Conformational Preferences of Isolated ... - ACS Publications

Jan 4, 2019 - Vasyl Yatsyna , Ranim Mallat , Tim Gorn , Michael Schmitt , Raimund Feifel , Anouk M. Rijs , Vitali Zhaunerchyk. Physical Chemistry Chem...
0 downloads 0 Views 3MB Size
Subscriber access provided by - Access paid by the | UCSB Libraries

A: Spectroscopy, Molecular Structure, and Quantum Chemistry

Conformational Preferences of Isolated Glycylglycine (Gly-Gly) Investigated with IRMPD-VUV Action Spectroscopy and Advanced Computational Approaches Vasyl Yatsyna, Ranim Mallat, Tim Gorn, Michael Schmitt, Raimund Feifel, Anouk M. Rijs, and Vitali Zhaunerchyk J. Phys. Chem. A, Just Accepted Manuscript • DOI: 10.1021/acs.jpca.8b10881 • Publication Date (Web): 04 Jan 2019 Downloaded from http://pubs.acs.org on January 4, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Conformational Preferences of Isolated Glycylglycine (Gly-Gly) Investigated with IRMPD-VUV Action Spectroscopy and Advanced Computational Approaches †,‡

Vasyl Yatsyna,



Ranim Mallat,

Tim Gorn,

Anouk M. Rijs,

∗,‡





Michael Schmitt,

and Vitali Zhaunerchyk

Raimund Feifel,



∗,†

†Department of Physics, University of Gothenburg, 412 96 Gothenburg, Sweden ‡Radboud University, Institute for Molecules and Materials, FELIX Laboratory, Toernoovield 7-c, 6525 ED Nijmegen, The Netherlands

¶Heinrich-Heine-Universität,

Institut für Physikalische Chemie I, D-40225 Düsseldorf, Germany

E-mail: [email protected]; [email protected]

Abstract

1

In this paper we report the results of gas-phase

Gas-phase infrared (IR) spectroscopy of small

IR spectroscopy of neutral glycylglycine (Gly−1 Gly) in the 700 − 1850 cm frequency range.

and medium-sized peptides

A combination of laser desorption, jet-cooling,

level.

and IR multiple-photon dissociation vacuum-

techniques to transfer intact biomolecules into

ultraviolet (IRMPD-VUV) action spectroscopy

the gas-phase, IR spectroscopy became a rou-

is employed, together with extensive quantum-

tine tool for studies of biomolecular systems of

chemical calculations that assist in the analy-

increasing size and complexity.

sis of the experimental data.

Introduction 18

into protein folding forces

allows an insight

912

at the atomic

Thanks to the development of various

13

It has the ad-

As a result, we

vantage of a high sensitivity of molecular vi-

determined that the most favorable conformer

brations to the backbone structure and non-

in the low-temperature environment of the su-

covalent interactions such as hydrogen bonding.

personic jet is the nearly-planar structure with

Since low sample density in the gas phase re-

two C5 hydrogen bonding interactions. Calcu-

stricts direct IR absorption measurements, ac-

lations clearly show that this conformer is fa-

tion spectroscopy techniques are employed in-

vored due to its exibility (considerable entropy

stead.

stabilization), as well as ecient conformer re-

absorption is detected through a change in ion-

laxation processes in the jet.

ization or uorescence yield, electron detach-

To gain more

In action spectroscopy the IR photon

14

understanding into the relative stability of the

ment or fragmentation.

lowest-energy Gly-Gly conformers, the relative

UV spectroscopy is a powerful method that al-

strength of hydrogen bonding and steric inter-

lows recording IR spectra of individual molec-

actions is analyzed using the non-covalent in-

ular conformers.

teractions (NCI) approach.

only be applied to molecules with an aromatic

Double-resonance IR-

This method, however, can

UV chromophore. Among the twenty standard amino acids only three have such a UV chromophore, which signicantly limits the scope

ACS Paragon Plus Environment 1

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 20

of the peptides which can be studied. For the

lated in our experimental conditions, and com-

spectroscopy of charged species several tech-

pare them to theoretical calculations in order

niques are available, such as tagging (messen-

to identify the most abundant conformers. Al-

ger) technique and IR multiple-photon disso-

though the recent rotational spectroscopy study

ciation (IRMPD), whereas IR spectroscopy of

of Gly-Gly observed three dierent conform-

neutral isolated peptides without an aromatic

ers,

ring remains challenging. One way to circum-

leaves some open questions, for instance, about

vent this problem is to attach an aromatic moi-

the relative stability of the observed folded and

ety, either chemically

planar structures, especially in view of favor-

interactions.

16

15

or through non-covalent

While this approach has found

26

their abundances were not reported. This

able entropic stabilization

28,29

in the latter. In

wide application, the attached aromatic group

this respect, the current work will complement

can modify the intrinsic properties of the pep-

the previous rotational spectroscopy results.

tide by introducing extra non-covalent interac-

Secondly, by studying Gly-Gly we intend to

tions which may also alter its conformational

explore the use of the IRMPD-VUV approach

landscape.

for structural analysis of exible peptides with

Recently, combination

we

have

of

demonstrated

IRMPD

that

spectroscopy

a large degree of conformational heterogene-

the a

ity. Thirdly, we aim to shed some light on the

molecular beam with single-photon vacuum-

strength and importance of non-covalent inter-

ultraviolet (VUV) ionization can be used to

actions that are present in the Gly-Gly con-

record vibrational spectra of cooled neutral

formers. For this purpose, we will employ elec-

molecules of arbitrary structure, without an

tron density topological analysis within atoms

aromatic chromophore in particular.

in molecules (AIM)

17

of

26

In the

current work we apply this approach to IR

3032

and non-covalent inter-

actions (NCI) approaches.

3335

spectroscopy of chromophore-free glycyl-glycine (Gly-Gly).

An

initial

challenge

in

obtain-

2

ing gas-phase vibrational spectra for Gly-Gly

Methods

and other thermolabile molecules is how to

2.1

volatilize an intact molecule. For the studies of charged Gly-Gly species electrospray ionization

The experiments were performed at the free

methods were employed yielding protonated or deprotonated molecular targets.

1822

electron laser FELIX Laboratory at Radboud

In this

University, Nijmegen, The Netherlands, using

case, however, one has to disentangle the ef-

the laser desorption molecular beam set-up.

fects of the proton on the peptide's intrinsic

from Sigma Aldrich (purity of 98%). The sam-

Laser ablation method is commonly

ple was mixed with carbon black and applied

used for rotational spectroscopy of neutral ther-

onto a graphite bar that was placed in front

molabile molecules, though in many cases its

of a pulsed valve nozzle (0.5 mm orice, Jor-

application to peptides and amino-acids is as-

dan TOF Products, Inc) in a source vacuum

sociated with signicant photofragmentation eects.

2325

chamber. The sample molecules were laser des-

Only recently a rotational spec-

orbed from the bar using a Nd:YAG laser (1064

troscopy study of laser-ablated Gly-Gly has been reported.

26

nm, 5 ns, 1−2 mJ/pulse, Polaris II, New Wave

In our work, we employ a

Research) and then seeded into a supersonic-

soft laser desorption process, which imparts minimal

internal

molecules,

14,27

energy

into

the

14

A Gly-Gly sample was obtained commercially

structural properties and its vibrational signatures.

Experiment and data analysis

jet created by a pulsed valve with argon (3

desorbed

bar) as a carrier gas.

followed by collisional cooling

The supersonic-jet ex-

pansion allowed the desorbed molecules to ef-

in an argon supersonic jet.

ciently cool down to the rovibrational ground

The purpose of the current work is three-

state.

fold. Firstly, we aim to obtain the IR signatures

The molecules in the central and cold-

est part of the expansion region were colli-

of the ensemble of Gly-Gly conformers popu-

ACS Paragon Plus Environment 2

Page 3 of 20

mated by a skimmer and delivered to the in-

Ar+

teraction chamber, where the molecular beam

Intensity, arb. units

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

was crossed with a pulsed IR beam from FELIX free electron laser (FEL). The IR frequency of FELIX was scanned in the range of 700-1850 −1 cm , corresponding to IR pulse energies of 30-80 mJ. The exposure time to the IR FEL radiation within the interaction region is di-

IR off IR on

30

30.2

132 132.2 132.4

rectly proportional to the mass of the carrier gas; the heavier the gas the longer the exposure time.

20 30 40 50 60 70 80 90 100 110 120 130 140

mass, amu

With a typical FELIX macropulse

duration of 6-8

µs,

the sample molecules car-

Figure 1:

ried by argon had an estimated residence time of 3

µs

desorbed Gly-Gly molecules under IR on (red

in the interaction region. If a molecu-

curve) and IR o  (black curve) conditions.

lar vibrational transition was resonant with the

The inset plots show how the fragment peak

scanned IR light frequency, the molecule dis-

at 30 amu and parent peak at 132 amu change

sociated via InfraRed Multiple Photon Dissociation (IRMPD) process.

VUV (10.5 eV) mass spectra of laser-

their intensities when the IR FELIX wavelength

At the end of the

is resonant with a vibrational transition of Gly-

FELIX macropulse, the IRMPD fragments and

Gly.

parent molecules were ionized by 118 nm VUV laser pulses of ∼2-3 ns duration (hν =10.49 eV, th 3+ 9 harmonic of a Nd :YAG laser, Spectra-

Assuming a constant interaction volume and

The created ions were detected by

a constant time prole of macro-pulses upon

means of a reectron-type time-of-ight mass

scanning the FELIX wavelength, a relative ab-

spectrometer. By measuring the IRMPD frag-

sorption cross section versus IR frequency can

mentation yield versus FELIX wavelength, the

be expressed through the experimentally mea-

vibrational IRMPD-VUV spectrum of Gly-Gly

sured quantities as

Physics).

was obtained. The FELIX laser was operated

σrel

at 5 Hz, while the desorption laser, the pulsed valve and the VUV ionizing laser were operated at 10 Hz.

This enabled measuring refer-



 Poff , Pon

(1)

Nph denotes the number of photons in the pulse, P denotes the intensity of the parent

where

ence mass spectra without IR radiation at every

IR

second VUV laser pulse (see Fig. 1).

ion, and the on and o  subscripts refer to

The VUV laser light was generated by focus3+ ing 355 nm light (third harmonic of Nd :YAG

the measurements performed with FELIX being on and o, respectively. If all IRMPD frag-

laser) into the center of a gas cell containing a 1:10 mixture of Xe and Ar gases.

1 ln = Nph

ments

The result-

equal

ing 118 nm laser beam was refocused into the

P Fi,on are detected, P to Fi,on + Pon , and

the

Poff

signal is

Eq. 1 can be ex-

pressed as

molecular interaction region through a MgF2 lens. The residual 355 nm beam remained di-

σrel

vergent and was not focused in the interaction

1 = ln Nph

P

 Fi,on + Pon . Pon

(2)

region since the focal distance of the MgF2 lens was larger for 355 nm. To further separate the

Eq. 2 is more preferable over Eq. 1 since it only

118 nm and 355 nm beams we used o-axis

involves quantities measured with FELIX be-

alignment through the MgF2 lens.

The esti-

ing on, and hence allows eliminating shot-to-

mated pulse energy of the generated VUV ra-

shot uctuations of the laser desorption source.

≤ 1 µJ, based on the conversion 10−4 in the phase-matched Xe-Ar

To implement Eq. 2 into the current measure-

diation was eciency of mixture.

36,37

ments, it should be corrected for a fraction of fragments produced solely by VUV in the ab-

ACS Paragon Plus Environment 3

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

H11

sence of the FELIX pulses. The corrected Eq.

Page 4 of 20

H12 N1

2 takes the form

O'

1 2 1 , 1

σrel

1 ln = Nph

P

 (1 − α)

Fi,on + Pon Pon



C1

(3)

1

ω1

α

C2

T 2

H''

θTC

N O

N2

C1 where

C

H2

O''

C2

H

denotes the fraction of the parent

O1

molecules that undergo fragmentation when the molecules are not irradiated with FELIX. This

Figure 2:

fraction can be found as

the cis-trans backbone conguration at the N-

P

Fi,off α= P Fi,off + Poff We note that Eq.

Structure of the Gly-Gly molecule in

and C-termini, respectively (ψ1 close to

(4)

φ2

close to

180°).



and

3 neglects the multiple-

photon nature of the IRMPD process.

pulsion in genuine PM6.

Nev-

The resulting struc-

ertheless, it is valid for our measurements as-

tures were sorted according to their relative en-

suming that IRMPD signals are only observed

ergies, and duplicate structures were discarded.

when the IR frequency is resonant with a molec-

The geometries of obtained unique structures

ular vibration and the IR intensity is suciently

(≈80) with relative energies below 1600 kJ/mol

high to dissociate the molecule.

were

optimized

with

higher

accuracy

DFT

methods using the quantum chemistry package

The Gly-Gly raw spectra were obtained by

Gaussian 16.

42

The selection of DFT function-

scanning the FELIX IR laser in the range of −1 700-1850 cm , and measuring the relative in-

als was based on their performance for sim-

tensities of the IRMPD fragments and the par-

ilar dipeptide systems.

ent molecule. The nal spectrum (Fig. 4) was

ods (ω B97X-D/6-311++G(d,p),

obtained by applying Eq. 3 to the data and tak-

31+G(d),

45

43

The selected meth-

44

M06-2X/6-

MP2/aug-cc-pVTZ and B3LYP-

46,47

ing into account the FELIX power prole across

D3BJ/N07D

the measured range.

As there were no varia-

paring theoretical predictions with the Gly-

(see Eq. 4) when scanning FELIX,

Gly experimental rotational constants available

tions in

α

a constant

α

elsewhere

factor was used in the analysis

of each FELIX scan.

≈ 1.6

(see Table S1 of Supporting Infor-

mation (Supp.

The mean width of the

measured IR bands (FWHM of

26

) were further assessed by com-

Info)).

determined that the

% of the

As a result, we have

ω B97X-D

functional with

center frequency) was slightly larger than the

the 6-311++G(d,p) basis set is the most reli-

bandwidth of the FELIX laser (0.5−1 %). This

able for structure calculations of the Gly-Gly

is related to the multiple-photon excitation in-

peptide. The lowest energy Gly-Gly conformers found

volved in the IRMPD process that is known to broaden the vibrational transitions.

38

with our conformational search are shown in Fig. 3.

2.2

Their energies and dihedral angles,

denoted in Fig. 2, are presented in Table 1.

Theoretical treatment

We found that the backbone of most of the conformers deviates from planarity by at least

To explore the conformational space of GlyGly a random search method was applied.

39

10°, with one exception of

A

sc-cis-trans

struc-

large number of structures (≈657000) with ran-

ture that is nearly planar.

dom torsional angles for rotatable bonds was

non-planar conformer exists as a pair of enan-

generated and optimized with a semi-empirical

tiomeric structures with identical energy.

quantum-chemical method PM6-D3H4

40

Moreover, each As

imple-

a result, all structures shown in Fig. 3 have a

program package.

degeneracy factor of 2, except for the planar

PM6-D3H4 corrects for dispersion, hydrogen

sc-cis-trans

conformer which has a degener-

bonding, and the too small hydrogen steric re-

acy factor of 1.

mented in the MOPAC

41

ACS Paragon Plus Environment 4

Page 5 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

C5-trans

C5g-(s)

trans-trans

C5g-(a)

C5g+(s)

C5C7-

C5C7+

sc-cis-trans

g-trans-trans

C5g+(a)

−1 Figure 3: The structures of the lowest energy conformers of Gly-Gly (∆(E +ZP E) below 840 cm , 10 kJ/mol) optimized with the

ω B97X-D/6-311++G(d,p)

method.

Table 1: Zero-point energy corrected electronic energies ∆(E + ZP E), Gibbs free energies ∆G at 400 K, and dihedral angles of the lowest-energy conformers of Gly-Gly shown in Fig. 3, as obtained from the ωB97X-D/6-311++G(d,p) optimized geometries. See Fig. 2 for the description of the dihedral angles. Energies are calculated using the G4MP2 method based on the ωB97X-D/6-311++G(d,p) geometries and harmonic frequencies, and are presented with respect to the C5-trans energies in cm−1 (values in kJ/mol are shown in brackets). ∆(E + ZP E), cm−1 (kJ/mol) 0 (0) 90 (1.1) 179 (2.1) 22 (0.26) 12.5 (0.15) 408 (4.9) 592 (7.1) 617 (7.4) 664 (7.9) 755 (9.0)

C5-trans C5g-(s) C5g+(s) C5C7+ C5C7trans-trans C5g-(a) C5g+(a) g-trans-trans sc-cis-trans

∆G at 400 K,

cm−1 (kJ/mol) 0 (0) 177 (2.1) 199 (2.4) 485 (5.8) 474 (5.7) 378 (4.5) 613 (7.3) 809 (9.7) 876 (10.5) 506 (6.1)

ψ1

ω1

φ2

ψT

θT

φ11

φ21

-13.2 -20.1 -10.5 -10.2 -11.4 -170.1 -18.0 -11.2 -151.5 -0.01

-179.1 170.3 -169.7 175.9 -173.1 -177.5 175.1 -172.6 -176.8 -180.00

-173.5 -91.8 90.3 74.1 -74.6 179.5 -92.6 89.6 -179.1 -179.99

-0.9 -2.0 2.6 123.1 -122.8 -0.1 -176.5 177.6 0.5 0.00

179.7 -179.7 179.7 1.0 -1.2 180.0 178.5 -178.5 -179.8 180.00

150.4 156.0 147.9 142.5 144.0 -52.4 152.6 149.4 -155.1 -64.1

-88.4 -83.4 -91.1 -96.1 -94.7 63.6 -86.9 -89.9 -35.3 64.1

For the assignment of the experimental IR bands,

spectra

of

all

low-energy

of Gly-Gly were calculated.

formers upon laser desorption and subsequent

structures

supersonic-jet expansion is well reected by

For this pur-

a relative Gibbs free energy distribution at

pose, harmonic frequencies and intensities were

300 − 500

calculated with the B3LYP-D3BJ functional

signments based on electronic energies with

(Becke, three-parameter, Lee-Yang-Parr with

zero-point energy (ZPE) corrections should be

D3 version of Grimme's dispersion with Becke-

avoided.

Johnson damping

Gibbs

46

) and the N07D basis set.

47

K, while the conformational as-

2729,5254

free

To calculate highly-accurate

energies

we

used

methods

modication described in Ref. 43:

structures using Vibrational second-order Per-

single-point energy calculations were based on

turbation

molecular geometries and frequency calcula-

48

The

B3LYP-

the

ω B97X-D/6-

harmonic and anharmonic frequency calcula-

311++G(d,p) and B3LYP-D3BJ/N07D meth-

tions as it provides high accuracy within rea-

ods.

sonable computational time.

more reliable for geometry and frequency cal-

These methods were chosen as they are

culations

43

ACS Paragon Plus Environment 5

both

a

tions

It was shown that the population of con-

using

with

high-level

D3BJ/N07D method was chosen both for the

4851

obtained

G4MP2

composite

56

calculated for the most important low-energy (VPT2).

and

the

Anharmonic frequencies and intensities were

Theory

CBS-QB3

55

than the default methods used in

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

CBS-QB3 and G4MP2. The

lowest-energy

Page 6 of 20

critical points (RCPs) associated with ring-type

vibrational

modes

are

structures correspond to second order saddle points (λ1

approximation,

which can have an adverse

the AIM analysis, useful for the identication

eect on the accuracy of calculated Gibbs free

of strong hydrogen bonds, we have also ap-

energies.

plied the NCI (non-covalent interactions) ap-

57,58

Treatment of some low-energy tor-

33

< 0, λ2

λ3 > 0). 32

known to be poorly described by the harmonic

and

Following

sional vibrations as hindered rotations provides

proach.

a better description.

Therefore, in our analy-

ration of NCI index isosurfaces

sis we applied an automatic procedure available

colormap from sign(λ2 )·ρ data.

in Gaussian

dex corresponds to the reduced density gradient

59

59

to identify hindered rotations,

and corrected the calculated free energy using

(RDG):

This approach is based on the explo-

where

r) =

RDG(

hind hind (∆Evib + ∆Evib ) − T (∆Svib + ∆Svib )

hind ∆Evib

and

hind ∆Svib

where

are the hindered ro-

tor approximation corrections (McClurg

60

lled with a The NCI in-

33

the following relation:

∆Gvib =

33

ρ(r)

|∇ρ(r)| 1 , 2 1/3 2(3π ) ρ(r)4/3

is an electron density map that can

be obtained from DFT calculations.

) to

In our

the vibrational thermal energy and entropy, re-

work we have used the electron density data

spectively.

from the

ω B97X-D/6-311++G(d,p)

method as

this method has been found to provide the

For the calculation of relaxation barriers between the conformers, a transitions state (TS)

best description for Gly-Gly geometries.

search was performed using the QST2 and

is also worth noting that any other DFT or

QST3 methods

as implemented in Gaussian.

wavefunction-based methods with sucient ba-

In the cases where QST2 failed to nd a TS,

sis sets are expected to be equally reliable for

the QST3 method was applied, for which an

the NCI analysis.

61

It

35

initial guess for the TS structure was obtained from the relaxed PES scan connecting two conformers.

3

To obtain the barriers to relaxation,

accurate electronic and Gibbs free energy calcu-

3.1

lations were performed by applying the G4MP2

Results and Discussion IR spectroscopy: experiment,

method to the optimized TS structures and

theory and Gly-Gly conform-

the corresponding conformers.

ers

The geome-

try optimization and hindered rotor frequency analysis were performed with the

The

ω B97X-D/6-

experimental

IRMPD

spectrum

of

jet-

cooled Gly-Gly is presented in Fig. 4 (the black

311++G(d,p) method. The identication of hydrogen bonding and

trace in each plot). The IRMPD-VUV method

other weak non-covalent interactions was per-

is not conformer-selective so all conformers suf-

formed using the electron density

topol-

ciently populated in the molecular beam con-

ogy analysis with the help of the Multiwfn soft-

tributed to the measured spectrum. For com-

ware.

In the rst step, the search of bond

parison, Fig. 4 presents the calculated scaled

critical points (BCPs) within the AIM (atoms

harmonic frequency spectra of several charac-

in molecules)

theory was undertaken. BCPs

teristic lowest-energy Gly-Gly conformers (the

are found as saddle points in the electron den-

62

30

ρ(r)

representing a minimum in the

colored bar spectra). As vibrational transitions −1 at frequencies below 1100 cm have lower in-

bonding direction and maximum in all other di-

tensities, the two dierent frequency ranges are

rections.

Mathematically, this is described by

plotted in panels (a) and (b) of Fig. 4. Com-

λ1 , λ2 , λ3 of the secondorder derivative matrix of ρ(r). At BCPs, λ1 and λ2 are negative while λ3 is positive. Ring

parison with other conformers that are higher

sity map

ρ(r),

32

the sign of eigenvalues

in energy or have similar spectra can be found in Fig. S1 of the Supp.

ACS Paragon Plus Environment 6

Info.

It can be seen

Page 7 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

from Fig. 4 that most of the experimental fea-

spectral features. For example, the calculated

tures are readily reproduced by the calculated Moreover, the mea-

frequency of the NH2 scissoring vibration for trans-trans (1638 cm−1 ) is blue-shifted with −1 respect to the experiment (1620 cm ). More-

sured spectrum shows considerable dierences

over, the CC stretching vibration has a more

with the calculated spectra of all other low-

delocalized character in the

energy conformers (see also Fig. S1 of the Supp.

former than in

Info). This implies that the measured spectrum

quency for

spectrum of

C5-trans

(red bars), which is the

lowest energy conformer.

C5-trans .

is dominated by

Such a conclusion

trans-trans conC5-trans . Its calculated fretrans-trans (843 cm−1 ) does not

agree well with the experimental value of 863 −1 −1 cm , whereas C5-trans (857 cm ) does. The

is elaborated in more detail in what follows. The calculated spectra of all conformers pre-

NH2 wagging (inversion) vibration involves a

dict that the strongest experimental bands at −1 1782 and 1704 cm correspond to the C=O

larger extent of NC and CC stretching in

stretching fundamental transitions. The latter

quency (895 cm

trans-

trans ,

resulting in a higher calculated fre−1 ) that does not match the −1 observed transition at 795 cm . In summary,

band, known as Amide I, originates from amide (peptide link) C=O stretching. Its position −1 at 1704 cm indicates that the the observed

the calculated peak positions and intensities of

spectrum is dominated by conformers with free

ment with the experimental spectrum whereas

amide C=O group, implying no hydrogen bond-

other conformers do not match to various de-

ing at this site. This allows us to conclude that

grees.

the

C5C7-

and

C5C7+

the

conformers, that have

C5-trans

conformer are in favorable agree-

Predominance of the

C5-trans

conformer

a hydrogen bonded amide C=O group, do not

can be further veried by accounting for an-

contribute signicantly to the spectrum.

harmonicity using the VPT2 approach (Fig. 5).

The

question

as

to

what

extent

planar

The VPT2 calculation for the

trans-trans ) and non-planar (C5g-(s) and C5g-(a) ) conformers are popu-

C5-trans

con-

(C5-trans and

former

lated can be assessed using spectral features in

band positions, yielding a mean absolute error −1 (MAE) as low as 6 cm . The calculated over-

accurately

reproduces

experimental

the so-called ngerprint frequency range below −1 1400 cm . For example, the measured strong −1 peak at 1393 cm reects the COH bending

tone and combination bands (Fig. 5) also elim-

vibration of planar Gly-Gly conformers. In the case of non-planar conformers the calculated

For example, a relatively strong band observed −1 at 1224 cm was not predicted by the har-

transition for this vibration is red-shifted and

monic frequency calculations.

signicantly reduced in intensity with respect

a strong band at this frequency corresponding

to the planar counterparts. The normal mode

to the overtone of OH wagging+COH bending

analysis also shows that two modes including

vibration.

both CN and CO stretching vibrations, ob−1 served at 1153 and 1120 cm , are delocalized

of the OH wagging vibration

inate some discrepancies between the scaled harmonic frequency and experimental spectra.

VPT2 predicts

The reported high anharmonicity

50,63

supports this

assignment. The observation of a single conformer,

and are sensitive to the planarity of Gly-Gly.

C5-

The intensities and the frequency spacing be-

trans ,

tween the calculated transitions of these modes

disagrees with theoretical abundances of the

in the non-planar conformers dier from the ex-

lowest-energy conformers (Fig. 6(a)), calculated

periment. The same holds for the CC stretching

from Gibbs free energy data at dierent tem-

vibrations.

peratures.

Based on these considerations, we

C5g-(s)

and

C5g-(a) ,

are not so

500 K,

spectroscopy

study

27

where the calculated population of conformer is less than 40%.

In or-

C5-trans

der to explain this seeming contradiction, con-

can be distinguished by other

formational relaxation processes occurring in

The remaining planar conformers

trans-trans

IRMPD

The temperature of laser-desorbed

C5-trans

important under our experimental conditions. and

our

molecules is expected to be in the range of 350-

can conclude that the non-planar conformers, such as

in

ACS Paragon Plus Environment 7

The Journal of Physical Chemistry NH2 CC wagg. str.

100

CN str. (N term.)

CH2 wagg.

CN,CO str. Am III

400

Experimental C5-trans

0 100

0 400

Experimental C5C7-

NH2 Am I C=O str. (COOH) sciss.

E ( G) [cm-1]

Experimental C5-trans

0 (0)

13 (474) Experimental C5C7-

200

50 0 100

0 400

Experimental C5g-(s)

Experimental C5g-(s)

90 (177)

Experimental trans-trans

408 (378)

Experimental C5g-(a)

592 (613)

200

50 0 100

0 400

Experimental trans-trans

200

50 0 100

0 400

Experimental C5g-(a)

200

50 0 700

0 800

900

1000 1100

1100 1200 1300 1400 1500 1600 1700 1800

wavenumber, cm -1

wavenumber, cm -1

(b)

(a) Figure 4:

Am II

200

50

Relative intensity, arb. un.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 20

The experimental IRMPD spectra (the black trace in each plot) compared with the

scaled harmonic frequency spectra (B3LYP-D3BJ/N07D, scaling factor 0.976) of several lowest energy conformers of Gly-Gly (C5-trans is shown in red,

trans

in purple, and

C5g-(a)

C5C7- in green, C5g-(s)

in blue,

trans-

in orange, respectively). Note that the y-axis range in panel (a) is

shorter than in panel (b) to facilitate comparison between experimental and calculated vibrational transitions. Abbreviations: Am - Amide, str. - stretching, wagg. - wagging.

52,64

the supersonic jet need to be taken into acIndeed, it is known that if the barri-

have barriers lower than the critical value −1 of 800 cm . This suggests that conformer re-

ers to interconversion are low, ecient confor-

laxation via collisional cooling can occur along

mational relaxations to the lowest energy con-

these pathways in our molecular beam.

formers take place during supersonic-jet expan-

barriers calculated at higher temperatures, e.g.

sion.

Table 2 presents the calculated bar-

at 300 K were found to be much higher than

rier heights for the possible relaxation path-

those calculated at 15 K, with a single exception

ways of Gly-Gly conformers (see also Fig. S4

of

in Supp. Info). The data were obtained using

cold environment of the supersonic jet is hence

count.

52,6468

∆G

∆Eel ,

relaxation.

The

and Gibbs free ener-

favorable for collisional relaxation which would

at 15 K and 300 K. The barriers es-

not take place at higher temperatures. We also

electronic energies gies

sc-cis-trans →C5-trans

The

timated from

∆G

at 15 K are known to ade-

note that the small negative value of

∆G(15 K)

quately describe relaxation for the experimen-

calculated for pathway I is most likely due to a

tal conditions similar to ours,

and therefore

calculation limitation aecting the intrinsic ac-

can be used to assess the likelihood of relax-

curacy of the Gibbs free energy, which becomes

ation. As illustrated in Table 2, pathways I-VI

apparent when structures with very similar ge-

68

ACS Paragon Plus Environment 8

Page 9 of 20

200

C5-trans C5C7+ C5C7C5g-(s) C5g+(s) trans-trans C5g-(a) C5g+(a) sc-cis-trans g-trans-trans

50

Boltzmann population, %

VPT2 fundamental bands VPT2 overtones+combinations VPT2 total spectrum

180 160 140 Relative intensity

120 100

40

30

20

10

80

VPT2 total spectrum Experiment GlyGly

60

0 100

200

300

40

400 500 600 700 Temperature, K

0

800

800

1000

1200 1400 1600 wavenumber, cm−1

1800

80 C5-trans

70

Figure 5:

trans

900 1000

(a)

20

Anharmonic IR spectra for

C5-

Boltzmann populaion, %

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

conformer calculated with VPT2, com-

pared with the experimental IRMPD spectrum of Gly-Gly. The calculated spectra were convoluted with a Gaussian prole with FWHM of 1.5 % of central frequency to match the experimental conditions.

60 50 40 30 20

trans-trans

10

C5C7

0 100

ometries and energies are compared. According

200

300

to the electronic energy calculation, the barrier

400 500 600 700 Temperature, K

for this pathway clearly exists though it is very −1 low (∆Eel = 98 cm ).

800

900 1000

(b)

Our calculations also suggest that the in-

Figure 6: (a) The theoretical Boltzmann popu-

terconversion processes between the lowest en-

lations of the lowest energy conformers of Gly-

ergy conformers

C5C7±

C5-trans , trans-trans

and

Gly, estimated from Gibbs free energies calcu-

are not possible (see pathways VII-IX

lated at dierent temperatures. The Gibbs free

in Table 2 and Fig.

S5 of Supp.

Info).

Such

energy data is taken from G4MP2 calculations

ω B97X-D

processes require several dihedral angles (re-

which employed

laxation coordinates) to be altered and involve −1 high relaxation barriers (>800 cm ), which

quencies.

signicantly reduces the probability of conver-

relaxation processes in the argon jet. The con-

sion.

formers

Assuming

complete

relaxation

geometries and fre-

(b) The calculated Boltzmann pop-

ulations that take into account the conformer

52,64,6971

C5-trans , trans-trans

and

C5C7±

cannot undergo relaxation, but gain population

through pathways I-VI, the populations of the

from other conformers due to relaxation pro-

conformers that do not undergo relaxation, i.e.,

cesses I-VI (Table 2).

C5-trans , trans-trans

and

C5C7±,

temperature are presented in Fig. 6(b). that since

C5C7+

and

C5C7-

versus Note

C5C7 . From this C5-trans structure

as one structure denoted as

are very simi-

gure, it follows that the

lar in structure and energy, they are considered

is expected to be the most abundant under our

ACS Paragon Plus Environment 9

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 20

Table 2: The relaxation barriers between the lowest energy conformers of Gly-Gly calculated as the dierences between energies of the corresponding conformers and their transition states, based either on the electronic energy values, ∆Eel (G4MP2 method), or based on Gibbs free energy values at 15 and 300 K, ∆G(15 K) and ∆G(300 K) (G4MP2 method, harmonic frequency and hindered rotation analysis from ωB97XD/6-311++G(d,p)). Note that int1 is the intermediate conformer between the C5C7and C5-trans , and is similar to C5-trans except for the θT dihedral that diers by 180°. Relaxation pathway I. C5g-(s) → C5-trans II. C5g+(s) → C5-trans III. C5g-(a) → C5g-(s) IV. C5g+(a) → C5g+(s) V. sc-cis-trans → C5-trans VI. g-trans-trans → trans-trans VII. C5C7- → int1 VIII. int1 → C5-trans IX. trans-trans → sc-cis-trans

experimental conditions. and

C5C7

Barrier to relaxation, cm−1 ∆G(15 K) ∆G(300 K) 98 -27 410 94 19 327 592 599 980 583 572 825 699 402 374 144 51 228 2023 1857 1999 3763 3466 3605 913 911 1334

∆Eel

φ2 φ2 ψT ψT

NH2 inversion φ1 θT , ψT , φ2 θT , ψT ψ1

3.2

The predicted ratio

between the populations of

trans

Relaxation coordinate

C5-trans , trans-

Energy dierence ∆(Eel +ZPE) 90 179 502 438 755 256 -1481 1494 -347

Hydrogen bonding in Gly-Gly conformers

conformers is 78:10:12 at 400

K. This result is in a good agreement with our

There are some debates in the literature about

previous conclusion, drawn from the measured

the presence and importance of hydrogen bond-

IRMPD spectrum, that

C5-trans

ing in glycine and alanine structures.

is the most

31,7274

The

functional groups of Gly-Gly structures are sim-

abundant. The three conformers that do not relax in

ilar to glycine and alanine and can show similar

the supersonic jet (Fig. 6(b)) were observed

weak non-covalent interactions. Therefore, we

in the recent rotational spectroscopy study of

employed AIM (atoms in molecules) and NCI

Gly-Gly,

(non-covalent interactions) approaches to iden-

26

though relative abundances of these

conformers were not reported.

tify weak interactions in the Gly-Gly conform-

The spectro-

ers.

scopic observation of presumably low abun-

C5C7-

conformers (see

Two types of relatively strong hydrogen bond-

Fig. 6(b)) is most probably related to the

ing interactions were identied using the AIM

high resolution and sensitivity of the microwave

approach (see Fig. S7 of Supp.

dant

and

trans-trans

(MW) spectroscopy technique.

Info).

The

rst type corresponds to H2 N· · · HN hydrogen

Although it

is not possible to directly compare conformer

bonding (r (N· · · H) = 2.2 Å) that was found in

abundances between the MW experiment and

most of the conformers with cis conguration at

ours, one can expect the two methods to yield

the N-terminus (see e.g. structures

dierent populations of the conformers for two

C5g-(s)

reasons.

action leads to the formation of a ring involv-

pulses,

The laser ablation method (ps laser

λ = 355

and

C5C7+

C5-trans ,

in Fig. 3).

Such inter-

ing ve atoms, a so-called C5 interaction.

nm) employed in the MW

If

study is known to result in evaporated molec-

such a hydrogen bond is absent, the cis cong-

ular ensemble with higher temperatures than

uration is less stable than trans. For example,

the one achieved with laser desorption method

the

employed by us (ns pulses,

λ = 1064

nm).

sc-cis-trans trans-trans (see

27

Table 1). Another strong in-

teraction identied by the AIM approach cor-

Moreover, the MW study used neon (15 bar) as

responds to OH· · · O=C hydrogen bonding at

a carrier gas, which is known to result in less ef-

69,71

structure is less stable than

compared

the C-terminus (r (H· · · O) = 1.7 Å) that forms

to argon that is used in the current measure-

a seven-membered ring which was denoted with

ments.

C7. This interaction stabilizes the

cient conformational relaxation

ACS Paragon Plus Environment 10

C5C7+

and

Page 11 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

C5C7-

conformers. The relative distances be-

character

(e.g.

sc-cis-trans and transC7 character (C5C7+ ). The

tween the BCPs and RCPs identied by the

trans ),

AIM analysis indicates that the C7 interaction

NCI analysis showed that the formation of the

is stronger than C5.

C5 ring is associated with considerable steric

or strong

Weaker hydrogen bonding interactions that

repulsion eects which weaken the structure

could not be identied with the AIM method

stabilization by the hydrogen bond. It was also

were

found that the C5 interactions at the N- and

found

with

the

help

of

the

NCI

ap-

proach introduced by Jonson and co-workers.

33

C-termini show some competition due to shar-

Fig. 7(a,c) illustrates the identied hydrogen

ing the same donor (amide NH group).

bonding interactions (blue and green colored is-

example, the C5 interaction at the N-terminus

lands) for the

C5C7+

and

trans-trans

struc-

of the

C5-trans

conformer (ρ

tures. Each interaction corresponds to a min-

slightly weaker than in the

imum in the plots of RDG versus sign(λ2 )·ρ

tures (ρ

shown in Fig. 7(b,d). In such plots, the strength

ing.

of non-covalent interactions is related to the

terminus of

sharpness of the minima and the values of

ρ

transC5C7+.

clearly shows that hydrogen bonds in are much weaker than those in

= 0.020)

C5g±(s)

is

struc-

that do not have donor shar-

Similarly, the hydrogen bond at the C-

than those (ρ

at the minima. For example, the NCI analysis

trans

= 0.022)

For

C5-trans (ρ = 0.015) is weaker in trans-trans and sc-cis-trans

= 0.017).

Finally, as can be seen from Table 3, the

cis-trans

sc-

conformer is associated with a weak

The results of the NCI analysis applied to dif-

interaction at the N-terminus, H2 N· · · HN. It

ferent Gly-Gly conformers are summarized in

is almost twice weaker than the corresponding

Table 3 (see also Fig. S8 of Supp.

C5 interaction in the structurally similar

Info).

For

C5-

each interaction found at the N- and C-termini,

trans

Table 3 lists the RDG values in dierent minima

spectroscopy. If we neglect the weak H2 N· · · HN

and the corresponding sign(λ2 )·ρ values. These

interaction in

two numbers provide information on the type

lization due to the C5 interaction at the N-

and strength of the non-covalent interaction.

terminus of

35

Based on the data from Table 3, several conclu-

conformer, observed by IRMPD-VUV

sc-cis-trans ,

C5-trans

the energy stabi-

can be estimated from

the the energy dierence between the two con−1 formers that is equal to 750 cm (9 kJ/mol).

sions can be drawn. Firstly, the planar conformers ordered by increasing strength of their N-terminus hydro-

4

transtrans < sc-cis-trans < g-trans-trans < C5trans (Table 3). Interestingly, the interaction of relatively stable trans-trans conformer gen bonding attractive interaction are:

is the weakest (see also Fig. 7(c,d)).

IRMPD-VUV spectroscopy of a cold molecular beam of laser-desorbed Gly-Gly revealed that the most dominant conformer of this dipep-

It is

tide is

even weaker than CH· · · O=C interactions in the

C5g±(s)

structure.

the contribution of the

On

trans-trans

trum.

with reduced entropy contributions

drance contributes to the relatively high stabilhydrogen

bonding

trans .

interac-

com-

C5-

It is worth noting that Gibbs free en-

ergy calculations (Fig. 6(a)) clearly capture the

tions at the C-terminus can have either weak CH· · · O=C character (C5g±(s) ),

28,29

pared to more exible structures such as

trans-trans . the

The explanation can be found in the

rigidity of this structure which is associated

Fig. 7(a,c)). Therefore, the reduced steric hin-

Secondly,

conformer, sta-

found to be negligible in the measured spec-

are less pro-

nounced than in the case of cis structures (see

ity of

C5C7

Interestingly,

bilized by a strong C7 hydrogen bond, was

the other hand, the steric repulsion eects at the N-terminus of

with the nearly planar cis-

hydrogen bonding interactions.

does not contribute signicantly to the sta-

trans-trans

C5-trans ,

trans backbone structure stabilized by two C5

structures and most probably it

bilization of the

Conclusions

stabilization eect due to entropy, predicting

weak C5

ACS Paragon Plus Environment 11

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 20

Table 3: The minimum values of RDG and the corresponding sign(λ2 )·ρ values associated with dierent attractive interactions at N- and C-termini of Gly-Gly structures. The values denoted with ∗ correspond to the weak CH· · · O=C interaction. Note that the interactions at the N- and C-termini of the g-trans-trans conformer have nearly identical strength, which hindered nding the individual RDG and sign(λ2 )·ρ values. Conformer

C5-trans C5C7+ C5g-(s) C5g+(s) trans-trans g-trans-trans sc-cis-trans

sign(λ2 )·ρ

RDG(min) N-terminus

C-terminus

N-terminus

C-terminus

0

0.145

-0.0202

-0.0151

0

0

-0.0219

-0.038

0

0.137∗

-0.022

-0.0148∗

0

0.153∗

-0.0216

-0.0147∗

0.312

0.115

-0.012

-0.0165

0.081

0.081

-0.0165

-0.0165

0.08

0.089

-0.0139

-0.0169

C5-trans structure is more favorable than C5C7 at non-zero temperatures. that

molecule.

14

The current study shows that the

structure of the most stable conformers of pep-

Conformational relaxation in the supersonic

tides without aromatic moiety can successfully

jet is another eect that was found to play

be studied with IRMDP-VUV spectroscopy in

an important role in the observation of the

combination with the jet-cooling method.

C5-trans

we observed, jet-cooling is particularly ecient

conformer as the most dominant.

As

The accurate determination of relaxation barri-

for conformational relaxation.

ers between dierent conformers indicated that

ables reducing the number of conformers pop-

C5-trans

The latter en-

can gain population from a number

ulated, signicantly simplifying the measured

of low-energy conformers, making it the most

spectra. This is highly advantageous for spec-

abundant in a wide range of temperatures. It is

troscopy of exible peptides with a large degree

worth noting that owing to extremely low bar−1 riers to interconversion (