key: cord-0313184-5y8tdk8j
authors: Kopicki, Janine-Denise; Saikia, Ankur; Niebling, Stephan; Günther, Christian; Garcia-Alai, Maria; Springer, Sebastian; Uetrecht, Charlotte
title: Mapping the peptide binding groove of MHC class I
date: 2021-08-13
journal: bioRxiv
DOI: 10.1101/2021.08.12.455998
sha: 38b933409309e8cd074c50598bbec68d88e5c964
doc_id: 313184
cord_uid: 5y8tdk8j

An essential element of adaptive immunity is the selective binding of peptide antigens by major histocompatibility complex (MHC) class I proteins and their presentation to cytotoxic T lymphocytes on the cell surface. Using native mass spectrometry, we here analyze the binding of peptides to an empty disulfide-stabilized HLA-A*02:01 molecule. This novel approach allows us to examine the binding properties of diverse peptides. The unique stability of our MHC class I even enables us to determine the binding affinity of complexes, which are suboptimally loaded with truncated or charge-reduced peptides. Notably, a unique erucamide adduct decouples affinity analysis from peptide identity alleviating issues usually attributed to clustering. We discovered that two anchor positions at the binding surface between MHC and peptide can be stabilized independently and further analyze the contribution of other peptidic amino acids on the binding. We propose this as an alternative, likely universally applicable method to artificial prediction tools to estimate the binding strength of peptides to MHC class I complexes quickly and efficiently. This newly described MHC class I-peptide binding affinity quantitation represents a much needed orthogonal, confirmatory approach to existing computational affinity predictions and has the potential to eliminate binding affinity biases and thus accelerate drug discovery in infectious diseases autoimmunity, vaccine design, and cancer immunotherapy. Teaser Native mass spectrometry determines both specific binding properties and affinities of different peptides to an empty disulfide-stabilized HLA-A*02:01 molecule in an unbiased fashion.

two anchor positions at the binding surface between MHC and peptide can be stabilized 23 independently and further analyze the contribution of other peptidic amino acids on the binding. 24 We propose this as an alternative, likely universally applicable method to artificial prediction tools 25 to estimate the binding strength of peptides to MHC class I complexes quickly and efficiently. This 26 newly described MHC class I-peptide binding affinity quantitation represents a much needed 27 orthogonal, confirmatory approach to existing computational affinity predictions and has the 28 potential to eliminate binding affinity biases and thus accelerate drug discovery in infectious 29 diseases autoimmunity, vaccine design, and cancer immunotherapy. GM or GL dipeptide, which is then removed via SEC. Afterwards, full-length peptides can bind into the 108 empty binding groove. Raw (C) and deconvoluted (D) native mass spectra of peptide-free dsA2 recorded at 109 an acceleration voltage of 25 V. The empty dsA2 is the predominant species (yellow). In addition, 110 dsA2/erucamide (turquoise stripes) and dissociated β2m (pink) and heavy chain (green) can be seen. 111 112 In the following, it is examined whether native MS can differentiate the binding of high-affinity 113 and low-affinity peptides by comparing the A2 epitope NV9 from human cytomegalovirus pp65 114 (sequence NLVPMVATV; theoretical dissociation constant, Kd,th = 26 nM predicted by NetMHC 115 (9)) with the irrelevant YF9 (YPNVNIHNF; Kd,th = 27 µM) and with GV9 (GLGGGGGGV; 116 Kd,th = 2.7 µM), a simplified NV9 derivative that retains the anchor residues, L and V. 10 µM dsA2 117 is incubated with 50 µM peptide for ten minutes prior to native MS, where the acceleration voltage 118 across the collision cell is increased incrementally (10, 25, 50, 75, and 100 V) to observe the 119 dissociation behavior. At 75 and 100 V, β2m heavily dissociates from the hc (data not shown), and so in the following, only results for 10 V, 25 V and 50 V are shown. The signal is quantified by 121 determining the area under the curve (AUC) over the entire spectrum for each mass species (Fig. 122 2; raw spectrum in Fig. S2 ). While at 10 V and 25 V, the distribution of the different mass species 123 is very similar, the ratios change significantly at 50 V. This is a frequent observation with the 124 electrospray ionization (ESI) process, where non-covalent hydrophilic bonds such as those between 125 dsA2 and high affinity peptides are retained (16, (27) (28) (29) , but hydrophobic interactions are 126 weakened. By increasing the acceleration voltage, the dissociation of a protein-ligand complex 127 usually does not occur gradually but spontaneously beyond a certain threshold, at which an 128 energetic state is encountered that denatures the complex, in our case between 25 and 50 V. 129 Therefore, the measurements at 10 V are used to calculate the dissociation constants. 130 In the presence of low-affinity YF9, the empty dsA2 molecule (43,733 ± 4 Da) generates the 131 highest signal (56 ± 3 % at 10 V; Fig. 2A ). Most of the remainder of dsA2 carries only the 132 erucamide adduct (44.071 ± 5 Da; 39 ± 2 %). There is very little dsA2/YF9 complex (4 ± 2 %), and 133 because of the very low binding affinity of YF9, it can be assumed that this signal does not 134 correspond to real binding events but rather to an artifact of the electrospray process known as non-135 specific clustering (17, (30) (31) (32) (33) . Assuming that the other tested peptides cluster to the same extent, 136 all native MS data is therefore corrected for the clustering determined with YF9. Since the data 137 measured for the peptides of interest are therefore netted with YF9, no affinities are calculated for 138 this control peptide. Corresponding raw data for the negative control are listed in the supplement 139 ( Table S 2) . 140 For NV9, in contrast, very efficient binding is observed, with 64 ± 3 % for dsA2/NV9 at 10 V and 141 40 ± 4 % at 50 V. Here, the dsA2/erucamide complex is completely absent, which suggests that 142 erucamide is displaced by NV9. Erucamide either binds into the peptide groove, or it binds 143 elsewhere and is displaced by a conformational change caused by peptide binding. A small amount 144 of another mass species (45,624 ± 4 Da) that corresponds to dsA2 with two molecules of NV9 145 (dsA2/NV9/NV9) with an abundance of 4 ± 4 % (10 V) and 1 ± 2 % at 50 V is also observed. The 146 latter is likely the result of unspecific clustering, as the abundance correlates with the intensity of 147 the first binding event and is similar to the intensity of YF9 binding. Within NV9, the leucine in 148 the second position and the C-terminal valine bind into the B and F pocket of the binding groove, 149 respectively (12, 34, 35) . The proportion of dsA2 occupied with GV9 at 10 V (43 ± 2 %, and 150 1.5 ± 0.4 % for dsA2/GV9/GV9) is significantly lower than the proportion of dsA2/NV9. This 151 clearly shows that the minimal binding motif cannot support the same affinity as NV9, suggesting 152 that other amino acids contribute significantly to the binding. At 50 V, however, the abundance of 153 the dsA2/pep complex is the same for GV9 and NV9. This indicates that the strong B and F pocket 154 side chain interactions together with the binding of the termini determine pMHC gas phase stability. for the respective mass species at 10 V, 25 V, and 50 V. The mean value of the AUC in absence or presence 160 of the different peptides (protein-peptide ratio 1:5) from at least three independent measurements is depicted 161 along with error bars that represent the corresponding standard deviation. "dsA2" (yellow bars) corresponds 162 to the empty HLA-A*02:01(Y84C/A139C) disulfide mutant complex, "dsA2/pep" (blue bars) to dsA2 163 bound to one peptide, "dsA2/pep/pep" to dsA2 bound to two molecules of this peptide (purple bars), and 164 "dsA2/erucamide" to dsA2 bound to erucamide (turquoise-striped bars), respectively. The negative control 165 YF9 barely associates with dsA2, whereas the positive control shows a high proportion of dsA2/NV9, 166 indicating high affinity. GV9, which contains only the two anchor residues Leu-2 and Val-9 of NV9, still 167 shows a high affinity, and at 50 V, their dsA2/pep proportions are very similar, showing that their gas phase 168 stability is comparable. (B) Representative charge-deconvoluted spectra of the distinct protein and protein-169 peptide complex species recorded at 25 V. Pink and dark green correspond to the free β2m domain and heavy 170 chain, respectively. Light green corresponds to a free heavy chain still attached to a peptide. The different 171 complexes are assigned in yellow (empty dsA2), turquoise stripes (dsA2/erucamide), blue (dsA2/pep) and 172 purple (dsA2/pep/pep).

Despite very efficient binding, the obtained Kd for NV9 is only 8 ± 2 µM, while NV9 has nM 175 affinity (36). Hence, a fully occupied peptide binding pocket (predicted > 99 %) is expected in our 176 measurements ( Fig. 2A) . Protein denaturation due to storage or other physicochemical stress (data 177 not shown) is excluded, and thus ISD, which occurs in the source region of the mass spectrometer 178 and results in reduction of the dsA2/NV9 complex, is considered next. Here, lower cone voltages 179 can increase the proportion of the occupied complex (37, 38). Indeed, the occupancy of the peptide binding groove with both the peptide and erucamide is higher at lower cone voltages. This linear 181 relationship is most evident with dsA2 alone, to which erucamide is the only binding partner (Fig.   182 3A). The data suggest that at a cone voltage of ≤ 36 V, which is unfeasible experimentally, dsA2 is 183 100 % occupied with erucamide. This observation allows us to use erucamide as a reference species 184 for peptide binding measurements, assuming that in solution, all free dsA2 protein is initially bound 185 to erucamide as suggested by the zero cone voltage extrapolation, and that the peptide replaces it 186 in the binding region. From the erucamide-bound fraction measured at a cone voltage of 150 V in 187 native MS, the fraction of dsA2/erucamide corresponding to the fraction of dsA2 not bound to 188 peptide at the end of the binding reaction can be recalculated using the correction factor of 2.2 189 originating from the linear function's slope. From this, the fraction of dsA2/peptide is inferred, 190 resulting in an apparent Kd that reflects the in-solution environment (see the Materials and 191 Methods). This is highly advantageous, since for individual pMHCs, the ISD is naturally influenced 192 by peptide size and sequence, precluding compensation, while the ISD of the MHC-erucamide 193 complex is invariant and peptide-independent. Therefore, for each peptide both a dissociation 194 constant for the high cone voltage (150 V) based on clustering-corrected MHC-peptide signal 195 (Kd,high) and another for the theoretical low cone voltage of 36 V based on the MHC-erucamide 196 signal (Kd,low) is described (Fig. 3A) . While the experimentally determined Kd,high is overestimated 197 due to ISD, the Kd,low is a more reliable approximation. binding (protein-peptide ratio: 1:10, Fig. 4B ). Conformational stabilization corresponds to an 242 increase in the midpoint of thermal denaturation (Tm) above that of the empty dsA2, which is 243 measured by tryptophan nanoscale differential scanning fluorimetry (nDSF (40, 42)) to be 244 35.7 ± 0.6 °C. While the negative control YF9 clearly shows no (∆Tm = 1 ± 1 K), the positive 245 control NV9 shows a high degree of stabilization (∆Tm = 23.4 ± 0.6 K) in agreement with published 246 data (12). The other peptides show excellent correlation between affinity and stabilization ability. 247 All peptides identified as strong binders by their apparent dissociation constant exhibit a ∆Tm of at 248 least 6.4 ± 0.6 K (GV9), while the ∆Tm for the low-affinity peptides ranges between ≈ 0 K and ≈ 3 K 249 ( Fig. 4B Neutralizing the terminal charges of the peptide reduces binding efficiency 305 Next, the influence of the charged termini of the peptide upon the binding affinity is analyzed. For 306 this purpose, three variants of NV9 are designed: Ac-NV9-NH2 has an acetylated N-terminus and 307 an amidated C-terminus, whereas Ac-NV9 and NV9-NH2 each carry only one of these 308 modifications. For the dsA2/pep fraction at 10 V, Ac-NV9 and NV9-NH2 show only a small 309 difference to the unmodified NV9, resulting in comparable apparent Kd values (Kd,low = 0.11 ± 0.05 µM and 0.004 ± 0.003 µM). Further, no increase of dsA2/erucamide is 311 observed (Fig. 5A) . For both peptides, the protein-ligand complex is still stable at 50 V (Fig. S5 ).

For Ac-NV9, the proportion of the occupancy is even higher than for NV9 itself (57 ± 2 % and 313 8 ± 1 % vs. 40 ± 4 % and 1 ± 2 %, Table S 2) . Remarkably, all modified peptides have an increased 314 double occupancy. For the previously discussed peptides, dsA2/pep/pep is significantly lower with 315 the result that correction for non-specific clustering reduces it to a value below threshold. For Ac-316 NV9, this effect is significant, since even at 50 V the proportion of dsA2/pep/pep (purple bars) is 317 still 8 ± 1 %. However, the stabilization effect on dsA2 in a 1:10 thermal denaturation approach is 318 rather moderate for Ac-NV9 (∆Tm = 7.5 ± 0.8 K), while it is very strong for NV9-NH2 319 (∆Tm = 14.6 ± 0.6 K), indicating that the N-terminus has more relevance for tight peptide binding 320 than the C-terminus. measurements is depicted along with error bars that represent the corresponding standard deviation. "dsA2" 326 (yellow bars) corresponds to the empty complex, "dsA2/pep" (blue bars) to dsA2 bound to one peptide, 327 "dsA2/pep/pep" to dsA2 bound to two molecules of this certain peptide (purple bars) and "dsA2/erucamide" 328 to dsA2 bound to erucamide (turquoise-striped bars) respectively. By modifying only one terminus (Ac-329 NV9 and NV9-NH2), the affinity of the peptide to dsA2 changes only marginally, but if the charges on both 330 termini are neutralized (Ac-NV9-NH2), the peptide binding is greatly reduced. The nDSF measurements likewise show that the stabilization effect with ∆Tm = 2.7 ± 0.6 K is rather 346 weak. Although NV9-NH2 is far better rated within our scheme (Fig. 4A) Therefore, none of the examined tetra-to heptapeptides are strong binders. In direct comparison, 375 VV7, which still contains the C-terminal anchor residue Val-9, performs worse than NA7, which 376 in contrast carries Leu-2. NA7 displaces more of the erucamide, and at 50 V, significantly more 377 NA7 than VV7 is observed (Fig. S5) . Thus, Leu-2 appears to possess more binding strength than 378 Val-9. Curiously enough, the N-terminally truncated pentapeptide MV5 performs significantly 379 better than the corresponding hexa-(PV6) and heptapeptides (VV7). The increased proportion of 380 dsA2/MV5, which is even more evident at 50 V (Fig. S5 ) and the simultaneous occurrence of dsA2/MV5/MV5 (purple bars, Fig. 6 ) strongly suggest that this peptide occupies an additional 382 binding site within the peptide groove. Since this effect is not apparent for the smaller and therefore 383 less spatially restricted VV4, the terminal methionine seems to be the decisive factor here. Since no 384 increased dual occupancy is observed for NM5 and since binding at 50 V is weaker in direct 385 comparison to MV5, it seems that the terminal methionine as such is not determinant on its own, 386 but much more the relative position within the peptide. Considering the tetrapeptides, again the N-387 terminally truncated peptide, VV4, performs better than the C-terminally truncated NP4 at all 388 acceleration voltages, which points to the positioning of the erucamide within the peptide pocket. 389 While NP4 carries only the anchor peptide Leu-2, it seems that VV4 has a higher chance to bind to 390 the dsA2 peptide groove due to its two terminal valines. "dsA2/pep/pep" to dsA2 bound to two molecules of this certain peptide (purple bars), "dsA2/pep2" to dsA2 440 bound to another peptide when two different peptides where present (light green), "dsA2/pep2/pep2" to 441 dsA2 bound to two molecules of the second peptide (dark green), "dsA2/pep1/pep2" to dsA2 bound to one 442 molecule of each of both peptides (red bars) and "dsA2/erucamide" to dsA2 bound to the erucamide 443 (turquoise-striped bars) respectively. able to synergistically stabilize the complex. The melting temperature of dsA2 exposed to only one 453 of the truncated NV9 variants (Fig. S3) While only wildtype pMHC were studied by native MS so far (46), we have recently demonstrated 470 that peptides added to empty disulfide-stabilized class I molecules can be detected as well (14). 471 This study shows that this method, which is fast and amenable to high-throughput approaches, can 472 be used to measure MHC class I peptide binding affinities, and it is used to map the contributions 473 of parts of the peptide to high-affinity binding. 474 Our data affirms the key interactions between dsA2 and its high-affinity ligand NV9 that were 475 described previously (47) (48) (49) . It is therefore scarcely surprising that Leu-2 and the C-terminal Val- (14), the F pocket can be occupied by GM with methionine as anchor residue. 495 Methionine in the C-terminal position is not preferred but only tolerated by HLA-A*02:01, as 496 reported before (50). Compared to MV5, which can bind with either valines, the second 497 pentapeptide, NM5, therefore shows the same strong binding behavior in our experiments. For 498 larger peptides like NV9, which were found to be double bound to dsA2, it is more likely that the 499 signal originates from nonspecific clustering of NV9 with the abundant dsA2/NV9 complex rather 500 than from the actual binding of two peptides to the dsA2 peptide binding groove due to spatial 501 reasons. 52), and amidation, providing the same functional group as an additional amino acid, provide more realistic and biologically relevant modifications. Since N-terminal acetylation is one of the most 527 prevalent modifications in pro-and eukaryotes (51, 53), the observed occupancy for Ac-NV9 in the 528 native MS experiment, which is even higher than for the non-modified NV9 might even be a real 529 effect. Nevertheless, this work confirms that one of the termini needs to be intact to form strong 530 binding to either of the outer pockets. In turn, N-terminal peptides from acetylated proteins will 531 likely not bind as decapeptides with a C-terminal overhang. Moreover, this explains why 532 undecapeptides with putative overhang on both sides have reduced affinity, as neither of the termini 533 is intact. According to our data, the short, truncated peptides, which are unable to bind using both 534 termini due to spatial limitations, can also stabilize their respective binding pocket individually, 535 which once again supports our thesis concerning the independent stabilization of A and F pocket. 536 Binding of lipophilic small molecules into the class I binding groove has been shown several times, Instruments) were pulled into closed capillaries in a two-step program using a squared box filament 610 (2.5 mm × 2.5 mm) within a micropipette puller (P-1000, Sutter Instruments). The capillaries were 611 then gold-coated using a sputter coater (5.0 × 10 -2 mbar, 30.0 mA, 100 s, 3 runs to vacuum limit 612 3.0 × 10 -2 mbar argon, distance of plate holder: 5 cm; CCU-010, safematic). Capillaries were 613 opened directly on the sample cone of the mass spectrometer. In regular MS mode, spectra were 614 recorded at a capillary voltage of 1.45 kV and a cone voltage of 150 V. Protein species with 615 quaternary structure were assigned by MS/MS analysis. These experiments were carried out using 616 argon as collision gas (1.2 × 10 -2 mbar). The acceleration voltage ranged from 10 V to 100 V. 617 Comparability of results was ensured as MS quadrupole profiles and pusher settings were kept 618 constant in all measurements. The instrument settings of the mass spectrometer were optimized for 619 non-denaturing conditions. A spectrum of cesium iodide (25 g/L) was recorded on the same day of 620 the particular measurement to calibrate the data. (39)).

Guidelines for the production and quality control of synthetic peptide 659 vaccines, Annex 1

Mass spectrometry-based antigen discovery for cancer 662 immunotherapy

Human 664 endogenous retroviruses and their implication for immunotherapeutics of cancer

The three-dimensional structure of peptide-MHC complexes

The Protein Data Bank

Specificity 670 pockets for the side chains of peptide antigens in HLA-Aw68

Importance of peptide amino and carboxyl termini to the stability 672 of MHC class I molecules

FLEXamers: A Double Tag for Universal Generation of Versatile 674

NetMHCpan-4.0: Improved Peptide-MHC Class I Interaction Predictions 676

Integrating Eluted Ligand and Peptide Binding Affinity Data

Dipeptides promote folding and peptide binding of MHC class I molecules

Peptide-independent stabilization of MHC class I molecules breaches cellular 681 quality control

Empty peptide-receptive MHC class I molecules for efficient detection of 683 antigen-specific T cells

High-throughput peptide-MHC complex generation and kinetic screenings 685 of TCRs with peptide-receptive HLA-A*02:01 molecules

Structures of peptide-free and partially loaded MHC class I molecules 687 reveal mechanisms of peptide selection

Advances in Virus Research

Quantitative determination of noncovalent binding 691 interactions using automated nanoelectrospray mass spectrometry

Method for distinguishing specific from 694 nonspecific protein-ligand complexes in nanoelectrospray ionization mass spectrometry

Quantitative determination 697 of noncovalent binding interactions using soft ionization mass spectrometry

Hydrophobic protein-ligand 700 interactions preserved in the gas phase

Determination of ligand-protein dissociation constants by 702 electrospray mass spectrometry-based diffusion measurements

Which electrospray-based 705 ionization method best reflects protein-ligand interactions found in solution? a comparison 706 of ESI, nanoESI, and ESSI for the determination of dissociation constants with mass 707 spectrometry

Software Requirements for the Analysis and Interpretation of Native 709

Ion Mobility Mass Spectrometry Data

Interferences and contaminants 711 encountered in modern mass spectrometry

Bioactive contaminants leach from disposable laboratory 713 plasticware

Extraction, identification, and functional characterization of a bioactive 715 substance from automated compound-handling plastic tips

UCSF Chimera--a visualization system for exploratory research and 718 analysis

Mass spectrometry of protein complexes: from 720 origins to applications

Application of electrospray mass spectrometry in probing protein-protein 722 and protein-ligand noncovalent interactions

Mass spectrometry of protein-ligand complexes: enhanced gas-724 phase stability of ribonuclease-nucleotide complexes

Nonspecific protein-carbohydrate complexes 727 produced by nanoelectrospray ionization. Factors influencing their formation and stability

A deconvolution method for the separation of 730 specific versus nonspecific interactions in noncovalent protein-ligand complexes analyzed 731 by ESI-FT-ICR mass spectrometry

A Comparative Study for Sonic Spray

Electrospray Ionization Methods to Determine Noncovalent Protein-Ligand Interactions

Protein Secondary Structure Affects Glycan Clustering in Native Mass 736

Antigenic peptide binding by class I and class II histocompatibility 738 proteins

Structural bases for the affinity-driven selection of a public TCR against a 740 dominant human cytomegalovirus epitope

The Immune Epitope Database (IEDB): 2018 update

Using accurate mass 744 electrospray ionization-time-of-flight mass spectrometry with in-source collision-induced 745 dissociation to sequence peptide mixtures

Protein-protein binding affinities in solution determined by 747 electrospray mass spectrometry

FoldAffinity: binding affinities from nDSF experiments

Not all empty MHC class I molecules are molten globules: tryptophan 751 fluorescence reveals a two-step mechanism of thermal denaturation

Peptide-MHC I complex stability measured by nanoscale differential 754 scanning fluorimetry reveals molecular mechanism of thermal denaturation

Peptide-MHC I complex stability measured by nanoscale differential 757 scanning fluorimetry reveals molecular mechanism of thermal denaturation

Emerging principles for the 760 recognition of peptide antigens by MHC class I molecules

Conformational flexibility of the MHC class I alpha1-alpha2 762 domain in peptide bound and free states: a molecular dynamics simulation study

Class I-765 restricted cytotoxic T cell recognition of split peptide ligands

Determining the topology of virus assembly intermediates using ion mobility spectrometry-769 mass spectrometry

Allele-specific motifs 771 revealed by sequencing of self-peptides eluted from MHC molecules

Sequence motifs important for peptide binding to the human MHC class 774 I molecule, HLA-A2

Scheme for ranking potential HLA-A2 binding 776 peptides based on independent binding of individual peptide side-chains

Majority of peptides binding HLA-A*0201 with high affinity crossreact 779 with other A2-supertype molecules

Control of protein degradation by N-783 terminal acetylation and the N-end rule pathway

N-terminal acetylation: an essential protein modification emerges as 786 an important regulator of stress responses

The role of HLA genes in pharmacogenomics: 789 unravelling HLA associated adverse drug reactions

Crystal structures of lysophospholipid-bound MHC class I molecules

Crystal structure of the N-myristoylated lipopeptide-bound MHC class I 793 complex

Erucamide from Radish Leaves Has an Inhibitory Effect Against 795

Acetylcholinesterase and Prevents Memory Deficit Induced by Trimethyltin

13-Docosenamide release by bacteria in 798 response to glucose during growth-fluorescein quenching and clinical application

Rapid 801 identification of MHC class I-restricted antigens relevant to autoimmune diabetes using 802 retrogenic T cells

High-Throughput Identification of MHC Class I Binding Peptides Using 804 an Ultradense Peptide Array

Real-time, high-throughput measurements 807 of peptide-MHC-I dissociation using a scintillation proximity assay

Advances in Protein Chemistry and Structural Biology, R. Donev

More than one reason to rethink the use of 812 peptides in vaccine design

Thermal stability of self-assembled peptide vaccine materials

Improving the performance of a quadrupole time-of-flight 816 instrument for macromolecular mass spectrometry

mMass data miner: an open source 819 alternative for mass spectrometric data analysis

Bayesian deconvolution of mass and ion mobility spectra: from binary 822 interactions to polydisperse ensembles