA.https://doi.org/10.17632/nwjcn4fy87.1
-
PRIDEID PXD036786. High-throughput proteomics of nanogram-scale samples with Zeno SWATH MS.
References
-
The human plasma proteome: history, character, and diagnostic prospectsMolecular & Cellular Proteomics 1:845–867.https://doi.org/10.1074/mcp.r200007-mcp200
-
Dissociation of biomolecules by an intense low-energy electron beam in a high sensitivity time-of-flight mass spectrometerJournal of the American Society for Mass Spectrometry 32:1964–1975.https://doi.org/10.1021/jasms.0c00425
-
In-Depth comparative proteomic analysis of yeast proteome using iTRAQ and SWATH based MSMolecular BioSystems 11:2135–2143.https://doi.org/10.1039/c5mb00234f
-
Analysis of 1508 plasma samples by capillary-flow data-independent acquisition profiles proteomics of weight loss and maintenanceMolecular & Cellular Proteomics 18:1242–1254.https://doi.org/10.1074/mcp.RA118.001288
-
BookProteomicsIn: Lindon John C, Tranter George E, Koppenaal David W, editors. Encyclopedia of Spectroscopy and Spectrometry (Third Edition). Oxford: Academic Press. pp. 766–773.https://doi.org/10.1016/B978-0-12-803224-4.00061-3
-
Biomarker discovery and verification of esophageal squamous cell carcinoma using integration of SWATH/MRMJournal of Proteome Research 14:3793–3803.https://doi.org/10.1021/acs.jproteome.5b00438
-
A novel ion trap that enables high duty cycle and wide m/z range on an orthogonal injection TOF mass spectrometerJournal of the American Society for Mass Spectrometry 20:1342–1348.https://doi.org/10.1016/j.jasms.2009.03.018
-
Ultra-Fast proteomics with scanning SWATHNature Biotechnology 39:846–854.https://doi.org/10.1038/s41587-021-00860-4
-
A multicenter study benchmarks software tools for label-free proteome quantificationNature Biotechnology 34:1130–1136.https://doi.org/10.1038/nbt.3685
-
The cross-sectional association between snacking behaviour and measures of adiposity: the fenland study, UKThe British Journal of Nutrition 114:1286–1293.https://doi.org/10.1017/S000711451500269X
-
The pride database resources in 2022: a hub for mass spectrometry-based proteomics evidencesNucleic Acids Research 50:D543–D552.https://doi.org/10.1093/nar/gkab1038
-
Protein Z: a putative novel biomarker for early detection of ovarian cancerInternational Journal of Cancer 138:2984–2992.https://doi.org/10.1002/ijc.30020
-
Deep and highly sensitive proteome coverage by LC-MS/MS without prefractionationMolecular & Cellular Proteomics 10:M110.https://doi.org/10.1074/mcp.M110.003699
-
UniProt: the universal protein knowledgebaseNucleic Acids Research 45:D158–D169.https://doi.org/10.1093/nar/gkw1099
Decision letter
-
Benoît KornmannSenior and Reviewing Editor; University of Oxford, United Kingdom
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
[Editors' note: this paper was reviewed by Review Commons.]
https://doi.org/10.7554/eLife.83947.sa1Author response
We thank the reviewers for their constructive feedback. We have addressed their comments and conducted additional experiments and analyses. We believe their input has helped us to greatly improve our manuscript, and we hope all edits are satisfactory.
Reviewer #1 (Evidence, reproducibility and clarity (Required)):
In this manuscript Wang et al., benchmark the new ZenoTOF with analytical and micro-flow set up and show impressive numbers of proteins identified and quantified.
The paper is well written, and I have only a few minor comments:
We thank the reviewer for the positive and constructive feedback.
1. Figure 1. many of the panels are hard to read. Especially 1a.
We apologise, and agree in retrospect that original Figure 1 and specifically 1a were overloaded with information not necessary for the manuscript. Based on the reviewer’s suggestion, we simplified the instrument structure scheme and made it a separate Figure (Figure 1 – figure supplement 1 in the revised manuscript). We hope the edits improved the readability of the figure.
2. Figure 1d. can the human amount not be normalised to log2=0?
We modified Figure 1d (Figure 1c in the revised manuscript), normalised human to log2 = 0, and made it easier to read.
3. Please provide in the legend the bin size for the figure in 1e.
Thanks for pointing this out. As suggested, we included the bin size in the legend of Figure 1e (Figure 2c in the revised manuscript).
4. Page 10 top: did SWATH identify more proteins than Zeno SWATH in plasma? There is something wrong as the figure shows something else. Also: That sentence in brackets is confusing.
We apologise if our Figure was confusing. With Zeno SWATH we obtained a better identification number in both precursor and protein numbers compared to SWATH in plasma (as in the other matrices). We hope that the updated versions of the figures are now clear.
Moreover, we added a sentence to the text to explicitly spell out the numbers:
“In human neat plasma, with Zeno SWATH acquisition, we identified 3,947 precursors from 245 proteins, while with SWATH acquisition, 2,516 precursors from 156 proteins were identified with same 62.5 ng sample load”.
5. Typo: page 10: respectively.
Thanks for noticing, the typo was corrected.
6. Please add raw data to PRIDE or a similar repository.
We uploaded the data to PRIDE, and the accession number is # PXD036786. All data will be made public with the publication of the manuscript.
Reviewer #1 (Significance (Required)):
This is an impressive new technology that has been benchmarked by the Ralser group. It outperforms current state-of-the-art approaches.
The primary audience is the proteomics community.
My expertise is proteomics and quantitative mass spectrometry. I am well qualified to review this paper.
We greatly thank the reviewer for the time they spent reviewing our article and their input.
Reviewer #2 (Evidence, reproducibility and clarity (Required)):
Summary:
Wang et al. present an evaluation of a new generation of time-of-flight-based mass spectrometer that improves on the fraction of ions factually used for detection of peptide analytes, thus boosting the sensitivity of the Zenotof 7600 system when compared to the same instrument with the duty-cycle-enhancing Zenotrap module disabled and also when compared to the previous generation instrument of the same vendor in some of the comparisons.
The authors position the MS acquisition technique as particularly suitable in combination with medium (micro-) and high ('analytical') flow and throughput methods where higher flow rates (vs. conventional nanoflow-LCMS) allow rapid sample turnover and high throughput, yet limit the efficiency of electrospray and ion transfer into the MS system, thus being in dire need for enhanced sensitivity of the MS system employed for detection. The competency of such an MS system for very low input materials as e.g. encountered in emerging single-cell proteomic workflows, typically employing nanoflow chromatography, was thus not part of the study.
Accordingly, a medium- (micro-flow) and very high ('analytical'-flow) throughput LC method were screened on the three MS (parameter) setups using human cell lysate digests typically utilized in such technical evaluations. Well-received, the authors further extended their analysis (for the new instrument) across additional sample types of clinical and extended biological interest and spanning different levels of complexity and dynamic range of contained protein analytes.
In addition, the authors also performed a controlled ratio 2-species mixture experiment which allows detailed benchmarking of proteome coverage as well as the quality of protein quantification in a known differential comparison for the medium throughput (micro-flow) method.
The data quite convincingly demonstrate an increased sensitivity of the instrument based on similar identification performance in DIA bottom up proteomics from ca. 3- to 8-fold lower input peptide mass. However, I see a number of shortcomings mainly in the presentation and in part the completeness of the work, with specific comments below.
We thank the reviewer for the accurate summary of our manuscript, and we addressed his specific requirements, outlined below in a point to point reply.
We further would like to comment that we concentrated on medium-to-high flow rates and microflow chromatography, as we see the main application of the new acquisition technology in medium- and large-scale screening approaches in systems biology and biomedicine. We totally agree with the reviewer that single-cell biology is another exciting new application for proteomics too, yet in its earlier stages, and we would choose another setup (i.e. nano flowrate chromatography and not high flowrates) for this application. That said, we agree with the reviewer that, plausibly, Zeno trapping can be useful in single-cell proteomics as the sensitivity increases. We added a final paragraph to the Discussion section and highlight this potential application.
Major comments:
– Are the key conclusions convincing?
1. The concluded 10x sensitivity increase is overstating the observed numbers (x5-x8). In addition, the authors should at least discuss other changes than the Zeno trap incurred in the Zeno SWATH vs non-Zeno-SWATH DIA setups, particularly changes in accumulation times per m/z range, with Zeno Swath accumulating ~42 % longer per cycle spanning the same m/z range (85 vs 60 windows with 11ms per window) in the uflow method set and ~ 18 % longer in the high-flow method set (same window number but 13 ms vs. 11 ms dwell time per window). This should be discussed as one of the optimizations/factors contributing to the increased sensitivity observed in Zeno Swath measurements vs conventional SWATH.
We apologise for having used the placative 10-fold increased sensitivity from the precursor ID increases obtained with lower sample amounts. We revised the manuscript now considerably and express gains in IDs with specific numbers, rather than giving an ‘order of magnitude’ range.
Benchmarking two different proteomic methods has a typical caveat—that one may not be able to use the exact same settings on the different methods if one does not want to discriminate against one method. Expressed in other words – for our specific case – do I want to quantify the pure impact of the Zeno trap itself, or do I want to evaluate the performance of a workflow that uses the Zeno Trap, versus a workflow that does not use the Zeno Trap? Both are legitimate questions, but we believe the second is more relevant for most of the readers. In our specific case, several method parameters differ between SWATH and Zeno SWATH for the reason that due to the higher ion transmission, the detector saturates earlier in Zeno SWATH than it does in SWATH mode, which results in slightly different acquisition schemes being optimal for either SWATH or Zeno SWATH, respectively.
On our specific platform, when using the 60 window acquisition scheme in Zeno SWATH, using our chromatography, and when more than 62.5 ng of sample is injected, the number of ions that hit the detector exceeds a threshold set to protect the detector from damage. Once this threshold is exceeded, the instrument firmware stops acquiring. This is why we cannot inject very high amounts of digest on our instrument when running Zeno SWATH acquisition. To keep the ion counts below the ion striking threshold in Zeno SWATH, we used a smaller window size (resulting in lower ion transmission per unit time), and in order to compensate for that, we increased the window number to 85 at 62.5 ng load as well as all dilution series.
In order to address the reviewer’s point, we tested to quantify the impact of the different windows. We conducted an experiment using SWATH, injected 100 ng of sample, and acquired the data with both 40, 60, and 85 windows. Of note, we observed no strong effect on the number of SWATH windows, and the number of precursors or proteins identified, particularly with respect to 60 and 85 windows. We now clearly state differences in the number of SWATH windows between methods, and their effect in the manuscript (Figure 5).
Illustrated here is the comparison of different SWATH-MS acquisition schemes (SWATH window number 40, 60, and 85). With 100 ng of K562 standard injected, in the 40 window SWATH method, 26,115 unique precursors from 3,904 proteins were identified; 25,884 precursors from 4,025 proteins were identified by 60 window SWATH-MS; and 25,945 precursors from 3,995 proteins were identified by 85 window SWATH scheme.
On a similar note, it is also correct that the accumulation time is different between the acquisition schemes that we used for analytical flowrate and for micro flowrate chromatography, running with SWATH and Zeno SWATH MS. Also this is intentional, as the different chromatographic regimes create very different requirements on the acquisition schemes to perform optimally.
On that note, it was unclear to me when and where the 40 variable window SWATH method mentioned in the methods was used and where the settings can be found.
The 40 variable window SWATH method was used on the SCIEX 6600 TripleTOF system for comparison of identification numbers and protein coverage between the two instruments in Figure 2a-c. We apologise that due to the focus on Zeno SWATH, we kept the part about the 6600 too brief. We expanded the Methods section for the 6600:
“TripleTOF 6600:
For inter-platform comparison, K562 dilution series were also acquired on a nanoAcquity UPLC System (Waters) coupled to a SCIEX TripleTOF 6600 mass spectrometer. The peptides were separated with the same 20-min gradient on a Waters HSS T3 column (300 µm × 150 mm, 1.8 µm) using a flow rate of 5 µl/min. The SWATH-MS/MS acquisition scheme was used as described in previous work (40 variable size windows and 35 ms accumulation time [8]).”
2. Since injected material is a critical parameter here, it would be good if it was mentioned also with the key conclusion on the increased number of confidently quantified peptides in microflow (based on the 2-species controlled quantity experiment).
We thank the reviewer for the suggestion. We have added a new 2-species deconvolution experiment that shows the performance of Zeno SWATH versus SWATH with 50% lower sample amount. (See below). Furthermore, we now explain better how the injection amounts were chosen. Based on previous works featuring similar benchmarks (Navarro et al. 2016; Meier et al. 2020), we justified the total injection amount to the maximum injectable amount of the MS system and defined the 2-species mixtures ratio for the LFQbench test.
Methods “to generate the hybridisation of different species samples, tryptic peptides were combined following previous works (Navarro et al. 2016; Meier et al. 2020) featuring similar benchmarks, justified to the maximum injectable amount of MS system to avoid ion striking, the sample mixtures were in the following ratios…”
– Regarding the confidently quantified peptides, we added LFQbench based on peptides. At both peptide and protein levels, Zeno SWATH shows a strong increase in the number of confident identification (with same injection amount, SWATH quantifies 11,227 precursors from 1,741 human proteins and 4,755 precursors from 776 yeast proteins; Zeno SWATH greatly increases the quantification number to 32,946 precursors from 3,282 human proteins and 13,983 precursors from 1,692 yeast proteins) (Figure 1—figure supplement 2b,c).
– The additional experiment where we inject 50% of the sample of 2-species sample preparations is included. We find that for both species, a similar strong increase of precise quantification number of proteins from different species is observed (SWATH method: around 1.6-fold increase with double the injection amount; Zeno SWATH: around 1.3-fold increase for both species). Also, even with double the injection amount, SWATH acquisition quantifies fewer proteins compared to Zeno SWATH (Zeno SWATH precisely quantifies (valid ratios with at least two replicates in each of A and B) 2,714 human and 1,299 yeast proteins, while with double the injection SWATH quantifies 1,741 human and 776 yeast proteins). With the same injection amount, Zeno SWATH aquantifies 3,282 human and 1,692 yeast proteins, increasing the confidence quantification number around 2-fold. In general, Zeno SWATH allows more precise quantification of mixed species compared to SWATH acquisition (Figure 1 —figure supplement 2f,g).
3. Conclusions 'increasing protein identification numbers through the use of analytical-flow-rate chromatography' does not capture the observed data; the use of analytical-flow-rate does not convey an increase in protein identification numbers but enhanced sensitivity rather enables the maintenance of high protein identification numbers / proteome coverage despite/concurrent with analytical-flow chromatography.
We apologise for poor phrasing; we never aimed to suggest that high-flow-rate liquid chromatography would give higher ID numbers compared to lower flow rates – the sentence should have said that Zeno SWATH performs better than SWATH on the fast chromatography. We corrected the statement. In addition to further avoiding ambiguity, we give the examples. “For instance, with 250 ng of K562 human cell line standard injected, around 3,000 proteins were quantified with Zeno SWATH MS while a similar protein quantification number can only be reached by SWATH MS with 2µg injection on the same instrument and chromatography”.
4. In titration curve experiments like these, probing proteome coverage from relatively small sample amounts, special care.
Although the sentence is somewhat cut off, we believe the reviewer refers to the difficulties and considerations to be taken when working with small amounts of samples. We agree. The way we do these experiments is by first injecting several blanks, to exclude numbers that are explained by carryover, and then start with the highest dilution/lowest concentrated sample. The caveat is that like in any proteomic experiment, the values obtained with the lowest sample amounts are the least reliable. In part, this is mitigated in our manuscript by obtaining relatively similar results across the replicates (K562) included in our study. Nonetheless, we outlined this potential caveat in the manuscript.
Methods: “Before all the dilution series is prepared (as described below), we first inject replicates of blanks (50/50 v/v ACN/H2O) to exclude the identification numbers that are explained by carryover, then start with the highest dilution (lowest concentration), and performed triplicate injections to prove similar results among replicates.”
Results: “The concentration series was acquired starting with the lowest concentrations injected after several blank measurements with a 20-min, 5 µl/min chromatographic gradient (Methods), covering sample amounts ranging from near 1 ng to 500 ng (Methods).”
5. Should the authors qualify some of their claims as preliminary or speculative, or remove them altogether?
We carefully revised the paper so that we clearly separate claims from preliminary/speculative results. All key results are backed up experimentally.
5.1 'Zeno SWATH increases protein identification in complex samples 5- to 10-fold when compared to current SWATH acquisition methods on the same instrument'.
5.2 At no point this is shown, a decrease of required input amounts by 5-8-fold (increase in sensitivity) is shown by the data, not a multiplication of protein identification rates by that factor.
We apologise for the poor phrasing. We have been too hand-waving in descrbing the performance gains as ‘an order of magnitude’ ‘10 fold’ as estimated by the number of IDs gains obtained in the dilution-series experiments. We adjusted the numbers as suggested by the reviewer. We now strictly avoid hand-waving statements and give precise values instead, throughout the text.
For instance:
“Interpreting the data in a different way, Zeno SWATH MS identified a similar number of precursors and proteins from 62.5 ng of K562 human cell line standard as SWATH MS did from 500 ng of the same sample on the same instrument (SWATH identified 64,316 precursors or 5,242 proteins from 500 ng)”.
6. Would additional experiments be essential to support the claims of the paper? Request additional experiments only where necessary for the paper as it is, and do not ask authors to open new lines of experimentation.
6.1 Figure 1f, Supplemental Figure 1b, Figure 3 and Supplemental Figure 3 lack data for the Zeno SWATH method's performance at higher concentration. Given the fact that there is a clear, continuous trend of significant enhancement of proteomic depth in the highest 3 concentrations sampled by the Zeno SWATH method, I lack an assessment of the upper limit of proteome coverage achievable by the new platform when input material is not limited, or at least learn why injecting more is not advisable on the ZenoTOF 7600 system. It is clear that the region of interest is the lower loads where sensitivity gains are most pronounced, but with the strong trend in IDs per ng injected in the sampled range and discrepant range sampled by the non-Zeno method I feel there is a gap in the dataset and the upper ceiling of proteome coverage could be mapped out more thoroughly (At least for human cell lysate and possibly human plasma where trends appear most (log2-)linear).
6.2 Similarly, unless constrained for technical or practical reasons, I would suggest to find the ceiling for achievable proteome depth in analytical flow (4, 8 ug?)
We thank the reviewer for the comment. As described above, we have already injected up to the maximum limit. On our 7600 instrument, when running. Zeno SWATH MS, cannot analyse higher sample amounts as a detector ion-count threshold is reached (‘ion striking’). We improved the discussion of this instrument limitation in the manuscript, and apologise that this technical limit of the hardware was not sufficiently discussed in the previous manuscript version. However, high amounts can be injected on the 6600 and the 7600 instruments when both are operated with SWATH-MS. This data is included, and we now discuss the ion striking limit much more predominantly.
6.3 Are the suggested experiments realistic in terms of time and resources? It would help if you could add an estimated cost and time investment for substantial experiments.
All requests by the reviewer were totally reasonable, and we conducted all experiments that were technically feasible (i.e. all but the high-sample-amount injections using Zeno SWATH MS, where the maximum was already shown in the previous data).
6.4. All these should be re-injections of existing samples on these MS setups and a minor effort provided instrument availability (<1w) and rapid re-analysis via DIA-NN.
Please see 6.1.
7. Are the data and the methods presented in such a way that they can be reproduced?
All method details are provided.
7.1 The raw data have not been deposited to a public repository. Reproducibility of the study would benefit significantly by raw data (including search results and spectral libraries with log files of creation) upload/sharing e.g. via ProteomeXchange/PRIDE.
The raw data were uploaded to PRIDE, ID#PXD036786, and will be released with the publication of the manuscript.
7.2 If any software versions or firmwares on the hardware are required to perform the measurements on the ZenoTOF on the market today, these versions and prospective release dates should be included or the accessibility of these settings commented on.
We added the version of SCIEX OS and DIA-NN. Sciex OS is commercial, while DIA-NN is free to use.
“All raw data from ZenoTOF 7600 system were acquired by SCIEX OS (v2.1.6) (SciexOS 3.0, which supports Zeno SWATH, is now commercial software available on the market) and processed with DIA-NN (v1.8 β 20, provided in the Supplementary Materials).”
8. Are the experiments adequately replicated and statistical analysis adequate?
8.1 Figures 3 and Supplemental Figure 3 need a clarification in the legend as to the nature and origin of ID numbers (mean? Number of replicates? Add error bars if possible)
We thank the reviewer for the suggestion. We modified the legends accordingly. All critical benchmarks were conducted in replicates and, if applicable, also in serial dilution series. Only the validation experiments in which we confirm the gains of Zeno SWATH MS versus SWATH-MS on additional biological matrices (i.e. plant, yeast, bacterial, plasma proteome extracts) are conducted using serial dilution series only. We have provided the ID numbers in Supplementary table 1.
We have added explanation sentence in legend: “Shown data were the identification number of respective species and concentration, detailed information of numbers can be found in Supplementary Table 1.”
8.2 The usage of DIA-NN for data analysis is somewhat unclear, in particular the in Methods/Spectral libraries "For the analysis of plasma samples, a project-independent public spectral library [29] was used as described previously [15]. The Human UniProt [30] isoform sequence database (UP000005640, 19 October 2021) was used to annotate the library and the processing was performed using the MBR mode in DIA-NN."
We added more detailed settings of DIA-NN and subsequent filtering steps in Methods. We also included a pipeline of DIA-NN in Supplementary Materials.
“All raw data from ZenoTOF 7600 system were acquired by SCIEX OS (v2.1.6) (note that SCIEX OS 3.0, which supports Zeno SWATH MS, is the commercial software available on the market), processed with DIA-NN (v1.8 β 20) (20) using mass accuracy of 20 and 12 ppm at the MS2 and MS1 level, respectively, scan window of 7, protein inference disabled (spectral library as described above) or relaxed (library-free, with additional command --relaxed-prot-inf), quantification strategy of "Robust LC (high precision)", and library generation set as “IDs, RT& IM profiling”. All other settings were kept default.
The high-flow K562 human cell line lysate benchmark data was analysed using the DDA-based library described above, with MBR disabled.
The plasma acquisition analysis was performed as described previously (15). Specifically, the ‘Deep learning-based spectra, RTs and IMs prediction’ option was activated in DIA-NN, to replace all spectra and retention times in the public spectral library (29) with ones predicted in silico. Further, the protein annotation in the library was replaced using the ‘Reannotate’ function in DIA-NN with the annotation from the Human UniProt (33) isoform sequence database (UP000005640, 19 October 2021). The processing was performed using the MBR mode in DIA-NN—a two-step analysis wherein a spectral library (empirical or predicted in silico) is first refined based on the DIA experiment in question, and subsequently used to reanalyse it (20).
For the analysis of micro-flow K562 human cell line lysate acquisition as well as samples from other species, the respective FASTA databases from UniProt (human: UP000005640 (10 March 2022) yeast: UP000002311_559292 (28 September 2021) E. coli: UP000000625 (19 October 2021); chickpea: UP000087171_3827 (22 October 2021)) were used by DIA-NN in library-free mode. Specifically, following a two-step MBR approach described previously (20), an in silico spectral library is first generated by DIA-NN from the FASTA file(s); this library is then refined based on the DIA dataset and subsequently used to reanalyse the dataset, to obtain the final results.
The data were filtered in the following way. First, a 1% run-specific q-value filter was automatically applied at the precursor level by DIA-NN. Further, to compare the protein identification numbers in the dilution series between different injection amounts and LC-MS modes, a 1% global protein group q-value filter (Lib.PG.Q.Value column in the report) was also applied. For the LFQbench-type two-species benchmarks for quantification accuracy, a 1% global protein group q-value filter and 1% run-specific protein group q-value was used instead (Lib.PG.Q.Value and PG.Q.Value column in the report). We note that in any experiment processed using the MBR mode in DIA-NN, 1% global precursor q-value filtering is also applied automatically (20).
Coefficients of variation (CV) were calculated for each protein or precursor as its empirical standard deviation divided by its empirical mean, and are reported in percentages. CV values were calculated for proteins or precursors identified in all 3 replicate measurements.
We included the DIA-NN pipeline file in PRIDE (Perez-Riverol et al., 2022) PXD036786, which fully specifies the DIA-NN configuration in processing of each of the datasets analysed.”
8.3 The authors should address in a revised version whether the identification numbers reported stem from two-pass or single-pass analysis (i.e. when the feature termed Match-between-runs) implemented since DIANNv1.8 was enabled and whether all runs, spanning different injection amounts were co-analyzed and data-re-queried for a targeted library containing precursors identified in high load samples in first pass analysis and then queried in low-load samples. In other words, are the low-load IDs independent of high load IDs? If not (i.e. the different loads were co-analyzed with MBR), what proteome coverage to the low sample loads reach bona fide, without the 'guidance' of high-load IDs?
We clarified this in the revision, thank you for the suggestion. We are using double-pass analysis: MBR was enabled for a fair comparison in the dilution series. The library-free analysis by DIA-NN is generally supposed to be used with a two-pass strategy, originally described (Demichev et al. 2020) and now termed ‘MBR’ within DIA-NN. That is, ‘bone fide’ IDs during library-free search are supposed to be only used to ‘refine’ the spectral library, which will then be used for the second pass. Otherwise, the data analysis (with DIA-NN or other software) would struggle with lower injection amounts—these are typically challenging when the search space is large due to the library-free mode. We would like to highlight that although the ‘MBR’ in DIA-NN shares the name with the respective algorithm in MaxQuant, the principle of operation here is different and simpler, i.e. in DIA-NN ‘MBR’ just refers to the two-pass analysis strategy with library refinement after the first pass (Demichev et al. 2020). This way, indeed, the numbers reported for each of the runs in the dilution series are increased through the MBR strategy, by virtue of runs being analysed together. Effectively, for the lower injection amounts, this works like analysing those with a project-specific DIA-based spectral library. In order to address the reviewer’s point, we now also added a benchmark of bona fide first-pass ID numbers (Figure 3 —figure supplement 1b), which indicates an even greater advantage for Zeno SWATH over SWATH.
8.4 Side note: Turning this around, could a high-load injection e.g. from a pool of limited-amount samples serve as a guiding element in a MBR-enabled analysis of a large cohort with limited sample amounts available per biological condition?
We agree with the reviewer's point of view, thank you for the input. Indeed, this is a main intention of MBR. For example, in the original DIA-NN release (Demichev et al. 2020), we manually performed the two-step ‘MBR’ procedure to improve identifications in mixed-species runs of the LFQbench test: the injections in which yeast or E. coli digests were at high concentration helped to increase the IDs in the injections in which the respective digests were at low concentration. However, the concept will be expanded further and explained in a future publication that will detail the new developments inside DIA-NN: we believe a future development of the MBR feature in DIA-NN is out of the scope of this manuscript, which focuses on Zeno SWATH technology.
Minor comments:
1. Specific experimental issues that are easily addressable.
1.1 The authors state the impact on dynamic range of identification when comparing ID sets against an external dataset with presumable cellular concentration numbers. I would in addition suggest comparing the dynamic range of the quantitative values observed from the available data which should provide a direct assessment of the dynamic range of quantification of the two methods.
We thank the reviewer for the suggestion. We created a correlation plot for directly comparing SWATH and Zeno SWATH. Overall, the shared proteins quantified in both methods are highly correlated, with an r² of 0.93. The plot also reflects that the quantities agree extremely well correlated in the upper four orders of magnitudes of the concentration range, and that the noise largely comes from the one order of magnitude that is close to the LOD (in SWATH), were Zeno SWATH performs better. Furthermore, the correlation analysis confirms that most proteins quantified by Zeno SWATH and not SWATH are in the low abundant range (below 10E3). (Figure 2d)
2. Are prior studies referenced appropriately?
2.1 The statement that conventional DIA methods rely on nanoflow chromatography (p3, paragraph 3) is not accurate as there is previous implementations of data-independent acquisition MS of microflow separations, in part the group's work and referred to later in the text.
We apologise if our text was misleading; indeed it was, to our knowledge, our own lab that published the first coupling of microLC and SWATH-MS, in a preprint in 2016 (https://doi.org/10.1101/073478), eventually published in 2018 (Vowinckel et al. 2018). Another important study was by Bian et al. 2020 (reference 16) in this section, which shows micro-flow can overcome nanoflow sample limitation with moderate loss of sensitivity.
Another important study we quote is by Bruderer et al., Analysis of 1508 plasma samples by capillary-flow data-independent acquisition profiles proteomics of weight loss and maintenance. Mol. Cell Proteom. 18, 1242-1254 (2019), which, next to our own manuscript by Zelezniak et al., Cell Systems, 2018, are among the first two studies that used a combination of microLC and DIA for processing large proteomic sample numbers. These references have been included in the manuscript.
2.2 P.3 paragraph 3 'Moreover, the increased sensitivity of DIA methods has facilitated applications in large-scale proteomics, including system-biology studies in various model organisms, disease states, and species [5-9]' Include Ref 4 where improved sensitivity of DIA was demonstrated (at proteomic breadth.)
We thank the reviewer for the suggestion, and included the reference.
3. Are the text and figures clear and accurate?
3.1 Text and Figures need to be edited for typos, language, and clarity/accuracy.
We carefully checked and amended all figures and the text as suggested.
1) Abstract 'Zeno SWATH increases protein identification in complex samples
5- to 10-fold when compared to current SWATH acquisition methods on the same instrument' – At no point this is shown, a drop of required input amounts by 5-8-fold (increase in sensitivity) is shown by the data, not a multiplication of protein identification rates.
We apologise for the placative language and poor phrasing; obviously the increase in ID numbers can not be used as numeric value to describe a fold-gain in sensitivity. Please see the point major comment 5.2 above. We now use the precise numbers throughout the text.
2) P. 4 paragraph 3: Use terms 'consensus' or 'shared' identifications or similar to refer to the proteins identified in all 3 replicates, rather than 'reproducible' when discussing the reproducibility of peptide and protein quantification (as contrast to reproducibility of identification).
We thank the reviewer for the comment. We rephrased accordingly.
3) P.3 paragraph 2 'selects and fragments multiple charge ions' -> multiply charged (?)
We rephrased the sentence to make it more clear.
“In bottom-up proteomics, data-dependent acquisition (DDA) selects precursors from full-scan MS1 mass spectrum for fragmentation, generating tandem (MS/MS) mass spectra that can be matched to spectra in a database.”
4) P. 4 p. 1 'leading to under-detection' please clarify (leading to partial ion usage and limited sensitivity?)
We changed the wording to make it more clear.
“leading to a proportion of ions not being able to be detected, resulting in low sensitivity.”
5) P. 6 paragraph 3 'The gain in identification number of Zeno SWATH versus SWATH is mostly explained by an increased dynamic range: i.e. more low-abundance proteins are detected' – Reformulate/clarify: Is increased dynamic range of identifications against external quantities an explanation or perhaps simply the increased sensitivity with improved duty cycle?
We added a correlation plot (Figure 4) and rephrased this section to make it easier understood.
“improved duty cycle with Zeno SWATH increases intensity at MS/MS level and enhances the identification rate and quantification precision of lower abundance proteins (protein quantity in reference less than 10E3), thereby enlarging/widening the concentration range of analyte identification.”
6) Term 'active gradient' unclear. An inactive gradient is isocratic flow. Omit 'active'. Isocratic/other portions are overhead.
We rephrased the respective term, and instead of active gradient used gradient time versus ‘total runtime’, which includes washing and equilibration steps.
7) Figure 1 panel a) iteration scheme a-d) is redundant with the rest of the figure; use alternative iter scheme within panel a). Panel a) is further contains illegibly small fonts and should be edited for legibility
We thank the reviewer for the suggestion. We separated Figure 1a into a larger figure (Supplementary Figure 1 in the revised manuscript) for better visualisation and more clear explanation.
8) Revisit y-axis labels. Example: Figure 1f) 'Precursors Identificaiton' -> Precursors identified/Precursor identifications. Correct throughout manuscript.
We thank the reviewer, and have corrected the Figures accordingly.
9) ID bar graphs in all Figures: Cumulative IDs shade of grey is not properly visible, suggest alternate color scheme or add a black color outline to the bars.
We agree, and have added a black outline to make it more visible.
10) Figure 1 e) legend 'along gradient length' -> gradient time / retention time.
We changed the legend (current figure 2b) accordingly.
11) Figure 1 d) too small, trend lines mentioned in text invisible in graph. Boxplots very small.
We changed our figures accordingly.
12) There is three different terms used for the high throughput method (analytical-flow, high-flow, and another one. Please align where possible for clarity (i.e. choose 2 names for the 2 methods throughout the manuscript).
We thank the reviewer for the suggestion, and we unified the term to analytical-flow-rate.
4. Do you have suggestions that would help the authors improve the presentation of their data and conclusions?
4.1 They authors may consider adding a short explanation of the term 'dynamic range coverage of identification' to contrast this from a direct assessment of dynamic range of quantitative values observed in this study.
We agree, and we apologise that apparently a confusion emerges, as both the analytical method as well as the proteome span over a broad dynamic range. We addressed this point by rewording the respective paragraphs; see minor comment 3.1 5.
4.2 2-species controlled experiment: The discrepancy of observed vs true mixing ratios suggests the data were scaled during the analysis which, with these mixture ratios, tends to distort the accuracy (i.e. generates offset of observed from true ratio. That's very likely not a pipetting error on a log scale). In other words, you may want to evaluate the raw quantitative ratios (w/o any normalization/scaling applied) which should be more reflective of true/manual pipetting ratios in light of normalization strategy incompatibily with certain species mix scenarios (compare Supplementary Figure 1 a).
Note to the editor(s): This will not affect the clear benefit of Zenotrap usage demonstrated by the 2-species benchmark as is but can be considered a minor yet recommended improvement (thus here).
We thank the reviewer for the suggestion. Indeed the data normalisation might cause some bias. We also plotted the non-normalised value for LFQbench. Please find the non-normalised values below and also in (Figure 1 —figure supplement 2d,e).
Compared to the LFQ data presented in the manuscript (Figure 1c Zeno SWATH, Supplementary Figure 2a SWATH; using PG.MaxLFQ for calculation, median of human normalised to log2 = 0), the non-normalised LFQbench shown below did not have an obvious worse quantification in both species. Moreover, the confidently quantified proteins in both species (valid ratios with at least two replicates in each of A and B) are exactly the same with or without normalisation.
4.3 The 2-species controlled experiment can reveal more information than currently extracted and I would recommend to show Zeno Swath and Swath xy scatters, including count-scaled density distributions of the observed ratios, side-by side. This would give deeper understanding of the large impact of the Zeno SWATH method. Also, I believe I haven't seen any instrument to date delivering precise quantification over as broad a dynamic range as surmisable from Figure 1d which might be worth wile highlighting.
We thank the reviewer for the suggestion. Indeed the scatter plot turned out to be an excellent visualisation of the increased performance of Zeno SWATH versus SWATH. We created a correlation plot of the protein quantity (PG.MaxLFQ) SWATH and Zeno SWATH data in 2 species (Figure 1d–f). Overall, the proteins quantified in both methods are highly correlated, with an r² of 0.94 (human) and 0.95 (yeast). The plot also reflects that the quantities agree extremely well in the upper three orders of magnitudes of the concentration range, and only become noisier in the one order of magnitude that is close to the LOD in SWATH. We also followed the reviewer’s suggestion and created a scatter plot comparing the ratio between the protein quantity (PG.MaxLFQ) of mixture A & B. In general, Zeno SWATH shows better separation and less interference between two species.
Reviewer #2 (Significance (Required)):
Wang et al. describe a technical advance in ion usage and sensitivity based on an ion-trap device storing and focusing ions for TOF-based bottom-proteomics measurements. The study demonstrates improved sensitivity relative to previous generation instrumentation and also explores the impact of the specific trap device relative to the general improvements of the remaining MS system. The work outlines a route towards high coverage proteomics at very high throughput and robustness, as desirable in clinical proteomics and prospective personalized medicine approaches. While not all sample types of interest are limited to the amounts where the strongest improvements are seen in the presented data, large scale studies across expansive cohorts will likely be rendered more practical and realistic due to reduced instrument contamination at reduced loads and also further applications beyond those discussed in the manuscript will be rendered feasible on the newer generation instrument.
The improved ZenoTOF system and SWATH method follows a series of innovations in the mass spectrometry instrumentation, most notably and related the drastic improvement of ion utilization by storage e.g in a trapped ion mobility device earlier in the ion stream where, beyond an accumulation-based boost of sensitivity, ion mobility as a further biophysical properties is assessed in addition to the conventional m/z, as reviewed recently (doi: 10.1016/j.mcpro.2021.100138.). While these developments culminated and have been targeting low-flow, ultra-high sensitivity applications such as single-cell proteomics, the present study takes a different angle towards higher throughput measurements from significantly larger than single cell, but also significantly lower than historically required sample amounts that were prohibitive to a range of applications that are now easier to accomplish thanks to this and related work of the authors and others.
The presented research appears of broad relevance and interest to the scientific community interested in protein abundance pattern analysis, in particular in larger (clinical) cohorts. Furthermore, the performance metrics on proteomic depth from human cell lysate digests will likely allow researchers with analytical quests other than those exemplified in the manuscript to extrapolate the ZenoTOF and Zeno SWATH suitability for their respective analytical targets.
Reviewer Field of expertise/background:
Quantitative proteomics. DIA mass spectrometry method & algorithm development & heavy usage. Protein Biochemistry. Molecular Biology.
References
Bruderer, Roland, Jan Muntel, Sebastian Müller, Oliver M. Bernhardt, Tejas Gandhi, Ornella Cominetti, Charlotte Macron, et al. 2019. “Analysis of 1508 Plasma Samples by Capillary-Flow Data-Independent Acquisition Profiles Proteomics of Weight Loss and Maintenance.” Molecular & Cellular Proteomics: MCP 18 (6): 1242–54.
Demichev, Vadim, Christoph B. Messner, Spyros I. Vernardis, Kathryn S. Lilley, and Markus Ralser. 2020. “DIA-NN: Neural Networks and Interference Correction Enable Deep Proteome Coverage in High Throughput.” Nature Methods 17 (1): 41–44.
Messner, Christoph B., Vadim Demichev, Daniel Wendisch, Laura Michalick, Matthew White, Anja Freiwald, Kathrin Textoris-Taube, et al. 2020. “Ultra-High-Throughput Clinical Proteomics Reveals Classifiers of COVID-19 Infection.” Cell Systems 11 (1): 11–24.e4.
The UniProt Consortium, Alex Bateman, Maria Jesus Martin, Claire O’Donovan, Michele Magrane, Emanuele Alpi, Ricardo Antunes, et al. 2016. “UniProt: The Universal Protein Knowledgebase.” Nucleic Acids Research 45 (D1): D158–69.
Vowinckel, Jakob, Aleksej Zelezniak, Roland Bruderer, Michael Mülleder, Lukas Reiter, and Markus Ralser. 2018. “Cost-Effective Generation of Precise Label-Free Quantitative Proteomes in High-Throughput by microLC and Data-Independent Acquisition.” Scientific Reports 8 (1): 4346.
https://doi.org/10.7554/eLife.83947.sa2Article and author information
Author details
Funding
Bundesministerium für Bildung und Forschung (031L0220)
- Markus Ralser
Bundesministerium für Bildung und Forschung (161L0221)
- Vadim Demichev
Cancer Research UK (FC001134)
- Markus Ralser
Medical Research Council (FC001134)
- Markus Ralser
Wellcome Trust (200829/Z/16/Z)
- Markus Ralser
European Research Council (ERC-SyG-2020 951475)
- Markus Ralser
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.
Acknowledgements
We thank the Charité Core Facility High Throughput Mass Spectrometry for technical support, and advice in experimental design, sample preparation, data analysis, and interpretation; we thank Fatma Amari, Spyros Vernardis, and Poppy Adkin for data acquisition from TripleTOF 6600 system; Lisa Juliane Kahl and Sreejith J Varma for providing E. coli and C. arietinum seedling stocks; and Hezi Tenenboim as well as our team members for proofreading and commenting on the manuscript. This work was supported by the Ministry of Education and Research (BMBF), as part of the National Research Node ‘Mass Spectrometry in Systems Medicine’ (MSCoreSys), under grant agreements 031L0220 (to MR) and 161L0221 (to VD). Work in the MR lab is further supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001134), the UK Medical Research Council (FC001134), the Wellcome Trust (IA 200829/Z/16/Z), and the European Research Council (ERC-SyG-2020 951475). ZW is a member of the International Max Planck Research School (IMPRS) for Infectious Diseases and Immunology. ZW’s PhD studies are part of the Sonderforschungsbereich (SFB) TRR 186.
Senior and Reviewing Editor
- Benoît Kornmann, University of Oxford, United Kingdom
Version history
- Preprint posted: April 14, 2022 (view preprint)
- Received: October 4, 2022
- Accepted: November 8, 2022
- Version of Record published: November 30, 2022 (version 1)
Copyright
© 2022, Wang et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 2,136
- Page views
-
- 245
- Downloads
-
- 10
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Biochemistry and Chemical Biology
- Cancer Biology
Identification oncogenes is fundamental to revealing the molecular basis of cancer. Here, we found that FOXP2 is overexpressed in human prostate cancer cells and prostate tumors, but its expression is absent in normal prostate epithelial cells and low in benign prostatic hyperplasia. FOXP2 is a FOX transcription factor family member and tightly associated with vocal development. To date, little is known regarding the link of FOXP2 to prostate cancer. We observed that high FOXP2 expression and frequent amplification are significantly associated with high Gleason score. Ectopic expression of FOXP2 induces malignant transformation of mouse NIH3T3 fibroblasts and human prostate epithelial cell RWPE-1. Conversely, FOXP2 knockdown suppresses the proliferation of prostate cancer cells. Transgenic overexpression of FOXP2 in the mouse prostate causes prostatic intraepithelial neoplasia. Overexpression of FOXP2 aberrantly activates oncogenic MET signaling and inhibition of MET signaling effectively reverts the FOXP2-induced oncogenic phenotype. CUT&Tag assay identified FOXP2-binding sites located in MET and its associated gene HGF. Additionally, the novel recurrent FOXP2-CPED1 fusion identified in prostate tumors results in high expression of truncated FOXP2, which exhibit a similar capacity for malignant transformation. Together, our data indicate that FOXP2 is involved in tumorigenicity of prostate.
-
- Biochemistry and Chemical Biology
- Cell Biology
Pancreatic a-cells secrete glucagon, an insulin counter-regulatory peptide hormone critical for the maintenance of glucose homeostasis. Investigation of the function of human a-cells remains a challenge due to the lack of cost-effective purification methods to isolate high-quality a-cells from islets. Here, we use the reaction-based probe diacetylated Zinpyr1 (DA-ZP1) to introduce a novel and simple method for enriching live a-cells from dissociated human islet cells with ~ 95% purity. The a-cells, confirmed by sorting and immunostaining for glucagon, were cultured up to 10 days to form a-pseudoislets. The a-pseudoislets could be maintained in culture without significant loss of viability, and responded to glucose challenge by secreting appropriate levels of glucagon. RNA-sequencing analyses (RNA-seq) revealed that expression levels of key a-cell identity genes were sustained in culture while some of the genes such as DLK1, GSN, SMIM24 were altered in a-pseudoislets in a time-dependent manner. In conclusion, we report a method to sort human primary a-cells with high purity that can be used for downstream analyses such as functional and transcriptional studies.