A Robust and Tunable Halogen Bond Organocatalyzed 2-Deoxyglycosylation Involving Quantum Tunneling ✉ Chunfa Xu 1,2, V

ARTICLE

https://doi.org/10.1038/s41467-020-18595-2 OPEN A robust and tunable halogen bond organocatalyzed 2-deoxyglycosylation involving quantum tunneling ✉ Chunfa Xu 1,2, V. U. Bhaskara Rao1,2, Julia Weigen1,2 & Charles C. J. Loh 1,2

The development of noncovalent halogen bonding (XB) catalysis is rapidly gaining traction, as isolated reports documented better performance than the well-established hydrogen bonding 1234567890():,; thiourea catalysis. However, convincing cases allowing XB activation to be competitive in challenging bond formations are lacking. Herein, we report a robust XB catalyzed 2-deoxyglycosylation, featuring a biomimetic reaction network indicative of dynamic XB activation. Benchmarking studies uncovered an improved substrate tolerance compared to thiourea- catalyzed protocols. Kinetic investigations reveal an autoinductive sigmoidal kinetic proﬁle, supporting an in situ ampliﬁcation of a XB dependent active catalytic species. Kinetic isotopic effect measurements further support quantum tunneling in the rate determining step. Fur- thermore, we demonstrate XB catalysis tunability via a halogen swapping strategy, facilitating 2-deoxyribosylations of D-ribals. This protocol showcases the clear emergence of XB catalysis as a versatile activation mode in noncovalent organocatalysis, and as an important addition to the catalytic toolbox of chemical glycosylations.

1 Abteilung Chemische Biologie, Max Planck Institut für Molekulare Physiologie, Otto-Hahn-Straße 11, 44227 Dortmund, Germany. 2 Fakültät für Chemie und ✉ Chemische Biologie, Technische Universität Dortmund, Otto-Hahn-Straße 4a, 44227 Dortmund, Germany. email: [email protected]

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

oncovalent catalysis constitutes one of the major pillars of challenging bonds over thiourea catalysis in biologically relevant organocatalysis1–4, by capitalizing intermolecular inter- molecules can be developed. In line with this aim, we have N fi actions, such as hydrogen bonds (HB) to activate specific identi ed the biologically relevant 2-deoxyglycosylation as an functional groups under mild conditions. Indisputably, HB ideal reaction model for the discovery for unknown XB catalytic catalysis5,6 dominates the majority of noncovalent organocata- mechanisms46–59. The unique poly-oxygenated nature of glycosyl lyzed protocols, which is largely spearheaded by thiourea cata- substrates and products provides numerous XB acceptor moieties lysis7–14. Furthermore, thiourea organocatalysis is widely dubbed for the in situ catalytic establishment of dynamic noncovalent as biomimetic2, as it derives inspiration from how enzymes activations. Moreover, the synthetic importance of 2- capitalize noncovalent interactions to catalyze biochemical reac- deoxyglycosides as a privileged and biologically useful com- tions1–4. Hence, biomimetic thiourea catalysis is well recognized pound class is well exemplified by the continual intense interest as a powerful synthetic tool spanning from asymmetric catalysis by many different research groups46–59, due to its prevalence in to natural product synthesis5–14. glycosidic natural products, such as digitoxin and saccharomicin In contrast, while the utility of the more directional halogen B (ref. 46). bond (XB) is increasingly establishing its prominence in non- On the organocatalytic front60–64, seminal studies by Galan covalent organocatalysis15–21, mechanistic understanding and and McGarrigle et al. demonstrated the utility of thiourea cata- utility of unique XB activation modes is highly limited. While lysis in accessing 2-deoxyglycosides via mild organocatalytic nature also exploits XBs in catalytic processes22, for instance in conditions47,56. However, significant limitations still exist in such thyroid hormones, the development of biomimetic XB catalytic protocols. For instance, the sole application of thiourea catalysis strategies in constructing biologically important molecules is limits the donor scope to galactals47, and expansion to challen- surprisingly underexplored in the literature. ging glycosyl donors, such as glucals and rhamnals required In recent years, despite the development of excellent proof-of- harsher conditions, i.e., the usage of either a conventional strong concept XB-catalyzed reactions by multiple research groups23–42, Brønsted acid48, or a judiciously matched combination of protocols whereby XB catalysis shows clear catalytic advantages thiourea and an appropriate enantiomer of a strong chiral in over conventional thiourea catalysis is lacking21. Moreover, a Brønsted acid49. A caveat of directly employing strong Brønsted conceivable configurational advantage pertaining to XB catalysis acid activators lies in the product decomposition due to greatly would involve the ease of σ-hole tunability via halogen swapping enhanced hydrolytic cleavage, of up to 2000-fold elevated acid to accommodate challenging substrates—an advantage not labilities of 2-deoxyglycosides (Supplementary Note 14)65, which well adoptable in thiourea catalysis due to strict requirements necessitates rigorous water exclusion procedures, such as exten- of hydrogen atoms for dual HB activation. This property is ded pre-vacuum suctioning of substrates47–49,56. As the critical however not yet successfully exploited. In addition, the majority mechanistic influence of activators and catalysts in glycosylation of currently reported XB catalytic strategies are limited to pathways is well recognized but largely understudied66, the recent monofunctional group activation, such as solely on carbonyls, realization of unconventional XB catalytic mechanisms offers imines, quinolines, iodonium ylides, or halides15–21. Biomimetic huge potential to expand the glycosylation activator toolbox15–21 complex kinetic behavior involving in situ dynamic XB activation and to accommodate improved substrate scopes, while retaining on multiple reaction species offers great promise in unraveling mild conditions required for broad substrate utility and cir- unknown XB mechanisms and reactivity, but little is currently cumventing decomposition. known. Furthermore, the applicability of XB activation in advancing In an elegant seminal benchmarking case, Huber et al. chemical glycosylations is still in its infancy. Apart from our demonstrated a XB-catalyzed Diels–Alder reactions between above mentioned XB-catalyzed strain-release glycosylation44, cyclopentadiene and methyl vinyl ketone (Fig. 1a) through XB- Huber and Codée et al. demonstrated a seminal stoichiometric ketone activation27. A comparison of the kinetic profiles between proof-of-principle XB activation in Koenigs–Knorr glycosyla- XB and thiourea catalysts illuminated superior XB catalytic per- tion67, and recently Takemoto et al. described a powerful XB-co- formance. An extension of the same activation concept was catalytic role to elevate the Brønsted acidity in thiourea for N- exemplified by Toy et al., wherein they demonstrated sub- glycofunctionalizations and N-glycosylations68,69. stantially improved effectiveness of a bidentate XB catalyst over We herein report a remarkable XB-catalyzed 2-deox- thiourea catalysis in a bis-Friedel–Crafts-type reaction of indoles yglycosylation of glycals 1 (Fig. 1e), featuring robustness, mild- on ketones41. ness, substrate broadness, and mechanistic complexity via a Furthermore, Takemoto and coworkers showcased in 2017 a switchable intricate reaction network. Noncovalent catalytic powerful cross-enolate coupling reaction catalyzed by a mono- benchmarking studies are conducted between our protocol and dentate XB catalyst, which furnished superior yield over thiourea widely established thiourea-catalyzed protocols47,56, and our catalysis (Fig. 1b)32. Recent DFT calculations by Wong et al. offer protocol furnishes overall superior donor and acceptor substrate theoretical basis into the potential competitiveness of XB catalysis generality, noteworthily in silylated galactals, glucals, rhamnals, with respect to thiourea catalysis43. Lately, Yeung and coworkers and pentose-derived donors. We introduce a halogen swapping also reported that XB catalysis enables addition of silylated C- strategy for challenging D-ribal substrates, and this was found to nucleophiles to N-acyliminium ions (Fig. 1c), which was unac- be effective in tolerating challenging 2-deoxypyranoribosylations. hievable using thiourea catalysis35. Our group also demonstrated In situ NMR monitoring reveal a sigmoidal kinetic profile char- in 2019 that a multistage XB catalysis gave broadly superior acteristic of autoinduction, suggesting an amplificative formation anomeric selectivity in strain-release glycosylations44, (Fig. 1d) of an in situ catalyst. Kinetic isotopic effect (KIE) experiments superceding that of our earlier reported thiourea-catalyzed pro- suggest that quantum tunneling is operative in the proton transfer tocol45. These very limited prior examples provide pressing rate-determining step of the mechanism. impetus for more experimental proof and deeper mechanistic understanding of XB catalysis, as a high-performance noncovalent catalytic tool in challenging reactions. Results Specifically, we aim to harness and unravel noncovalent Establishment of an XB-catalyzed 2-deoxyglycosylation.We mechanisms unique to XB organocatalysis, so that enabling initiated our investigation on a model glycosylation by selecting methodologies, which facilitates the preferential construction of the D-glucal substrates 1a–1b as our glycosyl donor, and 2a as the

2 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

Limited prior reports of competitiveness of XB over thiourea catalysis

a Huber et al. (2014) b Takemoto et al. (2017) Diels–Alder reaction Cross-enolate coupling O O

2 1 2 O R Bidentate XB Cat. O R R Monodentate XB Cat. O R3 O (20 mol%) I (15 mol%) O Ph Z + O + Base R1 R3 n Z Superior temporal kinetic performance Superior yield over thiourea over thiourea n

c Yeung et al. (2019) d Loh et al. (2019) Addition to N-acyliminium ions Multistage XB catalyzed strain-release glycosylation O O O Monodentate O Bidentate XB Cat. O BnO BnO XB Cat. O (0.5–5 mol%) R NR (10 mol%) NR BnO BnO TMSCl, NuX Superior anomeric selectivity across O OAc Nu a broad substrate range over thiourea Superior XB reactivity over thiourea Mechanistic distinctiveness over thiourea

e Current work: Triple XB activation

A or Br-A - - OTf- OTf OTf N N N n N n O N nOct Oct N Oct PGO I 1 I / Br O I (3 mol%) PGO H H + ROH O O O H 23OR R R R Acceptor activation Oxyanion stabilization Overall broader substrate tolerances over thioureas PGO Mechanistic distinctiveness of XB catalysis over thiourea O Quantum tunneling in the rate determining step nOct H - O ON/OFF switchable kinetic modulated by XBs Glycosidic bond OTf N I XB catalyst tunability via halogen swapping activation R N

Fig. 1 Prior reports in enhanced XB over thiourea based HB catalysis. a XB-catalyzed Diels–Alder reaction. b XB-catalyzed cross-enolate coupling reaction. c XB-catalyzed Friedel–Crafts-type bis-indolylation. d XB-catalyzed strain-release glycosylation. e Current work. acceptor. Utilizing the benzylated D-glucal 1a (Fig. 2, entry 1–3) are added into the reaction to inhibit the XB activation mechanism was deleterious in our reaction because of substantial formation by competition for the σ-hole site, due to the extremely high affinity of Ferrier side products. Encouragingly, when silylated D-glucal of halide ions to the XB catalyst (Fig. 2, entries 9 and 10)25,29.Both 1b (Fig. 2, entry 4) was employed48, we observed a yield elevation halide competition experiments gave negligible yield (<5%) of 3a, of 3a to 92% with excellent anomeric selectivity. A screen of further augmenting the presence of XB catalysis. various XB catalysts A–D revealed that A is optimal for our As parallel controls to ascertain TBAB poisoning effects on reaction32. We eventually arrived at our optimized protocol conventional Brønsted acids without XB influence, the glycosyla- (Fig. 2, entry 18), whereby only 3 mol% of catalyst A is required tion was carried out twice in catalytic amounts of HCl (5 mol%), a for the glycosylation, furnishing 3a with 89% yield and excellent surrogate for HI, in the absence and presence of the competition anomeric selectivity (>20:1). reagent TBAB (Fig. 2, entries 11 and 12). The similarly good yields of 3a in both control experiments (80–86% NMR yield) signify Control experiments for confirming XB catalysis. To better that TBAB competition has no effect on fortuitous acid catalysis, understand the possible interference effects due to trace acidic HI further solidifying the cruciality of XB catalysis in our protocol. In generation, a series of control experiments were conducted21.While control experiments simply reacting A with isopropanol, the the non-iodinated control catalyst E gave good yields of 3a (Fig. 2, absence of O-acceptor substituted benzimidazolium side products – 1 entry 8), the benzimidazolium hydrogen is known to be acidic and in LC MS and H NMR analysis of the reaction mixture further could function as either a nonclassical HB or a Brønsted acid cat- supports the absence of trace acid catalysis (Supplementary – alyst (Supplementary Note 15)70,71, and does not constitute exclu- Figs. S5 S9 and Supplementary Note 16). sionary proof of XB catalysis by A. To better investigate the Furthermore, the addition of catalytic amounts of organic criticality of XB catalyst influences on our system, TBAB and TBAI base F (Fig. 2, entries 13 and 14), or inorganic base K2CO3 (Fig. 2,

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

OR1

O R2O OR1 O OH R3O O [a,b] O R2O O Conditions + O O R3O O O O 1a–b 2a O 3a O O

Cat. Additive 3a Entry Solvent Time/h R1 R2 R3 (mol%) (mol%) Yield% (:)

[c] 1 A (5%) CH2Cl2 – 10 Bn Bn Bn 49% (n.d.) 2 A (5%) PhF – 10 Bn Bn Bn 27% (n.d.)[c] 3 A (5%) THF – 10 Bn Bn Bn 35% (n.d.)[c] 4 A (5%) CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 91% (>20:1) B (5%) 5 CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 86% (90:10) C (5%) 6 CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 79% (81:19) D (5%) [d] 7 CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 7% (>20:1) E (5%) 8 CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 83% (>20:1) A (5%) 9 CH2Cl2 TBAB (10%) 10 TIPS iPr2SiOSiiPr2 <5% (n.d.) A (5%) 10 CH2Cl2 TBAI (10%) 10 TIPS iPr2SiOSiiPr2 <5% (n.d.) 11 HCl (5%) CH2Cl2 – 10 TIPS iPr2SiOSiiPr2 86% (>20:1) 12 HCl (5%) CH2Cl2 TBAB (10%) 10 TIPS iPr2SiOSiiPr2 80% (>20:1) A (5%) 13 CH2Cl2 F (5%) 10 TIPS iPr2SiOSiiPr2 <5% (n.d.) A (5%) 14 CH2Cl2 F (10%) 10 TIPS iPr2SiOSiiPr2 <5% (n.d.) 15 A (5%) CH Cl 2 2 K2CO3 (10%) 10 TIPS iPr2SiOSiiPr2 <5% (n.d.) A (5%) [e] 16 CH2Cl2 – 24 TIPS iPr2SiOSiiPr2 92% (93:7) 17 - CH2Cl2 – 24 TIPS iPr2SiOSiiPr2 <5% (n.d.) [f] A (3%) [e,g] 18 CH2Cl2 – 48 TIPS iPr2SiOSiiPr2 89% (>20:1)

OTf- OTf- N N N N N n N n Oct Oct N CF3 N I I n n I I Oct 2 OTf- Oct A CF3 B C

Dodec tBu N - - OTf OTf N N nOct N I F t t D E Bu N Bu CF3

Fig. 2 Optimization of XB-catalyzed 2-deoxyglycosylation. [a] 2a (0.1 mmol), 1a–b (0.15 mmol), catalyst, 50 °C, solvent (0.2 M), time, argon. [b] Yield and α:β ratio were determined by crude 1H NMR spectra analysis using 1,3,5-trimethoxybenzene as an internal standard. [c] 40–50% NMR yields of Ferrier rearrangement side-products detected. [d] 92% 1b remaining. [e] 2a (0.2 mmol), 1b (0.30 mmol), catalyst, 50 °C, CH2Cl2 (0.2 M), time, argon. [f] Conducted at 40 °C. [g] Isolated yields. n.d. not determined, iPr isopropyl, TBAB tetrabutylammonium bromide, TBAI tetrabutylammonium iodide. entry 15) terminated the glycosylation (<5% yield). This (Fig. 3). For the hexoses-based donors 1, natural and nonnatural observation is in contrast with our previously reported XB- D- and L-glucal substrates, D- and L-galactal, and L-rhamnal- catalyzed strain-release glycosylation44, whereby the reaction still based substrates are very well tolerated in this methodology. proceeded with relatively good yields in the presence of base. A Generally, a range of different O-acceptors bearing the free control experiment in the absence of catalyst A did not result in hydroxyl moiety at multiple positions are very well tolerated to reaction (Fig. 2, entry 17). Evaluation of these control experi- yield the target glycosides 3 with excellent anomeric selectivity. ments revealed a strong dependency on XB catalysis in this This include the protected monosaccharides acceptors to generate protocol. The presence of base deactivation does not necessarily 3a–3j, 3s–3ac, and 3an–3ar. Steroidal acceptors such as choles- conﬁrm the presence of trace acidity, if XB activation is crucial in terol, testosterone, and methyltestosterone are also well accom- initiating a proton transfer step, and the base serves as quencher modated within the different hexoses donors (3k–3m, 3ad–3ag, of the proton transfer. These observations point us toward a and 3as). Interestingly, protected amino acid residues, including possible complex interplay of mechanisms through dynamic XB L-serine, L-threonine, and L-tyrosine work well in our protocol to interactions in our methodology. generate glycopeptide-type derivatives (3n–3q, 3ah–3aj, and 3at). Simple primary and secondary alcohols, such as propargyl alcohol Substrate scope. With an optimized protocol in hand, we pro- and isopropanol also gave the target glycosides with generally ceeded to establish the substrate scope of the protocol. We good to excellent yields, and very good to excellent selectivity (3r determined that this protocol tolerates a wide range of hexose- and 3ak–3al). based donors 1b–1h, furnishing glycosides 3a–at with good to Furthermore, we sought to investigate whether our protocol is excellent yields and generally excellent anomeric selectivity robust enough to accommodate less facile pentoses-based donors

4 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

Hexoses donor scope 1 i OTIPS OR Pr R1O iPr Si O O TIPSO TBSO O iPr OTBS 1 O R O i R2 O O Si Pr O Si O A (3 mol%) O O 1 OTBS 2 i i R -OH Pr Pr 1b 1c: R = TBS or O O + PGO i Si CH Cl , 40 °C O Pr 1 2 2 2 PGO iPr 1d: R = Bn 1g iPr i 1h O O Si Pr 48 h R2 O 1e: R1 = Me O O Si Non-natural sugars 3 i Natural sugars 1f Pr iPr D- and L-glucal substrates OTIPS OTIPS OTIPS iPr iPr i i OTIPS i Pr OTIPS Pr i Ph Pr i OTIPS i OTIPS i Pr Pr iPr Pr Pr Si O Ph iPr Si O Si O i i i O O O O Pr Pr Pr Si O Si O OMe Si O Si O O O O BnO O O O O O O O O OMe OBn Si O O O Si Si O O O O O O O O O i i O Si i i i i O Si Si Si Pr Pr Pr Pr O Pr Pr O i i O O O i O i O Pr Pr i iPr i Pr O iPr Pr MeO O Pr Pr O O BzO BnO O 3d O 3e BnO O BzO 3c O OMe BnO [a] [a, b] 3f O 3g 3a O 3b BzO [a, b] 71% 80% MeO [a] O [a] OMe 71% : >20:1 [a] [a] 89% 77% : >20:1 : >20:1 78% 75% : >20:1 : >20:1 : >20:1 :>20:1

O i OTIPS O Pr OTIPS i O Ph O iPr i OTIPS Pr O O i H Pr Si O O Pr O iPr O Si O O O O TIPSO Si O O O O O H H O TIPSO BnO O O O Si TIPSO OMe O O O O Si H O i i O i Si Pr Pr O O i O Pr O O H O O iPr Pr Si i i iPr 3l Pr Pr 3m i Si O 3j 3k H H iPr Si i O H H 3h Pr Si O 3i O Pr [a] [a] i Si i [a] [a] 62% O i [a] Pr O Pr 68% iPr 67% H [a, b] O Pr O Si iPr 81% 73% 72% iPr : >20:1 : 91:9 : >20:1 iPr : >20:1 : >20:1 : 79:21 O

O OTIPS OTIPS iPr OTIPS iPr iPr i i OTIPS iPr Pr Pr iPr Si O Si O O OMe Si O i O O TIPSO O Pr NHBoc Si O O O NHBoc O O NHBoc O O O O Si Si O Si O O O OMe i O i O OMe i i Pr i i Si iPr Pr 3n Pr Pr Si O Pr Pr NHBoc 3r O 3o 3p 3q i i iPr Si i Pr Pr [a, b] O [a, b] O O Pr [a, c] [a] 77% i 68% [a, d] 89% 71% Pr 59% CO2Me : >20:1 : >20:1 : >20:1 : >20:1 : >20:1 D- and L-galactal substrates OTBS OBn OMe TBSO BnO MeO OTBS TBSO O O O O OTBS O TBSO OTBS TBSO O O OTBS Ph TBSO BnO MeO TBSO Ph TBSO O O O O O O O O BnO OBn O O O TBSO TBSO TBSO O TBSO O O O O O O O O O O O O O O O O O BzO OTBS BnO TBSO MeO O O O BzO O 3z BnO 3s O 3tO 3u O BzO 3w OMe [a, e] 3v OMe [a, g] 3x 3y [a, g] 91% [a, f] [a, f] [a, g] 44% BnO O 61% 55% 59% 80% [a, g] [a, f] MeO : 91:9 : 85:15 : >20:1 82% 73% : >20:1 : 83:17 : >20:1 : >20:1 : >20:1 OTBS OTBS TBSO Ph O OTBS OTBS OTBS TBSO TBSO TBSO O TBSO O O O O O OMe O O TBSO O TBSO TBSO BnO TBSO TBSO O TBSO OMe H O O O O H O O O H 3ae H OTBS O O 3ad H H H 3af H 3aa O 3ac [a, h] H TBSO 3ab [a, g] [a, h] 80% [a, g] [a, g] 75% 54% H 72% 74% [a, g] O : >20:1 77% : >20:1 O : >20:1 : 80:20 : 89:11 : >20:1 O O OTBS OTBS OTIPS TBSO OTBS TBSO OTBS OTBS BnO TBSO TBSO TBSO O H O O O TBSO O O O TBSO TBSO O BnO O H NHBoc TBSO TBSO TBSO H NHBoc 3am O OMe O O OTBS 3ag 3ah 3ai O OMe 3aj OMe 3ak O 3al O 85%[a, f] TBSO NHBoc [a, f] [a, f] [a, g] [a, f] O [a, f] [a, i] 84% O : >20:1 54% 75% 76% O 75% 88% BnO : >20:1 BnO : 93:7 O : >20:1 : >20:1 : >20:1 : >20:1 BnO OMe

L-rhamnal substrates O O OBn OBn OMe O O OBn H O O Ph O O O OMe O BnO O BnO O O H H BnO BnO O NHBoc O O O O i O BnO O O O OBn Pr i O O Si O Pr O OMe i O i Si O Pr OBn Pr i i O O O Si O i O O Si O Pr Pr i Pr i O Si i O i O Pr O Pr Si O 3an Pr Pr O Si Si 3as iPr i 3ao 3ap 3aq iPr 3ar i 3at Pr Si O Si O iPr Pr [j] i O Si [j, k] iPr [j] iPr [j, l] [j, m] O Si [j] 94% Pr [j, n] iPr 66% O Si 88% O Si 83% iPr 51% iPr 88% 58% i i i i : 79:21 Pr : 95:5 Pr : >20:1 Pr :>20:1 : 86:14 Pr : 88:12 : 88:12 iPr iPr

Pentoses donor scope i iPr iPr Pr R i i iPr O i Pr O Pr i Pr Si O i O O Si Pr O Pr Si O A (5 mol%) Si O O O O i Si O or O O + R-OH Pr O O i O Si O O Si O O Pr 2 CH2Cl2, 40 °C, 24 h Si or Si O or O i i Si 1j i i i O O Pr Pr 1i 1k Pr Pr i O Pr iPr Pr R O Si Si R iPr i iPr Pr i i Natural sugars Non-natural sugar 4 Pr Pr iPr

D- and L-xylal substrates iPr i Pr iPr iPr i Si O i i i Pr iPr O Pr Pr Pr iPr i Si O Ph i Pr Si O O O Pr Si O O O Si O O BnO OBn Si O O Si O O O NHBoc O O O O i O Si O O O iPr Pr Si O O O Si O i O OMe Si Si i Pr i iPr O i O O H Pr Pr O iPr Pr i i O iPr iPr Pr Pr 4b O OBn 4f H H 4a O O 4c 4d 4e [aa] [aa, ab] [aa, ac] 90% H [aa] O 79% 68% [aa] [aa] 86% O O 59% 67% : 72:28 : 83:17 : 84:16 : 85:15 BnO : 91:9 : 84:16 MeO i D-ribal substrates i Pr O Pr i iPr i Pr iPr Pr Si O O Si O O O Ph O H O O BnO OBn O O iPr Si O O iPr Si O O O O O Si O H Si H O O O H O O O i i H BnO O O O i i Pr Pr O Si O Pr Pr O OMe OBn Si H H O O O O O O H O O iPr i O 4g 4h H O i i Pr iPr i [aa, ad] iPr Pr 4jPr 4k Pr [aa] Si O Si O Si O 4l 4m O O 75% 81% i 4i iPr [aa] iPr [aa] Pr O Si O Si O : 79:21 O Si [aa] i 68% 74% [aa, af, ag] [aa, af, ah] O : 80:20 i Pr iPr 55% 64% Pr 92% i : 91:9 i : 80:20 iPr Pr Pr : >20:1 O : 85:15 : >20:1

Fig. 3 Substrate scope of the XB-catalyzed glycosylation. Hexoses donor scope: [a] 2 (0.2 mmol), 1b–1h (0.3 mmol) and 3 mol% catalyst A,inCH2Cl2 (1 mL), argon, 40 °C, 48 h; α:β ratio was determined by crude 1H NMR analysis. [b] 5 mol% catalyst A, 50 °C. [c] 5 mol% catalyst A, 50 °C, 72 h. [d] 5 mol % catalyst A, 50 °C, 108 h. [e] RT, 24 h. [f] 30 °C. [g] 30 °C, 72 h. [h] 30 °C, 24 h. [i] 10 mol% catalyst A, 40 °C, 96 h. [j] 2 (0.2 mmol), 1f(0.4 mmol) and 5 mol% catalyst A,inCH2Cl2 (1 mL), argon, 30 °C, 48 h. [k] 108 h. [l] 40 °C. [m] 96 h. [n] 50 °C, 72 h. Pentoses donor scope: [aa] 2 (0.2 mmol), 1i–k 1 (0.3 mmol) and 5 mol% catalyst A,inCH2Cl2 (1 mL), argon, 40 °C, 24 h; α:β ratio was determined by crude H NMR analysis. [ab] 96 h. [ac] 30 °C, 48 h. [ad] 48 h. [ae] 3 mol% catalyst A, 30 °C, 48 h. [af] 1j (0.4 mmol). [ag] 3 mol% catalyst A, 35 °C, 48 h. [ah] 3 mol% catalyst A, 30 °C. RT = room temperature 23 °C, PG = Protecting Group.

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

1i–1k, which are challenging in thiourea catalysis47,56. Gratify- (Fig. 5 and Supplementary Fig. S44) were conducted. In a first ingly, natural (1i, 1j) and nonnatural pentose-based donors (1k) control experiment (Fig. 5, Equation 1), the deuterated 2- work generally well in our XB catalysis protocol, to generate the propanol-OD 5 was instead used as the acceptor in the XB- target glycosides (4a–4m) with good to excellent yields and good catalyzed glycosylation. We observed the formation of both the anomeric selectivity. In the case of the D-xylal substrate 1i cis-addition product 6 and the trans-addition product 7 in a ratio (Fig. 3), while there is a clear anomeric preference toward the α- of 9.5:1 (6:7). anomer, a diminishment of anomeric selectivity was observed In a second confirmatory experiment, a galactal donor 8 with a compared to the hexoses, probably due to absence of steric deuterium label on C2 was synthesized and subjected to the hindrance on the β-face from the C5 substituent. The nonnatural standard XB catalytic conditions with isopropanol (Fig. 5, L-xylal was tolerated in the XB catalytic protocol (4i–4k), Equation 2). In this case, we detected the formation of both the furnishing anomeric selectivity in the similar range as the natural cis-addition product 7 and the trans-addition product 6,ina D-congeners. product ratio of 13.5:1 (7:6). The scrambling of the deuterium We also investigated the use of D-ribal 1j to access the 2- labels on C2 in both experiments and the appearance of cis- and deoxyribose scaffold in 4l–4m. Glycosidic linkages to 2- trans-addition products, suggests that the reaction proceeds deoxypyranoribose can be successfully constructed using glycosyl through a step-wise reaction, indicating that protonation of the acceptors bearing free primary or secondary alcohol groups, with C2 of the glycal from the acceptor –OH proton constitute the first generally moderate to good yields and excellent anomeric step of the mechanism. selectivity (Fig. 3). Intermolecular competition experiments using equimolar The overall substrate scope indicated broader donor and amounts of isopropanol and deuterated 2-propanol-OD (Fig. 5, acceptor tolerance over established thiourea catalysis, particularly Equations 3–4, and Supplementary Figs. S39 and S40) were in accessing 2-deoxyglycosides from challenging glucals, rhamnals, conducted to determine the presence of primary KIEs. We first and pentose-derived donors. It must be mentioned, however, that conducted two experiments, each terminated at high and low thiourea catalysis offers better tolerance on D-galactal donors with conversions, respectively. In the former, a KIE value of 5.7 is ether protecting groups, while XB catalysis is superior when obtained and in the latter, a KIE value of 13.3 is obtained. challenging secondary and phenolic acceptors are used on silylated We postulate that this difference in primary KIE attained D-galactal donors. Data on benchmarking studies comparing under different conversions might be attributed to a mechanistic these substrates and protecting group influences against thiourea shift in the overall mechanism as time proceeded, i.e., the initial catalysis will be explored in a later section. significance of tunneling in the proton transfer rate-determining To further ascertain the presence of XB activation on a model step is less pronounced, when a more predominant dynamic glycosyl acceptor 2al,a13C NMR titration with the addition of acceptor exchange cycle sets in the later part of the reaction (see increasing quantities of isopropanol 2al to A was conducted mechanistic explanation in the later section of the manuscript), (Fig. 4a). A downfield shift of the C–I carbon resonance in A of which can involve further proton transfer elementary steps. This ~1.9 p.p.m. was observed. The direction of the NMR shift is in relatively larger KIE values beyond the semiclassical limit (>9) accordance with known literature reports of catalytic XB supports the involvement of quantum tunneling in the rate- activation27,29,44,68. Fitting of the host–guest titration to a 1:1 determining step72–78, where the conversion from reactants to fi −1 isotherm curve yielded an excellent twithaKa value of 0.45 M products tunnels directly through a kinetic barrier without (Fig. 4b, Supplementary Fig. S2 and Supplementary Table S2). The traversing an energetic maxima. TBAB and TBAI competition experiments were corroborated by While such quantum tunneling phenomena were previously – 13C NMR resonance shifts of the C–I carbon of catalyst A when known in proton transfer reactions72 78, as well as in enzymatic an equimolar amount of either TBAB or TBAI was added catalysis77,79,80, this is hitherto not well understood in the context (Supplementary Figs. S24 and S30)25. In the case of a 1:1 molar of noncovalent organocatalysis. A further primary KIE value was ratio of A:TBAB, a downfield shift of 15.45 p.p.m. of the 13C determined using a nonpolar solvent mixture of 10:1 PhCH3: resonance was observed. When a solution of a 1:1 molar ratio of CH2Cl2, and a substantial elevation of primary KIE to 23.4 was A:TBAI was measured, an analogous downfield 13C resonance observed (Fig. 5, Equation 5 and Supplementary Fig. S43). This shift of 14.34 p.p.m. was observed. The distinctively larger solvent dependence is in line with previously reported proton downfield shifts arising from interaction with TBAB and TBAI, transfer reactions involving quantum tunneling81,82, as polar compared to the 1:1 adduct formed between catalyst A and solvents increase the effective mass of the proton via coupling of isopropanol (Fig. 4a and Supplementary Table S2) further solvent dipoles, which reduces tunneling and hence decreases the support that halides, such as bromide and iodide bind much measured KIE. tighter to catalyst A than isopropanol25. Isothermal titration Furthermore, competition experiments to determine secondary calorimetry (ITC) experiments between TBAB or TBAI with A KIE were carried out (Fig. 5, Equations 6–7, and Supplementary −1 yielded Ka values of 7050 and 2210 M , respectively (Supple- Figs. S45 and S46). When equimolar amounts of isopropanol and mentary Figs. S3 and S4), confirming that halide ion binding to A 9 were used, a secondary KIE value of 1.16 was obtained. In is of significant orders of magnitude higher than an isopropanol to addition, when galactals 1c and 8 were employed in an analogous A interaction, reaffirming the value of halide competition experiment, a secondary KIE of 0.994 was obtained. The large experiments (Fig. 2) in ascertaining XB influences by poisoning primary KIEs and the close to unity values attained for secondary the catalyst. A 13C NMR experiment obtained by mixing a 1:1 KIE experiments strongly support the possibility of OH bond molar ratio of donor 1c with catalyst A gave an almost negligible breaking as the rate-determining step. 13C NMR shift of the C–I resonance (Supplementary Fig. S36). This supports the hypothesis that noncovalent interactions between the glycosyl donor and A are insignificant in the reaction Mechanistic experiments via in situ NMR spectroscopy.We mechanism38. then undertook an in situ NMR monitoring experiment (Fig. 6) under standard conditions. Surprisingly, the experiment displayed a counter-intuitive sigmoidal kinetic profile83–85, differing Mechanistic investigation by deuterated experiments. For a from saturation kinetics observed our previously reported XB- deeper insight into the mechanism, deuterated experiments catalyzed strain-release glycosylation44. Superimposition of the

6 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

OH OTf 13C NMR shift of C–I carbon on XB cat. A N N 2al Molar Ratio I A (2al to cat. A) 4

50:1

10:1

2 1:1

1 only cat. A

114.8 114.6 114.4 114.2 114.0 113.8 113.6 113.4 113.2 113.0 112.8 112.6 f1 (ppm)

b 13C NMR titration Fitting of titration curve to a 1:1 binding isotherm 3.0

2.5

2.0

1.5

1.0 Chemical shift change

0.5

0.0 01234 [iPrOH]/M

Fig. 4 NMR titrations and isothermal titration calorimetry to ascertain XB activation. a 13C NMR titration (13C resonance of C–I bond denoted in arrows). b Fit of the 13C NMR titration to a 1:1 binding isotherm. product formation profile of 3al with the depletion profiles of symmetrical profile is indicative of a slow in situ generation of a both reactants (Fig. 6a), isopropanol 2al (limiting reagent) and 1c critical active catalytic intermediate83. Spurred by this finding, we uncovered the symmetrical nature of the product vs substrates wanted to delineate the mechanistic differences between an XB- profiles. Previous literature precedence corroborate that such a catalyzed case with a true strong Brønsted acid catalyst.

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

OTBS OTBS OTBS TBSO TBSO OTBS TBSO OTBS O D O TBSO TBSO TBSO TBSO OH Catalyst O O D + O + TBSO A (3 mol%) 6 O 7 O TBSO 1c TBSO CD Cl , 30 °C Equatorial D Axial D 2 2 OD CH2Cl2, 30 °C 2al 3al O + 48 h 1c Total yield Deuterated product ratio 5 (ratio of 6+7:3al = 8.6:1) of 6:7= 9.5:1 (1) 98% deuteration 71% a OTBS Catalyzed by A (3 mol%) TBSO 0.30 O Arising from TBSO + minute non-deuterated 5 1c Deuterium labels in bold: D 3al O 0.25 2al 3al OTBS OTBS OTBS TBSO TBSO TBSO H 0.20 O A (3 mol%) D O O TBSO TBSO + TBSO (2) CH2Cl2, 30 °C H D 8 O 6 O D 48 h 7 0.15 OH Total yield Axial D Equatorial D

+ (6+7) Conc/M Induction 59% 2al Deuterated product ratio of 7:6= 13.5:1 0.10 period

Determining primary KIE OTBS at high conversion TBSO O 0.05 OTBS TBSO TBSO D O OH A (3 mol%) 3al O O + TBSO + 0.00 OTBS CH2Cl2, 30 °C TBSO 1c 5 2al 48 h (3) 0 200 400 600 800 1000 1200 0.2 mmol O 0.2 mmol 0.2 mmol 84% conversion 280 + TBSO Time/min D 6 O

68% total isolated yield b Catalyzed by HCl (3 mol%) Ratio of 3al to 6 = 5.7 0.30 Determining primary KIE OTBS at low conversion TBSO 1c OTBS O TBSO TBSO 2al OD OH A (3 mol%) 0.25 O + + 3al O TBSO 3al CH2Cl2, 30 °C 5 2al OTBS 1c 12 h TBSO (4) 0.2 mmol 0.2 mmol 0.20 0.3 mmol 10% conversion O + TBSO D 6 O 0.15 10% total isolated yield

Ratio of 3al to 6 = 13.3 Conc/M (indicative of quantum tunneling) 0.10 Determining primary KIE OTBS in non-polar media TBSO O OTBS 0.05 TBSO TBSO OD OH A (3 mol%) O + + 3al O TBSO 10:1 PhCH3:CH2Cl2 5 OTBS 1c 2al 30 °C, 12 h TBSO (5) 0.00 0.2 mmol 0.2 mmol 0.3 mmol 11.6% conversion O 0 50 100 150 200 250 300 350 400 450 500 + TBSO D 6 O Time/min Ratio of 3al to 6 = 23.4 (solvent dependence of KIE consistent with tunneling) Fig. 6 XB vs strong acid kinetic proﬁle comparison. Comparison of

Determining secondary KIE OTBS a sigmoidal kinetics under standard conditions with b saturation kinetics TBSO O with a strong Brønsted acid. OTBS TBSO TBSO OH OH A (3 mol%) 3al O O + + TBSO OTBS (6) D CH2Cl2, 30 °C TBSO Comparing the profile of catalyst A with 3 mol% HCl (Fig. 6b) 24 h 1c 2al 9 O 0.2 mmol 60% conversion 0.2 mmol 0.2 mmol + TBSO showed clear divergences. True Brønsted acid catalysis adheres to 10 O saturation kinetics, and no induction time was observed. This D further indicate that in our XB catalytic method, a mechanistically Ratio of 3al to 10 = 1.16 different route from a true Brønsted acid catalyst is operative. To OTBS TBSO ascertain the postulate of in situ generation of an active catalytic O TBSO OTBS OTBS intermediate during the induction period, we conducted a TBSO TBSO OH fi 1c A (3 mol%) O D O sequential control experiment. 2al is rst reacted with catalyst TBSO 0.075 mmol + + TBSO (7) fi 2al CH2Cl2, 30 °C 3al A at 30 °C for 6 h in the absence of 1c (Fig. 7a), to rst generate OTBS O H TBSO 0.1 mmol 16 h 7 O more of the in situ catalyst in a time period analogous to O 90% conversion + OTBS TBSO TBSO the induction period. Subsequently, 1c was added with the H O 8 D + Ratio of 3al to 7+6 = 0.994 immediate commencement of NMR monitoring. We observed 0.075 mmol TBSO D fi 6 O then the disappearance of the induction period and a signi cantly accelerated reaction time of ~400 min. Fig. 5 Deuterium-labeling studies. Deuterated experiments confirming The acceleration arising by conducting the same experiment in step-wise mechanism through deuterium scrambling and ascertaining sequential order supports the generation of an in situ generated source of C2 proton originates from the glycosyl acceptor. active catalyst from 2al and A (Fig. 7b depicts accelerated kinetic trace with respect to standard conditions)83. The observation of a Intermolecular competition experiments to understand primary and fi fi secondary kinetic isotopic effects (KIEs), and KIE elevation to support concave shaped pro le indicates continual ampli cation (active quantum tunneling. catalyst increasing at an increasing rate) of the in situ catalyst, as the catalytic cycle proceeds upon 1c addition86. In addition, the

8 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

a OH 1. A (3 mol%) 2. 1c added In situ generated 3al CD2Cl2, 6 h catalyst Immediate begin 2al 30 °C oil bath of NMR monitoring 0.30 1c 2al 0.25 3al

0.20

0.15 Conc/M 0.10

0.05

0.00 0 200 400 600 800 Time/min

b 0.20

Standard conditions Sequential conditions 0.15 Acceleration ON switching by preforming more 0.10 in situ catalyst [3al]/M

0.05

0.00 0 200 400 600 800 1000 1200 Time/min

c OTBS 2. TBAI (6 mol%) TBSO OH 1. A (3 mol%) + 1c added In situ generated O TBSO CH2Cl2, 6 h catalyst CH2Cl2, 48 h 2al 30 °C oil bath 30 °C oil bath 3al O < 5% NMR yield d OTBS TBSO O TBSO 1c OH + 30 °C 0.15 mmol <5% conversion of 3al OTBS CH Cl , 48 h 0.019 mmol 3al remained TBSO 2al 2 2 Absence of A O 0.1 mmol + TBSO

3al O 0.02 mmol

Fig. 7 Sequential and control experiments. a Sequential kinetic experiment unraveled enhanced acceleration of product formation. b Comparison of 3al formation kinetics of standard vs sequential conditions. c Poisoning using TBAI to ascertain XB influence on the in situ catalyst. d Control experiment to support the absence of autocatalysis. persistently accelerating kinetic behavior also corroborates that In an orthogonal competition experiment to test if this in situ the amount of in situ active catalyst present during the entire catalytic species involves XB activation, 2al is first reacted with experiment is significantly lower than the 3 mol% of A added at catalyst A at 30 °C for 6 h (Fig. 7c), followed by the sequential the onset. This acceleration is analogous to an ON switching of addition of TBAI as an in situ XB catalyst poison and 1c resulted the catalytic network. in negligible formation of 3al. This is consistent with the

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 hypothesis that the in situ active catalyst depends crucially on XB and displayed saturation kinetics. Moreover, this control experi- participation for catalytic activity. ment is indicative of a steady-state dynamic exchange equilibrium To investigate if our sigmoidal kinetic at standard condition is between unreacted acceptor in the reaction and the aglycone representative of an autoinductive or an autocatalytic profile86–88, component of 3al. Intriguingly, evaluation of all NMR spectra we conducted a control reaction between 1c, 2al, and product 3al along the time coordinate revealed the detection of only α- as a putative catalyst in the absence of A. This experiment resulted glycosides 3al and 3ak. The absence of β-glycosides corroborate an in negligible conversion (<5%, Fig. 7d), and further supports the SN1-type exchange mechanism, which involves the cleavage and hypothesis of an autoinductive mechanism over autocatalysis83,86. reformation of the α-glycosidic linkage. A separate control Unexpectedly, an in situ NMR monitoring experiment adding 20 experiment evaluating transacetalization on benzyl protected D- mol% of product 3al to the standard XB catalytic conditions galactal revealed that this dynamic acceptor exchange process is revealed retardation of catalysis (Fig. 8a). While this observation not operative when ether protecting groups are used (Supple- supports the absence of autocatalytic behavior, it points toward a mentary Fig. S55 and Supplementary Discussion 3). possible catalyst deactivation by 3al. It is essential to note that To test the dependence of this dynamic exchange on XB addition of 3al at the onset of the reaction enables the direct activation, additional 5 mol% TBAI was added into the acceptor deactivation of A before the formation of any in situ catalyst in the exchange experiment conditions to poison A (Fig. 8d). This reaction mixture, hence hampering reaction progression. This experiment failed to generate 3ak, supporting crucial XB catalytic retardation effect is analogous to an OFF switching of the catalytic influences in this dynamic equilibration. A 13C NMR experiment network (Fig. 8b, compared with standard conditions). For a mixing a 1:1 ratio of A:3al (Fig. 8e) yielded a 0.158 p.p.m. confirmatory test to ascertain XB catalytic influences between 3al downfield shift of the 13CC–I resonance of A, evidencing an XB and A, an acceptor exchange experiment mixing propargyl alcohol catalytic interaction on the glycosidic bond. 2r with 3al in the presence of A was conducted (Fig. 8c, Supplementary Figs. S49 and 50, and Supplementary Table S10). Concentration dependence of the reaction profile. In order to The temporal 3ak formation and 3al depletion profiles indicated understand the mechanistic role of the glycosyl donor, the gly- that a XB-catalyzed transacetalization-type reaction is operative, cosyl acceptor and the XB catalyst on the reaction profile, as well

OTBS abOTBS TBSO TBSO 0.20 A (3 mol%) O Standard conditions O + OH TBSO Addition of 20 mol% 3al at the onset (Figure 6a conditions) TBSO 3al (20 mol%) O 1c 2al CD2Cl2, 30 °C 3al 0.15 0.30

0.25 0.10 Retardation 0.20 [3al]/M OFF switching by adding 3al 0.05 0.15 1c

Conc/M 2al 0.10 3al 0.00 0 200 400 600 800 1000 1200 0.05 Time/min

d OTBS 0.00 TBSO 0 200 400 600 800 1000 1200 1400 A (5 mol%) O TBAI (5 mol%) Time/min TBSO + HO No acceptor exchange CD2Cl2, No 3ak formed 3al O 2r 3h RT, then c OTBS OTBS TBSO TBSO OH 12 h 30 °C A (5 mol%) O O + TBSO + HO TBSO e CD Cl , 30 °C 2al 2 2 3ak 13 3al O 2r O OTf C NMR shift of C-I carbon on XB cat. A N N

A 2 I 112.872 ppm 0.08 [3ak] [3al] Only A 0.06

0.04 Conc/M

1 113.030 ppm 0.02

Non-steady state Steady state 1:1 A:3al 0.00

0 400 800 1200 1600 2000 2400 2800 114.9 114.7 114.5 114.3 114.1 113.9 113.7 113.5 113.3 113.1 112.9 112.7 112.5 112.3 112.1 111.9 111.7 111.5 Time/min f1 (ppm)

Fig. 8 Dynamic acceptor exchange. a Addition of 20 mol% product at the onset retarded XB catalysis. b Comparison of 3al formation kinetics of standard vs conditions in a. c Acceptor exchange experiment to ascertain dynamic acceptor exchange. d Competition experiment with TBAI showed XB catalyst poisoning halted the dynamic acceptor exchange. e 1:1 adduct of catalyst A to 3al revealed a downﬁeld 13C NMR shift.

10 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

a Varying catalyst A concentration b Varying donor 1c concentration c Varying acceptor 2al concentration 0.15 0.16 0.20 13 mol% 0.3 M 0.2 M 8 mol% 0.4 M 0.14 0.3 M 3 mol% 0.12 0.2 M 0.4 M 0.15 0.12 0.1 M 0.19 0.10 0.10 0.08

[3al]/M 0.06 [3al]/M [3al]/M 0.06 0.04 0.05 0.03 0.02 0.00 0.00 0.00 0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 200 400 600 800 Time/min Time/min Time/min

d O PGO High acceptor H H H O O O OTf concentration 3 H Reservoir of R R R n O N R off-cycle N nOct HB aggregation prevents catalyst A activation catalyst + ROH 2 Adduct 11 -A A I -2 +TBAI H H Catalyst A H O O O poisoning R R 2 R (Figure 2, Entry 10) Quantum Proton transfer tunneling rate-determining step

XB-catalyzed H H PGO H dynamic acceptor + O O OTf +TBAI In situ catalyst 12 H δ exchange R N 12 poisoning 13 N n X Oct Solvent separated (Figure 7c) ion pair I PGO (in situ catalyst) O O O R PGO nOct H OTf O 1 N I 3 Oxyanion–catalyst Catalyst A +TBAI R complex = X Poisoning N (Figure 8d) 14 Amplificative XB catalytic cycle H O Amplification chemical signal 12, ON switch R 2 PGO Retardation chemical signal 3, OFF switch H H O + O PG = protecting group H δ R 2 X 13

Fig. 9 Reaction mechanism. a Varying catalyst A concentration on the kinetic profile, b varying donor 1c concentration on the kinetic profile, c varying acceptor 2al concentration on the kinetic profile, d proposed mechanism of XB-catalyzed 2-deoxyglycosylation through an XB-modulated intricate reaction network. as on the induction period, we varied the concentration of each retardation sets in at concentrations from 0.3 to 0.4 M. The substrate and conducted an in situ 1H NMR monitoring for the observation of a positive order at the lower concentration window reaction between 1c and 2al catalyzed by A. suggests the involvement of the glycosyl acceptor in the rate- When the concentration of catalyst A was varied (Fig. 9a), we determining step at standard conditions (0.2 M acceptor). observed that increase in catalyst loadings resulted in a We speculate that the observed retardation at higher acceptor corresponding decrease in the induction period, and the left concentrations beyond the standard conditions could be due shifting of the sigmoidal profile, which is consistent with a to the formation of a more extensive hydrogen-bonded positive order with respect to XB catalyst, and is indicative of the isopropanol network (Fig. 9d), which might inhibit catalyst participation of the catalyst in the rate-determining step. activity by stabilizing the OH bonds in isopropanol multimeric When the concentration of the glycosyl donor 1c was varied aggregates89,90, hence hindering the establishment of productive (Fig. 9b), we observed that the reaction profiles were relatively catalyst–glycosyl acceptor XB interactions for catalytic activation. invariant to the concentration changes and overlapped rather well with each other. The induction periods also remained relatively Proposed intricate XB-modulated network mechanism. Based constant, suggesting that the glycal donor is not involved in the on the entirety of our mechanistic data, we propose that an rate-determining step. intricate reaction network, comprising of multiple dynamic XB Interestingly, when the glycosyl acceptor’s concentration was catalytic influences is operative in our methodology (Fig. 9d). permutated, a more complex kinetic behavior was observed This mechanistic explanation hinges upon the presence of an (Fig. 9c). We observed a positive correlation at lower concentra- off-cycle reservoir of A, which shuttles between an amplificative tions, i.e., in the concentration window between 0.1 to 0.2 M, cycle and a dynamic exchange cycle in response to ON/OFF concentration increases resulted in shortening of induction chemical signals. period. However, at larger concentrations beyond the 0.2 M The mechanism starts with a slow formation of a critical threshold, we observed the onset of a negative correlation where amount of an in situ catalytic intermediate, which we postulate is

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 the actual catalyst responsible for product amplification. While Subsequently, the XB catalytic effect of A in 14 results in a we made rigorous efforts to detect this intermediate in situ, dynamic acceptor exchange evidenced by our acceptor exchange however, due to minute amounts of this intermediate (pre- experiment (Fig. 8c), TBAI poisoning, and 13C NMR shift sumably much less than the 3 mol% of A inferred from (Fig. 8d, e), which involves firstly cleavage of the newly formed continuous amplification in standard and sequential kinetic glycosidic linkage in 3 to reform oxocarbenium intermediate 13. studies) beyond the NMR detection limit (5%), we are unable to Next, a new molecule of 2 will first initiate a proton transfer with directly detect and characterize the intermediate. the original alkoxide ion in X− to form a new alkoxide, which will Despite this, based on holistic analysis of numerous informa- attack the oxocarbenium ion to regenerate 3 coupled with the exit tive indirect experiments, such as in situ sequential experiments, of the original protonated alkoxide as 2, releasing catalyst A back base addition, and poisoning of the in situ generated species by into the off-cycle reservoir. Intriguingly, our proposed mechan- TBAI, we deduced that this intermediate possesses Brønsted ism seems to constitute a simplified synthetic mimetic of complex acidic characteristics, with concurrent XB catalytic dependency. noncovalent interactions modulated biochemical mechanisms Significantly, TBAI poisoning characteristics were not observed in catalyzed by enzymes, operating in tightly regulated mechanisms. control true Brønsted acid-catalyzed experiments, such as HOTf, Unexpectedly, we determined that dynamic XB-modulated which performed similarly regardless of the presence or absence mechanistic complexity94 could be constructed simply from a of TBAI, hence excluding non-XB effects in our mechanism, such ternary chemical mixture comprising of a glycosyl donor, an as trace acid catalysis21. acceptor, and a XB catalyst. Hence, we propose based on these data that the catalytic intermediate in our protocol is a solvent-separated ion-pair 12, comprising of a solvated XB-stabilized oxyanion–catalyst com- Benchmarking noncovalent organocatalytic robustness between plex (X−) and a solvated protonated acceptor, arising from the XB and thiourea catalysis. To evaluate the advances in non- rate-determining proton transfer step, which involves quantum covalent catalyst robustness of benzimidazolium XB catalyst A, tunneling (Fig. 9, and Supplementary Notes 17 and 18). X− will and we sought to benchmark our XB catalytic protocol with further serve as an oxocarbenium neutralizing counteranion in conventional thioureas47,56,95,96. subsequent steps. Our experiments ascertaining the presence of In an effort to understand broadness and robustness across acceptor OH bond cleavage (Fig. 5), the crucial involvement of glycal donors beyond widely reported galactals, we observed a Brønsted acidity (Fig. 2, entries 13–15), and successful TBAI general superiority trend in using XB catalysis over thiourea on poisoning of the in situ catalyst (Fig. 7c) collectively corroborate these non-galactal substrates. First, within the hexose family, the intermediacy of XB-dependent ion-pair 12, as the true in situ donors such as silylated D-glucals and L-rhamnals work catalyst through a proton transfer rate-determining step. This excellently with catalyst A, but not with 15 and 16. Recognizing in situ solvent-separated ion-pair hypothesis is also further the possible caveat that could arise from protecting group supported by primary KIE elevation due to increase in the influences, we performed the comparative glycosylation also with tunneling effect81,82, when the reaction was conducted in a more benzylated glucals and rhamnals. Significantly, literature known nonpolar 10:1 PhCH3:CH2Cl2 solvent mixture. This proton ether protecting group preference using thiourea catalysis on 2- transfer arising from the Lewis acidity of the XB donor on the deoxygalactosylation47 holds neither for glucosylation nor alcoholic glycosyl acceptor is reminiscent of the Lewis acid- rhamnosylation. assisted Brønsted acid concept proposed by Yamamoto et al.91.In Second, analogous trends favoring XB over thiourea catalysis addition, this concept has also been applied by Schmidt et al. in can also be observed from the pentose family of glycal donors. glycosylations using gold complexes or boronates as Lewis Using silylated D-xylal 1i (Fig. 10c), the benchmarking experi- acids92,93. ments revealed that catalyst A enabled smooth glycosylation, but 12 can subsequently protonate C2 of glycal 1 to form no reaction was observed with 15 or 16. Using a challenging oxocarbenium intermediate 13. This step-wise mechanism is sterically hindered tertiary alcohol, such as 2af, we noticed that evidenced by scrambling of deuterium labels originating from the XB catalysis still furnished the desired glycosylation product acceptor on C2 (Fig. 5, Equations 1 and 2). As 12 gets consumed in smoothly, whereas catalysis via 15 or 16 do not generate any the initial catalytic cycle, the reversible 11 to 12 reaction will be observable product. In the case of silylated D-ribal 1j (Fig. 10d), increasingly shifted toward 12 as reaction time progresses, in situ both catalyst A and in particular the brominated version of amplifying the active catalyst 12, resulting in an accelerating catalyst A (Br-A in parenthesis) gave superior yields superceding concave shaped kinetic in the initial phase of the sigmoidal profile the thiourea congeners. This halogen swapping advantage by XB (Fig. 7b). Subsequently, the alkoxide in 13 will attack the catalysis will be further discussed in a later section. As a control oxocarbenium species to form 2-deoxyglycoside 3 with the release experiment to clarify specificity of protecting group effects on of A, coupled with a sequential in-cycle intermolecular proton noncovalent catalysis, we noticed that benzylated xylals and ribals transfer between two molecules of acceptor 2 under the activation remained lowly reactive, regardless whether thiourea or XB of the released catalyst A to regenerate the in situ catalyst 12, catalysis are employed. analogous to an 11 to 12 conversion. On the other hand, the To provide fairer comparison with catalyst 15 and 16, and a formation of more glycoside 3 through the amplification cycle more holistic understanding across these noncovalent catalysts, (blue circle, Fig. 9d) activates the dynamic exchange mechanism, we reproduced the exact conditions reported by Galan and which siphons off-cycle A, arising from XB interactions with A and McGarrigle with catalyst A on benzylated galactals, a known the glycosidic bond oxygen in 3 to form 14. This lowering of the strength of these prior methodologies47,56. Our comparative data off-cycle concentration of A due to the dynamic acceptor exchange confirmed that ether protected galactals are better suited for will accordingly deplete the concentration of A required to form thiourea catalysis, although XB catalysis can also activate increasing 12 in the amplificative cycle. This effect is registered by benzylated galactals reasonably (Fig. 10e). However, we have the rate deceleration after the inflection point in the standard demonstrated that employing silyl protecting group in conjunc- sigmoidal kinetic profile (Fig. 6a), but attenuated in the sequential tion with XB catalysis gave consistently excellent performance in profile (Fig. 7b) due to pre-generation of more 12. However, this yields and anomeric selectivity on the galactal donor family effect is strengthened when we directly introduced the glycoside at (Fig. 3). In cases where more challenging acceptors, such as the the onset of the reaction, resulting in retardation (Fig. 8b). phenolic group in protected tyrosine and isopropanol are used,

12 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

a Benchmarking glucals OPG b Benchmarking rhamnals O O OPG OH PGO O O OH O PGO O O O A (3 mol%) A (3 mol%) PGO O O OPG O O PGO + O O OPG + O O or 15 (1 mol%) or 15 (1 mol%) O O O O 1 or 16 (1 mol%) O 16 PGO 2a O 1 O or (1 mol%) 2a PGO 3 O 3 O

CF3 CF3

S PG = Silyl (donor 1b) Bn PG = Silyl (donor 1f) Bn CF3 N N CF3 H H 15 [a] [a] [a] [a] Cat. 15 <5% 44%, α:β 76:24 H Cat. 15 <5% 78%, α:β 83:17 O N S [a] α β77:23[a] 16 [a] [a] Cat. 16 <5% 64%, : NH Cat. 16 <5% 39%, α:β 87:13 [b] [a] A 89%, α:β>20:1 64%, α:β 75:25 A 88%, 88:12 α:β[b] 83%, α:β 86:14[a]

c Benchmarking xylals d PGO O Benchmarking ribals OH PGO O HO H OH PGO O PGO O A (3 mol%) O O O O O PGO + H PGO A or Br-A O or H O O OPG or 15 (1 mol%) + (3 mol%) O O O O O O or 16 (1 mol%) O 1 2a O OPG O or 15 (1 mol%) 2af 4 O 4 O 1 2a or 16 (1 mol%) O O O O

PG = Silyl (donor 1i+2a) Silyl (donor 1i+ tertiary alcohol 2af) Bn (with 2a or 2af) PG = Silyl (donor 1j+2a) Bn

[a] Cat. 15 <5% <5%[a] <5%[a] Cat. 15 44%, >20:1 α:β[b] <5%[c] Cat. 16 <5%[a] <5%[a] <5%[a] Cat. 16 52%, >20:1 α:β[b] <5%[c] [b] [b] A 86%, α:β 83:17 81%, α:β 80:20 <5%[c] A (Br-A) 64% (81%), >20:1 α:β[b] <5%[c] e f Benchmarking exact 2-deoxygalactosylation substrates Benchmarking challenging 2-deoxygalactosylation acceptors reported by Galan and McGarrigle OTBS OTBS OBn TBSO TBSO OBn OTBS BnO BnO OH A (3 mol%) TBSO A (3 mol%) O + O O OH O TBSO or O + TBSO R BnO R BnO or 15 (1 mol%) TBSO or 15 (1 mol%) 1c 2 or 16 (1 mol%) 3 O 1d 2 or 16 (1 mol%) 3 O R 1c R

1d as glycosyl donor 1c as glycosyl donor R = O

O OMe R = O Ph O R = NHBoc O O O O O O BnO BnO O O BnO [b] [b] O BnO O 15 < < OMe OMe O Cat. 5% 5% Cat. 16 <5%[b] <5%[b] Cat. 15 96% Cat. 15 72%[b] 98% 96% α β A α β [b] α:β [b] (lit values) α:β >20:1 α:β >20:1 α:β >20:1 (lit values) : >20:1 75%, : >20:1 88%, >20:1 Cat. 15 94%[b] 93%[b] 72%[b] Cat. 15 68%[b] (reproduced) α:β >20:1 α:β >20:1 α:β >20:1 (reproduced) α:β >20:1 [d] Cat. 16 95% Not 89% Cat. 16 11% (lit values) α:β >20:1 reported α:β >20:1 (lit values) α:β >20:1 Cat. 16 96%[b] [b] Cat. 16 72% <5%[a,d] (reproduced) α:β >20:1 - α:β >20:1 (reproduced) [b] [b] [c] 91%[b] A 55% 70% 46% A α:β 85:15 α:β >20:1 α:β >20:1 α:β 91:9

Fig. 10 Benchmarking XB and thiourea organocatalysis. [a] 1H NMR yield determined with 1,3,5-trimethoxybenzene as an internal standard. [b] Isolated yields after flash column chromatography. [c] A mixture with inseparable Ferrier-type product impurity was obtained after flash chromatography purification, the yield was calculated by 1H NMR analysis. [d] Reaction time was 48 h, literature gave 68% yield with 2-methyl THF as solvent at 83 °C refluxing temperature. a Benchmarking glucals, b Benchmarking rhamnals, c Benchmarking xylals, d Benchmarking ribals, e Benchmarking reported 2- deoxygalactosylation substrates, f Benchmarking challenging 2-deoxygalactosylation acceptors.

XB catalysis gave clear superiority, while thiourea/thiouracil donors, when conventional thiourea catalytic methods are not protocols generated negligible product (Fig. 10f). optimal. In all, we note that XB catalysis offered distinct advantages over thiourea/thiouracil catalysis in activating the entire spectrum of different sugar donors beyond galactals. Pure thiourea catalysis is Applicability of switchable mechanism to elevate yields.To useful in 2-deoxygalactosylation with ether protecting groups, as demonstrate further applicability of the XB-dependent “switch- previously demonstrated by Galan and McGarrigle et al.47,56,95,96, able” characteristic, we performed sequential experiments ana- but reactivity do not scale well across donors, as well as logous to Fig. 7 in an attempt to elevate yields of representative challenging acceptors, and the limitations are evident when other rhamnosides 3aq and 3at (Fig. 11a) that gave moderate yields in non-galactal glycosyl donors are employed. our reaction scope under standard conditions. To our delight, the These benchmarking experiments reinforce immense value of pre-generation of in situ catalysts had a positive effect in our exploiting XB catalytic protocols as an enhanced organocatalytic protocol to substantially improve chemical yields of sluggish tool for accommodating a broader variety of non-galactal glycosyl substrates.

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

a O OMe O OBn O OMe O O NHBoc O O iPr O OBn O i OBn Si Pr iPr Si O O Si iPr i O Si 3aq Pr 3at i iPr Pr 70% 51% 75% 58% iPr : 91:9 : 86:14 : 88:12 : 88:12

b iPr iPr O i Si O Pr O O iPr Si O O NHBoc O - Si O O OTf O O O N O Si N n iPr i Oct Pr i Pr iPr O O 4n O Br–A 4m O Br Cat. A: 64%, : >20:1 Cat. A: complex mixture Cat. Br-A: 81%, : >20:1 Cat. Br-A: 71%, : >20:1 iPr iPr O O O O iPr Si iPr Si

O O O O O O Si H Si H iPr i iPr i 4p Pr 4o H Pr H H H

Cat. A: complex mixture Cat. A: complex mixture Cat. Br-A: 55%, : 90:10 Cat. Br-A: 64%, : >20:1 O O

Fig. 11 XB tunability. a Elevation of yields on rhamnosides using the sequential procedure, yields, and selectivity from standard conditions indicated in parenthesis. b Iodine to bromide halogen swapping on A resulted in signiﬁcantly expanded 2-deoxyribosylation scope.

Tunability of XB catalysis via halogen swapping. Despite the controlled reaction network. This report showcases a marked broadly improved performance of catalyst A, we noticed certain advancement in noncovalent organocatalysis, where the use of limitations of our protocol utilizing the iodide derivative A, lesser explored XB catalysis resulted in mechanistic divergences, particularly in 2-deoxyglycosylation reactions with D-ribals. and significantly elevated substrate scope compared to conven- Meticulous experiments revealed that A is very sensitive to the tional thiourea catalysis, particularly in the formation of chal- acceptors employed, furnishing complex mixtures when amino- lenging glycosidic linkages. Detailed mechanistic studies were also acid-derived acceptors or more sterically hindered secondary and illuminating and instructive of unique and hitherto unknown tertiary alcohol containing steroidal acceptors were employed intricacies and dynamism of XB activation, such as the involve- (Fig. 11b, 4n–4p). ment of quantum tunneling. To address this issue, we reasoned that the σ-hole on the iodine In situ temporal kinetic experiments and comparison revealed atom on A could be fine-tuned by employing an iodine to a sigmoidal kinetic profile characteristic of autoinductive reac- bromine halogen swapping strategy. Surprisingly, using the tions. KIE measurements also revealed unusually high KIE values bromide derivative Br-A as catalyst showed marked improvement beyond the semiclassical limit, as well as KIE elevation in non- (Fig. 11b) to access such challenging glycosidic linkages in 4n–4p polar solvents, which are indicative of quantum tunneling and the with very good to excellent anomeric selectivity, which were not formation of an in situ ion-pair intermediate in the rate- accessible via A. determining proton transfer step. Intriguingly, deeper studies Gratifyingly, even 2-deoxyribosylation to 4m gave substantially unraveled a biomimetic reaction network embedded with improved yield (81%) using Br-A. These observations support switchable ON/OFF mechanism, modulated by dynamic XB that postulate that stronger iodide XB donors might be overly interactions between catalyst, substrates, and products. Further- strong Lewis acids for sensitive ribal substrates, which resulted in more, we capitalized on this switchable property, to elevate yields undesired decomposition pathways. As such, tuning XB donor of 2-deoxyrhamnosides through a sequential protocol. We also strength by employing the weaker Br-A donor could provide an demonstrate that tuning σ-hole properties via a halogen swap on attenuated, milder catalytic activation route which circumvents the catalyst is beneficial in elevating XB catalyst robustness on decomposition, but increase catalytic performance. This success glycosyl donors, such as D-ribals, particularly when more chal- of our halogen swapping method depicts the power of XB lenging glycosyl acceptors are employed. catalysis as a high-performance-tunable noncovalent organoca- We are optimistic that the development of this protocol has talytic tool, especially when sensitivity of substrates is pivotal. promising wider implications in noncovalent organocatalysis, in deconvoluting differences of mechanism and substrate tolerances Discussion between XB and thiourea catalysis, and will contribute to devel- In conclusion, we demonstrate a robust XB organocatalyzed 2- oping mild XB catalysis as a competitive new generational tool in deoxyglycosylation, operating through an intricately XB catalyst challenging, but biologically relevant bond forming reactions.

14 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

Methods 10. Takemoto, Y. Development of chiral thiourea catalysts and its application to General techniques. Unless otherwise stated, all reactions were set up under inert asymmetric catalytic reactions. Chem. Pharm. Bull. 58, 593–601 (2010). atmosphere (argon) utilizing glassware that were oven-dried and cooled under 11. Serdyuk, O. V., Heckel, C. M. & Tsogoeva, S. B. Bifunctional primary amine- argon purging. Silica gel flash column chromatography was performed on deac- thioureas in asymmetric organocatalysis. Org. Biomol. Chem. 11, 7051–7071 tivated silica gel Merck 60 (particle size 40–63 μm). Starting materials were pro- (2013). cured from commercial sources and used without purifications. Solvents were dried 12. Tsakos, M. & Kokotos, C. G. Primary and secondary amine-(thio)ureas and according to reported procedures or procured from commercial suppliers. Mon- squaramides and their applications in asymmetric organocatalysis. itoring of reactions was done using thin layer chromatography (TLC) on Merck Tetrahedron 69, 10199–10222 (2013). silica gel aluminum plates with F254 indicator. TLC visualization of plates was 13. Nishikawa, Y. Recent topics in dual hydrogen bonding catalysis. Tetrahedron performed under UV light (254 nm) or employing KMnO4 stain or H2SO4-EtOH Lett. 59, 216–223 (2018). (10% H2SO4 v/v). 14. Odagi, M. & Nagasawa, K. Recent advances in natural products synthesis NMR spectra were collected at 300 K on a Bruker DRX400 (400 MHz), Bruker using bifunctional organocatalysts bearing a hydrogen-bonding donor moiety. DRX500 (500 MHz), INOVA500 (500 MHz), or Bruker DRX700 (700 MHz) Asian J. Org. Chem. 8, 1766–1774 (2019). spectrometers, using CDCl3,CD2Cl2, acetone-d6, CD3CN, or benzene-d6 as 15. Beale, T. M., Chudzinski, M. G., Sarwar, M. G. & Taylor, M. S. Halogen 1 δ deuterated solvents. Data for H NMR are reported as follows: chemical shift ( p. bonding in solution: thermodynamics and applications. Chem. Soc. Rev. 42, p.m.), multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, – = 1667 1680 (2013). and br broad), coupling constant (Hz), NMR spectra were internally referenced – δ = 1 δ = 16. Cavallo, G. et al. The halogen bond. Chem. Rev. 116, 2478 2601 (2016). to the following residual solvent signals (CDCl3: 7.26 p.p.m. for H, 77.16 fi 13 δ = 1 δ = 13 17. Bul eld, D. & Huber, S. M. Halogen bonding in organic synthesis and p.p.m. for C; CD2Cl2: 5.32 p.p.m. for H, 54.00 p.p.m. for C; acetone- – δ = 1 δ = 13 δ = organocatalysis. Chem. Eur. J. 22, 14434 14450 (2016). d6: 2.05 p.p.m. for H, 29.92 p.p.m. for C; CD3CN: 1.94 p.p.m. for 1 δ = 13 δ = 1 δ = 18. Breugst, M., von der Heiden, D. & Schmauck, J. Novel noncovalent H, 1.32 p.p.m. for CofCD3,C6D6: 7.16 p.p.m. for H, 128.06 p.p.m. π 13 interactions in catalysis: a focus on halogen, chalcogen, and anion- bonding. for C). Synthesis 49, 3224–3236 (2017). High-resolution mass spectra were recorded on an LTQ Orbitrap mass 19. Zhao, Y., Cotelle, Y., Sakai, N. & Matile, S. Unorthodox interactions at work. J. spectrometer coupled to an Accela HPLC-System (HPLC column: Hypersyl Am. Chem. Soc. 138, 4270–4277 (2016). GOLD, 50 mm×1 mm, particle size 1.9 μm, ionization method: electron spray 20. Tepper, R. & Schubert, U. S. Halogen bonding in solution: anion recognition, ionization) and Bruker ultrafleXtreme MALDI-TOF–TOF (three decimal templated self-assembly, and organocatalysis. Angew. Chem. Int. Ed. 57, accuracy). Optical rotations were measured in a Schmidt + Haensch Polartronic 6004–6016 (2018). HH8 polarimeter equipped with a sodium lamp source (589 nm), and are reported as follows: [α]DT°C (c = g/100 mL, solvent). ITC experiments were performed on 21. Sutar, R. L. & Huber, S. M. Catalysis of organic reactions through halogen bonding. ACS Catal. 9, 9622–9639 (2019). a MicroCal VP-ITC device. fi The anomeric ratio was determined by 1H NMR spectroscopy on the crude 22. Schol eld, M. R., Vander Zanden, C. M., Carter, M. & Ho, P. S. Halogen – reaction mixture and the anomeric configuration was determined by 2D NOESY. bonding (X-Bonding): a biological perspective. Protein Sci. 22, 139 152 Chemical yields refer to combined isolated yields of all anomers after flash column (2013). chromatography. NMR yields were determined using appropriate internal 23. Bruckmann, A., Pena, M. A. & Bolm, C. Organocatalysis through halogen- – standards. bond activation. Synlett 6, 900 902 (2008). 24. Coulembier, O., Meyer, F. & Dubois, P. Controlled room temperature ROP of L-lactide by ICl : a simple halogen-bonding catalyst. Polym. Chem. 1, 434–437 General procedure for XB-catalyzed 2-deoxyglycosylation 3 . A mixture of cata- (2010). lyst A (3–10 mol%), glycosyl acceptor 2 (0.2 mmol), and glycal 1 (0.3–0.4 mmol) 25. Walter, S. M., Kniep, F., Herdtweck, E. & Huber, S. M. Halogen-bond-induced was dissolved in anhydrous CH Cl (1 mL, 0.2 M for 2), and sealed in a dry tube 2 2 activation of a carbon-heteroatom bond. Angew. Chem. Int. Ed. 50, 7187–7191 under argon. The mixture was stirred at a specific temperature (room temperature (2011). to 50 °C depending on the substrate). Subsequently, the solvent was removed under 26. Kniep, F. et al. Organocatalysis by neutral multidentate halogen-bond donors. reduced pressure and the residue was first analyzed by crude 1H NMR to determine Angew. Chem. Int. Ed. 52, 7028–7032 (2013). anomeric selectivity and then subjected to flash column chromatography (dry loading) to give the 2-deoxyglycoside 3 or 4. 27. Jungbauer, S. H. et al. Activation of a carbonyl compound by halogen bonding. Chem. Commun. 50, 6281–6284 (2010). 28. He, W., Ge, Y.-C. & Tan, C.-H. Halogen-bonding-induced hydrogen transfer Data availability to C=N bond with Hantzsch ester. Org. Lett. 16, 3244–3247 (2014). The authors declare that the data supporting the findings of this study are available 29. Jungbauer, S. & Huber, S. M. Cationic Multidentate halogen-bond donors in within the article and its Supplementary Information files. Additional data are available halide abstraction organocatalysis: catalyst optimization by preorganization. J. from the corresponding author upon reasonable request. Am. Chem. Soc. 137, 12110–12120 (2015). 30. Takeda, Y., Hisakuni, D., Lin, C.-H. & Minikata, S. 2-Halogenoimidazolium salt catalyzed aza-diels-alder reaction through halogen-bond formation. Org. Received: 14 January 2020; Accepted: 26 August 2020; Lett. 17, 318–321 (2015). 31. Saito, M., Tsuji, N., Kobayashi, Y. & Takemoto, Y. Direct dehydroxylative coupling reaction of alcohols with organosilanes through Si-X bond activation by halogen bonding. Org. Lett. 17, 318–321 (2015). 32. Saito, M., Kobayashi, Y., Tsuzuki, S. & Takemoto, Y. Electrophilic activation of iodonium ylides by halogen-bond-donor catalysis for cross-enolate coupling. References Angew. Chem. Int. Ed. 56, 7653–7657 (2017). 1. Knowles, R. R. & Jacobsen, E. N. Attractive noncovalent interactions in 33. Chan, Y.-C. & Yeung, Y.-Y. Halogen bond catalyzed bromocarbocyclization. asymmetric catalysis: links between enzymes and small molecule catalysts. Angew. Chem. Int. Ed. 57, 3483–3487 (2018). Proc. Natl Acad. Sci. USA 107, 20678–20685 (2010). 34. Benz, S., Poblador-Bahamonde, A., Low-Ders, N. & Matile, S. Cataysis with 2. Marchetti, L. & Levine, M. Biomimetic catalysis. ACS Catal. 1, 1090–1118 pnictogen, chalcogen, and halogen bonds. Angew. Chem. Int. Ed. 57, (2011). 5408–5412 (2018). 3. Wheeler, S. E., Seguin, T. J., Guan, Y. & Doney, A. C. Noncovalent interactions 35. Chan, Y.-C. & Yeung, Y.-Y. Halogen-bond-catalyzed addition of carbon-based in organocatalysis and the prospect of computational catalysis design. Acc. nucleophiles to N-acylimminium ions. Org. Lett. 21, 5665–5669 (2019). Chem. Res. 49, 1061–1069 (2016). 36. Kuwano, S., Suzuki, T., Hosaka, Y. & Arai, T. A chiral organic base catalyst 4. Mattson, A. E. Tricks for noncovalent catalysis. Science 358, 720 (2017). with halogen-bonding-donor functionality: asymmetric mannich reactions of 5. Schreiner, P. R. Metal-free organocatalysis through explicit hydrogen bonding malononitrile with N-Boc aldimines and ketimines. Chem. Commun. 54, interactions. Chem. Soc. Rev. 32, 289–296 (2003). 3847–3850 (2018). 6. Doyle, A. G. & Jacobsen, E. N. Small-molecule H-bond donors in asymmetric 37. Heinen, F., Engelage, E., Dreger, A., Weiss, R. & Huber, S. M. Iodine(III) catalysis. Chem. Rev. 107, 5713–5743 (2007). derivatives as halogen bonding organocatalysts. Angew. Chem. Int. Ed. 57, 7. Takemoto, Y. Recognition and activation by ureas and thioureas: 3830–3833 (2018). stereoselective reactions using ureas and thioureas as hydrogen-bonding 38. Kuwano, S., Suzuki, T., Yamanaka, M., Tsutsumi, R. & Arai, T. Catalysis based donors. Org. Biomol. Chem. 3, 4299–4306 (2005). on C-I…π halogen bonds: electrophilic activation of 2-alkenylindoles by 8. Connon, S. J. Organocatalysis mediated by (thio)urea derivatives. Chem. Eur. cationic halogen-bond donor for [4+2] cycloadditions. Angew. Chem. Int. Ed. J. 12, 5418–5427 (2006). 58, 10220–10224 (2019). 9. Zhang, Z. & Schreiner, P. R. (Thio)urea organocatalysis-what can be learnt 39. Guillier, K. et al. A halogen-bond donor catalyst for templated from anion recognition. Chem. Soc. Rev. 38, 1187–1198 (2009). macrocyclization. Angew. Chem. Int. Ed. 58, 14940–14943 (2019).

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2

40. Dreger, A. et al. A halogen-bonding-catalysed nazarov cyclisation reaction. 68. Kobayashi, Y., Nakatsuji, Y., Li, S., Tsuzuki, S. & Takemoto, Y. Direct N- Chem. Commun. 55, 8262–8265 (2019). glycofunctionalization of amides with glycosyl trichloroacetimidate by 41. Liu, X., Ma, S. & Toy, P. H. Halogen bond-catalyzed Friedel-Crafts reactions thiourea/halogen bond donor co-catalysis. Angew. Chem. Int. Ed. 57, of aldehydes and ketones using a bidentate halogen bond donor catalyst: 3646–3650 (2018). synthesis of symmetrical bis(indolyl)methanes. Org. Lett. 21, 9212–9216 69. Li, S., Kobayashi, Y. & Takemoto, Y. Organocatalytic direct α-selective N- (2019). glycosylation of amide with glycosyl trichloroacetimiate. Chem. Pharm. Bull. 42. Kaasik, M. et al. Halo-1,2,3-triazolium salts as halogen bond donors for the 66, 768–770 (2018). activation of imines in dihydropyridinone synthesis. J. Org. Chem. 84, 70. Mumcu, A. & Küçükbay, H. Determination of pKa values of some novel 4294–4303 (2019). benzimidazole salts by using a new approach with 1H NMR spectroscopy. 43. Kee, C. W. & Wong, M. W. In silico design of halogen-bonding-based Magn. Reson. Chem. 53, 1024–1030 (2015). organocatalyst for Diels-Alder reaction, claisen rearrangement, and cope-type 71. Dunn, M. H., Konstandaras, N., Cole, M. L. & Harper, J. B. Targeted and hydroamination. J. Org. Chem. 81, 7459–7470 (2016). systematic approach to the study of pKa values of imidazolium salts in 44. Xu, C. & Loh, C. C. J. A multi-stage halogen bond catalyzed strain-release dimethyl sulfoxide. J. Org. Chem. 82, 7324–7331 (2017). glycosylation unravels new hedgehog signaling inhibitors. J. Am. Chem. Soc. 72. Bell, R. P. The Tunnel Effect in Chemistry (University Press, Cambridge, 1980). 141, 5381–5391 (2019). 73. Kwart, H. Temperature dependence of the primary kinetic isotopic effect as a 45. Xu, C. & Loh, C. C. J. An ultra-low thiourea catalyzed strain-release mechanistic criterion. Acc. Chem. Res. 15, 401–408 (1982). glycosylation and a multicatalytic diversification strategy. Nat. Commun. 9, 74. Schreiner, P. R. Tunneling control of chemical reactions: the third reactivity 4057 (2018). paradigm. J. Am. Chem. Soc. 139, 15276–15283 (2017). 46. Bennett, C. S. & Galan, M. C. Methods for 2-deoxyglycoside synthesis. Chem. 75. Ley, D., Gerbig, D. & Schreiner, P. R. Tunneling control of chemical Rev. 118, 7931–7985 (2018). reactions – the organic chemist’s perspective. Org. Biomol. Chem. 10, 47. Balmond, E. I., Coe, D. M., Galan, M. C. & McGarrigle, E. M. α-Selective 3781–3790 (2012). organocatalytic synthesis of 2-deoxygalactosides. Angew. Chem. Int. Ed. 51, 76. Meisner, J. & Kästner, J. Atom tunneling in chemistry. Angew. Chem. Int. Ed. 9152–9155 (2012). 55, 5400–5413 (2016). 48. Balmond, E. I. et al. A 3,4-trans-fused cyclic protecting group facilitates α- 77. Nagel, Z. D. & Klinman, J. P. Tunneling and dynamics in enzymatic hydride selective catalytic synthesis of 2-deoxyglycosides. Angew. Chem. Int. Ed. 53, transfer. Chem. Rev. 106, 3095–3118 (2006). 8190–8194 (2014). 78. More O’ Ferrall, R. A. A pictorial representation of zero-point energy and 49. Palo-Nieto, C., Sau, A., Williams, R. & Galan, M. C. Cooperative Brønsted tunneling contributions to the kinetic isotopic effect. J. Phys. Org. Chem. 23, acid-type organocatalysis fort the stereoselective synthesis of deoxyglycosides. 572–579 (2010). J. Org. Chem. 82, 407–414 (2017). 79. Reyes, A. C., Amyes, T. L. & Richard, J. P. Primary deuterium kinetic isotope 50. Sau, A. et al. Palladium-catalyzed direct stereoselective synthesis of effects: a probe for the origin of the rate acceleration for hydride transfer deoxyglycosides from glycals. Angew. Chem. Int. Ed. 56, 3640–3644 (2017). catalyzed by glyerol-3-phosphate dehydrogenase. Biochemistry 57, 4338–4348 51. Palo-Nieto, C., Sau, A. & Galan, M. C. Gold(I)-catalyzed directed (2018). stereoselective synthesis of deoxyglycosides from glycals. J. Am. Chem. Soc. 80. Truhlar, D. G. et al. The incorporation of quantum effects in enyzme kinetics 139, 14041–14044 (2017). modeling. Acc. Chem. Res. 35, 341–349 (2002). 52. Das, S., Pekel, D., Neudörfl, J.-M. & Berkessel, A. Organocatalytic 81. Caldin, E. F. & Mateo, S. Kinetic isotope effects in various solvents for the glycosylation by using electron-deficient pyridinium salts. Angew. Chem. Int. proton-transfer reactions of 4-nitrophenylnitromethane with bases. J. Chem. Ed. 54, 12479–12483 (2015). Soc. Chem. Commun. Issue 22, 854–855 (1973). 53. Zhao, G. & Wang, T. Stereoselective synthesis of 2-deoxyglycosides from 82. Caldin, E. F. & Mateo, S. Kinetic isotope effects and tunnelling in the proton- glycals by visible-light-induced photoacid catalysis. Angew. Chem. Int. Ed. 57, transfer reaction between 4-nitrophenylnitromethane and 6120–6124 (2018). tetramethylguanidine in various aprotic solvents. J. Chem. Soc. Faraday Trans. 54. Sherry, B. D., Loy, R. N. & Toste, F. D. Rhenium(V)-catalyzed synthesis of 2- 1 71, 1876–1904 (1975). deoxy-α-glycosides. J. Am. Chem. Soc. 126, 4510–4511 (2004). 83. Mower, M. P. & Blackmond, D. G. Mechanistic rationalization of unusual 55. Ghosh, T., Mukherji, A. & Kancharla, P. K. Sterically hindered 2,4,6-Tri-tert- sigmoidal kinetic profiles in the Machetti-De Sarlo cycloaddition reaction. J. butylpyridinium salts as single hydrogen bond donors for highly steroselective Am. Chem. Soc. 137, 2386–2391 (2015). glycosylation reactions of glycals. Org. Lett. 21, 3490–3495 (2019). 84. Sharma, K. et al. Cation-controlled enantio-and diastereoselective synthesis of 56. Bradshaw, G. A. et al. Stereoselective organocatalyzed glycosylations- indolines: an autoinductive phase-transfer initiated 5-endo-trig process. J. Am. thiouracil, thioureas and monothophthalimide act as Brønsted acid catalysts at Chem. Soc. 137, 13414–13424 (2015). low loadings. Chem. Sci. 10, 508–514 (2019). 85. Keske, E. C., West, T. H. & Lloyd-Jones, G. C. Analysis of autoinduction, 57. Nakatsuji, Y., Kobayashi, Y. & Takemoto, Y. Direct addition of amides to inhibition and auto-inhibition in a Rh-catalyzed C-C cleavage: mechanism of glycals enabled by solvation-insusceptible 2-haloazolium salt catalysis. Angew. decyanative aryl-silylation. ACS Catal. 8, 8932–8940 (2018). Chem. Int. Ed. 58, 14115–14119 (2019). 86. Blackmond, D. An examination of the role of autocatalytic cycles in the 58. Palo-Nieto, C. et al. Copper reactivity can be tuned to catalyze the chemistry of proposed primordial reactions. Angew. Chem. Int. Ed. 48, stereoselective synthesis of 2-deoxyglycosides from glycals. Org. Lett. 22, 386–390 (2009). 1991–1996 (2020). 87. Bissette, A. J. & Fletcher, S. P. Mechanisms of autocatalysis. Angew. Chem. Int. 59. Liu, M. et al. Stereoselective electro-2-deoxyglycosylation from glycals. Angew. Ed. 52, 12800–12826 (2013). Chem. Int. Ed. 59, 15204–15208 (2020). 88. Plasson, R., Brandenburg, A., Jullien, L. & Bersini, H. Autocatalyses. J. Phys. 60. Williams, R. & Galan, M. C. Recent advances in organocatalytic Chem. A 115, 8073–8085 (2011). glycosylations. Eur. J. Org. Chem. 2017, 6247–6264 (2017). 89. Kononov, L. O. Chemical reactivity and solution structure: on the way to a 61. Park, Y. et al. Macrocyclic bis-thioureas catalyze stereospecific glycosylation paradigm shift? RSC Adv. 5, 46718–46734 (2015). reactions. Science 355, 162–166 (2017). 90. Kononov, L. O. et al. Concentration dependence on glycosylation outcomes: a 62. Levi, S. M., Li, Q., Rötheli, A. & Jacobsen, E. N. Catalytic activation of glycosyl clue to reproducibility and understanding the reasons behind. Eur. J. Org. phosphates for stereoselective coupling reactions. Proc. Natl Acad. Sci. USA Chem. 2012, 1926–1934 (2012). 116,35–39 (2019). 91. Yamamoto, H. & Futatsugi, K. “Designer acids”: combined acid catalysis for 63. Mayfield, A. B., Metternich, Trotta, A. H. & Jacobsen, E. N. Stereospecific asymmetric synthesis. Angew. Chem. Int. Ed. 44, 1924–1942 (2005). furanosylations catalyzed by bis-thiourea hydrogen bond donors. J. Am. 92. Peng, P. & Schmidt, R. R. An alternative reaction course in O-glycosidation Chem. Soc. 142, 4061–4069 (2020). with O-glycosyl trichloroacetimidates as glycosyl donors and Lewis acidic 64. Li, Q., Levi, S. M. & Jacobsen, E. N. Highly stereoselective β-mannosylations metal salts as catalyst: acid-base catalysis with gold chloride-glycosyl acceptor and β-rhamnosylations catalyzed by bis-thiourea. J. Am. Chem. Soc. 142, adducts. J. Am. Chem. Soc. 137, 12653–12659 (2015). 11865–11872 (2020). 93. Kumar, A., Kumar, V., Dere, R. T. & Schmidt, R. R. Glycoside bond formation 65. Overend, W. G., Rees, C. W. & Sequeira, J. S. Reactions at position 1 of via acid-base catalysis. Org. Lett. 13, 3612–3615 (2011). carbohydrates. Part III. The acid-catalysed hydrolysis of glycosides. J. Chem. 94. Whitesides, G. M. & Ismagilov, R. F. Complexity in chemistry. Science 284, Soc. 3429–3440 (1962). 89–92 (1999). 66. Ranade, S. C. & Demchenko, A. V. Mechanism of chemical glycosylation: 95. Schreiner, P. R. & Wittkopp, A. H-bonding additives act like Lewis acid focus on the mode of activation and departure of anomeric leaving groups. J. catalysts. Org. Lett. 4, 217–220 (2002). Carbohydr. Chem. 32,1–43 (2013). 96. Kotke, M. & Schreiner, P. R. Generally applicable organocatalytic 67. Castelli, R. et al. Activation of glycosyl halides by halogen bonding. Chem. tetrahydropyranylation of hydroxy functionalities with very low catalyst Asian J. 9, 2095–2098 (2014). loading. Synthesis Issue 5, 779–790 (2007).

16 NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-18595-2 ARTICLE

Acknowledgements Additional information We are grateful to Fonds der Chemischen Industrie for generous funding to C.C.J.L. Supplementary information is available for this paper at https://doi.org/10.1038/s41467- through a Liebig fellowship. Prof. Herbert Waldmann is greatly acknowledged for his 020-18595-2. generous assistance and infrastructural support. C.X. acknowledges the Max Planck Society for postdoctoral funding and the Alexander von Humboldt Foundation for a postdoctoral Correspondence and requests for materials should be addressed to C.C.J.L. fellowship. V.U.B.R. acknowledges the Max Planck Society for postdoctoral funding. We are grateful for infrastructural support from the Technical University of Dortmund and the Peer review information Nature Communications thanks the anonymous reviewer(s) for Max Planck Institut für Molekulare Physiologie, Dortmund. Robert Lott is acknowledged their contribution to the peer review of this work. for initial preliminary contributions to the project. Dr. Raphael Gasper-Schoenenbruecher of the Central Unit for Crystallography and Biophysics of the Max Planck Institut für Reprints and permission information is available at http://www.nature.com/reprints Molekulare Physiologie is gratefully acknowledged for the ITC expertise and assistance. Jens Warmer is gratefully acknowledged for MALDI-TOF mass spectrometry assistance. Bern- Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in hard Griewel is gratefully acknowledged for NMR expertise and generous assistance in the published maps and institutional afﬁliations. in situ NMR monitoring experiments. The NMR department of the faculty of Chemistry and Chemical Biology of the Technical University of Dortmund is gratefully acknowledged for providing NMR timeslots for the NMR monitoring experiments. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, Author contributions adaptation, distribution and reproduction in any medium or format, as long as you give C.C.J.L. conceived the idea and supervised the project. C.X. and V.U.B.R. conducted the appropriate credit to the original author(s) and the source, provide a link to the Creative majority of the experimental work. C.C.J.L, C.X., and V.U.B.R. prepared this manuscript. Commons license, and indicate if changes were made. The images or other third party C.C.J.L., C.X., V.U.B.R., and J.W. contributed to data analysis and commented on the material in this article are included in the article’s Creative Commons license, unless manuscript. indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory Funding regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Open Access funding enabled and organized by Projekt DEAL. licenses/by/4.0/. Competing interests The authors declare no competing interests. © The Author(s) 2020

NATURE COMMUNICATIONS | (2020) 11:4911 | https://doi.org/10.1038/s41467-020-18595-2 | www.nature.com/naturecommunications 17