Ellagitannin–Lipid Interaction by HR-MAS NMR Spectroscopy

molecules

Article Ellagitannin–Lipid Interaction by HR-MAS NMR Spectroscopy

Valtteri Virtanen * , Susanna Räikkönen, Elina Puljula and Maarit Karonen

Natural Chemistry Research Group, Department of Chemistry, University of Turku, FI-20014 Turku, Finland; [email protected] (S.R.); [email protected] (E.P.); maarit.karonen@utu.ﬁ (M.K.) * Correspondence: vtjvir@utu.ﬁ; Tel.: +358-29-450-3205

Abstract: Ellagitannins have antimicrobial activity, which might be related to their interactions with membrane lipids. We studied the interactions of 12 different ellagitannins and pentagalloylglucose with a lipid extract of Escherichia coli by high-resolution magic angle spinning NMR spectroscopy. The nuclear Overhauser effect was utilized to measure the cross relaxation rates between ellagitannin and lipid protons. The shifting of lipid signals in 1H NMR spectra of ellagitannin–lipid mixture due to ring current effect was also observed. The ellagitannins that showed interaction with lipids had clear structural similarities. All ellagitannins that had interactions with lipids had glucopyranose cores. In addition to the central polyol, the most important structural feature affecting the interaction seemed to be the structural ﬂexibility of the ellagitannin. Even dimeric and trimeric ellagitannins could penetrate to the lipid bilayers if their structures were ﬂexible with free galloyl and hexahydroxydiphenoyl groups.

Keywords: E. coli; HR-MAS-NMR; interaction; lipid membrane; tannins; UPLC-DAD-MS

Citation: Virtanen, V.; Räikkönen, S.; Puljula, E.; Karonen, M. 1. Introduction Ellagitannin–Lipid Interaction by HR-MAS NMR Spectroscopy. Tannins are a group of specialized plant metabolites, which, when included in the di- Molecules 2021, 26, 373. etary feed of ruminants, have been shown to induce many beneficial effects such as increas- https://doi.org/10.3390/ ing their effective amino acid absorption, lowering their methane production, and acting as molecules26020373 anthelmintics [1–6]. Additionally, tannins have been recently shown to inhibit the growth of several bacteria more effectively than what they were previously thought capable [7]. Academic Editors: Many of these favorable effects are traditionally thought to be governed by the protein Teresa Escribano-Bailón, affinity/protein precipitation capacity of tannins or their oxidative activity, which have Ignacio García-Estévez and been extensively studied in the literature [8–10]. However, the possible interactions be- Encarna Gómez-Plaza Received: 15 December 2020 tween lipids and tannins have not been widely considered even though they might play an Accepted: 7 January 2021 important role in understanding the capability and the mechanisms with which tannins Published: 12 January 2021 inhibit, for instance, the growth of different bacteria and their possibilities as antimicrobial agents in general [11]. Publisher’s Note: MDPI stays neu- High-resolution magic angle spinning (HR-MAS) NMR spectroscopy has revolution- tral with regard to jurisdictional clai- ized and opened all new possibilities to study lipids, lipid membranes, and their potential ms in published maps and institutio- interactions with other compounds [12–15]. A notable benefit of the HR-MAS probe is nal affiliations. that it tolerates the kind of semisolid emulsion type samples that ellagitannins (ETs) and lipids form in a solution, while still enabling normal liquid-state NMR experiments with reasonable resolution. Useful experiments include methods such as nuclear Overhauser effect spectroscopy (NOESY) to detect correlations between specific parts of the lipids with Copyright: © 2021 by the authors. Li- other molecules [12]. We studied the interactions of 12 ellagitannins (Figure1) with a lipid censee MDPI, Basel, Switzerland. extract of Escherichia coli (E. coli) by HR-MAS NMR. The ETs were selected to represent This article is an open access article different branches of the ET biosynthetic pathway and based on their studied hydropho- distributed under the terms and con- bicity [16–18]. In addition, the lipid interactions of pentagalloylglucose, the biosynthetic ditions of the Creative Commons At- tribution (CC BY) license (https:// precursor of ETs, were studied. The main aims were to study whether there are interactions creativecommons.org/licenses/by/ between ETs and lipids and whether these interactions can be studied by HR-MAS NMR. 4.0/).

Molecules 2021, 26, 373. https://doi.org/10.3390/molecules26020373 https://www.mdpi.com/journal/molecules Molecules 2021, 26, x FOR PEER REVIEW 2 of 13

Molecules 2021, 26, 373 2 of 13 interactions between ETs and lipids and whether these interactions can be studied by HR- MAS NMR.

R O 5 HO OH HO OH 4 O R O R1 R3O OR2 HO OH 2 O 1 R1=OH, R2=R3=G, R4~R5=(S)-HHDP 3 R1= OG, R2=R3=(S)-HHDP, R4~R5=(S)-HHDP O O O 4 R1= OG, R2=R3=G, R4~R5=(S)-HHDP GO OH 5 R O 1 GO O O O H OR OG O OR3 GO HO O OH O OG O O 5 OR4 OR2 HO OH 1 2 4 3 5 OH OH NHTP group 6 R =G, R ~R =DHHDP, R ~R =(R)-HHDP HO OH 1 2 4 3 5 7 R =G, R ~R =modified DHHDP, R ~R =(R)-HHDP OH 8 R1=R3=R5=G, R2~R4=modified DHHDP OH 9 OH HO HO OG gallagyl group HO 12 O HO O OG O HO OG O O OH O O HO O HO O OH HO O O HO O O O 1 HO GO R O O OH HO O OG GO OH HO O 10 O OH HO O O OH O OH OH

O HO OH OH HO O O O O OH HO OG O HO O O OH HO O OH HO O HO OH OH OH HO O R3O HO O R2O 1 OR G= O HO O O O HO HO HO O O O GO OH O GO O HO OHHO OH HO OH GO HO O HO OH O GO O G G=HHDP= GO OH 11 R1=G, R2~R3=(S)-HHDP GO O O O OH 13 R1=G, R2~R3=(S)-HHDP

chebuloyl= O O DHHDP = O O O O

HOOC OH O OH O HO OH HO HO OH O O OH OH OH HO O OH O

Figure 1. Chemical structures of 12 ellagitannins and pentagalloylglucose studied for their interaction with lipids: tellima- tellima- grandin I 1, vescalagin 2, casuarictin 3, tellimagrandin II 4, pentagalloylglucose 5, geraniin 6, chebulagic acid 7, chebulinic grandin I 1, vescalagin 2, casuarictin 3, tellimagrandin II 4, pentagalloylglucose 5, geraniin 6, chebulagic acid 7, chebulinic acid 8, punicalagin 9, oenothein B 10, sanguiin H-6 11, oenothein A 12, and lambertianin C 13. DHHDP = dehydrohexahy- acid 8, punicalagin 9, oenothein B 10, sanguiin H-6 11, oenothein A 12, and lambertianin C 13. DHHDP = dehydrohex- droxydiphenoyl, G = galloyl, HHDP = hexahydroxydiphenoyl, chebuloyl = modified dehydrohexahydroxydiphenoyl, ahydroxydiphenoyl,NHTP = nonahydroxytriphenoyl. G = galloyl, HHDP = hexahydroxydiphenoyl, chebuloyl = modiﬁed dehydrohexahydroxydiphenoyl, NHTP = nonahydroxytriphenoyl.

Molecules 2021, 26, x FOR PEER REVIEW 3 of 13

Molecules 2021, 26, 373 3 of 13

2. Results and Discussion 2.1.2. Results Characterization and Discussion of the Lipids in E. coli Lipid Extract 2.1. CharacterizationThe protons in the of the E. Lipidscoli lipid in E. extracts coli Lipid were Extract assigned mainly based on the 2D-corre- lationThe spectra protons measured in the withE. coli thelipid 600 extractsMHz inst wererument assigned (Chapter mainly 3.3), basedas the oncorrelation the 2D- spectracorrelation measured spectra with measured the 400 with MHz the HR-MAS 600 MHz instrument instrument (Sectiondid not achieve 3.3), as thegood correlation enough resolutionspectra measured even after with parameter the 400 MHzoptimization. HR-MAS Attempts instrument were did also not made achieve to analyze good enough a sam- pleresolution of pure evenL-α-phosphatidylethanolamine after parameter optimization. (PE) Attemptslipid in D were2O to alsoverify made the toassignations analyze a madesample from of pure the L-400α-phosphatidylethanolamine MHz HR-MAS measurements, (PE) lipidbut, because in D2O to of verify its poor the solubility assignations in water,made fromthese themeasurements 400 MHz HR-MAS did not measurements,produce the desired but, because outcome. of The its poor solubility solubility of the in purewater, PE these lipid measurements was not markedly did not increased produce the even desired after outcome.the addition The solubilityof 0.1 M ofphosphate the pure bufferPE lipid or wasafter not mixing markedly the pure increased PE lipid even with after different the addition ratios of of the 0.1 E. M coli phosphate lipid extract. buffer or afterThe mixing assigned the pure lipid PE protons lipid with are different displayed ratios in Figure of the E.2 colifromlipid spectra extract. measured with both Thea 600 assigned MHz instrument lipid protons equipped are displayed with a inCryo-Probe Figure2 from (a) spectraand a 400 measured MHz instrument with both equippeda 600 MHz with instrument an HR-MAS equipped probe with (b), a and Cryo-Probe their chemical (a) and shifts a 400 are MHz as instrumentfollows: (a) equipped 1H NMR 1 (MeOD-with and HR-MAS4, 600 MHz) probe δ (0.90b), and(m, theirH-CH chemical3), 1.29 (m, shifts H-CH are as2), follows:1.60 (m, ( aH-C3),) H NMR 2.03 (MeOD-(m, H- CHCHd4, 6002), MHz) 2.33 (m,δ 0.90 H-C2), (m, H-CH3.16 (m,3), H- 1.29β), (m,4.00 H-CH(m, H-G3),2), 1.60 4.05 (m, (m, H-C3), H-α), 2.034.19 (m,(m, H-CHCHH-G1), 4.442), (m,2.33 H-G1), (m, H-C2), 4.57 3.16(s, H- (m,γ), H-5.23β), (m, 4.00 H-G2), (m, H-G3), 5.35 (m, 4.05 H-CH); (m, H- αand), 4.19 (b) (m,1H NMR H-G1), (D 4.442O, (m,400 H-G1),MHz) 1 δ δ4.57 0.95 (s, (m, H- γH-CH), 5.233), (m, 1.35 H-G2), (m, H-CH 5.35 (m,2), H-CH);1.63 (m, and H-C3), (b) 2.08H NMR (m, (DH-CHCH2O, 4002 MHz)), 2.44 (m,0.95 H-C2), (m, H- 3.34CH3 (m,), 1.35 H-β (m,), 4.29 H-CH (m,2), H-G1), 1.63 (m, 4.51 H-C3), (m, H-G1), 2.08 (m, 5.37 H-CHCH (m, H-G2/H-CH).2), 2.44 (m, H-C2),Lipid protons 3.34 (m, H-G3, H-β), H-4.29α, (m,and H-G1),H-γ could 4.51 not (m, be H-G1), accurately 5.37 (m,assigned H-G2/H-CH). from the 400 Lipid MHz protons measurements H-G3, H- αowing, and toH- lowerγ could resolving not be accuratelypower and assigned the water from suppression the 400 MHz method measurements used (described owing in toChapter lower 3.3),resolving which power masked and the the H- waterγ signal suppression entirely. method Additionally, used (described the protons in Section H-G2 and 3.3), H-CH which γ formedmasked a the single H- peaksignal rather entirely. than Additionally, resolving into the two protons separate H-G2 peaks and H-CH in the formed 400 MHz a single HR- MASpeak ratherspectrum. than resolving into two separate peaks in the 400 MHz HR-MAS spectrum.

1 Figure 2. H1H NMR NMR spectra spectra of of the the E. Ecoli. coli lipidlipid extract extract with with assigned assigned proton proton signals signals at 25 at °C: 25 (◦aC:) measured (a) measured in MeOD- in MeOD-d4 withd4 600 MHz, (b) measured in D2O with 400 MHz HR-MAS probe along with (c) a highlighted ppm range of 3.7–4.7. Example with 600 MHz, (b) measured in D2O with 400 MHz HR-MAS probe along with (c) a highlighted ppm range of 3.7–4.7. structures of the lipids with labels are presented in Figure 5. * Possible distortion in the HR-MAS spectra caused by water Example structures of the lipids with labels are presented in Figure 5. * Possible distortion in the HR-MAS spectra caused presaturation masks some signals detected in the 600 MHz spectra. by water presaturation masks some signals detected in the 600 MHz spectra.

Molecules 2021, 26, 373 4 of 13

2.2. ET–Lipid Interaction Measurements by HR-MAS NMR Initially, the HR-MAS measurements were done as described in Section 3.3 for the whole tannin series 1–13. Preliminary selections were made based on these results as to which tannins showed the highest levels of interaction established by the magnitude of the chemical shift changes seen in the 1H spectra (Figures3 and A1) as well as whether the aromatic protons of tannins showed any measurable correlations to the lipid protons in the NOESY spectra. From these tests, the highest level of lipid interaction was detected with tellimagrandin I 1, casuarictin 3, tellimagrandin II 4, pentagalloylglucose 5, sanguiin H-6 Molecules 2021, 26, x FOR PEER REVIEW 5 of 13 11, and lambertianin C 13. The rest of the ellagitannins (2, 6–8, 10, and 12) did not show detectable levels of correlation in the NOESY experiments, which is why the subsequently measured repetitions (n = 4) were made only to the aforementioned six tannins.

0.10 0.09 1 0.08 0.07 3 0.06 4 0.05 (ppm) 5 0.04 Δδ 0.03 11 0.02 13 0.01 0.00

CH2 C3 CH C2 β G1 G1 CH CH G2 2 FigureFigure 3. 3. 1H1H NMR NMR chemical chemical shift shift deltas deltas (Δδ (∆δ, ,ppm) ppm) of of E.E. coli coli lipidlipid extracts extracts in in the pres presenceence ofof tellimagrandintellimagrandin II1 1,, casuarictincasuarictin3 , 3tellimagrandin, tellimagrandin II II4, pentagalloylglucose4, pentagalloylglucose5, sanguiin 5, sanguiin H-6 H-611, and11, lambertianinand lambertianin C 13. C The 13 tannin. The tannin numbering numbering refers torefers Figure to1 Figureand lipid 1 and proton lipid assignationsproton assignations refer to Figurerefer to 5. Figure Values 5. areValues presented are presented as ∆δ (average as Δδ (average values and values standard and standard error, n =error, 4). n = 4). 2.2.1. H Chemical Shift Deltas of E. coli Lipid Extract in the Presence of Tannins Measured tannins showed the highest cross relaxation rates against lipid protons H- 1 E. coli C3, H-C2,We determined H-G1, and howH-CH/H-G2, much the whichH signals again of would lipidindicate extracts that shiftedthey effectively in the presence pen- of the added ellagitannins or pentagalloylglucose. The shifting of lipid signals is a result etrate into the lipid bilayer at least enough to correlate with the lipid protons closer to the of the ring current effect from the aromatic ring structures of the ETs. The magnitude lipid headgroups. From these protons, the most prominent cross relaxation was measured of the signal shift informed how much the added tannin had an effect on the spatial against H-G1 for all the tannins. Tellimagrandin II 4 and pentagalloylglucose 5 showed surrounding of the lipid proton, thus indicating how far into the lipid bilayer the tannin the highest cross relaxation rates out of the selected tannins, which is most likely due to can penetrate [14]. Figure3 shows the chemical shift delta values for the selected tannins 1, their rather flexible structures caused by the many freely rotating galloyl groups and rel- 3, 4, 5, 11, and 13. It seems that the chemical shifts of H-C2, H-β, and H-G1 were the most atively high hydrophobicity, enabling them to enter the lipid bilayer structure effectively. affected by the addition of the tannins. This indicates that the tannins can penetrate into ETs 1 and 3 exhibited measurable cross relaxation rates, but they were the lowest out of the lipid bilayer at least until H-C2, but the following fatty acid chain beginning with H-C3 the selected tannins, which is probably due to their reasonably rigid structures when com- and the subsequent CH -chain, H-CHCH , and H-CH are not measurably affected. pared with 4 and 5. The 2larger ETs dimeric2 11 and trimeric 13 displayed moderate cross Mainly, the monomeric ETs (1, 3 and 4) and pentagalloylglucose (5) induced larger relaxationchanges thanrates, the with oligomeric the trimeric ETs 13 ( 11beingand almost13). Additionally, as high as the within most effective the monomeric monomers. ETs, Bothit seemed of the ET that oligomers the more that hydrophobic showed NOE ones co causerrelations more with change the lipid in the protons chemical had shifts multi- of plelipid free protons galloyl thangroups. the In less comparison, hydrophobic dimeric ones. oenothein Tellimagrandin B 10 and II (trimeric4) caused oenothein the largest A 12change did not in display the lipid any proton measurable chemical NOE shifts correlations. out of all theBoth ETs. 10 Thisand 12 was have exceeded only one only free by galloylpentagalloylglucose group per monomeric (5) in regard unit toand some a macroc of theyclic lipid structure protons. caused These trendsby two followed oligomeric the linkagesassumption between that morethe monomeric hydrophobic units, compounds which ma wouldkes their interact structures more rigid. with lipids,Similarly, and ves- thus calaginpenetrate 1 and more punicalagin into the lipid 9 did bilayer not show structure. NOE It correlations is worth noting with that the the lipid, structural which ﬂexibil-is ex- plainedity of the by studiedtheir rigid monomeric structures, tannins with the increases former inhaving the same an HHDP order asgroup the hydrophobicityand an NHTP group,(1→3→ while4→5 ),the and latter probably has an also HHDP contributes group an signiﬁcantlyd a gallagyl to group. the rate However, of lipid interaction the lack of of correlationthe tannin. from However, geraniin hydrophobicity 6, chebulagic aloneacid 7 does, and not chebulinic determine acid whether 8 was surprising an ET interacts be- causemeasurably they have with beenE. shown coli or to of be the rather magnitude hydrophobic of the interaction.and, additionally, This canall three be seen of them from haveAppendix varyingA Figureamounts A1 of, wherefree galloyl geraniin groups6, chebulagic in their structure, acid 7, making and chebulinic them moderately acid 8 had flexible.almost noIt might effect be on that the chemicalthe DHHDP shifts group of the of lipid6 and protons, the chebuloyl although group allthree of 7 and of these 8 make ETs themhave less been likely shown to penetrate to be highly into hydrophobic the lipid bilayer [18]. when compared with the other monomeric tannins (1, 3, 4, and 5) shown to interact with the lipid protons. For the monomeric ETs 1, 3, and 4, cross relaxation rates could be calculated separately for some of the aromatic protons, which are defined in Chapter 3.3. The differences in the cross relaxation rates between these aromatic protons informs us how the studied ET is oriented in the lipid bilayer. It seems that the cross relaxation rates of the cross peaks correlating from the ETs’ galloyl group protons (1 cross peak 1, 3 cross peak 1, 4 cross peaks 1 and 2) are higher than the cross peaks from the ETs’ HHDP group protons (1 cross peaks 1 and 2, 3 cross peak 2, 4 cross peaks 3 and 4), which suggests that the galloyl groups of ETs are oriented more towards the lipid than their HHDP groups.

Molecules 2021, 26, 373 5 of 13

2.2.2. NOESY Cross Relaxation Rates of the Aromatic Protons of Tannins The NOESY experiment is particularly useful for determining if the ET and lipid are spatially close to each other, because typically, an NOE correlation is only detected when the correlating protons are spaced closer than 5 Å apart. NOESY cross relaxation rates for tellimagrandin I 1, casuarictin 3, tellimagrandin II 4, pentagalloylglucose 5, sanguiin H-6 11, and lambertianin C 13 were calculated based on Equation described in Section 3.3, and are presented in Figure4 for the mixing times of 0.1 s ( a) and 0.3 s (b). Theoretically, a higher cross relaxation rate indicates that the correlating protons are closer to each other, and thus correlate more intensely. The higher mixing time of 0.3 s proved to be more effective in reaching a higher measureable cross relaxation rate with the exception of 13, which had equally high rates with both measured mixing times. This is probably because of the relatively high molecular weight of 2805.90 Da of 13. Already with 11 having the molecular weight of 1871.27 Da, the difference between mixing times was not as distinct as with the smaller tannins. Measured tannins showed the highest cross relaxation rates against lipid protons H-C3, H-C2, H-G1, and H-CH/H-G2, which again would indicate that they effectively penetrate into the lipid bilayer at least enough to correlate with the lipid protons closer to the lipid headgroups. From these protons, the most prominent cross relaxation was measured against H-G1 for all the tannins. Tellimagrandin II 4 and pentagalloylglucose 5 showed the highest cross relaxation rates out of the selected tannins, which is most likely due to their rather flexible structures caused by the many freely rotating galloyl groups and relatively high hydrophobicity, enabling them to enter the lipid bilayer structure effectively. ETs 1 and 3 exhibited measurable cross relaxation rates, but they were the lowest out of the selected tannins, which is probably due to their reasonably rigid structures when compared with 4 and 5. The larger ETs dimeric 11 and trimeric 13 displayed moderate cross relaxation rates, with the trimeric 13 being almost as high as the most effective monomers. Both of the ET oligomers that showed NOE correlations with the lipid protons had multiple free galloyl groups. In comparison, dimeric oenothein B 10 and trimeric oenothein A 12 did not display any measurable NOE correlations. Both 10 and 12 have only one free galloyl group per monomeric unit and a macrocyclic structure caused by two oligomeric linkages between the monomeric units, which makes their structures rigid. Similarly, vescalagin 1 and punicalagin 9 did not show NOE correlations with the lipid, which is explained by their rigid structures, with the former having an HHDP group and an NHTP group, while the latter has an HHDP group and a gallagyl group. However, the lack of correlation from geraniin 6, chebulagic acid 7, and chebulinic acid 8 was surprising because they have been shown to be rather hydrophobic and, additionally, all three of them have varying amounts of free galloyl groups in their structure, making them moderately flexible. It might be that the DHHDP group of 6 and the chebuloyl group of 7 and 8 make them less likely to penetrate into the lipid bilayer when compared with the other monomeric tannins (1, 3, 4, and 5) shown to interact with the lipid protons. For the monomeric ETs 1, 3, and 4, cross relaxation rates could be calculated separately for some of the aromatic protons, which are defined in Section 3.3. The differences in the cross relaxation rates between these aromatic protons informs us how the studied ET is oriented in the lipid bilayer. It seems that the cross relaxation rates of the cross peaks correlating from the ETs’ galloyl group protons (1 cross peak 1, 3 cross peak 1, 4 cross peaks 1 and 2) are higher than the cross peaks from the ETs’ HHDP group protons (1 cross peaks 1 and 2, 3 cross peak 2, 4 cross peaks 3 and 4), which suggests that the galloyl groups of ETs are oriented more towards the lipid than their HHDP groups. Molecules 2021, 26, x FOR PEER REVIEW 6 of 13 Molecules 2021 , 26, 373 6 of 13

1 3 4

) a a a 1 ) ) − 1

0.06 1 − − 0.10 0.06 0.05 0.08 0.04 cross peak 1 0.04 0.06 cross peak 1 0.03 cross peak 1 cross peak 2 0.04 0.02 cross peak 2 0.02 cross peak 2 cross peak 3 cross peak 3 0.01 0.02 cross peak 4 0.00 cross relaxation rate (s 0.00 0.00 cross relaxation rate (s rate cross relaxation cross relaxation rate (s rate cross relaxation

CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH

CH2 G2 CH2 G2 CH2 G2 b b b ) ) )

1 0.06 1 1 − − 0.08 − 0.20 0.05 0.16 0.06 0.04 cross peak 1 cross peak 1 0.12 0.04 0.03 cross peak 1 cross peak 2 cross peak 2 cross peak 2 cross peak 3 0.02 0.08 0.02 cross peak 3 cross peak 4 0.01 0.04 0.00

cross relaxation rate (s rate cross relaxation 0.00 crossrelaxation rate (s 0.00 crossrelaxation rate (s CH CH C3 CH C2 β G1 CH CH CH C3 CH C2 G1 CH CH3 CH2 C3 CH C2 β G1 CH 3 2 3 2 β CH G2 CH G2 CH2 G2 2 2

5 11 13 a a a ) ) )

1 0.24 1 0.08 1 0.20 − − − 0.20 0.06 0.16 0.16 0.12 0.04 0.12 cross peak 1 cross peak 1 cross peak 1 0.08 0.08 0.02 0.04 0.04 0.00 0.00 0.00 crossrelaxation rate (s crossrelaxation rate (s crossrelaxation rate (s

CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH

CH2 G2 CH2 G2 CH2 G2

) b ) b ) b

1 0.20 1 0.16 1 0.20 − − 0.14 − 0.16 0.12 0.16 0.12 0.10 0.12 0.08 cross peak 1 cross peak 1 cross peak 1 0.08 0.06 0.08 0.04 0.04 0.04 0.02 0.00 0.00 0.00 cross relaxation rate (s cross relaxation rate (s cross relaxation rate (s

CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH

CH2 G2 CH2 G2 CH2 G2

Figure 4. Cross relaxation rates of the aromatic protons (cross peaks 1–4 refer to Section 3.3 for deﬁnitions) of tellimagrandin I 1, casuarictin 3, tellimagrandin II 4, pentagalloylglucose 5, sanguiinFigure H-6 11 ,4. and Cross lambertianin relaxation Crates13 against of the aromatic different protons lipid protons (cross with peaks mixing 1–4 refer times to ofChapter (a) 0.1 s3.3 and for ( bdefinitions)) 0.3 s. The of tannin tellimagrandin numbering I refers1, casuarictin to Figure 3,1 tellimagrandin and lipid proton II 4 assignations, refer to Figurepentagalloylglucose5. Cross peak labels 5, sanguiin are deﬁned H-6 in11, Section and lambertianin 3.3. Values areC 13 presented against different as s −1 (average lipid prot valuesons with and standardmixing times error, of n ( =a) 4). 0.1 s and (b) 0.3 s. The tannin numbering refers to Figure 1 and lipid proton assignations refer to Figure 5. Cross peak labels are defined in Chapter 3.3. Values are presented as s−1 (average values and standard error, n = 4).

Molecules 2021, 26, x FOR PEER REVIEW 8 of 13

labels for all the protons that were assigned shown in the PE lipid as examples. Because of the unknown component of the lipid extract, we purchased three different batches of

Molecules 2021, 26, 373 the lipid extract with which we performed the replicate HR-MAS measurements in order7 of 13 to account for the possible biological variability in the lipid extract. 1H HR-MAS NMR spectra of the different batches used in the study are shown in Figure S2 in Supplementary Materials. Pure PE lipid was purchased from Avanti Polar Lipids (Alabaster, AL, USA).

Figure 5.5. ExampleExample structuresstructures of of the the known known components components of of the the commercial commercialE. coliE. colilipid lipid extract, extract, L-α -phosphatidylethanolamineL-α-phosphatidylethanola- (PE),mine L-(PE),α-phosphatidylglycerol L-α-phosphatidylglycerol (PG), and(PG), cardiolipin and cardiolipin (CA), (CA), with labelingwith labeling on the on PE the lipid PE tolipid illustrate to illustrate the NMR the assignedNMR assigned protons. The length and precise structure of the fatty acid chains (CH2) have not been determined. protons. The length and precise structure of the fatty acid chains (CH2) have not been determined.

3.2. Isolation of Ellagitannins and Pentagalloylglucose 2.2.3. Effect of ET Concentration on the ET–Lipid Interaction The purification effect of concentration of the selected of the ETs added and pentagalloylglucose tannins was tested withfollowed tellimagrandin our previously II 4 whilereported keeping extraction the amount and isolation of E. coli methodology,lipid extract constant which briefly (4.0 mg). goes The as weighed follows: amountsextraction, of Sephadex4 were 0.5 LH-20 mg, 1.0 funnel mg, and and 2.0 column mg. Appendix chromatography,B Figure A2 and shows preparative the chemical and semiprepara- shift deltas tiveof the HPLC lipid [5,6,8,10,16,19,20]. extracts protons in Purified the presence tannins of were the differentcharacterized amounts based of on4 in their the retention solution. Appendixtimes in reverse-phaseB Figures A3 LC,and UVA4 showspectra, the molecular cross relaxation ions, and rates characteristic of aromatic fragments protons of es-4 tablishedagainst different in our previous lipid protons work in [5]. the Monomeric same three tannins concentrations. tellimagrandin Both the I (1 chemical) and tellima- shift grandinchanges ofII the(4) lipidwere protonsisolated and from the meadowsweet cross relaxation inflorescence; rates of aromatic vescalagin protons (2 of) 4fromagainst the flowerslipid protons and leaves increase of purple when theloosestrife; concentration casuarictin of 4 increases.(3) from sea However, buckthorn it isleaves; noteworthy penta- galloylglucosethat the increase (5 in) from cross tannic relaxation acid ratespurchased when movingfrom J.T. from Baker 1.0 (Denventer, mg to 2.0 mg Holland); is drastically and geraniinlarger than (6), whenchebulagic moving acid from (7), 0.5chebulinic to 1.0 mg, acid so (8 the), and concentration punicalagin effect(9) from might Terminalia not be chebuladirectly fruits. linear. Oligomeric ellagitannins oenothein B (10) and oenothein A (12) were isolated from willow herb flowers and sanguiin H-6 (11) and lambertianin C (13) from rasp- berry2.2.4. Effectleaves. of The Lipid stereochemistries Batch on the ET–Lipid of acyclic Interaction ellagitannins vescalagin (2) and castalagin wereWe reinvestigated used three different by Matsuo batches et al. of2015, the lipidshowing extract that and the performedNHTP-group the exists replicate in (S,R) HR- configurationMAS measurements and vescalagin in order to2 is account presented for according the possible to biologicalthis revision variability [21]. 1H NMR in the spec- lipid traextract. of compounds1H HR-MAS 1–13 NMR are shown spectra in of Figures these different S3–S15 Ein. coliSupplementaryextract batches Materials. (a–c) measured ◦ in D2O at 25 C are shown in Figure S2 in Supplementary Materials. We noticed that the 3.3.interaction NMR Analyses between ETs and lipids was the strongest in those lipid extracts that showed less additionalNMR measurements unassigned were peaks performed in the 1H with spectrum either (Figurea Bruker S2a,c). Avance-III Most probably,spectrometer the equippedunknown componentwith a Prodigy in the TCI lipid (inverted extract CryoProb (Figure S2b)e) cooled also interacted via liquid with nitrogen, ETs and which affected was the interactions between ETs and lipids.

2.3. Analysis of Stability of ET–Lipid Solution with UPLC-DAD-MS The stability of the ellagitannins was monitored to verify that no metabolites or degra- dation products were formed in the solution with the lipid extract and that all the detected Molecules 2021, 26, 373 8 of 13

interactions were caused by the original ET. Sample preparation and measurements were performed as described in Section 3.4. The samples were made in H2O to resemble the conditions in which the HR-MAS NMR measurements were done, but some of the studied ETs and pentagalloylglucose are so hydrophobic that some decline in solubility was expected during the stability measurements. During the 40 h analysis, only traces of ET metabolite products were detected and example chromatograms of the analysis of tellimagrandin II (4) are shown in Figure S1 in Supplementary Materials. This conﬁrmed that ETs were stable in the conditions used and that the detected interaction was indeed caused by the studied initial ET.

3. Materials and Methods 3.1. Chemicals Technical grade acetone was purchased from VWR (Haasrode, Belgium). Analyti- cal grade, analytical grade methanol, HPLC gradient grade acetonitrile, HPLC gradient grade methanol, HPLC grade formic acid, HPLC grade phosphoric acid, and LC-MS grade formic acid were purchased from VWR International (Fontenay-Sous-Bois, Paris, France). LC-MS grade acetonitrile was purchased from Merck (KGaA, Darmstadt, Ger- many). D2O (99.90% purity), MeOD-d4 (99.80% purity, water < 0.03%), and acetone-d6 (99.80% purity, water < 0.02%) were purchased from Eurisotop, a subsidiary of Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA). Ultra-pure type I water was puriﬁed with Merck Millipore Synergy UV system. Commercial E. coli lipid extract was purchased from Avanti Polar Lipids (Alabaster, AL, USA). E. coli extract contained L-α-phosphatidylethanolamine (PE, 57.5 w-%), L-α- phosphatidylglycerol (PG, 15.1 w-%), and cardiolipin (CA, 9.8 w-%), and the remaining 17.6 w-% consisted of an unidentiﬁed lipid, according to the manufacturer. Figure5 shows the representative structures of PE, PG, and CA lipids in the extract mixture along with labels for all the protons that were assigned shown in the PE lipid as examples. Because of the unknown component of the lipid extract, we purchased three different batches of the lipid extract with which we performed the replicate HR-MAS measurements in order to account for the possible biological variability in the lipid extract. 1H HR-MAS NMR spectra of the different batches used in the study are shown in Figure S2 in Supplementary Materials. Pure PE lipid was purchased from Avanti Polar Lipids (Alabaster, AL, USA).

3.2. Isolation of Ellagitannins and Pentagalloylglucose The purification of the selected ETs and pentagalloylglucose followed our previously reported extraction and isolation methodology, which briefly goes as follows: extraction, Sephadex LH-20 funnel and column chromatography, and preparative and semipreparative HPLC [5,6,8,10,16,19,20]. Purified tannins were characterized based on their retention times in reverse-phase LC, UV spectra, molecular ions, and characteristic fragments established in our previous work [5]. Monomeric tannins tellimagrandin I (1) and tellimagrandin II (4) were isolated from meadowsweet inflorescence; vescalagin (2) from the flowers and leaves of purple loosestrife; casuarictin (3) from sea buckthorn leaves; pentagalloylglucose (5) from tannic acid purchased from J.T. Baker (Denventer, Holland); and geraniin (6), chebulagic acid (7), chebulinic acid (8), and punicalagin (9) from Terminalia chebula fruits. Oligomeric ellagitannins oenothein B (10) and oenothein A (12) were isolated from willow herb flowers and sanguiin H-6 (11) and lambertianin C (13) from raspberry leaves. The stereochemistries of acyclic ellagitannins vescalagin (2) and castalagin were reinvestigated by Matsuo et al. 2015, showing that the NHTP-group exists in (S,R) configuration and vescalagin 2 is presented according to this revision [21]. 1H NMR spectra of compounds 1–13 are shown in Figures S3–S15 in Supplementary Materials.

3.3. NMR Analyses NMR measurements were performed with either a Bruker Avance-III spectrometer equipped with a Prodigy TCI (inverted CryoProbe) cooled via liquid nitrogen, which was Molecules 2021, 26, 373 9 of 13

operated at 600.16 MHz for 1H and 125.76 MHz for 13C or a Bruker Avance-III spectrometer equipped with a high-resolution magic angle spinning (HR-MAS) probe, which was operated at 399.75 MHz for 1H and 100.52 MHz for 13C. Typical 1H and 13C spectra were recorded in addition to multiple 2D spectra including COSY (correlation spectroscopy), NOESY (nuclear Overhauser effect spectroscopy), HSQC (heteronuclear single quantum coherence), and HMBC (heteronuclear multiple bond correlation). Measurements were ◦ done at 25 C in either MeOD-d4 or D2O. The following sample preparation method was adapted from Grélard et al., 2010 [22]. For HR-MAS measurements, 4.0 mg of E. coli lipid extract was weighed into an eppendorf along with 1.0, 2.0, or 3.0 mg of the studied tannin. The amount of tannin utilized depended on the degree of oligomerisation of tannins; 1.0 mg was used for monomeric tannins (1–9), 2.0 mg for dimeric tannins (10–11), and 3.0 mg for trimeric tannins (12–13). The lipid extract/tannin mixture was dissolved in 100 µL of D2O and subsequently handled via a freeze–thaw method in order for the lipids to form a bilayer [22]. The method consisted of shaking the sample vigorously in room temperature, freezing it in liquid nitrogen, and heating it in a warm water bath. This cycle was then repeated four times until a hazy emulsion was formed. The emulsion was transferred into an HR-MAS insert (50 µL, Bruker), which was subsequently placed into a ZrO2 HR-MAS rotor (4 mm, Bruker). The rotor was placed in the instrument, the MAS unit was operated at 9 kHz rotational speed, and the temperature was set to 25 ◦C. A typical measurement set consisted of a standard 1H experiment with water presaturation (zgpr), followed by two NOESY experiments with mixing times of 0.1 s and 0.3 s, and ﬁnally a second 1H experiment with water presaturation. During HR-MAS NMR data processing, the 1H spectra were calibrated based on the CH3 peaks value (δ = 0.9445 ppm) for all the tannin samples as well as the referenced pure lipid sample against which the chemical shift deltas were calculated. The CH3 peak was used as a reference as the methyl group end of the fatty acid chain of the lipids (Figure5) resides deepest in the formed lipid bilayer structure, and should thus be the least exposed to the tannins’ inﬂuence, and the solvent signal of D2O was suppressed and thus unusable. Cross relaxation rates were calculated from the NOESY spectra based on the following equation [15]:

cross peak volume number of cross peak protons cross relaxation rate = diagonal peak volume ∗ mixing time

From the NOESY spectra, lipid signal volumes (diagonal peak volume, abs) and their correlation signals to the tannins’ aromatic protons (cross peak volumes, abs) were integrated for both used mixing times (0.1 s and 0.3 s). For tellimagrandin I 1, casuarictin 3, and tellimagrandin II 4, the cross peaks could be integrated individually for some of the aromatic protons because the aromatic protons could be identiﬁed and were resolved enough in the 1H spectra, and they are labeled in Figure4 as follows. For tellimagrandin I, cross peak 1 refers to the protons of galloyl group attached to O2 and O3 of glucose, cross peak 2 to the proton of HHDP group attached to O6 of glucose, and cross peak 3 to the proton of HHDP group attached to O4 of glucose. For casuarictin, cross peak 1 refers to the proton of galloyl group attached to O1 of glucose and cross peak 2 to the protons of HHDP group attached to O2, O3, O4, and O6 of glucose. For, tellimagrandin II cross peak 1 refers to the proton of galloyl group attached to O1 of glucose, cross peak 2 to the protons of galloyl group attached to O2 and O3 of glucose, cross peak 3 to the proton of HHDP group attached to O6 of glucose, and cross peak 4 to the proton of HHDP attached to O4 of glucose. For pentagalloylglucose 5, sanguin H-6 11, and lambertianin C 13, the aromatic protons could not be separated based on the resolution achieved with the 400 MHz HR-MAS instrument, so the cross relaxation rates were calculated for all the aromatic protons as a group and labeled as cross peak 1. Molecules 2021, 26, 373 10 of 13

3.4. UPLC-DAD-MS Analyses The UPLC-DAD used in all analyses was an Acquity UPLC (Waters Corp., Milford, MA, USA) instrument consisting of a binary solvent manager, a column, and a diode array detector. The column utilized was an Aquity BEH phenyl column (2.1 × 100 mm, 1.7 µm; Waters Corp., Wexford, Ireland). The mobile phase consisted of acetonitrile (A) and 0.1% aqueous formic acid (B) with a constant flow rate of 0.5 mL min−1 with the following gradient: 0–0.5 min: 0.1% A; 0.5–5.0 min: 0.1–30% A (linear gradient); 5.0–5.1 min: 30–90% A (linear gradient); 5.1–7.1 min: 90% A; 7.1–7.2 min: 90–0.1% A (linear gradient); and 7.2–8.5 min: 0.1 % A. Column temperature was 25 ◦C. The UPLC was connected to a Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific GmbH, Bremen, Germany) via a heated ESI (electrospray ionization) source. Sheath gas and auxiliary gas flow rates were set to 60 Au and 20 Au, respectively, and spray voltage was set to 3.0 kV. The DAD detector was set to collect UV-data from 190–500 nm and the Orbitrap was operated with full scan at a mass range of 150–2000 Da with a resolution of 70,000. For the stability measurements, 0.5 mg of tannin was weighed together with 2 mg of the E. coli lipid extract in an eppendorf and dissolved in 2.5 mL of H2O. A fivefold dilution of this sample was made and filtered through a 0.2 µM filter (PTFE). This sample was injected hourly for 40 h to the UPLC-DAD-MS instrument. After the 20th injection, a new sample was filtered from the originally prepared tannin–lipid mixture to account for the effect that the filtration might have had on the sample.

3.5. Data Analysis and Software UPLC measurements and quantitations were done with Thermo Xcalibur version 4.1.31.9 (Thermo Fisher Scientiﬁc Inc., Waltham, MA, USA). NMR data were measured and analyzed using TopSpin software versions 3.5 pl 7 and 3.5 pl 5 (Bruker, Billerica, MA, USA). Graph visualizations were done with Origin 2016 (64-bit) software version SR2 b9.3.2.303 (OriginLab, Northampton, MA, USA).

4. Conclusions The main aim of this study was to uncover if the possible interactions between ellagitannins and lipids can be detected and studied with the help of HR-MAS NMR. The results showed that some ETs were more inclined to penetrate into the lipid bilayer than other ETs. The main deciding structural factors based on the results of the ETs studied here were that ETs with a cyclic polyol and high structural ﬂexibility are more likely to interact with lipids. This result is well in line with the studied hydrophobicity of these compounds, i.e., more hydrophobic ETs were shown to interact more with lipids. Most of the detected interaction seemed to happen towards the headgroups of the lipids and less with the fatty acid chain. As a technique, HR-MAS NMR proved to be very suitable for the study of the ET–lipid mixtures owing to the fact that it tolerates the sort of semisolid emulsions these compounds form in an aqueous measurement environment.

Supplementary Materials: The following materials are available online. Figure S1: UPLC-DAD chromatograms at 280 nm from stability measurements. Figure S2: 1H HR-MAS NMR spectra of the different E. coli extract batches used in this study. Figures S3–S15: 1H NMR spectrum of compounds 1–13. Author Contributions: Conceptualization, M.K.; methodology, E.P. and M.K.; software, E.P., S.R., and V.V.; validation, E.P., S.R., and V.V.; formal analysis, S.R. and V.V.; investigation, V.V.; resources, M.K.; data curation, S.R. and V.V.; writing—original draft preparation, V.V.; writing—review and editing, E.P., M.K., S.R., and V.V.; visualization, V.V.; supervision, E.P. and M.K.; project administration, M.K.; funding acquisition, M.K. All authors have read and agreed to the published version of the manuscript. Funding: The study was funded by the Academy of Finland (grant number 310549 to Maarit Karonen). Molecules 2021, 26, x FOR PEER REVIEW 11 of 13 Molecules Molecules 20212021, 26, 26, x, 373FOR PEER REVIEW 11 11of of13 13

Institutional Review Board Statement: Not applicable. InstitutionalInstitutional Review Review Board Board Statement: Statement: NotNot applicable. applicable. Informed Consent Statement: Not applicable. InformedInformed Consent Consent Statement: Statement: NotNot applicable. applicable. Data Availability Statement: The data presented in this study are available on request from the Data Availability Statement: The data presented in this study are available on request from the correspondingData Availability author. Statement: The data presented in this study are available on request from the correspondingcorresponding author. author. Acknowledgments: Jari Sinkkonen is greatly acknowledged for his help regarding NMR analysis Acknowledgments: Jari Sinkkonen is greatly acknowledged for his help regarding NMR analysis inAcknowledgments: the writing of the fundingJari Sinkkonen application is greatlyfor the Ac acknowledgedademy of Finland for his (M.K.). help regardingJani Rahkila NMR is acknowl- analysis in the writing of the funding application for the Academy of Finland (M.K.). Jani Rahkila is acknowl- edgedin the for writing his kind of help the fundingwith NMR application instruments for an thed in Academy the HR-MAS of Finland NMR (M.K.).analysis. Jani Optifeed Rahkila (Acad- is ac- edged for his kind help with NMR instruments and in the HR-MAS NMR analysis. Optifeed (Acad- emyknowledged of Finland, for project his kind no help298177) with is NMRthanked instruments for supplying and insome the HR-MASof the purified NMR hydrolysable analysis. Optifeed tan- emy of Finland, project no 298177) is thanked for supplying some of the purified hydrolysable tannins(Academy used in of the Finland, study. projectAll the no personnel 298177) isand thanked students for supplyingin the Natural some Chemistry of the purified Research hydrolysable Group nins used in the study. All the personnel and students in the Natural Chemistry Research Group whotannins have used participated in the study. in the All collection the personnel of plant and ma studentsterials or in the the purifications Natural Chemistry of the compounds Research Group are who have participated in the collection of plant materials or the purifications of the compounds are thankedwho have for participatedtheir kind help. in the collection of plant materials or the purifications of the compounds are thanked for their kind help. thanked for their kind help. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the designConflicts of the of study; Interest: in Thethe collection, authors declare analyses, no conflict or interpretation of interest. Theof data; funders in the had writing no role of in the the manu- design design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript;of the or study; in the in decision the collection, to publish analyses, the results. or interpretation of data; in the writing of the manuscript; or script; or in the decision to publish the results. in the decision to publish the results. Sample Availability: Samples of the compounds are not available from the authors. Sample Availability: Samples of the compounds are not available from the authors. Sample Availability: Samples of the compounds are not available from the authors. Appendix A Appendix A Appendix A

0.05 0.05 2 0.04 2 0.04 6 6 0.03 7 0.03 7 8 (ppm) 8

(ppm) 0.02

Δδ 0.02 9

Δδ 9 0.01 10 0.01 10 12 0.00 12 0.00 CH C3 CH C2 β G1 G1 CH CH 2 C3 CH C2 β G1 G1 CH 2 CH G2 CH 2 G2 2 Figure A1. 1H NMR chemical shift deltas (Δδ, ppm) of E. coli lipid extracts in the presence of vescalagin 2, geraniin 6, FigureFigure A1. A1. 1H1H NMR NMR chemical chemical shift shift deltas deltas (Δδ (∆, δppm), ppm) of of E.E. coli coli lipidlipid extracts extracts in in the the presence presence of of vescalagin vescalagin 2,2 geraniin, geraniin 6,6 , chebulagic acid 7, chebulinic acid 8, punicalagin 9, oenothein B 10, and oenothein A 12. The tannin numbering refers to chebulagicchebulagic acid acid 7,7 chebulinic, chebulinic acid acid 8,8 punicalagin, punicalagin 9,9 oenothein, oenothein B B1010, and, and oenothein oenothein A A 1212. .The The tannin tannin numbering numbering refers refers to to Figure 1 and lipid proton assignations refer to Figure 5. Values are presented as Δδ (n = 1). FigureFigure 11 and and lipid lipid proton proton assignations assignations refe referr to to Figure Figure 55.. Values areare presentedpresented asas Δδ∆δ ((nn == 1). 1). Appendix B Appendix B Appendix B

0.10 0.100.09 0.090.08 0.080.07 0.5 mg 0.07 0.5 mg 0.06 1.0 mg 0.060.05 1.0 mg (ppm) 0.05 2.0 mg (ppm) 0.04 2.0 mg Δδ 0.04 Δδ 0.03

0.030.02 0.020.01 0.010.00 0.00 CH C3 CH C2 β G1 G1 CH CH 2 C3 CH C2 β G1 G1 CH 2 CH G2 CH 2 G2 2 FigureFigure A2. A2. 1H1H NMR NMR chemical chemical shift shift deltas deltas (Δδ (∆,δ ppm), ppm) of of E.E. coli coli lipidlipid extracts extracts in in the the presen presencece of of three three different different weighed weighed Figure A2. 1H NMR chemical shift deltas (Δδ, ppm) of E. coli lipid extracts in the presence of three different weighed amounts of tellimagrandin II 44. amountsamounts of of tellimagrandin tellimagrandin II II 4. .

MoleculesMoleculesMolecules2021 2021 2021, ,26 ,26 26,, 373 ,x x FOR FOR PEER PEER REVIEW REVIEW 1212 ofofof 13 1313

aa bb 0.320.32 0.200.20 0.18 0.28 ) )

1 0.18 1 0.28 ) ) − − 1 1 − 0.160.16 − 0.240.24 0.140.14 0.120.12 0.200.20 0.5 0.5 mg mg 0.5 0.5 mg mg 0.100.10 1.0 mg 0.160.16 1.0 mg 0.080.08 1.0 mg 1.0 mg 2.0 mg 2.0 mg 0.060.06 2.0 mg 0.120.12 2.0 mg 0.040.04 0.080.08 0.020.02 cross relaxation rate (s rate cross relaxation (s rate cross relaxation cross relaxation rate (s rate cross relaxation 0.000.00 (s rate cross relaxation 0.040.04 -0.02-0.02 0.000.00 CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH2 G2 CH2 G2 CH2 G2 CH2 G2

cc dd 0.28 0.28 0.280.28 ) ) 1 1

) 0.24 ) − − 1 0.24 1 0.24

− − 0.24 0.20 0.20 0.200.20 0.16 0.16 0.16 0.5 0.5 mg mg 0.16 0.5 0.5 mg mg 0.120.12 1.0 1.0 mg mg 0.120.12 1.0 1.0 mg mg 2.0 mg 2.0 mg 0.080.08 2.0 mg 0.080.08 2.0 mg 0.04 0.040.04 0.04 cross relaxation rate (s rate relaxation cross (s rate relaxation cross

cross relaxation rate (s rate relaxation cross (s rate relaxation cross 0.00 0.000.00 0.00 -0.04 -0.04-0.04 -0.04 CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH2 G2 CH2 G2 CH2 G2 CH2 G2

Figure A3. a b c FigureFigure A3. A3.Cross Cross Cross relaxationrelaxation relaxation rates rates rates of of of the the the aromatic aromatic aromatic protons protons protons (cross (cross (cross peak peak peak 1 1 1 ( ( (a),a),), cross cross cross peak peak peak 2 2 2 ( ( (b),b),), cross cross cross peak peak peak 3 3 3 ( ( (c),c),), and and and cross cross cross peak peak peak 444 ( d( (d))d)))) of of of three three three different different different weighed weighed weighed amounts amounts amounts of of of tellimagrandin tellimagrandin tellimagrandin II II II4 4 4against against againstdifferent different differentlipid lipid lipidprotons protons protonswith with with aa a mixingmixing mixing timetime time ofof of 0.10.1 0.1 s.s. s. TheTheThe lipid lipid lipid proton proton proton assignations assignations assignations refer refer refer to to to Figure Figure Figure5 and5 5 an and crossd cross cross peak peak peak labels labels labels are are are deﬁned defined defined in in Sectionin Chapter Chapter 3.3. 3.3. 3.3.

aa bb 0.55 0.350.35 0.55 0.500.50 ) ) 1 1 ) 0.30 ) − − 1 0.30 1 0.45 − − 0.45 0.40 0.250.25 0.40 0.350.35 0.20 0.5 0.5 mg mg 0.5 0.5 mg mg 0.20 0.300.30 1.0 1.0 mg mg 1.0 1.0 mg mg 0.250.25 0.150.15 2.0 mg 2.0 mg 2.0 mg 0.200.20 2.0 mg 0.100.10 0.150.15 0.10 cross relaxation rate (s cross relaxation rate (s 0.10 cross relaxation rate (s 0.050.05 cross relaxation rate (s 0.050.05 0.000.00 0.000.00 CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH2 G2 CH2 G2 CH2 G2 CH2 G2

cc dd 0.35 0.35 0.400.40 ) ) 1 1 ) 0.30 ) − − 1 0.30 1 0.35 − − 0.35 0.30 0.250.25 0.30 0.5 mg 0.25 0.5 mg 0.200.20 0.5 mg 0.25 0.5 mg 1.0 1.0 mg mg 0.20 1.0 1.0 mg mg 0.15 0.20 0.15 2.0 2.0 mg mg 2.0 2.0 mg mg 0.150.15 0.100.10 0.100.10 cross relaxation rate (s rate cross relaxation (s rate cross relaxation

cross relaxation rate (s rate cross relaxation 0.05 (s rate cross relaxation 0.05 0.050.05 0.000.00 0.000.00 CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH3 CH2 C3 CH C2 β G1 CH CH2 G2 CH2 G2 CH2 G2 CH2 G2

FigureFigureFigure A4. A4. A4.Cross Cross Cross relaxationrelaxation relaxation rates rates rates of of of the the the aromatic aromatic aromatic protons protons protons (cross (cross (cross peak peak peak 1 1 1 ( a( (a),a),), cross cross cross peak peak peak 2 2 2 ( b( (b),b),), cross cross cross peak peak peak 3 3 3 ( c( (c),c),), and and and cross cross cross peak peak peak 444 ( d( (d))d)))) of of of three three three different different different weighed weighed weighed amounts amounts amounts of of of tellimagrandin tellimagrandin tellimagrandin II II II4 4 4against against againstdifferent different differentlipid lipid lipidprotons protons protonswith with with aa a mixingmixing mixing timetime time ofof of 0.30.3 0.3 s.s. s. TheTheThe lipid lipid lipid proton proton proton assignations assignations assignations refer refer refer to to to Figure Figure Figure5 and5 5 an and crossd cross cross peak peak peak labels labels labels are are are deﬁned defined defined in in Sectionin Chapter Chapter 3.3. 3.3. 3.3.

ReferencesReferencesReferences 1.1.1. Waghorn,Waghorn,Waghorn, G.C.; G.C.;G.C.; McNabb, McNabb, W.C.W.C. ConsequencesConsequences Consequences ofof of plantplant plant phenolicphenolic phenolic compoundscompounds compounds forfor for productivityproductivity productivity andand and healthhealth health ofof ruminants.ruminants. of ruminants. Proc.Proc. Proc.Nutr.Nutr. Nutr. Soc. Soc. Soc.2003 20032003, ,62 62, ,,383–392, 383–392,62, 383–392. doi:10.1079/pns2003245. doi:10.1079/pns2003245. [CrossRef] 2.2.2. Mueller-Harvey,Mueller-Harvey,Mueller-Harvey, I.; I.; I.; Bee, Bee, Bee, G.;G.; G.; Dohme-Meier,Dohme-Meier, Dohme-Meier, F.;F.; F.; Hoste,Hoste, Hoste, H.;H.; H.; Karonen,Karone Karonen,n, M.;M.; M.; Kölliker,Kölliker, Kölliker, R.;R.; R.; Lüscher,Lüscher, Lüscher, A.;A.; A.; Niderkorn,Niderkorn, Niderkorn, V.;V.; Pellikaan,Pellikaan, Pellikaan, W.F.;W.F. W.F.; ; Salminen,Salminen,Salminen, J.P.;J.P.; J.P.; et et et al. al. al. Beneﬁts Benefits Benefits of of of condensed condensed condensed tannins tannins tannins in in in forage forage forage legumes legumes legumes fed fed fed to to to ruminants: ruminants: ruminants:Importance Importance Importanceof of ofstructure, structure, structure,concentration, concentration concentration, , andandand diet diet diet composition. composition. composition.Crop Crop Crop Sci. Sci. Sci.2019 2019 2019,,59 ,59 59,, 861–885. ,861–885, 861–885, [doi:10.2135/cropsci2017.06.0369. CrossRefdoi:10.2135/cropsci2017.06.0369.] 3.3.3. Baert,Baert,Baert, N.; N.;N.; Pellikaan, Pellikaan,Pellikaan, W.F.; W.F.;W.F.; Karonen, Karonen,Karonen, M.; Salminen, M.;M.; Salminen,Salminen, J.-P.P. A J.-P.P.J.-P.P. study A ofA theststudyudy structure-activity ofof thethe structure-activitystructure-activity relationship ofrelationshiprelationship oligomeric of ellagitanninsof oligomericoligomeric onellagitanninsellagitannins ruminal fermentation on on ruminal ruminal in fermentation fermentation vitro. J. Dairy in in Sci.vitro. vitro.2016 J. J. Dairy ,Dairy99, 8041–8052. Sci. Sci. 2016 2016, ,99 [99CrossRef, ,8041–8052, 8041–8052,][PubMed doi:10.3168/jds.2016-11069. doi:10.3168/jds.2016-11069.] 4.4.4. Hoste,Hoste,Hoste, H.; H.; H.; Jackson, Jackson, Jackson, F.; F.; F.; Athanasiadou, Athanasiadou, Athanasiadou, S.; S.; S.; Thamsborg, Thamsborg, Thamsborg, S.M.; S.M.; S.M.; Hoskin, Hoskin Hoskin, S.O. ,S.O. S.O. The The The effects effects effects of of of tannin-rich tannin-rich tannin-rich plants plants plants on on on parasitic parasitic parasitic nematodes nematode nematodess ininin ruminants. ruminants. ruminants.Trends Trends Trends Parasitol. Parasitol. Parasitol.2006 2006 2006, ,22 ,22 22,, 253–261. ,253–261, 253–261, [ CrossRefdoi:10.1016/j.pt.2006.04.004. doi:10.1016/j.pt.2006.04.004.]

Molecules 2021, 26, 373 13 of 13

5. Karonen, M.; Ahern, J.R.; Legroux, L.; Suvanto, J.; Engström, M.T.; Sinkkonen, J.; Salminen, J.-P.; Hoste, H. Ellagitannins Inhibit the Exsheathment of Haemonchus contortus and Trichostrongylus colubriformis Larvae: The Efﬁciency Increases Together with the Molecular Size. J. Agric. Food Chem. 2020, 68, 4176–4186. [CrossRef] 6. Engström, M.T.; Karonen, M.; Ahern, J.R.; Baert, N.; Payré, B.; Hoste, H.; Salminen, J.P. Chemical Structures of Plant Hydrolyzable Tannins Reveal Their in Vitro Activity against Egg Hatching and Motility of Haemonchus contortus Nematodes. J. Agric. Food Chem. 2016, 64, 840–851. [CrossRef] 7. Puljula, E.; Walton, G.; Woodward, M.J.; Karonen, M. Antimicrobial Activities of Ellagitannins against Clostridiales perfringens, Escherichia coli, Lactobacillus plantarum and Staphylococcus aureus. Molecules 2020, 25, 3714. [CrossRef] 8. Karonen, M.; Oraviita, M.; Mueller-Harvey,I.; Salminen, J.P.; Green, R.J. Ellagitannins with Glucopyranose Cores Have Higher Affinities to Proteins than Acyclic Ellagitannins by Isothermal Titration Calorimetry. J. Agric. Food Chem. 2019, 67, 12730–12740. [CrossRef] 9. Engström, M.T.; Arvola, J.; Nenonen, S.; Virtanen, V.T.J.; Leppä, M.M.; Tähtinen, P.; Salminen, J.P. Structural Features of Hydrolyzable Tannins Determine Their Ability to Form Insoluble Complexes with Bovine Serum Albumin. J. Agric. Food Chem. 2019, 67, 6798–6808. [CrossRef] 10. Moilanen, J.; Salminen, J.-P. Ecologically neglected tannins and their biologically relevant activity: Chemical structures of plant ellagitannins reveal their in vitro oxidative activity at high pH. Chemoecology 2008, 18, 73–83. [CrossRef] 11. Liu, X.L.; Hao, Y.Q.; Jin, L.; Xu, Z.J.; McAllister, T.A.; Wang, Y. Anti-Escherichia coli O157:H7 properties of purple prairie clover and sainfoin condensed tannins. Molecules 2013, 18, 2183–2199. [CrossRef] 12. Gawrisch, K.; Eldho, N.V.; Polozov, I.V. Novel NMR tools to study structure and dynamics of biomembranes. Chem. Phys. Lipids 2002, 116, 135–151. [CrossRef] 13. Polenova, T.; Gupta, R.; Goldbourt, A. Magic angle spinning NMR spectroscopy: A versatile technique for structural and dynamic analysis of solid-phase systems. Anal. Chem. 2015, 87, 5458–5469. [CrossRef] 14. Scheidt, H.A.; Pampel, A.; Nissler, L.; Gebhardt, R.; Huster, D. Investigation of the membrane localization and distribution of flavonoids by high-resolution magic angle spinning NMR spectroscopy. Biochim. Biophys. Acta Biomembr. 2004, 1663, 97–107. [CrossRef] 15. Scheidt, H.A.; Huster, D. The interaction of small molecules with phospholipid membranes studied by 1H NOESY NMR under magic-angle spinning. Acta Pharmacol. Sin. 2008, 29, 35–49. [CrossRef] 16. Salminen, J.P.; Karonen, M. Chemical ecology of tannins and other phenolics: We need a change in approach. Funct. Ecol. 2011, 25, 325–338. [CrossRef] 17. Tanaka, T.; Zhang, H.; Jiang, Z.H.; Kouno, I. Relationship between hydrophobicity and structure of hydrolyzable tannins, and association of tannins with crude drug constituents in aqueous solution. Chem. Pharm. Bull. 1997, 45, 1891–1897. [CrossRef] [PubMed] 18. Virtanen, V.; Karonen, M. Partition coefﬁcients (logP) of hydrolysable tannins. Molecules 2020, 25, 3691. [CrossRef][PubMed] 19. Salminen, J.P.; Ossipov, V.; Loponen, J.; Haukioja, E.; Pihlaja, K. Characterisation of hydrolysable tannins from leaves of Betula pubescens by high-performance liquid chromatography-mass spectrometry. J. Chromatogr. A 1999, 864, 283–291. [CrossRef] 20. Salminen, J.-P.; Ossipov, V.; Haukioja, E.; Pihlaja, K. Seasonal variation in the content of hydrolysable tannins in leaves of Betula pubescens. Phytochemistry 2001, 57, 15–22. [CrossRef] 21. Matsuo, Y.; Wakamatsu, H.; Omar, M.; Tanaka, T. Reinvestigation of the Stereochemistry of the C-Glycosidic Ellagitannins, Vescalagin and Castalagin. Org. Lett. 2015, 17, 46–49. [CrossRef][PubMed] 22. Grélard, A.; Loudet, C.; Diller, A.; Dufourc, E.J. NMR spectroscopy of lipid bilayers. Methods Mol. Biol. 2010, 654, 341–359. [CrossRef][PubMed]