&Carbohydrates | Hot Paper |

Sweet Structural Signatures Unveiled in Ketohexoses

Celina Bermffldez, Isabel PeÇa, Santiago Mata, and Jos8 L. Alonso*[a]

Abstract: The conformational behaviour of naturally occur-

ring ketohexoses has been revealed in a supersonic expan-sion by Fourier transform microwave spectroscopy coupled

with a laser ablation source. Three, two and one conformersof d-tagatose, d-psicose and l-sorbose, respectively, have

been identified by their rotational constants extracted from

the analysis of the spectra. Singular structural signatures in-

volving the hydroxyl groups OH(1) and OH(2) have been dis-entangled from the intricate intramolecular hydrogen bond

networks stabilising the most abundant conformers. Thepresent results place the old Shallenberger and Kier sweet-

ness theories on a firmer footing.

Introduction

Human beings feel attracted to food owing to its flavours and

the pleasure experienced when tasting them. The sense of

taste, located in the oral cavity, is able to recognize up to fivedifferent tastes in food: salty, sour, bitter, sweet and umami

(from the Japanese umai; meaning delicious). Perceiving saltyand sour tastes is due to the presence of protons and ions.

Thus, there is no relation with the structure of the substance.In contrast, sensing the latter three flavours (bitter, sweet and

umami) is due to a response induced by chemoreceptors in

the taste bud which recognize specific molecular structures.Enquiry into these taste receptors has identified the structures

responsible for perceiving each flavour:[1] taste receptors (TRB),which are members of G-protein-coupled receptors (GPCRs)

are present in tongue and palate epithelia.[2] Particularly, withinthis TRB group, type T2R taste receptors recognize bitternesswhereas sweetness and umami are detected by type T1R;

T1R1 and T1R3 are assigned to umami and T1R2 and T1R3 tosweetness.[3–5] Therefore, only two receptors are responsible for

recognising all of the many and diverse sweeteners that exist.In order to account for all of them, these receptors have been

conjectured to contain different binding sites, each responsiblefor detecting sweeteners of similar sizes. Thus, all small-size

sweeteners, like monosaccharides or amino acids, are due tointeractions with the same receptor site.[6] Nonetheless, not allsimilar size molecules are equally sweet. This therefore raises

the question concerning which structural characteristicsendow sweeteners with their flavour, while other substances

with similar structures are not sweet, in other words, what is

the relation between sweetness and structure?Since early last century, abundant research has addressed

the link between sweetness and the structure of sweeteners.[7]

Several studies have assigned the sweet properties of a mole-cule to the presence of particular pairs of functional groups,[8, 9]

while others have assigned it to a specific disposition of anatom.[7, 10] Yet, none of these theories has been able to offer

a unified explanation regarding the sweetness–structure rela-tion until Shallenberger and Acree’s proposal.[11, 12] They ob-

served that the hydrogen bonds of sweet molecules strongly

affected their sweetness. Under these circumstances, theystated that the degree of sweetness depends on the strength

of two H-bonds by which the sweetener is bound to the sweetreceptor (Figure 1 a). They established that one of the two elec-

tronegative atoms might act as proton donor (AH) in the hy-drogen bond interaction and the other as acceptor (B). These

two groups form what is called the glucophore, which general-

ly refers to the part of the sweetener that interacts with thesweet receptor. A third binding site was later proposed by Kier,

the g-site, forming the “sweetness triangle” (Figure 1 b).[13–15]

This new g-site might interact with the receptor via hydropho-bic or van der Waals’ interaction. Nonetheless, the g-site waspostulated as unnecessary. Indeed, it was stated that it merely

enhances the sweet flavour depending on the g-AH-B dis-tance.[16] The role of the hydrophobic interactions has beenalso discussed by Simons et al.[17] in terms of carbohydrate mo-

lecular recognition at aromatic protein binding sites by creat-ing molecular complexes between monosaccharides and tolu-

ene.Evolution of sweetness theories have continued with the in-

troduction of the multipoint attachment theory. Tinti et al. pro-

posed that, besides the complementary sites for the AH-B-g tri-angle, the sweet receptor must contain at least five other link-

ing sites where the sweetener can interact, making eight sitesin all.[18] However, the glucophore in the sweetener need not

contain the analogous eight binding sites, although the great-er the number of interactions, the sweeter the substance. To-

[a] Dr. C. Bermffldez, Dr. I. PeÇa, S. Mata, Prof. J. L. AlonsoGrupo de Espectroscop&a Molecular (GEM), Edificio QuifimaLaboratorios de Espectroscopia y BioespectroscopiaUnidad Asociada CSIC, Parque Cient&fico UvaUniversidad de Valladolid, Paseo de Bel8n 5, 47011 Valladolid (Spain)E-mail : jlalonso@qf.uva.esHomepage: www.gem.uva.es

Supporting information for this article is available on the WWW underhttp ://dx.doi.org/10.1002/chem.201603223.

Chem. Eur. J. 2016, 22, 16829 – 16837 T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16829

Full PaperDOI: 10.1002/chem.201603223

gether with other related theories, this concurs with Shallen-berger’s initial idea that in order to be sweet a molecule needs

to establish two H-bonds with the receptor.At the time all of these theories were proposed, experimen-

tal data regarding the structure of sweeteners were restrictedto information from condensed phases, and were unable to

offer much detail concerning either the spatial distribution of

hydrogen atoms or the intrinsic intramolecular H-bond interac-tions that stabilise the molecules. A 3D conformational study

of these sweet substances might provide such information,shedding some light on the discussion regarding the relation

between sweetness and structure.Spectroscopic studies of monosaccharides undertaken under

the isolation conditions of the gas phase have proven ideal for

exploring conformational behaviour.[19] Numerous studies onchromophore-tagged sugars have been reported by Simons’

group[20] using UV/UV and IR/UV laser spectroscopy techniques.However, laser spectroscopy provides only vibrational resolu-

tion and band assignment is not always unambiguous. Fouriertransform microwave (FTMW) spectroscopy combined with

laser ablation and supersonic expansions have emerged as an

alternative solution. FTMW spectroscopy requires no chromo-phore and provides very high resolution. The first conforma-tional characterisation of the archetypal monosaccharide d-glu-cose has recently been made possible by combining laser abla-

tion (LA) with molecular beam Fourier transform microwave(MB-FTMW) spectroscopy.[21] A new generation of broadband

microwave spectrometers has allowed the spectrum to be re-corded over a wide range of frequencies, making conforma-tional research more efficient.[22–24] In our laboratory, we recent-

ly combined the latest broad-band chirped-pulse Fourier trans-form microwave technology (CP-FTMW) introduced by Pate

et al.[22] with our laser ablation sources (LA)[25] and applied thesaid combined technique to the rotational studies of several

monosaccharides.[26–30] Even large solid biomolecules such as

uridine[31] have recently been placed in the gas phase by laserablation and characterised by the aforementioned techniques.

The study of ketohexoses is extremely appealing, since theyare as sweet as table sugar (sucrose) with the exception of d-

fructose, which is twice as sweet. Its conformational behaviourhas recently been revealed using the abovementioned experi-

mental approach.[26] From the analysis of the rotational spec-trum, the two pyranoside conformers of Figure 2 were detect-

ed in supersonic expansion. The spatial disposition of the mostabundant b 2C5 g@ cc conformer (98 %) stabilised by a network

of five cooperative[32] intramolecular H-bonds in a counter-clockwise (cc) arrangement seems to indicate that OH(1) and

OH(2) hydroxyl pairs are suitable candidates to be assigned asAH and B sites of the glucophore, respectively, as anticipatedby Shallenberger et al.[11, 12] and confirmed in later publica-tions.[33, 34]

Equipped with both the potential of the LA-CP-FTMW exper-

imental tool to unveil the conformational behaviour of mono-

saccharides and the first results concerning d-fructose, thepresent work addresses the challenge of exploring the confor-

mational behaviour of d-fructose epimers: d-tagatose, d-psi-cose and l-sorbose (Figure 3). In the following sections, we an-

alyse their rotational spectrum in order to reveal their moststable structures characterised in the isolation conditions of su-

personic expansion. The results of this conformational analysis

are used to evaluate the theories proposed to date concerningsweetness. Common structural signatures that might be corre-

lated to their sweetness can shed some light on identifyingthe glucophore.

The food industry is currently looking to design tasty newsubstances, not only maintaining a sweet taste but also avoid-

ing undesirable health effects. Consequently, there is keen in-terest in identifying the structural requirements that a sweetsubstance must contain in order to plan the synthesis of newartificial sweeteners.

Results

d-Tagatose

d-Tagatose, the C-4 epimer of d-fructose, is a healthy sweeten-

er with antidiabetic properties.[35] In its crystalline form, d-taga-tose exists in the a form with a 5C2 pyranoside ring configura-

tion (Figure 3).[36] As confirmed in our study of d-fructose[26]

and in the results of other related sugars,[21, 27–30] the intercon-

Figure 1. Representation of the sweetness theories. a) Shallenberger’stheory: circular area represents the glucophore of the sweetener. b) Diagramof the “sweetness triangle” in the sweetener.[13–15]

Figure 2. The two observed conformers of b-d-fructopyranose showing theintramolecular hydrogen-bond networks and the glucophore unit.{26}

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16830

Full Paper

version between the a and b anomers should not occurduring vaporisation,[37] since it is a solvent-mediated reaction.

Therefore, its 6 to 12 GHz broad-band rotational spectrum,shown in Figure 4, should be attributed to the a-d-tagatose

species as well as the photofragmentation products commonto other sugars.[26]

After removing these decomposition lines (Figure 4), pairs ofstrong a-type R-branch (J + 1)0,J + 1

!J0,J and (J + 1)1,J + 1

!J1,J rota-tional transitions were easily identified and ascribed to a firstrotameric species, labelled as I. Following iterative fittings andpredictions, b- and c-type R-branch transitions were assigned

and measured for this rotamer. A total of 63 transitions(Table S1 in the Supporting Information) were submitted to

rigid rotor analysis,[41] leading to the rotational constants listed

in the first column of Table 1.Once the strong lines of rotamer I were removed from the

broad-band spectrum, deeper insights into the low intensebackground led to the assignment of very weak rotational

transitions (see inset of Figure 4) belonging to rotamers II andIII. For rotamer II, 22 a- and c-type R-branch transitions were

measured whereas for rotamer III only 16 a-type were ob-served (Tables S2–S3 in the Supporting Information). Rigidrotor analysis[38] for each rotamer led to the rotational con-stants also listed in Table 1.

Experimental values of the rotational constants of rotamers Iand II nicely match with those predicted by ab initio calcula-

tions (see Table 1 and Table S4 in the Supporting Information)for conformers a 5C2 g + s and a 5C2 g + cc (see ref. [39] andfootnote of Table 1 for labelling). These are stabilised by intra-

molecular H-bonds between vicinal hydroxyl groups formingsmall networks, strongly reinforced by sigma-hydrogen bondcooperativity.[32] In both conformers, the axial orientation ofthe hydroxyl groups OH(2) and OH(3) disrupts any plausible H-

bond network over the entire molecule. Thus, as shown inFigure 5, two split H-bond networks stabilise both conformers.

They only differ in the clockwise (cl) or counter-clockwise (cc)

orientation of the network involving the OH(3), OH(4) and OH(5)

hydroxyl groups. This does not significantly affect their rota-

tional constant values but does drastically alter the mb dipolemoment component value, which changes from 1.7 D to near

zero when passing from the a 5C2 g + cl conformer to a 5C2

g + cc. On this basis, rotamer I must be conformer a 5C2 g + cland rotamer II conformer a 5C2 g + cc, since b-type transitions

are not observed for rotamer II. The values of the rotationalconstants of rotamer III allowed conclusive identification ofconformer a 5C2 t cc. It only exhibits a network of four cooper-ative OH(5)···OH(4)···OH(3)···OH(1)···O(ring) hydrogen bonds in a coun-

ter-clockwise (cc) arrangement favoured by the trans configu-ration (t) of the hydroxymethyl group.

As proof of the consistency of the overall assignment,

a scale factor ranging from 0.991 to 0.997 makes the theoreti-cal values for conformers a 5C2 g + cl, a 5C2 g + cc and a 5C2 t

cc concur with the experimental values for rotamers I, II and III,respectively (Table 1). The population distribution of these spe-cies, obtained from the relative intensities of rotational transi-tions,[22, 40] is estimated to be: a 5C2 g + cl, 72 %; a 5C2 g + cc,

22 %; and a 5C2 t cc, 6 %, in accordance with the predicted rela-tive energies.

d-Psicose

d-Psicose, also known as d-allulose or d-ribo-2-hexulose, is theC-3 epimer of d-fructose. Although it is a rare sugar, it has at-tracted much attention for its promising health properties as

a no-energy sweetener.[41] In crystalline phase, d-psicose is sta-bilised in its b form with a 2C5 pyranoside ring configuration(Figure 3).[42] Its laser ablated broad-band rotational spectrum(Figure 6) was analysed following the same procedure as de-

scribed above. After removing known photofragmentationlines, it was easy to identify intense pairs of b-type R-branch

progressions (J + 1)0,J + 1

!J1,J and (J + 1)1,J + 1

!J0,J (J ranging

from J = 4 to J = 7), as belonging to a first rotamer I. Neither a-nor c-type transitions were observed for this species.

As shown in the inset of Figure 6, a detailed inspection ofthe spectrum revealed the existence of other very weak b-type

R-branch progressions close to those of rotamer I. Once theirassignment through an isotopic species had been ruled out,

Figure 3. a) Fisher projection of ketohexoses indicating the carbon thatchanges configuration with regard to d-fructose (epimers). b) Haworth pro-jection of the a or b pyranose form of each ketohexose. The a or b nomen-clature is selected in agreement with IUPAC; thus, for ketohexoses, cis dispo-sition of OH(5) and OH(2) is labelled as a and trans as b. c) 5C2 and 2C5 chairconfigurations.

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16831

Full Paper

they were attributed to a second rotamer II. As with rotamer I,a- and c-type transitions were predicted but not found. Rigid

rotor analysis[38] of the measured lines for rotamers I and II,

listed in Tables S5–S6 in the Supporting Information, yieldedthe set of rotational constants displayed in the first and third

columns of Table 2. No other weak transitions attributable toother species remain in the broad-band spectrum.

When identifying the observed rotamers, we first took intoconsideration the very close values of their rotational con-

stants. As for rotamers I and II of d-tagatose, this fact stronglyindicates that both species must belong to conformers with

the same skeletal frame but with a different orientation of

their intramolecular H-bond network. On account of the valuesof their rotational constants and electric dipole moment com-

ponents, rotamers I and II are only consistent with those pre-dicted for the two lowest energy conformers b 2C5 g@ cc and

b 2C5 g@ cl (Table S7 in the Supporting Information), also listedin Table 2 for comparison. As shown in Figure 7, each confor-

Figure 4. Broad-band rotational spectrum of d-tagatose obtained by LA-CP-FTMW. Triangles indicate the common decomposition lines of sugars (formalde-hyde, acrolein, etc.) ruled out in the analysis. The a-type R-branch (J + 1)1,J + 1

!J1,J and (J + 1)0,J + 1

!J0,J pairs of rotational progressions ranging from J = 4 toJ = 8 of rotamer I are also indicated. Inset : small section of the spectrum that shows rotational transitions of the three rotamers observed for d-tagatose.

Table 1. Experimental and predicted[a] rotational parameters for the observed conformers of d-tagatose.

Rotamer I Rotamer II Rotamer III a 5C2 g + cl[h] a 5C2 g + cc a 5C2 t cc

A [MHz][b] 1627.89919(33)[f] 1614.28940(51) 1462.359(24) 1636.6 1618.6 1474.9B [MHz] 698.29606(15) 699.11441(25) 747.26266(56) 701.9 703.6 750.3C [MHz] 588.40040(10) 591.15555(28) 650.90140(53) 592.3 595.7 653.6jma j [D] observed[g] observed observed 1.9 1.0 2.7jmb j [D] observed – – 1.7 0.1 1.4jmc j [D] observed observed – 1.4 1.7 0.0s[c] [kHz] 6.5 6.3 9.6 – – –N[d] 63 22 16 – – –DE[e] [cm@1] – – – 0 91 288

[a] Ab initio calculations (MP2-6-311 + + G(d,p)). [b] A, B and C are the rotational constants; jma j , jmb j and jmc j are the absolute values of the electric dipolemoment components. [c] The rms deviation of the fit. [d] Number of fitted rotational transitions. [e] Relative electronic energies calculated at MP2/6-311 +

+ G(d,p) level of theory. [f] Standard error in parenthesis in the units of the last digit. [g] Experimental observation of the a-, b- and c-type spectrum. [h] a

or b denote the anomer type. 2C5 or 5C2 refer to pyranose ring configuration. The symbols g@ (@608), g + (+ 608) and t (1808) describe the values of thetorsion angle O(1)-C(1)-C(2)-O(2), which determines the orientation of the hydroxymethyl group. The last label “cl” and “cc” describes the clockwise and coun-ter-clockwise arrangements, respectively, of the intramolecular hydrogen bond networks involving hydroxyl groups OH(3), OH(4) and OH(5).

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16832

Full Paper

mer is stabilised by two different intramolecular hydrogenbond networks; one in common OH(2)···OH(1)···Oring and, as an-ticipated, a counter-clockwise (cc) OH(5)···OH(4)···OH(3)···OH(5) anda clockwise (cl) OH(3)···OH(4)···OH(5)···OH(3) cyclic H-bond network.

This does not appreciably affect the values of their rotationalconstants, and does not allow any discrimination betweenthem to be made. In addition, contrary to what was observed

in d-tagatose, the values of the dipole-moment componentsare not altered enough when passing from conformer b 2C5

g@ cc to b 2C5 g@ cl, in accordance with the experimental ob-servations of b-type R-branch transitions for both rotamers. It

could be a reasonable assignment to ascribe rotamer I to con-

former b 2C5 g@ cc and rotamer II to conformer b 2C5 g@ cltaking into account the relative intensities of the measured ro-

tational transitions which give rise to a population distributionof 60 and 40 %, respectively, in accord with ab initio calcula-

tions.

Figure 5. The three observed conformers of a-d-tagatose showing the intramolecular H-bonds with counter-clockwise (cc) or clockwise (cl) arrangements in-volving hydroxyl groups OH(3), OH(4) and OH(5).

Table 2. Experimental and predicted[a] rotational parameters for the ob-served conformers of d-psicose.

Rotamer I Rotamer II b 2C5 g@ cc b 2C5 g@ cl

A [MHz][b] 1626.7540(38)[f] 1626.75480(84) 1631.8 1635.8B [MHz] 723.03579(12) 724.68772(41) 728.3 728.3C [MHz] 660.98104(16) 661.03584(48) 665.3 664.9jma j [D] – – 0.4 0.2jmb j [D] observed[g] observed 1.4 0.7jmc j [D] – – 0.0 0.0s[c] [kHz] 5.3 9.4 – –N[d] 24 15 – –DE[e] [cm@1] – – 0 73

[a] Ab initio calculations (MP2-6-311 + + G(d,p)). [b] A, B and C are the ro-tational constants; jma j , jmb j and jmc j are the absolute values of the elec-tric dipole moment components. [c] The rms deviation of the fit.[d] Number of fitted rotational transitions. [e] Relative electronic energiescalculated at the MP2/6-311 + + G(d,p) level of theory. [f] Standard errorin parenthesis in the units of the last digit. [g] Experimental observationof the a-, b- and c-type spectrum.

Figure 6. Broad-band rotational spectrum of d-psicose obtained by LA-CP-FTMW. Triangles indicate all decomposition lines. The b-type R-branch(J + 1)0,J + 1

!J1,J and (J + 1)1,J + 1

!J0,J pairs of rotational progressions ranging from J = 4 to J = 7 of rotamer I are also indicated. Inset: small section of the spec-trum that shows the 321

!212 rotational transition of the two rotamers observed for d-psicose.

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16833

Full Paper

l-Sorbose

l-Sorbose, the C-5 epimer of d-fructose, is as sweet as sucrose.

Its main use is not as sweetener, whereas as intermediate inthe industrial synthesis of vitamin C.[43] In crystalline phase,

l-sorbose is found in the a anomer in the most stable 2C5

pyranoside ring configuration[44] (Figure 3). Its laser ablated

broad-band rotational spectrum was found to be dominated

by decomposition lines. After exhaustive detailed searches,a weak progression of a-type R-branch (J + 1)0,J + 1

!J0,J and

(J + 1)1,J + 1

!J1,J pairs of transitions ranging from J = 4 to J = 8was revealed (Figure 8).

Neither b- nor c-type transitions were observed for this spe-cies. No lines belonging to other species could be identified. A

total of 34 rotational transitions (Table S8 in the Supporting In-

formation) were analysed to derive the rigid rotor rotationalconstants values shown in Table 3. These are only compatible

with those predicted ab initio for conformer a 2C5 g@ cc (seeTable S9 in the Supporting Information), also listed in Table 3for comparison. Identification is reinforced by the fact thatonly weak a-type transitions are observed, in accordance with

the predicted values of the electric dipole moment compo-nents for this conformer.

The a 2C5 g@ cc conformer of l-sorbose is stabilised by

a five-cooperative intramolecular hydrogen bond network

OH(5)···OH(4)···OH(3)···OH(2)···OH(1)···O(ring) (Figure 9), in a counter-clockwise arrangement, spread over the entire molecule. The

observed network shows a strong analogy to that of d-fruc-tose,[26] with the sole exception of the disposition of the OH(5),

which is in axial disposition in d-fructose.

Discussion

In light of this new experimental information regarding theconformational behaviour of ketohexoses, theories explainingsweetness in terms of structure can be revisited in order to dis-

cuss their agreement with the present conformational results.Looking at the panel of the conformational panorama of keto-hexoses (Figure 10), it is easy to realize that all the most abun-dant species of ketohexoses: b 2C5 g@ cc (98 %) for d-fructose,

a 5C2 g + cl (72 %) and a 5C2 g + cc (22 %) for d-tagatose, b 2C5

g@ cc (60 %) and b 2C5 g@ cl (40 %) for d-psicose and a 2C5 g@cc for l-sorbose, show the same intramolecular H-bond net-

work OH(2)···OH(1)···Oring (see dashed lines in Figure 10). Despitetheir distinct 2C5 or 5C2 ring configurations, the anomeric OH(2)

is always in axial orientation while the OH(1) points towards theO(ring). All the above conformers share the common conforma-

tional signature OH(2)···O(1), which could be ascribed to the AHand B sites of glucophore, thus concurring with Shallenberg-

Figure 7. The two observed conformers of b-d-psicose showing the intramo-lecular H-bonds with counter-clockwise (cc) or clockwise (cl) arrangementsinvolving hydroxyl groups OH(3), OH(4) and OH(5).

Figure 8. Broad-band rotational spectrum of l-sorbose obtained by LA-CP-FTMW. Decomposition lines have been removed from the spectrum in orderto observe the very weak a-type R-branch (J + 1)1,J + 1

!J1,J and (J + 1)0,J + 1

!J0,J

pairs of rotational progressions ranging from J = 4 to J = 8 of the observedrotamer.

Table 3. Experimental and predicted[a] rotational parameters for the ob-served conformers of l-sorbose.

Rotamer I a 2C5 g@ cc

A [MHz][b] 1525.5027(80)[e] 1527.9B [MHz] 722.08428(33) 727.5C [MHz] 556.05153(30) 559.0jma j [D] observed[f] 0.8jmb j [D] – 0.1jmc j [D] – 0.1s[c] [kHz] 8.4 –N[d] 34 –

[a] Ab initio calculations (MP2/6-311 + + G(d,p)). [b] A, B and C are the ro-tational constants; jma j , jmb j and jmc j are the absolute values of the elec-tric dipole moment components. [c] The rms deviation of the fit.[d] Number of fitted rotational transitions. [e] Standard error in parenthe-sis in the units of the last digit. [f] Experimental observation of the a-, b-and c-type spectrum.

Figure 9. The observed conformer of a-l-sorbose showing the intramolecu-lar H-bond network in the counter-clockwise (cc) arrangement.

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16834

Full Paper

er’s proposal.[11, 12] The reverse identification of AH as OH(1) and

B as OH(2), proposed in the literature,[45] does not appear to beconsistent with the detailed structural information provided inthis study.

On the other hand, the presence of other hydroxyl groupsin ketohexoses might allow additional assignments to the glu-cophore. A priori, the other two vicinal hydroxyl pairsOH(3)···OH(4) and OH(4)···OH(5) in the most abundant conformersshow hydrogen-bond interactions that could easily be adaptedto the linkage to the receptor forming the AH-B glucophore,

as has been proposed in the literature.[46] However, neithertheir clockwise/counter-clockwise arrangement nor their axial/equatorial relative orientation allowed these hydroxyl groups

to constitute common conformational signatures that can beattributed to the glucophore. Therefore, according to our re-

sults, the glucophore sites responsible for the sweet flavourcan only be attributed to the hydroxyl groups OH(1) and OH(2).

Owing to their spatial disposition, OH(1) acts as proton donor

(AH) and OH(2) as acceptor (B), concurring with Shallenberger’stheory.[11, 12]

Taking account of Kier ’s proposal,[13] considering the idea ofa glucophore unit comprising an AH-B-g triangle, identifying

a third hydrophobic g site involved in the “sweetness triangle”proved relatively straightforward. Hence, all the observed con-

formations of Figure 10 show common surroundings of C(6)

free of any hydrophilic group. This site constitutes the main

hydrophobic environment and, consequently, is the most suita-ble for being considered as g in accordance with the afore-

mentioned sweetness theory.[13] Thus, based on the commonconformational signatures of ketohexoses, the “sweetness tri-

angle” formed by OH(2), OH(1) and C(6) (green triangles inFigure 10) is present in all stable conformers of ketohexoses inline with Kier’s theory.[13]

The degree of sweetness of ketohexoses can also be consid-ered in light of the present results. It is well known that d-fruc-tose is twice as sweet as l-sorbose. Surprisingly, the conforma-tional behaviour observed for both sugars, shown in Fig-ure 10 a and d, revealed extraordinary similarities. Hence, themost abundant b 2C5 g@ cc and a 2C5 g@ cc species of d-fruc-

tose and l-sorbose are stabilised by the same H-bond network

OH(5)···OH(4)···OH(3)···OH(2)···OH(1)···O(ring) spread over the entiremolecule. Some authors[33, 47] have rationalised this difference

in sweetness in terms of the existence of an intramolecular H-bond between the OH(5) and O(ring) in b-d-fructopyranose and

its absence in a-l-sorbopyranose. However, no OH(5)···O(ring) hy-drogen bond interactions are observed in any of the observed

conformers of the ketohexoses. Speculating with the multi-

point attachment theory,[18] which also assume the presence ofAH, B and g sites in the glucophore, the distinct axial/equatori-

al orientation of the OH(5) in the b 2C5 g@ cc and a 2C5 g@ ccconformers may add the possibility that the sweetener inter-

acts with the receptor through more places, altering its relativesweetness.

Conclusions

The conformational behaviour of ketohexoses was investigatedfor the first time by using a combination of Fourier transform

microwave spectroscopy and laser ablation. In the isolation

conditions of the gas phase, three, two and one rotamers weredetected for d-tagatose, d-psicose and l-sorbose, respectively.

All are found to be over-stabilised by cooperative networks ofintramolecular hydrogen bonds between vicinal hydroxyl

groups stretching throughout the whole molecule.The detailed structural information extracted from our ex-

periments provides information on the orientation of the OHgroups with respect to the molecular frame, and allows the in-

tramolecular interactions in which these functional groups areinvolved to be established. Hence, a singular structural signa-ture involving the hydroxyl groups OH(2) and OH(1) in the short

H-bond network OH(2)···OH(1)···Oring has been characterised forthe first time in the most abundant species of ketohexoses in-

dicating that anchoring to the sweet receptor occurs here, inaccordance with Shallenberger’s old proposal.[11, 12] In addition,

a hydrophobic g site surrounding the C(6) has been identified

in all conformers, supporting the AH-B-g “sweetness triangle”of Kier ’s proposal.[13] The high resolution reached by our LA-

CP-FTMW experiments has opened a new window to explorethe linkage between sweetness and structure.

Further investigation exploring the influence of water inconformational behaviour is being undertaken. In vacuum, the

Figure 10. Observed conformers for all ketohexoses. Dashed lines indicatethe H-bond interactions that share the most stable conformers of ketohexo-ses. The A-H, B and g sweetness triangle is indicated according to the sweet-ness theories in the literature.[11–13] In parenthesis, the abundance of eachconformer is shown.

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16835

Full Paper

isolated molecules can only make strained hydrogen bonds be-tween adjacent hydroxyl groups around the edge of the ring.

Incorporating environmental contributions such as solute–sol-vent interactions in aqueous solutions, which have the same

nature as glucophore–receptor interactions,[48] should contrib-ute to reinforcing our understanding of the structural features

involved in sweetness.

Experimental Section

LA-CP-FTMW spectroscopy

Commercial samples of d-tagatose, d-psicose and l-sorbose (m.p.:134 8C, 105 8C and 165 8C, respectively) were used without any fur-ther purification. Solid rods were prepared by pressing the com-pounds’ fine powder mixed with a small amount of commercialbinder and were placed in the ablation nozzle (Figure S1 in theSupporting Information). A picosecond Nd:YAG laser (10 mJ perpulse, 20 ps pulse width) was used as a vaporisation tool. Productsof the laser ablation (LA) were supersonically expanded by usingthe flow of carrier gas (Ne, 15 bar) and were characterised by chirp-ed pulse Fourier transform microwave (CP-FTMW) spectroscopy.

A Scheme of our LA-CP-FTMW spectrometer[25, 27] constructed atthe University of Valladolid is shown in Figure S1a. A chirped-pulseis directly generated by an arbitrary waveform generator (1) andamplified by the adjustable (3) travelling wave tube (2) amplifierwith 300 W maximum output power. Following amplification, a par-abolic reflector system comprising dual ridge horns (5–6) and twoparabolic reflectors (6–7) in a paraxial beam configuration is usedto broadcast the excitation pulse into the vacuum chamber and re-ceive the broad-band molecular emission. This molecular-free in-duction decay (FID) signal is directly digitised by using a digital os-cilloscope (8) after it has been amplified with a low-noise micro-wave amplifier (9). The rotational spectrum in the frequencydomain is obtained by taking a fast Fourier transformation of theFID, following the application of a Kaiser–Bessel window to im-prove baseline resolution. All frequency and trigger sources as wellas the digital oscilloscope are phase-locked to a 10 MHz Rb-disci-plined quartz oscillator (11).

The spectrometer operating sequence (Figure S1b) commenceswith a molecular pulse of 700 ms duration which drives the carriergas flow through the pulsed valve source (I). After a slight delay,a laser pulse hits the solid and vaporizes the sample (II). Four sepa-rate broadband rotational spectra are acquired in each injectioncycle (III). The four individual broadband chirped excitation pulses,of 4 ms duration, are spaced by 18 ms. Two ms after each excitationpulse ceases, the rotational free induction decay is then acquiredfor 10 ms (IV). Around 100 000 individual FIDs (four per each molec-ular expansion) at a 2 Hz repetition rate were averaged in the timedomain and Fourier transformed to obtain the rotational spectra ofd-tagatose, d-psicose and l-sorbose in the 6 to 12 frequencyrange.

Computational methods

Similar to other monosaccharides in condensed phase, ketohexo-ses present a six-member ring pyranose configuration. The carbonC(2) becomes an asymmetric centre, which results in two possiblea and b anomeric species. In aqueous solution, they mainly existas a mixture of a and b pyranoses,[49] adopting the energeticallyfavoured 5C2 and 2C5 chair configurations in which the hydroxy-methyl group is in equatorial orientation (Figure 3). In addition,

conformational analysis of ketohexoses entails considering intra-molecular hydrogen bonding between OH groups taking place foreach configuration. Plausible formation of hydrogen-bond net-works reinforces the stability of certain conformers due to hydro-gen bond cooperativity.[32] Knowledge of the subtle variation in hy-droxyl arrangement is relevant to distinguish between differentconformers and plays an important role in glucophore characterisa-tion.

Identifying the different conformers present in supersonic expan-sion involves a combination of the aforementioned experimentalapproach and theoretical calculations. A typical assignment strat-egy, similar to that previously described in our studies of aminoacids,[50] commences with the generation of the set of plausibleconformers by performance extensive semi-empirical calculations(AM1 and PM3) to find the lowest-energy conformers in the poten-tial energy surface. Those lower-energy conformers, which mightconceivably be populated under experimental conditions em-ployed in the supersonic expansion, are subsequently optimisedby using the Møller–Plesset second order perturbation (MP2)theory to include electron correlation and larger 6–311 + + G(d,p)basis set.[51] This level of theory has been found to behave satisfac-torily in previous studies of sugars. These provide the relativeenergy of conformers, the rotational constants (A, B, C) and electricdipole moment components (ma, mb, mc) relevant for our rotationalstudies. Each conformer has distinct values for these spectroscopicparameters, which means they will give rise to different rotationalspectra, enabling us to discriminate between them. Conformationalidentification can be achieved by comparing experimentally andtheoretically predicted values of the above molecular properties,which collectively leads to identifying the observed conformers.

Acknowledgements

This research was supported by Ministerio de Econom&a y

Competitividad (grant numbers CTQ 2013-40717-P and Consol-ider Ingenio 2010 CSD2009-00038) and Junta de Castilla y

Lejn (grant number VA077U16). C.B. wishes to thank the Minis-terio de Ciencia e Innovacijn for an FPI grant (BES 2011-

047695).

Keywords: conformational analysis · hydrogen bonding · laser

ablation · microwave spectroscopy · sugars

[1] M. Behrens, W. Meyerhof, C. Hellfritsch, T. Hofmann, Angew. Chem. Int.Ed. 2011, 50, 2220 – 2242; Angew. Chem. 2011, 123, 2268 – 2291.

[2] E. Adler, M. A. Hoon, K. L. Mueller, J. Chandrashekar, N. J. P. Ryba, C. S.Zuker, Cell 2000, 100, 693 – 702.

[3] X. Li, L. Staszewski, H. Xu, K. Durick, M. Zoller, E. Adler, Proc. Natl. Acad.Sci. USA 2002, 99, 4692 – 4696.

[4] G. Q. Zhao, Y. Zhang, M. A. Hoon, J. Chandrashekar, I. Erlenbach, N. J. P.Ryba, C. S. Zuker, Cell 2003, 115, 255 – 266.

[5] G. Nelson, M. A. Hoon, J. Chandrashekar, Y. Zhang, N. J. P. Ryba, C. S.Zuker, Cell 2001, 106, 381 – 390.

[6] G. Morini, A. Bassoli, P. A. Temussi, J. Med. Chem. 2005, 48, 5520 – 5529.[7] G. G. Birch, R. S. Shallenberger, Crit. Rev. Food Sci. Nutr. 1976, 8, 57 – 95.[8] G. Cohn, Die Organischen Geschmacksstoffe, F. Siemenroth, 1914.[9] E. Oertly, R. G. Myers, J. Am. Chem. Soc. 1919, 41, 855 – 867.

[10] T. Kaneko, J. Chem. Soc. 1938, 29, 433 – 439.[11] R. S. Shallenberger, T. E. Acree, Nature 1967, 216, 480 – 482.[12] R. S. Shallenberger, T. E. Acree, C. Y. Lee, Nature 1969, 221, 555 – 556.[13] L. B. Kier, J. Pharm. Sci. 1972, 61, 1394 – 1397.[14] A. van der Heijden, L. B. P. Brussel, H. G. Peer, Food Chem. 1978, 3, 207 –

211.

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16836

Full Paper

[15] T. Yamazaki, E. Benedetti, D. Kent, M. Goodman, Angew. Chem. Int. Ed.Engl. 1994, 33, 1437 – 1451; Angew. Chem. 1994, 106, 1502 – 1517.

[16] S. C. Ebeling, Food Chem. 1998, 61, 107 – 112.[17] E. Cristina Stanca-Kaposta, D. P. Gamblin, J. Screen, B. Liu, L. C. Snoek,

B. G. Davis, J. P. Simons, Phys. Chem. Chem. Phys. 2007, 9, 4444 – 4451.[18] C. Nofre, J.-M. Tinti, Food Chem. 1996, 56, 263 – 274.[19] J. L. Alonso, J. C. Ljpez, Top. Curr. Chem. 2014, 364, 335 – 401 and refer-

ences therein.[20] a) E. J. Cocinero, P. CarÅabal, T. D. Vaden, J. P. Simons, B. G. Davis, Nature

2011, 469, 76; b) J. P. Simons, R. A. Jockusch, P. CarÅabal, I. Henig, R. T.Kroemer, N. A. Macleod, L. C. Snoek, Int. Rev. Phys. Chem. 2005, 24, 489;c) J. P. Simons, Mol. Phys. 2009, 107, 2435.

[21] J. L. Alonso, M. A. Lozoya, I. PeÇa, J. C. Ljpez, C. Cabezas, S. Mata, S.Blanco, Chem. Sci. 2014, 5, 515 – 522.

[22] G. G. Brown, B. C. Dian, K. O. Douglass, S. M. Geyer, S. T. Shipman, B. H.Pate, Rev. Sci. Instrum. 2008, 79, 053103.

[23] J.-U. Grabow, S. Mata, J. L. Alonso, I. PeÇa, S. Blanco, J. C. Ljpez, C. Ca-bezas, Phys. Chem. Chem. Phys. 2011, 13, 21063 – 21069.

[24] I. PeÇa, A. M. Daly, C. Cabezas, S. Mata, C. Bermffldez, A. NiÇo, J. C.Ljpez, J.-U. Grabow, J. L. Alonso, J. Phys. Chem. Lett. 2013, 4, 65 – 69.

[25] S. Mata, I. PeÇa, C. Cabezas, J. C. Ljpez, J. L. Alonso, J. Mol. Spectrosc.2012, 280, 91 – 96.

[26] C. Bermffldez, I. PeÇa, C. Cabezas, A. M. Daly, J. L. Alonso, ChemPhysChem2013, 14, 893 – 895.

[27] I. PeÇa, S. Mata, A. Mart&n, C. Cabezas, A. M. Daly, J. L. Alonso, Phys.Chem. Chem. Phys. 2013, 15, 18243 – 18248.

[28] I. PeÇa, L. Kolesnikov#, C. Cabezas, C. Bermffldez, M. Berdakin, A. Simao,J. L. Alonso, Phys. Chem. Chem. Phys. 2014, 16, 23244 – 23250.

[29] I. PeÇa, E. J. Cocinero, C. Cabezas, A. Lesarri, S. Mata, P. Pcija, A. M. Daly,]. Cimas, C. Bermffldez, F. J. Basterretxea, S. Blanco, J. A. Fern#ndez, J. C.Ljpez, F. CastaÇo, J. L. Alonso, Angew. Chem. Int. Ed. 2013, 52, 11840 –11845; Angew. Chem. 2013, 125, 12056 – 12061.

[30] I. PeÇa, C. Cabezas, J. L. Alonso, Chem. Commun. 2015, 51, 10115.[31] I. PeÇa, C. Cabezas, J. L. Alonso, Angew. Chem. Int. Ed. 2015, 54, 2991 –

2994; Angew. Chem. 2015, 127, 3034 – 3037.[32] G. A. Jeffrey. W. Saenger, Hydrogen Bonding in Biological Structures,

Springer, New York 1991.[33] M. G. Lindley, G. G. Birch, J. Sci. Food Agric. 1975, 26, 117 – 124.[34] R. S. Shallenberger, Pure Appl. Chem. 1978, 50, 1409 – 1420.[35] a) Y. Lu, G. V. Levin, T. W. Donner, Diabetes Obes. Metab. 2008, 10, 347;

b)G. V. Levin, J. Med. Food 2002, 5, 23 – 36.

[36] S. Takagi, R. D. Rosenstein, Carbohydr. Res. 1969, 11, 156 – 158.[37] L. P. Guler, Y.-Q. Yu, H. I. Kentt-maa, J. Phys. Chem. A 2002, 106, 6754 –

6764.[38] W. Gordy, R. L. Cook, Microwave Molecular Spectroscopy, Wiley, New

York, 1984.

[39] A. D. McNaught, Pure Appl. Chem. 1996, 68, 919 – 2008.[40] a) G. T. Fraser, R. D. Suenram, C. L. Lugez, J. Phys. Chem. A 2000, 104,

1141 – 1146; b) S. Blanco, A. Lesarri, J. C. Ljpez, J. L. Alonso, J. Am. Chem.Soc. 2004, 126, 11675 – 11683.

[41] a) H. Oshima, I. Kimura, K. Izumori, Food Sci. Technol. Int. 2006, 12, 137 –143; b) W. Mu, W. Zhang, Y. Feng, B. Jiang, L. Zhou, Appl. Microbiol. Bio-technol. 2012, 94, 1461 – 1467; c) T. Matsuo, K. Izumori, Biosci. Biotechnol.Biochem. 2006, 70, 2081 – 2085; d) S. H. Baek, S. J. Park, H. G. Lee, J.Food Sci. 2010, 75, H49 – H53.

[42] a) A. Kwiecien, K. Slepokura, T. Lis, Carbohydr. Res. 2008, 343, 2336 –2339; b) K. Fukada, T. Ishii, K. Tanaka, M. Yamaji, Y. Yamaoka, K.-i. Koba-shi, K. Izumori, Bull. Chem. Soc. Jpn. 2010, 83, 1193 – 1197.

[43] a) A. H. , Nature 1934, 134, 724 – 725; b) T. Reichstein, A. Gressner, Helv.Chim. Acta 1934, 17, 311 – 328; c) P. De Wulf, W. Soetaert, E. J. Van-damme, Biotechnol. Bioeng. 2000, 69, 339 – 343.

[44] S. H. Kim, R. D. Rosenstein, Acta Crystallogr. 1967, 22, 648.[45] a) M. Mathlouthi, M. O. Portmann, J. Mol. Struct. 1990, 237, 327 – 338;

b) R. J. Woods, V. H. Smith, Jr. , W. A. Szarek, A. Farazdel, J. Chem. Soc.Chem. Commun. 1987, 937 – 939.

[46] G. G. Birch, S. Shamil, Z. Shepherd, Experientia 1986, 42, 1232 – 1234.[47] R. J. Woods, W. A. Szarek, V. H. Smith, J. Am. Chem. Soc. 1990, 112,

4732 – 4741.[48] M. Mathlouthi, Food Chem. 1984, 13, 1 – 16.[49] S. Kçpper, S. Freimund, Helv. Chim. Acta 2003, 86, 827 – 843.[50] I. PeÇa, M. E. Sanz, J. C. Ljpez, J. L. Alonso, J. Am. Chem. Soc. 2012, 134,

2305 – 2312.[51] Gaussian 09, Revision B.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel,

G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B.Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratch-ian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M.Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, J. E. Peralta, F.Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov,T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C.Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene,J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts,R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Och-terski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador,J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman,J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian, Inc. , Wallingford CT,2010.

Received: July 6, 2016

Published online on October 4, 2016

Chem. Eur. J. 2016, 22, 16829 – 16837 www.chemeurj.org T 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim16837

Full Paper