Mass spectrometry (MS) provides an impressive array of information about the structure, function and interactions of proteins. In recent years, many new developments have been in the field of native MS and these exemplify a new coming of age of this field. In this mini review, we connect the latest methodological and instrumental developments in native MS to the new insights these have enabled. We highlight the prominence of an increasingly common strategy of using hybrid approaches, where multiple MS-based techniques are used in combination, and integrative approaches, where MS is used alongside other techniques such as ion-mobility spectrometry. We also review how the emergence of a native top-down approach, which combines native MS with top-down proteomics into a single experiment, is the pièce de résistance of structural mass spectrometry's coming of age. Finally, we outline key developments that have enabled membrane protein native MS to shift from being extremely challenging to routine, and how this technique is uncovering inaccessible details of membrane protein–lipid interactions.
Mass spectrometry (MS) has evolved in recent years to become a pillar for integrative structural biology approaches. Structural MS provides a plethora of information, from protein primary structure and post-translational modifications (PTMs) to higher-order quaternary structure, protein dynamics and interactions. To achieve this breadth of information, different MS approaches have been separately developed with the proteins of interest in either denaturing or native conditions. In the first case, best known as the field of proteomics, the approach can be either peptide-centric (bottom-up) or protein-centric (top-down). Peptide-centric approaches typically analyse peptides derived from proteins in solution, while protein-centric approaches ionise individual protein subunits intact and generate fragments inside the mass spectrometer through gas-phase activation. In the second case, the protein's interactions are maintained in solution and during transfer to the gas phase of the mass spectrometer, hence the name ‘native’ . This approach preserves non-covalent interactions that are the hallmarks of a protein's structure and function and consequently offers access to insights otherwise lost through the other approaches.
We argue that while MS clearly came of age some time ago for structural biology, we are now entering a new age: membrane proteins that used to be almost intractable to study are now amenable to routine analysis, hybrid approaches are becoming the norm, and high-resolution instruments that enable proteomics and native MS analysis simultaneously on the same instrument are offering new experimental strategies.
Many of the latest advances, especially those concerning hybrid approaches are in the sphere of native MS. Therefore, we have chosen to focus on this area in this mini review. We will start with a brief introduction of native MS and hybrid MS-allied approaches, with an emphasis on the insights these give into protein structure. We will then draw attention to the emergence of native top-down MS as a new and highly exciting development in the field, before describing the advances made in native MS for the study of membrane proteins — the final frontier for this cornerstone approach.
Native mass spectrometry is at the heart of structural mass spectrometry
Native MS is the analysis of biomolecules from their ‘native’ state in solution and during transfer to the gas phase of the mass spectrometer . This is possible through the use of soft ionisation techniques, compatible buffers and careful optimisation of instrument parameters to balance adduction and disruption of structure. The advantage of this approach is that non-covalent interactions of proteins, essential for their fold, structure and interactions with other biomolecules, are preserved from solution throughout MS analysis. This unique capacity places native MS at the heart of structural MS, which is inherently interested in the structure and interactions of biomolecules.
Native MS, therefore, enables the identification of complexes, their exact stoichiometry, binding partners and thermodynamics of binding, complex topology, as well as their real-time solution formation . This approach is neatly complementary with classical proteomics-based MS, whose equally important methodologies are being combined in novel hybrid approaches.
Hybrid and integrative approaches for a wider structural and functional picture
By virtue of its strong complementarity, an increasing use of native MS is in hybrid approaches, where various MS-based techniques are used in combination to elucidate a wider picture of protein structure and function. While native MS informs about stoichiometry, co-occurring assemblies and ligand binding, proteomic approaches help to identify and quantify proteins, their different proteoforms, as well as detailed information about their PTMs (Figure 1). Combining both native MS and proteomic approaches can therefore be of great utility to reveal PTM micro-heterogeneities within biological samples. For example, by using high-resolution native MS, a large number of glyco-proteoforms for different proteins purified from human blood serum were recently detected and quantified [3,4]. Then, by combining this approach with peptide-centric proteomics, the heterogeneous glycosylations and their specific locations could be identified, revealing the presence of novel and previously unreported glycosylation sites. Application of such hybrid MS approaches for the assessment of glycosylation occurrence, number and sites in pharmaceutical products could greatly simplify quality control and stability assessment [5–7].
Overview of the different MS approaches for interrogating the structure and function of proteins.
Some PTMs can be lost using peptide-centric approaches, which means that complementary top-down proteomic approaches that retain those PTMs could be necessary to identify these modifications [8–10]. Recently, a three-tiered hybrid approach combining native MS, top-down as well as bottom-up proteomics was used to identify the stoichiometry of four different ribosomal particles, to determine the sites of their sequence processing and their PTMs . Hybrid approaches can therefore offer a comprehensive and extremely detailed view into the structure of proteins.
Chemical modifications of proteins in solution prior to proteomic analysis can add another structural dimension to hybrid MS-based approaches. For instance, combining cross-linking  with native MS enables the identification of interacting proteins while specifying protein–protein interaction sites and can be used to probe local conformational changes that occur within protein complexes [13–17]. Similarly, the use of hydrogen–deuterium exchange (HDX) to label backbone amides gives insights into protein secondary structure, conformational changes as well as ligand-binding sites [18–25]. Notably, many of these structural MS applications would not have been possible without the development of large-scale analysis software for data interpretation [26,27].
In addition to MS-based hybrid approaches, integrative approaches that combine MS with other techniques are also continuing to develop in combination with native MS. Ion mobility (IM), for instance, reveals the overall shape of protein complexes analysed by native MS and offers an orthogonal level of separation for complex samples [28–30]. This powerful approach uncovers conformational changes that might occur within proteins [31,32] and is also used to assess the stabilising effects of various ligands by generating collision-induced unfolding plots [33–36]. The coupling of IM with MS allows different proteoforms or bound states to be independently assessed, which is more difficult or impossible by other techniques. Recently, exploiting its ability to resolve subtly different structural forms, IM–MS has been used to monitor in-solution sequential unfolding of proteins . Native IM–MS can also be coupled to other hybrid MS approaches, for example, HDX [38,39] or top-down proteomics , giving an additional level of structural and functional information. However, and similarly to hybrid approaches, this is only feasible due to the parallel development of integrative modelling [29,41–43].
The emergence of native top-down
Many outstanding technological and methodological developments have given rise to the ability of native MS and proteomics to yield incredible insights into nearly all levels of protein structural orders. However, the ‘missing brick’ in this MS-based repertoire was the ability to reveal the primary structure information and identify unknown bound ligands within the context of a native MS experiment of a protein complex. In essence, this is the ability to perform top-down experimental approaches but from a native rather than denatured state (Figure 2). The limitation previously restricting the possibility of this native top-down approach was largely instrumental, and has been overcome by recent developments that enable multi-stage MS on native protein complexes.
Schematic of native mass spectrometry and native top-down mass spectrometry experiments.
Using the new extended mass range Orbitrap platform , multi-stage MS was made possible by adding a front-end collisional activation step to disassemble complexes that could then be isolated for further fragmentation with higher-energy collisional dissociation (HCD) . Similarly, in-source activation was also developed for a multi-stage MS approach using ion-mobility Q-ToF instruments [46,47], the latter being historically popular for native MS. In addition, an integrated approach was developed for use on a Fourier-transform ion-cyclotron resonance (FT-ICR) platform. This platform has the advantage of multiple activation methods, which allows for a high sequence coverage, and identification of labile PTMs, as well as interfacial and surface residues [48–52]. These developments now enable the identification, in a single experiment, of PTMs and sequence polymorphism as well as the characterisation of ligand binding within protein complexes.
Proteomics from native complexes
Protein assemblies can be very heterogeneous, and these variations can be related to different functional states arising in response to different cellular conditions. This heterogeneity is often difficult to assess using classical structural approaches that often require homogeneous sample or measure ensemble averages. Top-down and bottom-up proteomics can identify sequence polymorphisms and PTMs; however, these approaches cannot directly relate the various modifications to a specific protein complex. Therefore, an advantage of high-resolution mass spectrometers capable of multi-stage MS is the identification, in a single experiment, of distinct proteoforms and PTMs within protein complexes. This advantage was recently demonstrated by the identification of two mutually exclusive phosphorylation sites within the yeast homo-tetrameric fructose-1,6-biphosphatase 1 (FBP1) that are sensitive to variation in growth conditions .
The various activation methods already developed and used in top-down proteomics have been translated into native top-down applications. For instance, the integration of ultraviolet photodissociation (UVPD) in the HCD cell  has been found to offer higher sequence coverage compared with HCD alone as well as an improved retention of labile PTMs. This approach has been applied to the homodimeric branched-chain amino acid transferase 2, which was dissociated in source, and selected monomers activated either by HCD or by 293 nm photons, yielding an improved sequence coverage for the higher-energy fragmentation by UVPD and enabling the identification of a single amino acid mutation within the enzyme. Alternatively, electron-based activation methods — such as ECD and ETD — can be used to help maintain and thus detect labile PTMs. ECD was recently applied in combination with native MS to identify various isoforms of tau protein and to map its phosphorylation sites . In addition, electron-based activation methods can be of great structural use in native top-down experiments since these methods generate fragment ions correlating to the exposed surface of the studied native proteins .
These new methods have also been applied for the untargeted identification of protein complexes from cell lysates . In this research, 125 endogenous complexes were directly identified from mouse heart and human cancer cell lysates. This would not have been possible without the development of powerful data analysis software  and sample separation methods [57,58]. Increasing the scale of this untargeted identification, a recent study coupled two liquid-phase separation techniques to native top-down, size-exclusion chromatography and capillary zone electrophoresis, to analyse E. coli cell lysates . This strategy identified numerous protein complexes and proteoforms, half of which had not previously been reported. These new approaches are of high interest and have a strong potential to revolutionise interactomics studies.
Identifying ligands and their binding sites
The development of native top-down approaches offers the unique opportunity to identify directly small molecule ligands bound to protein complexes. This will remove any ambiguity that could be present when ligand identification is made using denaturing methods, in which direct evidence of the interaction is lacking. To date, only this approach has been used to identify lipids bound to membrane proteins, as we discuss later.
The difficult task of identifying ligand-binding sites within protein complexes can be uncovered by structural MS through careful choice of the activation technique used during native top-down. For instance, ECD cleaves polypeptide backbone bonds while preserving non-covalently bound ligands. Taking advantage of this action, the binding sites of manganese and cobalt to the intrinsically disordered protein α-synuclein were identified , and the binding site of a molecular tweezer assembly modulator to tau protein was determined . Electron-ionisation dissociation (EID) can also be used to probe ligand-binding sites . This fast activation method  gives higher sequence coverage compared with ECD and offers the advantage of dissociating folded gas-phase protein ions, yielding structural information for protein complexes that might be inaccessible by ECD. Another highly advantageous fast activation technique, UVPD, was recently used to determine ligand-binding sites in native top-down [54,63–66]. This activation method can generate higher sequence coverage, as mentioned above, while retaining non-covalent interactions, as well as identify ligand-binding sites challenging to identify by ECD or EID.
Advances in structural mass spectrometry of membrane proteins
Until recently, the study of soluble proteins by native MS has been easier and more successful than for membrane proteins. The key difference that makes membrane proteins challenging is that they start their journey in the instrument encapsulated in a membrane mimetic. This imposes an additional requirement; the membrane mimetic needs to be removed after ionisation and before detection (Figure 3). Most commonly, this mimetic has been detergent micelles, although some success has been made using different systems [67–71].
Schematic of important processes in native MS of membrane proteins.
The process of removing the protective membrane mimetic requires application of collisional activation, the levels of which require careful balancing to avoid the unwanted dissociation of interacting proteins/ligands. This selective dissociation became easier with the identification of detergents with relatively volatile properties requiring minimal input of collisional activation [72–74]. However, even with these ideal detergents, other parameters need to be carefully controlled, such as the solution-stability of the membrane protein in the favoured detergents [75,76] and the extent of delipidation of the sample as it is extracted from expression-host cell membranes and purified [75,77]. Alternatively, the strategic but very empirical use of micelles of mixed detergent composition is a promising method to balance these factors [74,78]. Notably, and very recently, membrane protein complexes encapsulated in vesicles were detected by native MS, a considerable advance that negates the need in some cases for a mimetic .
The other challenge for membrane protein native MS has been related to the instrument design, specifically to the location where the activation required to strip the membrane mimetic takes place. Ideally, this step should be as close to the source region as possible (Figure 3), so that the same repertoire of experiments is achievable as for soluble proteins. Overall, combinations of appropriate solution-preparation of membrane proteins and instrument modifications to increase the levels of in-source activation have worked to reduce this challenge [44,46,72,77,80,81].
Native MS to characterise membrane protein–ligand binding
A particular advantage of native MS is the ability to resolve different bound states of a protein — no longer is the measurement an average of the whole ensemble: the behaviour of the different bound states can be interrogated separately. In a seminal paper for this new age of membrane protein native MS, Laganowsky et al. showed how the folded states of membrane proteins are selectively stabilised by the binding of individual lipid molecules [73,82]. Of particular interest, the lipids identified to have strong stabilising effects were all associated with functional requirements , structural conformations and changes in activity of the proteins studied. Methodological improvements have also allowed full thermodynamic parameters of binding to be calculated, showing that there is an entropy–enthalpy compensation for the binding of lipids with different chain lengths, and a potential to differentiate lipid-binding sites based on different thermodynamic parameters of binding . Furthermore, allosteric effects on lipid binding were revealed, with the binding of one lipid affecting the binding of a second . These discoveries led to crystal structures of the ammonia channel bound to lipids [73,85], and the identification of a key residue that controls the allosteric-binding effects. These allosteric effects of lipid binding are not constrained to other lipid molecules; using the ammonia channel as a model system, it was shown that lipids affect the coupling of this membrane protein to its soluble protein partner, GlnK, thus suggesting a role of lipids in regulating nitrogen flux . Few techniques can match the ability of MS to investigate the effects of individual lipid-binding events on membrane protein behaviour, which is why this has become so powerful.
These advancements have led to many exciting new insights into membrane protein–ligand and lipid binding. For instance, off-target drug binding of HIV protease inhibitors to the human zinc metalloprotease ZMPSTE24, an intramembrane protease, was recently detected and the undesirable disease-linked inhibition of the protease observed directly . Antibiotic binding to the flippase MurJ was characterised and shown to be influenced by lipid binding . The functional importance of co-purified lipids was shown for the flippase TmrAB , and lipids were also proposed, in part, by native MS to play a role in the structural mechanism of Na+/H+ antiporters, by helping to facilitate conformational changes of the proteins within the membrane . The role of interfacial lipids in stabilising dimeric states of certain membrane proteins has also been revealed by native MS. This behaviour was correlated to the role of lipids in compensating for otherwise weak interfacial strengths between monomers, acting as a metaphorical glue . Lipids were also found to be important for the dimerisation of UapA by similar methods, with the dimeric entity determined to be the functional form of the protein . The development of membrane protein native MS has, therefore, created an unprecedented experimental strategy to explore the effects of individual lipid-binding events.
For different types of membrane proteins, the final frontier has been to find ways in which native MS can be used to study G protein-coupled receptors (GPCRs). These are the largest class of proteins in eukaryotes, and are the target for >40% of current drugs, making understanding their functional biology critically important. Using both the resolution afforded by new instrumentation, and approaches to finding volatile detergent combinations that keep proteins stable and functional, this protein class is now tractable. Ligand binding to the human purinergic receptor P2Y1 was detected directly, showing differentiated binding to proteoforms with different phosphorylation . And, most recently, exploiting the aptitude of native MS to investigate lipid interactions, it was demonstrated that GPCRs are sensitive to lipid binding, specifically the physiologically important lipid phosphatidylinositol-4,5-bisphosphate. This lipid stabilises the active states of GPCRs, and enhances the coupling to G-proteins, with which the GPCRs interact to communicate signalling information . The present study has further confirmed that GPCR regulation is complex, but also showcases how native MS now offers a way to uncover and study these effects in considerable detail.
Native top-down MS to identify membrane protein-bound ligands
The native top-down approach developed for soluble proteins was recently translated for the study of membrane proteins using a modified ion-mobility Q-ToF instrument, referred to as a ‘high-energy MS/MS platform’ [46,77]. By adding an in-source activation step, the energy required to strip the detergent micelle is administered in the front-end of the instrument which allows for subsequent ion selection in the quadrupole prior to activation in the collision cell (Figure 3). This makes it possible to directly identify endogenous membrane protein-bound lipids in a single experiment and avoids possible doubts from other approaches, where lipids detected by native MS bound to membrane proteins are identified using a chromatographic or an MS-based approach analysing the total lipid extract present in the sample [75,90,91]. This conventional method poses difficulties as detergent-purified membrane protein samples contain a large number of empty lipo-micelles and isobaric lipid species. Therefore, adding an in-source activation step that strips the detergent micelle makes it possible to select the membrane protein lipid-bound ion in the quadrupole and subsequently fragment it in the collision cell for lipid identification (Figure 3). Using this high-energy MS/MS platform, the additional mass found bound to the homodimeric leucine transporter could be identified as corresponding to two cardiolipin and six phospholipid species, one and three bound per monomer, respectively . In a recent study of three different class A GPCRs, endogenously bound lipids were also identified in a single experiment, revealing a preferential and specific binding to phosphatidylinositol-4,5-bisphosphate . This is an example of how advances in different areas of structural MS, in native and native top-down approaches, can combine to uncover new details of protein structure and function.
By virtue of the above-mentioned breakthroughs and advancements in instrumentation, MS continues to manifest itself as a key tool for integrative structural biology. Its powerful complementary role was recently epitomised by Snijder et al.  who, through a combination of native MS, cross-linking and cryo-EM, elucidated the dynamic assembly of the cyanobacterial circadian clock. The field has exciting challenges ahead that will further extend the capabilities of MS to investigate protein structure, function and interactions. For instance, the development of a high-resolution native top-down platform for analysing membrane proteins will offer the same level of discoveries that are now attainable for soluble proteins. Native top-down is now an open playground for integrating approaches such as IM, which has so far only been coupled ‘offline’ in a separate experiment [55,60]. Excitingly, the first step towards an ‘online’ coupling has been very recently made by Poltash et al.  who have described the development of a high-resolution native IM–MS using the Orbitrap MS platform. Overall, we expect that MS-based approaches, in parallel with their future developments, will continue to consolidate their powerful role in structural biology. It is clear, therefore, that this new age of MS will be pivotal in discovering the intricacies of protein structure and function.
electron capture dissociation
electron transfer dissociation
Fourier-transform ion-cyclotron resonance
G protein-coupled receptor
higher-energy collisional dissociation
Both authors contributed equally.
The Authors declare that there are no competing interests associated with the manuscript.