Formation of Neutral Peptide Aggregates as Studied by Mass‐Selective IR Action Spectroscopy

Abstract The spontaneous aggregation of proteins and peptides is widely studied owing to its relation to neurodegenerative diseases. To understand the underlying principles of peptide aggregation, elucidation of structure and structural changes upon their formation is key. This level of detail can be obtained by studying the peptide self‐assembly in the gas phase. Structural characterization of aggregates is mainly done on charged species, as adding charges is an intrinsic part of the technique to bring molecules into the gas phase. Studying neutral peptide aggregates will complement the existing picture. These studies are restricted to dimers due to experimental limitations. Herein, we present advances in laser desorption molecular beam spectroscopy to form neutral peptide aggregates consisting of up to 14 monomeric peptides in the gas phase. The combination of this technique with IR–UV spectroscopy allowed us to select each aggregate by size and subsequently characterize its structure.

Self-assembly of proteins and peptides into distinct ordered structures has triggered interest in awide variety of research fields ranging from biology [1] to thed evelopment of smart materials [2] and life-like nanosystems. [3] Amongst these examples,p rotein and peptide aggregation is mostly studied in relation to human disorders,s uch as the neurodegenerative diseases Alzheimersa nd ParkinsonsD isease,w here the formation of aggregates into amyloid fibrils is observed. [4] Multiple techniques have been used to provide insight into the aggregation process,s uch as transmission electron microscopy, [5] cryo-EM, [6] circular dichroism, [7] NMR spectroscopy, [8] and FTIR spectroscopy. [9] These methods were able to reveal the structures of full-grown fibrils,b ut lack the ability to determine the structures forming in the toxic early stage, such as dimers,trimers,a nd small oligomers,owing to ah igh degree of complexity and heterogeneity.
Mass spectrometry coupled with techniques such as infrared spectroscopy and/or ion mobility can shed light on the structure of these oligomers. [10] Electrospray ionization is commonly used to bring these aggregates as charged species into the mass spectrometer.T he initial steps of aggregate formation have been probed by mass spectrometry combined with IR spectroscopy and ion mobility,f ocusing on the amyloid-prone peptides VEALYL (insulin) and NFGAIL (human islet amyloid polypeptide). [11] An increase in b-sheet character in the IR signatures,i ndicative of the formation of fibrils,was observed in the more extended oligomers.Charge plays an important role,s ince it can induce unfolding and consequently alter the structure of the peptide.S tudying neutral peptides will bring insight into non-charge-driven structural preferences upon aggregation. However,t he formation of aggregates of neutral peptides is not straightforward. There are only ahandful examples of studies on peptide dimers,inw hich either ah eatable source or laser desorption was used to bring them into the gas phase.T he groups of Gerhards and Rijs/Gaigeot have studied the structure of neutral (Ac-Phe-OMe) 2 using IR-UV spectroscopy. [12] They concluded that these peptide dimers are oriented in an antiparallel b-sheet structure.S noek and co-workers investigated al arger peptide dimer:t he neutral and charged amyloidogenic peptide sequence VQIVYK. [13] They showed that the charge on the lysine prevents the formation of extended peptide backbone segments as observed in amyloid fibrils.I nc ontrast, the studied neutral dimer did form a bsheet structure in the gas phase.R ecently,w es tudied the competition between intra-and intermolecular interactions upon dimer formation of alanine-containing peptides. [14] Here,for the first time,wedemonstrate the formation of stable higher-order clusters of neutral peptides in the gas phase using laser desorption (the experiment is described in the Experimental Section and Supporting Information). The peptide used in our experiments is the capped dipeptide Ac-Ala-Ala-OBn (BioMatik, > 95 %purity,molecular weight of 292.31 amu). Its mass spectrum ( Figure 1) was recorded at the resonant ionization wavelength of the trimer (37 460 cm À1 ) and shows aprogression of peaks with increments of m/z 292 associated with the growth of (Ac-Ala-Ala-OBn) n=1-14 clusters.A ll aggregates have their absorption maximum around the trimer wavelength, except for the slightly blue-shifted dimer (see the Supporting Information). Therefore,i ti s possible to measure the IR spectra of all clusters simultaneously.T he formation and detection of these large aggregates is achieved by adjusting multiple experimental parameters,such as sample preparation (ratio sample/carbon black), gas pulse (density,l ength), desorption and excitation laser power and delay,excitation laser frequency, and time-of-flight settings.T he most dominant effect is obtained in the desorption step,r esulting in the enhancement of aggregate formation. Thedesorption laser (Figure 2a,picture not drawn to scale) is mildly focused on agraphite sample bar,resulting in ab eam diameter of about 1mma nd power of 1mJ/pulse. Thesample bar can be varied in height with respect to the axis of the molecular beam (blue arrow). Theo pening of the nozzle (grey) has ad iameter of 0.5 mm (black arrow). After desorption, which takes place inside the gas pulse,t he molecules are seeded in as upersonic argon expansion.
Thep osition of the sample with respect to the nozzle opening is key for optimal performance. [15] Changing the height of the sample bar changed the signal as shown in Figure 2b.T he normalized signal per multimer was monitored as function of the vertical distance of the desorption surface relative to the center of the nozzle.The optimal height for monomers was slightly below the center of the molecular beam axis,s uch that the opening of the nozzle was partly blocked. [16] Lowering the sample bar (negative x-axis values), that is,increasing the distance between the sample bar and the center of the molecular beam, significantly increased the amount of observed multimers.The multimer signal appeared to continue for about 0.1 mm below the nozzle orifice, whereas the monomer signal vanished below the nozzle opening ( Figure 2b;s ee also the Supporting Information). Theo bserved effects are not mass-dependent, that is,m onomers of any mass up to m/z 1500 were typically found just below the center of the nozzle orifice.I nc ontrast, clusters were always found near the lower edge of the nozzle opening, as can be seen in Figure 2c,w hich shows the difference between the rising edge at the full width at half maximum (FWHM) of each aggregate with respect to the monomer (see inset). Each point represents the average of multiple measurements taken at three different UV excitation wavelengths. Thezero position coincides with the center and 0.25 mm with the lower edge of the nozzle opening.
Thel onger pathway for the multimers allows for more collisions with present peptide molecules and argon atoms; molecules have more time to cool down and spend more time in ad enser environment undergoing collisions required for aggregate formation. This effect becomes more pronounced until the sample bar is moved too far away from the molecular beam path, resulting in asudden drop in the signal. Thehigh efficiency of the cooling and clustering process explains why all monomers that are present when partly blocking the nozzle are completely consumed in the aggregation process when the sample bar is lowered.
Them ass-selective IR spectra of the Ac-Ala-Ala-OBn clusters with n = 2-7 were measured in the fingerprint region from 1000 to 1800 cm À1 (Figure 3a). Theregion between 1600 and 1800 cm À1 comprises the C = Os tretching vibrations and shows two distinctive peaks.T he smaller peak above 1700 cm À1 results from the C=Og roup of the ester moiety present at the OBn-cap,whereas the peak between 1600 and 1700 cm À1 originates from the peptide C=Ogroups (amide I).  Every peptide monomer has one ester C=Oand two peptide C=Og roups.T he amide II band (NH bend) appears as ab road feature located between 1490 and 1570 cm À1 .O ther distinctive features include the large bands around 1200 cm À1 , originating from backbone and amide III motions.F igure 3b,c displays the normalized IR spectra for aggregates with n = 2-7 in the amide Iand II region, respectively.T hese regions are sensitive to hydrogen bonding,thereby providing structural details on intra-and intermolecular interactions.
Quantum chemical calculations were performed to structurally assign the experimental IR spectra. Fort he dimer, eight structural families were determined based on their hydrogen-bond patterns (see the Supporting Information). Thec alculated IR spectrum of the lowest energy structure shows the best agreement with the experimental spectrum. Thedimer was therefore assigned to this structure,composed of ap arallel b-sheet (Figure 4a). It is formed by two intermolecular hydrogen bonds,w hereby the backbones of the two peptides are aligned in ap arallel orientation. The ester C=Ogroups are not involved in any hydrogen bonding. Theindividual peptides present in the parallel b-sheet dimer retain their monomeric structures:t he linear and the g-turn conformer are both present in the dimer,inwhich their weak intramolecular hydrogen bonds (C5 hydrogen bond and NHp bond) are replaced by stronger intermolecular hydrogen bonds. [14,17] Thec alculated IR spectrum of an antiparallel dimer is also compared with the recorded IR spectrum in Figure 4a.This structure,like the parallel dimer,has no ester C=Og roups involved in hydrogen bonding;h owever, was discarded as ar esult of am ismatch in the amide II region together with its high energy.
In the IR spectra of the aggregates with n > 2, the peak between 1700 and 1800 cm À1 ,originating from the ester C=O moiety,d id not significantly change position or width. This indicates that for the higher-order clusters the ester C=O groups are also not involved in any hydrogen-bonding interaction. In contrast, the position of the amide I ( Figure 3b;see also Figure SI.7B in the Supporting Information) and amide II bands (Figure 3c)s hifted to the red and blue, respectively,indicative of an increased number of,and hence stronger,h ydrogen bonds. [18] These shifts became smaller as higher-order clusters were formed.
Calculations showed that the low-energy structures of the trimer were exclusively b-sheet-type structures.O ther structural families (globular) were at least 25 kJ mol À1 ,b ut often over 40 kJ mol À1 higher in energy.The experimental spectrum only matches to structures in which all three monomers are attached through b-sheet intermolecular hydrogen bonds (see the Supporting Information). In particular,conformers which arise from the attachment of the third peptide to the previously assigned dimeric structure P-T-1 are in good agreement. Thet hird peptide can aggregate on both sides of the dimer:a tt he C7 g-turn intramolecularly or weaker C5 intramolecularly hydrogen bonded side.O nt he basis of energetics and spatial orientation, it is expected that the weaker C5 hydrogen bond is broken to favor the formation of strong intermolecular b-sheet interactions. [14] Tw o b-sheetcontaining structures remain possible:a na ll-parallel structure and astructure in which the third monomer is attached in an antiparallel fashion to the dimer.T he added monomer peptide adopts either the observed linear conformer (for the all-parallel structure; Figure 4b,left) or the g-turn conformer for the antiparallel structure (Figure 4b,r ight). Thea ntiparallel structure shows ab etter overlap in the region around 1200 cm À1 ;h owever, the all-parallel structure has considerably (17 kJ mol À1 )l ower energy and reproduces the double peak in the amide II region better.O verall, the structure of  . Experimental (black) and calculated IR spectra (coloured) of the a) dimer,b)trimer,and c) tetramer.T he spectra of the all-parallel C12 families are shown on the left in blue;the best-matchingspectra for combinations with antiparallel b-sheets are shown on the right. The relative zero-point corrected energy for each aggregate is shown in the right-hand top corner together with the name given to the aggregate. Tetramer energies are not shown as they are calculated with different basis sets (see Table SI.1 in the SupportingInformation).
the trimer can be confidently assigned to an all-b-sheet structure,i nw hich the peptides are stacked on top of each other. Energetics and spectral features,such as the shoulder in the amide Iregion, point to the presence of parallel b-sheets.
Thet heoretical IR spectra of all-b-sheet tetramer structures show good agreement with the experiment (see the Supporting Information). Additionally,the energetics showed that, as was observed for the dimer and the trimer,t he allparallel structure has the lowest energy.T he spectra of two possible structures are shown in Figure 4c:a na ll-parallel conformer (6-31G* basis set) and amixed (anti)parallel lowenergy conformer (6-31 + G* basis set). Both conformers can originate from the addition of am onomer to one of the two discussed trimers,o rb ym erging two dimers.S pectrally,t he overlap for both structures is good in the amide Ir egion, although the all-parallel conformer was calculated on alower level of theory and therefore has as maller peak around 1800 cm À1 (see the Supporting Information). Ther emaining parts of the spectra show reasonable overlap for both b-sheet conformers.I nl ine with the dimer and trimer, the tetramer structure has all peptides stacked in b-sheets on top of each other. On the basis of energetics and structures of the dimer and trimer,t he formation of an all-parallel tetramer is expected.
Theo bserved red and blue shifts of the amide Ia nd II bands,respectively,inthe IR spectra continue for the higherorder clusters (n > 4). Forthe assignment of these clusters,we relied on size-related shifts in the spectrum and comparisons with the smaller aggregates rather than on DFT calculations (see the Supporting Information). As imilar trend for aggregation onwards from the tetramer is expected, in which the peptides aggregate into stacked b-sheet-containing structures,whereby the weak intramolecular hydrogen bonds are broken to favor stronger intersheet hydrogen bonds.T he observed red shift in the amide Ir egion for the larger aggregates continues towards the amide Ib and of the solidstate FTIR spectrum of Ac-Ala-Ala-OBn. This spectrum was recorded using aKBr pellet (see Supporting Information) and shows ac lear signature at 1629 cm À1 ,c orresponding to parallel b-sheets. [19] Thed ifference between the condensed phase and the gas phase is ascribed to interactions with the environment, in this case the other peptides in the matrix. [20] In conclusion, we have shown that it is possible to make aggregates of neutral peptides in the gas phase by allowing the laser-desorbed peptides more time to cool down in ad enser environment. Thea mount of neutral aggregated peptides (n = 14) is unprecedented, as peptide dimers were the highestorder clusters created previously.I R-UV ion dip spectroscopy and quantum chemical calculations allowed us to assign the structures of the aggregates (n = 2-4) to predominantly parallel b-sheets,w ith the peptides stacked on top of each other. Higher-order clusters (up to n = 7) showed ashift in the amide Ib and towards known b-sheet signatures observed in the solid phase,thus indicating that the structural preferences observed in the gas phase are related to those in the bulk. The presented results pave the way to complementary studies on neutral biologically active peptides,s uch as the hydrophobic amyloidogenic peptides.

Experimental Section
Theneutral peptide aggregateswere formed in alaser desorption molecularbeam time-of-flight mass spectrometer. [15] Conformational and mass-selective IR spectra were recorded by IR-UV ion dip spectroscopy using the free-electron laser FELIX. Thea mber force field and Gaussian09w ere used for our calculations. See the Supporting Information for details of the experiment.