ARTICLE

Molecular architecture of fungal cell walls revealedby solid-state NMRXue Kang 1, Alex Kirui 1, Artur Muszyński 2, Malitha C. Dickwella Widanage 1, Adrian Chen1,

Parastoo Azadi2, Ping Wang 3, Frederic Mentink-Vigier 4 & Tuo Wang 1

The high mortality of invasive fungal infections, and the limited number and inefficacy of

antifungals necessitate the development of new agents with novel mechanisms and targets.

The fungal cell wall is a promising target as it contains polysaccharides absent in humans,

however, its molecular structure remains elusive. Here we report the architecture of the cell

walls in the pathogenic fungus Aspergillus fumigatus. Solid-state NMR spectroscopy, assisted

by dynamic nuclear polarization and glycosyl linkage analysis, reveals that chitin and α-1,3-glucan build a hydrophobic scaffold that is surrounded by a hydrated matrix of diversely

linked β-glucans and capped by a dynamic layer of glycoproteins and α-1,3-glucan. The two-

domain distribution of α-1,3-glucans signifies the dual functions of this molecule: contributing

to cell wall rigidity and fungal virulence. This study provides a high-resolution model of fungal

cell walls and serves as the basis for assessing drug response to promote the development of

wall-targeted antifungals.

DOI: 10.1038/s41467-018-05199-0 OPEN

1 Department of Chemistry, Louisiana State University, Baton Rouge, LA 70803, USA. 2 Complex Carbohydrate Research Center, University of Georgia,Athens, GA 30602, USA. 3 Departments of Pediatrics, and Microbiology, Immunology and Parasitology, Louisiana State University Health Sciences Center,New Orleans, LA 70112, USA. 4National High Magnetic Field Laboratory, Tallahassee, FL 32310, USA. Correspondence and requests for materials should beaddressed to T.W. (email: tuowang@lsu.edu)

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 1

1234

5678

90():,;

Fungi are a group of eukaryotic microorganisms, some ofwhich are capable of causing superficial infections or serioussystemic diseases in humans. Superficial infections affect

nearly a quarter of humans, but more importantly, invasiveinfections caused by fungi such as the unicellular Candida speciesand filamentous Aspergillus fumigatus often result in fatality inindividuals with immunodeficiency. To date, life-threateningfungal infections affect more than two million people worldwide,with an exceptionally high-mortality rate of 20–95%1. As one ofthe most prevalent airborne fungi, A. fumigatus causes fatalinvasive aspergillosis in more than 200,000 patients annually,including a quarter of all leukemia patients, with an overallmortality rate of 50% for patients with treatment and nearly 100%for those left untreated2–6. The high-mortality rate is also coupledwith a substantial rise in occurrence due to a fast-growingpopulation with immunodeficiency and the wide application ofimmunosuppressive agents in medical treatments such as cancertherapy or organ transplantation.

Despite the above described medical significance, effectiveantifungal agents remain very limited. Most available antifungalstarget ergosterols in the cell membrane and therefore are toxic tohumans7,8. In addition, these antifungal drugs have limited effi-cacy. For example, Amphotericin B fails to prevent death in morethan half of the patients with invasive aspergillosis9. Moreover, asubstantial increase in drug resistance has been observed duringthe last decades6,8. Recent efforts have been devoted to developingagents that bind to the fungal cell wall since its polysaccharidesare absent in human cells10,11. Echinocandins are such newcompounds that disrupt glucan synthesis and perturb cell wallintegrity with reduced toxicity12–14. However, echinocandins arenot broad-spectrum drugs and are very expensive. All this makesit imperative to develop new compounds with better functionalmechanisms or different primary targets such as the poly-saccharides in the cell walls. One of the major challenges is thatthe fungal cell wall structure is poorly understood, placing abarrier to the development of cell wall-targeted antifungal agents.

Fungal cell walls typically contain, by weight, 50–60% glucans,20–30% glycoproteins, and a small portion of chitin, for example,10–20% for A. fumigatus10,15. Fungal glucans contain the pre-dominantly linear β-1,3-linkage and a small portion, typically10% or less, of β-1,6- and β-1,4-linkages. The supramolecularassembly of these biomolecules remains vague due to the lack of anon-destructive and high-resolution technique for characterizingthe insoluble, complex, and amorphous biomolecules within theintact cell wall16. To date, chitin microfibrils are thought to bedeposited next to the plasma membrane following the biosynth-esis of individual chains and the fibril formation process throughhydrogen bonding. These microfibrils may be covalently linked toβ-1,3-glucans17 that extend through the cell wall and tethermannoproteins on the cell wall surface through branched net-works with β-1,6-linked glucans18. The current understanding ofthe spatial packing has been shaped by the evidence from enzy-matic digestion, fractional solubilization, and isolation of cell wallcomponents followed by sugar analysis18,19. These chemical andenzymatic methods, however, are destructive and often fail toreveal the complicated polymer assembly generated by bio-synthesis machinery.

Recently, magic-angle spinning (MAS) solid-state NMR(ssNMR) spectroscopy has been employed to elucidate thestructure, spatial proximities, and ligand binding of native orgenetically modified polysaccharides in intact tissues, includingthe bacterial biofilm, plant biomass, and fungal pigment assem-blies20–28. Here, by integrating glucosyl linkage analysis, 2D 13C–13C/15N correlation ssNMR spectroscopy and the sensitivity-enhancing dynamic nuclear polarization (DNP)29–37 technique,we have successfully revealed the structural frame of the cell wall

in uniformly 13C, 15N-labeled A. fumigatus. We found that chitinand α-1,3-glucans closely interact to form a rigid and hydro-phobic scaffold that is surrounded by a soft and well-hydratedmatrix of β-glucans. Glycoproteins and a minor fraction of α-1,3-glucans form a highly dynamic shell coating the cell wall surface.In addition, we have revealed that fungal cell wall moleculesadopt polymorphic structures and heterogeneous dynamics inorder to perform versatile functions. Our findings also shed lightonto the machinery and mechanisms of cell wall componentbiosynthesis and their assembly. The methods presented in thisstudy enable the investigation of complex carbohydrates in intactcells and will allow the direct detection of fungal responses toantifungal agents through in-situ assessment of cell wallstructures.

ResultsChitin and glucans form the stiff fungal cell wall. Uniformly13C, 15N-labeled A. fumigatus samples were grown in a 13C/15Nliquid medium for 14 days. For ssNMR experiments, the A.fumigatus samples are analyzed in the intact and native state withminimal perturbation. We rely on the adequate sensitivity pro-vided by isotope labeling and the resolution from a series of two-dimensional (2D) 13C–13C and 13C–15N correlation spectra forassigning NMR resonances and analyzing the composition,mobility, intermolecular packing and site-specific water interac-tions of these complex carbohydrates in muro.

Glycosyl compositional analysis, assisted by ssNMR, demon-strated a major component of glucan (71%), chitin (9%), mannan(6%), and galactan (13%), as well as traces of chitosan andarabinan in A. fumigatus (Supplementary Table 1). A gaschromatography–mass spectrometry (GC-MS) glycosyl linkageanalysis of partially methylated alditol acetates (PMAA) (Supple-mentary Fig. 1) further revealed the highly diverse linkagepatterns of fungal glucans. The major form, 3-linked glucopyr-anosyl (3-Glcp), accounts for 73% of all neutral sugars and 86% ofGlcp residues, indicating the dominance of 1,3-glucans (Table 1).Another five types of Glcp linkages are also identified, comprising11% of all neutral sugars. Since glucans are better solubilized inthe linkage analysis than in the classical alditol acetate method ofthe compositional analysis, minor discrepancies between thesetwo methods are possible.

To assign the NMR resonances of cell wall polysaccharides, wemeasured 2D 13C–13C correlation spectra using 53-ms CORDmixing38,39 for through-space correlations (Fig. 1a) and refocused13C INADEQUATE pulse sequence40,41 for through-bondcorrelations (Fig. 1b). These 2D 13C–13C correlation experimentspreferentially detect the stiff cell wall due to the use of 1H–13Ccross polarization (CP). The A. fumigatus cell wall exhibitsexceptionally high resolution and the typical 13C full-width athalf-maximum (FWHM) linewidths range from 0.4 to 0.7 ppm.Major signals are from chitin, β-1,3-glucan, and α-1,3-glucan(Fig. 1a,b). This is consistent with the dominance of 3-Glcp in thelinkage analysis. The unique downfield 13C chemical shift of80–87 ppm at the linkage site of carbon 3 (C3) resolves the signalsof 1,3-glucans, and the C1 chemical shift tells the two anomersapart: 99–101 ppm and 102–105 ppm for α- and β-1,3-glucans,respectively. Weaker signals from β-1,4- and β-1,6-glucans havealso been identified, with downfield 13C chemical shifts of 85 and67 ppm at the linkage sites of C4 and C6, respectively. The lowintensity of β-1,4- and β-1,6-glucans is also in good agreementwith the glycosyl linkage analysis: only 7% of neutral sugars areglucans with linkages at C4 or C6 (Table 1). The representativestructures are shown in Fig. 1c and the 13C and 15N chemicalshifts are documented in Supplementary Table 2. These sugarunits may be covalently linked to form branched structures suchas the β-1,3/-1,6-glucan.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

2 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

Interestingly, the 2D 13C–13C correlation spectrum of thealkali-insoluble portion of cell walls from a related but non-pathogenic fungus, Aspergillus niger, also shows a comparablepattern to that of intact A. fumigatus, with signals primarilyfrom chitin, β-1,3-glucan and α-1,3-glucan (Supplementary

Fig. 2). Linkage analysis confirmed that more than 80% ofneutral sugars are 3-Glcp residues (Table 1). Therefore, thedominance of chitin, β-1,3-glucan, and α-1,3-glucan in thestiff core of cell walls is a conserved feature in Aspergillusspecies.

a

170180 30 20160 4013C single-quantum chemical shift (ppm)

60708090100110

130

140

150

160

170

180

50

B1a

B6a

B2a

B2aB3a

B3a B4a

B4aB5a

B5a

G1

G2

G3

G3

G2

G4

G6G5

G5G4

200

195

ChCH3ChCO

Ch2a

Ch1a Ch2a

Ch3a

Ch5a

Ch6a

Ch1c

A1aA2a

A3a

A5a

A6a

A2a

A4a

A4a

A5a

A3a

A1bA2b

A3bA3b

A2b

A4b

A6bA5b

A5b A4b

A1c

A2c

A3c

A3c

A2c

A4c

A5c

A4c

A5cA6c

H2H1

H2H3H4

H5

H5H6

b

13C

dou

ble-

quan

tum

che

mic

al s

hift

(ppm

)

c chitin and glucans

Chitin (Ch)α-1,3-Glucan (A)β-1,3-Glucan (B)β-1,4-Glucan (G)β-1,6-Glucan (H)

13C chemical shift (ppm)

13C

che

mic

al s

hift

(ppm

)

B5-4a

B1-3a

B2-4a

B4-6a

B3-4a

B1-5a

B1-2aB1-4a

B1-6a

B2-6a

B5-6a

B3-6a

B3-2a

B3-5a

B5-2a

H3-4H3-2

H5-6

H5-4

G1-4G1-6G1-2

G4-6

G5-6

G2-6

G4-5

G5-2

A1-6c

A3-2/5aA3-4a

A5/2-6a

A1-4a

A1-2/5a

A1-3a,c

A3-6a

A4-6a

A1-6aA1-2b A1-2c

A3-4c

A3-2c

A3-4bA3-2b

A5-4c

A4-2d

A5-6d

A5-6b

A4-6b

A5-6c

A1-5c

60708090100

60

70

80

90

100

50

110

Ch1-2a

Ch3-2a

Ch6-2a

Ch4-6a

Ch5-6a

Ch1-3aCh1-5a

Ch4-3aCh4-5a

Ch1-4a

β-1,3-glucan (B)

β-1,6-glucan (H)β-1,4-glucan (G)α-1,3-glucan (A)

Chitin (Ch)

OH

O

HOHO

OOH OH

OH

O O

OH

OHO

HOO

OH OH

OH

O O

OH

O

HO

O

O

OH

O

OH

OH

HO

OO

OH

OH

HO

HO

O

O

NH

CH3

O

NH

CH3

O

O

O

10170.1

84.5

69.6 71.7

60.5

103.6

74.4

86.4

68.7 77.1

61.3

103.3

69.4

71.7

85.3 74.3

63.4

103.655.0

72.9

83.0 75.760.9

22.6

174.8

116.2

102.669.4

79.0

71.676.9

67.3OH

OHHO

O

OO

O

HOHOHO

Fig. 1 Chitin and glucans form the rigid domain of intact A. fumigatus cell walls. a 2D 13C–13C correlation spectrum measured with 53-ms CORD mixingdetects all intramolecular cross peaks of chitin and four types of glucans. Abbreviations are used for resonance assignment and different polysaccharidesignals are color coded. b 13C CP J-INADEQUATE spectrum resolves the 13C through-bond connectivity for each polysaccharide. c Identifiedpolysaccharides and representative chemical shifts. All spectra were measured on an 800MHz solid-state NMR spectrometer

Table 1 13C-glycosyl linkages of fungal neutral sugars

Glycosyl residue A. fumigatus A. niger

Terminally linked mannopyranosyl (t-Manp) 1.4 0.1Terminally linked glucopyranosyl (t-Glcp) 2.6 2.5Terminally linked galactofuranosyl (t-Galf) 1.8 0.43-linked glucopyranosyl (3-Glcp) 72.8 81.14-linked glucopyranosyl (4-Glcp) 3.9 4.32,3-linked glucopyranosyl (2,3-Glcp) 2.3 4.02,6-linked glucopyranosyl (2,6-Glcp) 0.5 0.33,6-linked glucopyranosyl (3,6-Glcp) 2.1 2.82-linked mannopyranosyl (2-Manp) 1.7 0.86-linked mannopyranosyl (6-Manp) 0.8 0.54-linked galactopyranosyl (4-Galp) 10.0 3.4

The percentages reported are peak area from the relative EI detector response (%). The intact cell walls of A. fumigatus and the alkali-insoluble portion of A. niger cell walls are reported.

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 3

Fungal polysaccharides are highly polymorphic in structure.The decent 13C resolution of the A. fumigatus sample allows theunambiguous assignment of seven types of polysaccharides,including chitin, α-1,3-glucan, three types of β-glucans, mannan,and arabinan. In total 23 allomorphs have been identified forthese molecules. 1D 15N spectrum revealed multiple major typesof nitrogenated polysaccharides, with two peaks of chitin amidesat 123 and 128 ppm and a sharp amine peak at 33 ppm fromother amino sugars such as chitosan. Chitosan is an enzymaticalderivative of chitin and the degree of deacetylation (DD)42 is 10%as determined using the well-resolved 15N signals (Fig. 2a).

Chitin is found to be the most polymorphic molecule in fungalcell walls. 2D 15N–13C correlation spectrum resolves three majorforms of chitin (Fig. 2b), and just for the type-a chitin, we canresolve at least six C2-NH cross peaks, each representing adifferent sub-form. With another three minor types identifiedusing DNP, we have discovered 11 types of chitins (Supplemen-tary Table 2). This level of structural polymorphism is beyond ourcurrent knowledge obtained from X-ray and NMR studies onmodel chitins, and the 11 types of signals could not be explainedusing the known ways of packing: the antiparallel α-chitin,parallel β-chitin, and the mixed γ-chitin43,44. Signals from theknown structures, both the α and β allomorphs45, can be foundwithin the signals of type-a chitin, with a per-carbon chemicalshift difference as low as 0.3 ppm. In contrast, types b and c donot correlate with any known structures, and the chemical shiftdifference increased to at least 0.7 ppm and 1.2 ppm, respectively.Thus, chitin b and c belong to two structures that have never beenreported before. This unexpected level of structural polymorph-ism is potentially caused by the sophisticated pattern ofhydrogen-bonding through the N–H and C=O groups in chitinwhen multiple chains are put together outside the plasmamembrane after the biosynthesis.

It is striking that the three major forms of chitin identified inthis study are thoroughly mixed on the subnanometer scale in theintact A. fumigatus cell wall. This is revealed by the strong off-diagonal cross peaks between types a and b and between types aand c observed in the 15-ms 15N–15N proton-assisted recoupling(PAR) spectrum (Fig. 2c)46,47. For these interactions to happen in

the microfibrils, chitin allomorphs should coexist as tightlypacked chains in the fibrillar cross-section rather than asseparated domains associated longitudinally along the fibril. Thisorganization of chitin bears a resemblance to the assembly ofglucan chains in plant cellulose, in which seven types of glucanchains are found to coexist in the cross-section of a singlemicrofibril48. It should be noted that the presence of amide,methyl, and carbonyl groups substantially facilitates the reso-nance assignment of chitin allomorphs (Fig. 2d). This uniquechemical structure further serves as the basis for spectral editingto determine the chitin–glucan packing (vide infra).

MAS-DNP reveals a tight packing of chitin and α-1,3-glucan. Ithas been a long-standing question of how cell wall biomoleculesinteract to form the polymer network. To address this question,we measured a 15-ms 13C–13C PAR spectrum and discovered 23long range (5–10 Å) intermolecular cross peaks (Fig. 3a andSupplementary Fig. 3). Most of these cross peaks originate fromintermolecular interactions between chitin (Ch) and α-1,3-glu-cans (A), for instance, between type-b α-1,3-glucan carbon 1 andtype-b chitin carbon 4 (A1b–Ch4b) and between type-a/c α-1,3-glucan carbon 3 and type-a chitin carbon 4 (A3a,c–Ch4a). Inaddition, several cross peaks are also found between the chitin-α-1,3-glucan complex and β-glucans. Noteworthy examples includethe α-1,3-glucan carbon 1 to β-1,6-glucan carbon 3 (A1–H3)cross peak at (101, 79) ppm and α-1,3-glucan carbon 3 to β-1,3-glucan carbon 4 (A3–B4) cross peak at (85, 73) ppm. Therefore,despite the close packing of chitin and α-1,3-glucan, these twomolecules are still crosslinked by β-glucans.

To concurrently improve the sensitivity and resolution, wecombined the DNP technique49,50 with spectral-editing methodsand successfully identified another 35 long-range interactions.The magic-angle spinning DNP (MAS-DNP) enhances the NMRsensitivity by tens to hundreds of fold by transferring thepolarization from the electrons in radicals to NMR-active nucleiin biomolecules under microwave (µW) irradiation29. With anoptimized protocol ensuring homogeneous mixing of radicalswith cell wall biomolecules, a sensitivity enhancement (εon/off) of

a

a–ca–b

11012013C (ppm) 15N (ppm)

50607080170

100

120

105 100 25 20

a, NH-3 a, NH-2a, NH-4 a, NH-5

c, NH-4 b, NH-3 b, NH-6

b, NH-2

c, NH-3 c, NH-2

15N

(pp

m)

13C (ppm)

180

c 5 ms

d

15 ms

15N

(pp

m)

NH-CO

a

cb

NH-C1

a

cb

NH-C2/3/4/5/6 NH-CH3

a

b, c

OH

HO ONH

CH3

O

O

15N (ppm)

b

123

4 56

NH-C2

a

c b

5052545658OH

HO O

NH2

O

Chitin

Chitosan

NH-NH’

a–c

11013015N (ppm)

a–b

150 100 50 0

Amide

Amine

120

130

110

140

120

130

110

140130140

130

140

80

Fig. 2 Chitin is structurally polymorphic in intact A. fumigatus cell walls. a 1D 15N spectrum resolved multiple amide and amine signals. b High-resolution15N–13C correlations spectra resolved three major types of chitins. c DNP-assisted 15N–15N PAR spectra measured using 5ms (red) and 15 ms (black)mixing times revealed extensive cross peaks between different chitin allomorphs. d DNP-assisted 15N–13C correlation spectra measured using 3-mg A.fumigatus

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

4 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

30-fold has been achieved using the intact cells of 13C, 15N-labeled A. fumigatus on a 600MHz/395 GHz DNP spectrometer(Fig. 3c). The feasible sensitivity not only facilitates the detectionof long-range cross peaks with weak intensities but also allows usto employ spectral-editing methods to alleviate the signaloverlapping issue in intact cells. Briefly, the 15N magnetizationof chitin amide is first selected through a 15N–13C dipolarfilter51,52 and then transferred to spatially proximal glucans via aproton-driven spin diffusion (PDSD) mixing period (Supple-mentary Fig. 4). By subtracting two parent 15N–13C correlationspectra measured with short (100 ms) and long (1 s) mixingtimes, we can eliminate all intramolecular signals53 (Supplemen-tary Fig. 5). The resulting spectrum only contains long-rangeintermolecular cross peaks that are structurally important(Fig. 3d). Using this method, we have identified seven additionalcross peaks between different chitin allomorphs, between thechitin and α-glucan, such as the chitin nitrogen to α-1,3-glucancarbon 1 (ChNH-A1) cross peak, and between chitin amides andβ-glucan carbons such as the ChNH–H4 and ChNH–B4 crosspeaks. The same strategy is extended to measure chitin-edited2D 13C–13C correlation spectrum, which enables the identifica-tion of another 25 long-range cross peaks, for instance, betweenchitin carbonyl/methyl groups to β-1,3- or β-1,6-glucans(Fig. 3e).

Overall, these experiments have generated 65 long-range crosspeaks, among which 54 are intermolecular interactions and 11 areinter-allomorph cross peaks within chitin or α-1,3-glucans(Fig. 3b). The cross peaks are further categorized into 20 strong,21 medium, and 24 weak restraints according to the relativeintensity and the experimental methods (Supplementary Table 3).A total of 8 out of the 17 cross peaks between chitin and α-1,3-glucan are strong, constituting 40% of all strong restraints,supporting the proposed complex formed by tightly packed chitinand α-1,3-glucans. These two molecules further exhibit 25

through-space cross peaks to β-1,3-glucans, among which only7 are strong restraints, therefore, the β-1,3-glucan, assisted by β-1,4- and β-1,6-glucans, may serve as tethers between multiplechitin-α-1,3-glucans segments. This finding is further supportedby the high hydration level and mobility of β-1,3-glucan (videinfra).

Chitin and α-1,3-glucan form a hydrophobic core. To investi-gate carbohydrate–water interactions, we conducted the water-edited 2D 13C–13C correlation experiment54–56. This experimentrelies on a 1H-T2 relaxation filter to eliminate all original poly-saccharide signals and then transfers the water 1H magnetizationto carbohydrates so that only carbohydrates with bound watercan be detected. The water-edited signals are compared with theequilibrium intensities of a control spectrum (Fig. 4a, b), and theintensity ratios tell the polymer hydration in a site-specificmanner. Among all the complex carbohydrates, α-1,3-glucan andchitin are most hydrophobic and have the lowest water-transferred intensity. For instance, all α-1,3-glucan cross peaks,including A4-1 (carbons 4 to carbon 1), A4-3, A4-2, and A4-6,show substantial dephasing in the 2D water-edited spectrum(Fig. 4a) and its 69.5-ppm 13C cross-section (Fig. 4b). The resi-dual intensities are below 40% for chitin and α-1,3-glucan but arehigher than 60% for all the well-hydrated β-glucans (Fig. 4c andSupplementary Table 4). The hydration data dovetail nicely withthe long-range correlation results, collectively indicating thatchitin and α-1,3-glucan tightly pack to form the hydrophobicdomains that are surrounded by a well-hydrated matrix formed ofβ-glucans. The formation of this dehydrated domain might befacilitated by the assembly of chitin microfibrils, with α-1,3-glu-cans occasionally deposited on the fibrillar surface or betweenmultiple chains or several microfibrils to further reduce the wateraccessibility.

13C (ppm)

A1a,c-H3A1a,c-Ch2b,c,d

A1a,c-Ch4e

A1a,c-B2a/G5

Ch3d,e-A2c,d

60708090

60

70

80

90

100

110

50

13C

(pp

m)

a

A1b-G4 A1b-Ch6bA1b-Ch4b

A3a,c-Ch4c

A3a,c-B4aA3a,c-Ch4a

Ch4a-A2c,dCh3d,e-B5c

15N

(pp

m)

13C (ppm)

607080100110

A6b/Ch6a-Ch6c

13C (ppm)

ChNHb,c-A2b

ChNHb-B4a

ChNHb-H4

ChNHa,b-A4b/A2cChNH

c-4b

ChNHa-4b

ChNHc-A1a,c,d

150

130

110

65 long-range interactions dPARDARR

b

β-1,6-glucan

β-1,3-glucan

β-1,4-glucan

13C (ppm)

406080100 20

c

With DNP

No DNP, ×4

ε~30

Carbohydrate

Protein & lipids

2 (1,0)

2 (0,2)

17 (8,3) 6 (1,2)

2 (0,1)

12 (6,4)

13 (1,6)

10 (2,3)

Chitin

α-1,3-glucan1 (1,0)

20 1060708090 50

e

13C

(pp

m)

Ch2d,e-B6a

B5a-Ch2d

Ch4a-2dCh4a-B6b

Ch2d-4a

Ch2e-A3a,c

Ch2f-B6a

B5a-Ch4a

B5a-A3a,c

Ch4a-2f

180

70

70

80

60

70

80

90

5060

170

G6-ChCOa

A2b-ChCH3a,c

B5a-ChCH3a,c

B2b-ChCH3

a,c

Fig. 3 MAS-DNP solid-state NMR reveals the tight packing of chitin and α-1,3-glucan. a 15 ms 13C–13C PAR spectrum (black) reports many intermolecularcross peaks, mainly between chitin and α-1,3-glucans. A 100-ms DARR spectrum (orange) that primarily detects intramolecular correlations is overlaid forcomparison. b Summary of 65 long-range restraints. For each category, the number of all cross peaks and the number of strong and intermediate restraints(in parenthesis) are listed. c Sensitivity enhancement εon/off of 30-fold was obtained for A. fumigatus. A picture of a DNP sample is also included. d DNP-assisted intermolecular-only 15N–13C correlation spectrum unambiguously detected several chitin–glucan cross peaks. e DNP-assisted chitin-editedspectrum only shows signals from chitin itself or the glucans that are spatially proximal. The DNP experiments were conducted on a 600MHz/395 GHzspectrometer

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 5

Rigid chitin and α-glucan are enclosed by mobile β-glucans. Tosystematically examine the dynamics of cell wall biomolecules, wemeasured NMR relaxations and a series of 1D 13C spectra thatselect components with different mobilities. The 1D 13C spectrummeasured using 13C direct polarization (DP) and a long recycledelay of 35 s provides quantitative detection of all cell wallcomponents including polysaccharides, proteins, lipids and smallmolecules (Fig. 5a). This quantitative spectrum differs sub-stantially from the spectrum that favorably detects mobilemolecules through the combined use of 13C DP and short recycledelays of 2 s. The difference between these two DP spectra iscomparable to the 13C CP spectrum that favors the rigid com-ponents with stronger 1H–13C dipolar couplings (Fig. 5b, c). Themajor peaks in this difference spectrum are from chitin and α-1,3-glucan, unveiling the stiffness of these two molecules. At thesame time, signals from relatively mobile components, such asproteins, lipids, β-glucans, and small molecules, are substantiallysuppressed (Fig. 5b).

When compared with the quantitative detection, α-1,3-glucans have the highest intensity in the CP spectrum selectingstiff polymers, doubling that of quantitative DP, but the lowestsignals in the 2-s DP spectrum that favors mobile molecules(Fig. 5d). Accordingly, α-1,3-glucan is the most rigid poly-saccharides in A. fumigatus. Two subtypes, a and b, of chitin arefound to be the second most rigid molecules, with 50% highersignals in CP but 0–50% less intensity in 2-s DP spectrum thanthe quantitative spectrum. All the β-glucans, together withanother two types of chitins, fully retain their signals in 2-s DPspectrum, thus are dynamic (Fig. 5d and SupplementaryTable 5).

The molecular motional rates are quantified by measurementsof 13C-T1 relaxation that probes local reorientation and 1H-T1ρ

for larger-scale motions, such as cooperative movement ofmultiple sugar rings. Cell wall polysaccharides exhibit distinct1H-T1ρ relaxation times, signifying a highly heterogeneous profileof dynamics (Supplementary Fig. 6a and Supplementary Table 6).The average 1H-T1ρ is 4.5 ms, 2.8 ms, and 1.1 ms for α-1,3-glucans, chitin, and β-glucan, respectively (Fig. 5e). Consistently,α-1,3-glucans also have the slowest 13C-T1 relaxation among all

types of glucans (Supplementary Fig. 6b and SupplementaryTable 7). Therefore, α-1,3-glucan is the most rigid molecule,followed by chitins and then β-glucans.

Interestingly, chitin has the largest variance in relaxation times(Fig. 5e), which, together with the allomorph-specific 1D 13Cintensity, suggests a functional-relevant structural polymorphism:a subgroup of chitin allomorphs are responsible for forming rigidmicrofibrils whereas the remainders retain considerable disorderdue to unfavored conformations or unstable hydrogen-bondingpatterns. Type-a chitin has similar chemical shifts as the α and βchitins, and these allomorphs clearly bear high rigidity in theintact cell walls. It should be noted that for most α-1,3-glucanpeaks the satisfactory fit of 1H-T1ρ data can only be achievedusing double exponential equations. This suggests a two-domaindistribution: 70–90% of α-1,3-glucans interact with chitins toform a stiff and hydrophobic scaffold conferring rigidity to thecell wall (the long 1H-T1ρ component of 3.8–5.0 ms) while theother 10–30% have very short 1H-T1ρ relaxation of less than amillisecond due to the interactions with the mobile β-glucans(Fig. 5e and Supplementary Fig. 6a). Particularly, β-1,3-glucan isthe major binding target of α-1,3-glucans as they have 12intermolecular cross peaks, among which half are stronginteractions (Fig. 3b).

Glycoproteins and α-1,3-glucans form a highly mobile shell.While the previous 2D experiments were CP-based and primarilyfocused on the relatively rigid polysaccharides that aremechanically important, a 13C DP J-INADEQUATE spectrumwith a short recycle delay of 2-s is also measured to highlight themobile domain of cell walls (Fig. 6a). The primary signals havebeen attributed to proteins and some polysaccharides. The verysharp 13C linewidths of 0.3–0.5 ppm confirmed the rapidmotional averaging of these molecules. Unambiguous signals ofmannan and arabinan are observed (Fig. 6b). Since mannan is amajor component of fungal glycoproteins purposely forming anoutmost layer of fungal cell walls1,15, our results reveal that thisouter shell is highly dynamic and spatially separated from therelatively rigid inner domain of chitin and glucans. Despite the

60708090100110

× 2

69.5 ppm

A1a A3a

A2a/G3/H4

A6aG1 G4

G5H3 G6

A6b

A4a/G2/H2

× 2

Ch3a

Ch2a

G2B4a

G6

103.3 ppm

B6a

G5/B2a

B5aG4

B3a

G1/B1a/Ch1a

Ch4c

A2b

H6

B3b

5565758595105

60

70

80

90

100

13C

che

mic

al s

hift

(ppm

)

13C chemical shift (ppm) 13C chemical shift (ppm)

ba Water-editedFull

H2-3

G5-1/B2-1a

G1-5/B1-2a

G2-1 G2-4G2-5 G2-6

G5-4 G5-2 G5-6

G1-2

G1-6

G1-4

B6-1aB6-3a B6-5a

B6-2a

B6-4a

B2-3aB2-5a

B2-4a

B2-6a

B1-4a

B1-6a

B1-5a

B1-3a

Chitin (Ch)α-1,3-Glucan (A)β-1,3-Glucan (B)β-1,4-Glucan (G)β-1,6-Glucan (H)

A6-1a,cA6-3a,c

A6-2/5a

A6-4a

A4-1aA4-3a

A4-6aA4-6b

A6-4bA6-2b/A6-5c

Ch1-3aCh1-2a

Ch5-4c

1.0

0.8

0.6

0.4

0.2

0.0

Chitin

β-1,3-

gluca

n

α-1,3

-gluc

an

β-1,4-

gluca

n

β-1,6-

gluca

n

Nor

mal

ized

inte

nsity

, wat

er-e

dite

d/co

ntro

l

c

60708090100110

Fig. 4 Chitin and α-1,3-glucans form the hydrophobic core of A. fumigatus cell walls. a Overlay of 2D water-edited (red) and control (black) 13C–13Ccorrelation spectra and b 1D 13C cross sections. c Relative intensities of different polysaccharides in water-edited spectra. The water-edited spectrumpreferentially detects well-hydrated molecules. Error bars indicate standard deviations propagated from NMR signal-to-noise ratios. Chitin andα-1,3-glucans have the lowest intensities in water-edited spectra (shaded)

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

6 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

60708090

140

150

160

100110

170

190

7080

13C

dou

ble-

quan

tum

che

mic

al s

hift

(ppm

)

13C single-quantum chemical shift (ppm)

M1a M2a

M4a

M3aM4a

M2aM3a

M5aM6a

α-1,3-Glucan (A)β-1,4-Mannan (M)Arabinan (Ar)

Ar1a

Ar1b

Ar2a

Ar2b

Ar3a

Ar4a

Ar4a

Ar5a

Ar2a

A1a

A1b

A1c

A2a

A2b

A2c

A3a

A4a

A3bA3b

A2b

A4b

A5aA4a

A5a A4a

100110 7080

175

185

200 160 120 80 40

100

140

180

220

260

300

20

60

Polysaccharide

Proteinaliphatic

Protein, aromatic

Protein, CO Protein, Ca

Folded, protein sidechain

13C

dou

ble-

quan

tum

che

mic

al s

hift

(ppm

)

13C single-quantum chemical shift (ppm)

a b

Fig. 6 Glycoprotein and α-1,3-glucan form a highly mobile shell. a 2D 13C DP J-INADEQUATE spectrum selects only the very mobile molecules of proteinsand polysaccharides. b The polysaccharide region reveals the presence of β-1,4-mannan, arabinan and α-1,3-glucan

13C chemical shift (ppm)

100 80 60 40 20 0

A1A3

A2/5 A4

A6Ch1Ch4

Ch6

ChCH3

G/B1a/Ch1a,c,d,e

H1/B1a/Ch1bB3a,b

B5aH5

B2a/G2/Ch5b

Ch3a,b/B2c

A2/5a/G3

A4a,d/A5b/G2/H2/B4b,c/Ch5c

H6G6

B6a

A6a,b,c/Ch6a,b

Carbohydrate

Proteins and lipids

A1a,c,dA3a,c

A2c,d/A4b

A3b

Chitin (Ch)α-1,3-glucan (A)β-1,3-glucan (B)β-1,4-glucan (G)β-1,6-glucan (H)

H3

Ch5a

A2b/Ch3f

B4a

Ch1d/A1b

Ch4d,e

Ch4a

Ch3c,d,e

Ch2a,c

ChCH3a,c

Ch4c

Ch2e

Ch2f ChCH3b

a Mobile molecules

Quantitative

bDifference

c Rigid molecules

2.0

1.5

1.0

0.5Nor

mal

ized

Inte

nsity

dDistribution map of relative rigidity

β-1,3-gluca

n

β-1,4-gluca

n

β-1,6-gluca

n

α-1,3-gluca

n

Chitina,b

Lipids

Lipids

α-1,3-glucan

Chitin

β-glucans

5

4

3

2

1

0

I2s,DP/I35s,DP

ICP/I35s,DP

1 H T

1ρ (

ms)

e

7779103

86.52224515574101

60.7708571

13C chemical shift (ppm)

Ch2b

6

7561.3

Fig. 5 A. fumigatus cell walls are dynamically heterogeneous. a 1D 13C DP spectra of intact A. fumigatus cells with quantitative detection of all molecules(black) or selective detection of the mobile components (red). Abbreviations of carbohydrate names, carbon numbers, and subtypes (superscripts) areincluded in the assignment. b Difference spectrum obtained by subtraction of the two 1D DP spectra. c 13C CP spectrum that favors rigid molecules. d Peakintensity ratios between different DP spectra and the CP spectrum show that α-1,3-glucan and a subset of chitins are relatively rigid. Error bars are standarddeviations propagated from NMR signal-to-noise ratios. e 1H-T1ρ relaxation times. The open and filled bars represent the short and long-components of 1H-T1ρ relaxation. Error bars are standard deviations of the fit parameters. Dashlines indicate the average value of the longer 1H-T1ρ component for each type ofpolysaccharides

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 7

fact that most α-1,3-glucans participate in the formation of stiffand hydrophobic cores, a considerable amount of α-1,3-glucansignals remain in this dynamic domain (Fig. 6b), which supportsthe hypothesis that α-1,3-glucans may form the outmost layer ofcell walls to block the immune recognition of β-glucan receptorsin the host cells11,57. Thus, α-1,3-glucan is bi-functional: sup-porting cell walls through the formation of hydrophobic scaffoldsand potentially increasing fungal virulence by disabling the hostdetection of invading microbes.

In addition to this mobile domain, we have also identified arigid component of proteins (Supplementary Fig. 7a andSupplementary Table 8). These proteins are as hydrophobic asthe fatty acid chains of the membrane and are more hydrophobicthan any polysaccharides (Supplementary Fig. 7b and Supple-mentary Table 9). These rigid proteins are mainly membraneproteins but may also exist in the hydrophobin rodlet proteinlayer, a hydrophobic coating preventing the immunerecognition58,59.

DiscussionThis study presents a high-resolution and non-destructivemethod for determining the architecture of fungal cell walls.Solid-state NMR and MAS-DNP results of 65 intermolecular andinterallomorph interactions, site-specific hydration, and mole-cular mobility steadily indicate a two-domain distribution ofmolecules: glucans and chitins form a relatively rigid and innerportion of cell walls, while mannoproteins and α-1,3-glucan formthe extremely mobile outer shell. In the inner domain, α-1,3-glucan and chitin are tightly packed as the most rigid andhydrophobic cores that are embedded in a well-hydrated andrelatively mobile matrix formed by β-1,3-, β-1,4-, and β-1,6-glu-cans (Fig. 7).

Compared with previous biochemical analyses, this NMR-derived model has both consistency and revision. For decades,we have been solubilizing individual components of the cell wallusing chemical or enzymatical treatment, for example, alkalisolubilization, and then determine the composition and covalentlinkages of the extracted portions17,18,60–64. A conserved skele-ton of branched β-1,3 and 1,6-glucans with β-1,4-linkages to

chitins is found in the alkali-insoluble component of most fungi,while the other molecules vary substantially in mold andyeast16,60. Mannoproteins are found in the cell wall surface andconnected to the cell wall via either non-covalent connections orcovalent linkages to β-1,6-glucans65,66. These biochemical dataand our ssNMR analysis dovetail well considering the structuralroles of chitin and β-glucans, as well as the outer layer ofproteins.

The current model also shifts the prevailing paradigm offungal cell wall in three aspects. First, we have identified themolecules that determine cell wall rigidity. As the most abundantbuilding unit of the fungal cell wall, β-glucans, especially themulti-branched polymers of β-1,3/-1,6-glucan, have long beenproposed to form the rigid network15,16. However, the high levelof hydration and the intermediate mobility of β-glucans wefound have clearly excluded this structural role. Instead, wefound that the rigid scaffold are formed by chitin and α-1,3-glucan. The structural role of α-1,3-glucan is unexpected since itis usually defined as the major alkali-soluble polysaccharide inprevious studies16. Second, the current model reveals the brid-ging role of β-1,3-glucans to a great detail. At present, all β-glucans are indistinguishable in dynamics and hydration, butamong the three types of β-glucans, β-1,3-glucan absolutely hasmore pronounced interactions with other molecules, in parti-cular, with α-1,3-glucan. This can be seen from the 12 inter-molecular cross peaks between β-1,3-glucan and α-1,3-glucan inwhich half are strong restraints (Fig. 3b). β-1,3-glucan and chitinalso have a large number of cross peaks, 13 in total, most ofwhich are only weak or intermediate in strength. Thus, chitinmay serve as a secondary anchor, following α-1,3-glucan, for β-1,3-glucan to link the rigid and mobile domains (Fig. 7). Third,the dual functions of α-1,3-glucan have been emphasized. Thehigh rigidity of α-1,3-glucan has not been expected and wepropose that the stiffening by covalent or physical interactionswith β-1,3-glucans and chitin grant α-1,3-glucan with the cap-ability of performing structural roles. Linkages between α-1,3-and α-1,4-glucose units have been reported previously61, but it isunclear whether covalent interactions exist between α-1,3-glu-cans and chitin or β-1,3-glucans. Interestingly, the occurrence of

Proteins

Mannan

Arabinan

Chitin (Ch)

α-1,3-glucan (A)

α-1,3-glucan

β-1,3-glucan (B)

β-1,4-glucan (G)

β-1,6-glucan (H)

Highly mobileouter shell

Innerdomain

66%

62%

87% 1.1 ms

19%, 4.5 ms

30%, 2.8 ms

Hydrated& mobile

Hydrophobic& rigid

Ch-A17 (8)

A-B12 (6)

Membrane

Ch-B13 (1)

Ch-H6 (1)

A-G2 (0)

Ch-G2 (1)

A-H2 (0)

Fig. 7 Illustrative model of the supramolecular architecture of A. fumigatus cell walls. Dashline circles highlight the intermolecular interactions with the totalnumbers of NMR restraints indicated. The number of strong restraints is in parenthesis. The average hydration levels (percentage values) and the average1H-T1ρ relaxation times (in millisecond) of each polymer are also labeled

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

8 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

α-1,3-glucans in two distinct domains, the outer surface and theinner rigid cores, demonstrated the structural and functionalversatility of this molecule.

A. fumigatus produces a low amount of chitosan, a cationicmolecule with various biomedical and industrial applications dueto its antimicrobial, anti-tumor, wound-healing, and cholesterol-lowering properties67. Previously, structural roles have beenassumed for chitosan since the chitosan-deficient strains of thepathogenic fungus Cryptococcus neoformans have compromisedcell wall integrity in vitro, resulting in attenuated virulence68.However, in A. fumigatus, chitosan accounts for <1% of all car-bohydrates (Supplementary Table 1) and remains mobile asrevealed by the sharp 15N linewidth of 1.5 ppm (Fig. 2a).Therefore, chitosan contributes negligibly to cell wall rigidity inA. fumigatus.

The current study also provides insights into the poly-saccharide scaffold for pigment deposition. The insoluble pig-ment of melanin increases cell hydrophobicity and reduces cellwall porosity, which was proposed to be the cause of drugresistance in many fungi69–72. Recent ssNMR studies of melanin“ghosts” indicated that melanin may be integrated into the cellwall via association with the polysaccharides in C.neoformans27,28. However, the scaffold that holds the pigment inplace was unclear. Chitin has been proposed to be the supportingpolysaccharide, but our results revealed a stiff and hydrophobicframe of α-1,3-glucans and chitins that could accommodate thearomatic assembly of pigments, which, in turn, may provide anexplanation for the high hydrophobicity of these two poly-saccharides (Fig. 4c).

Notably, the fungal wall is significantly more dynamic than itscounterparts in plants. The narrow 13C linewidth of fungalpolysaccharides is comparable to that of the matrix poly-saccharides in the fast-growing primary plant cell walls, but isapparently narrower than that of rigid cellulose microfibrils48.Upon maturation, plant cell walls are further rigidified anddehydrated by the deposition of lignin and the coalescence ofcellulose microfibrils73. Therefore, despite the presence of arelatively rigid scaffold formed by chitin fibrils and α-1,3-glucans,the polymer network in fungi still retains considerable plasticity,which allows the fungal cell wall to reshape its molecular archi-tecture to survive through different external stress and to fulfilldiverse functions. The lack of need for vertical growth and thelimited size of microbes may also explain the high mobility offungal cell walls. In addition, the fungal cell wall also has asubstantially larger number of intermolecular cross peaks thanthe primary plant cell walls. This may be caused by the extensivecovalent cross-linking of glucans and chitin in fungi, while plantsprincipally rely on non-covalent interactions, such as van derWaals forces and electrostatic interactions, as well as the entan-glement and entrapment of polymers. Given the high mobility offungal biomolecules, it may be crucial to have chemical linkagesinstead of physical contacts as the primary interactions so thatthis dynamic structure could remain intact under external andinternal stress.

Although this study builds a frame for the fungal cell wallstructure, in particular, for the complex carbohydrates, morestructural details await systematic investigations. Emergingquestions include the correlation of linkage patterns withstructural or mechanical functions for β-glucans and thedevelopmental changes of cell wall architecture, for example,during conidial outgrowth. A detailed and dynamic under-standing of the supramolecular assembly of fungal cell wallsand future studies on drug response will substantially facilitatethe design of better antifungal compounds that inhibit abroader spectrum of invasive fungal infections with minimal orno side toxicity.

MethodsPreparation of fungal materials. Uniformly 13C,15N-labeled mycelium was pre-pared by IsoLife (Wageningen, The Netherlands) using the following protocol.Aspergillus fumigatus cultures (strain RL 578, a wild strain obtained from compost)were grown on a 13C/15N liquid medium (a modified Czapek-Dox medium) understill cultivation at 30 °C for 14 days in the dark on 50-mL medium in 250 mLcapacity Erlenmeyer flasks. At the end of the cultivation the mycelium was flash-frozen in liquid nitrogen and stored at −80 °C. The resulting materials were furtherdialyzed for six times over 3 days to remove the majority of small molecules fromthe isotope-labeled media and to reduce the ion concentration. For solid-stateNMR experiments, 30 and 100 mg of this whole-cell material was packed into 3.2-mm and 4-mm magic-angle spinning (MAS) rotors, respectively. Another 5 mgwas proceeded in the DNP matrix for DNP experiments and 3 mg was finallytransferred into a 3.2-mm sapphire rotor. The hydration level is 90% for the initialculture and has been decreased to around 50% in the NMR samples by multipletimes of compression.

To verify the composition of the rigid portion of fungal cell walls and compareit with the whole-cell, the alkali-insoluble polysaccharides were extracted by IsoLifefrom A. niger mycelium by deproteinization with 2% w/v sodium hydroxidesolution (30:1 v/w, 90 °C, 2 h), separation of alkali-insoluble fraction bycentrifugation (4000 × g, 15 min), extraction of chitosan under reflux (10% v/vacetic acid, 40:1 v/w, 60 °C, 6 h), separation of crude cell walls by centrifugation(4000 × g, 15 min). The crude was washed with water, ethanol and acetone andthen air-dried at 20 °C. A total of 70 mg of the extracted polysaccharides waspacked in a 4-mm rotor for solid-state NMR experiments.

Glycosyl composition analysis. Glycosyl composition analysis of neutral sugarsconstituting the cell wall was achieved after the dispersion of 600 µg dry samples in0.5 mL 2M TFA (v/v) in sealed reaction tubes, by 20 min sonication in an ultra-sound water bath at RT, followed by 2 h of hydrolysis at 121 °C, overnightreduction with NaBD4, and 1 h acetylation with acetic anhydride and pyridine at80 °C. Inositol was used as an internal standard. The glycosyl constituents wereassigned based on the retention time of the derivatives of original sugar standardsand unique EI-MS fragments of 13C alditol acetate derivatives identified in thefungal samples. We used the same GLC-MS equipment and temperature programas for linkage analysis.

Glycosyl linkage analysis. The glycosyl linkage of 13C, 15N uniformly labeledpolysaccharides was obtained by GC-MS analysis of partially methylated alditolacetates (PMMA)74 after 2 h hydrolysis with 2M (v/v) TFA at 121 °C, overnightreduction with NaBD4 and acetylation with acetic anhydride and pyridine. Theinositol was used as an internal standard. GLC-MS analysis was performed on anHP-5890 GC interfaced to a mass selective detector 5970 MSD using a SupelcoSP2330 capillary column (30 × 0.25 mm ID, Supelco) with the following tem-perature program: 60 °C for 1 min, then ramp to 170 °C at 27.5 °C/min, and to 235°C at 4 °C/min with 2 min hold and finally at to 240 °C at 3 °C/min with 12 minhold.

Because the organism was grown in media supplemented with 13C and 15N asthe sole carbon and nitrogen source, respectively, one could expect these isotopes tobe incorporated in the cell wall polysaccharides. Consequently, GC-MS analysis ofpartially methylated alditol acetates (PMAA) generated a set of unique EI-MSdiagnostic fragments that differed from fragments predicted for classical PMAAs(Supplementary Fig. 1). It should be noted that the complex glucan samplessolubilize better during the linkage analysis (DMSO solvent and permethylationstep prior to TFA hydrolysis) than in the classical alditol acetate method ofcompositional analysis (only TFA hydrolysis). Thus, we have a better detection ofglucans in the PMMA analysis.

Solid-state NMR experiments. Solid-state NMR experiments were conducted ona Bruker Avance 800MHz (18.8 Tesla) spectrometer and a 400MHz (9.4 Tesla)Bruker Avance spectrometer using 3.2-mm and 4-mm MAS HCN probes,respectively. Most experiments except those with MAS-DNP were collected under10–13.5 kHz MAS at 290 K. 13C chemical shifts were externally referenced to theadamantane CH2 signal at 38.48 ppm on the TMS scale. 15N chemical shifts werereferenced to the liquid ammonia scale either externally through the methionineamide resonance (127.88 ppm) of the model peptide N-formyl-Met-Leu-Phe-OH75

or using the ratio of the gyromagnetic ratios of 15N and 13C. Typical radio-frequency field strengths, unless specifically mentioned, were 80–100 kHz for 1Hdecoupling, 62.5 kHz for 1H CP and hard pulses, and 50–62.5 kHz for 13C and 41kHz for 15N.

To assign the 13C and 15N resonances of cell wall biomolecules, five types ofexperiments were measured: (1) 2D refocused 13C J-INADEQUATE40,41 spectrawith DP and short recycle delays of 2 s for selective detection of the very mobilephase in the outer layer of cell walls; (2) 13C CP J-INADEQUATE for the detectionof rigid molecules located inside the cell wall; (3) 13C–13C RFDR76 for the detectionof one-bond cross peaks; (4) 2D 13C–13C spectra using 53 ms CORD38,39 or 50 msDARR for the detection of intramolecular cross peaks; (5) 2D 15N–13C N(CA)CXheteronuclear correlation spectra77 to select the amide signals of chitins. The N(CA)CX is measured using a 0.6 ms 1H–15N CP contact time, a 5 ms 15N–13C CP

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 9

and a 100 ms 13C–13C DARR mixing period. The 15N and 13C spin-lock fieldstrengths for the 15N–13C CP were 20 kHz and 33 kHz, respectively. The 13C and15N carrier frequencies were 55 ppm and 70 ppm, respectively. A strong 1Hdecoupling of 100 kHz is applied during the 15N–13C CP.

To determine the spatial proximities of biomolecules in intact cell walls, wemeasured 15-ms 13C–13C PAR78 using 13C field strengths of 53 kHz and 1H fieldstrengths of 50 kHz. A total of 23 intermolecular cross peaks have been identified inthe PAR spectra, which, in combination with 7 long-range cross peaks in 3-s PDSDand 35 long-range cross peaks in DNP experiments, restrain the spatial packing ofmolecules in intact cell walls.

To site specifically determine the water accessibilities of differentpolysaccharides, we measured water-edited 2D 13C−13C correlation spectra54,55,79.This experiment initiates with 1H excitation followed by a 1H-T2 filter of 0.88 ms ×2 that eliminates 97% of polysaccharide signals but retains 80% of watermagnetization, a 4-ms 1H mixing period for water-to-polysaccharide transfer and a1 ms 1H–13C CP for 13C detection. A 50-ms DARR mixing period is used for boththe water-edited spectrum and the control 2D spectrum showing full intensity. Theintensity ratio between the water-edited spectrum and the control spectrum isquantified, which is further normalized by that of the C2–C3 cross peak ofβ-1,6-glucan, the highest value among all the peaks, to reflect the relative degree ofhydration.

To examine the dynamics of polysaccharides, we measured a series of 1D 13Cspectra with different methods for the creation of initial magnetization. 1D 13Cdirect polarization (DP) spectra were measured using a 35 s recycle delay to obtainthe quantitative signals and a 2 s recycle delay to selectively detect the dynamiccomponents. The difference spectrum was obtained by subtracting the 2 s DPspectrum from the 35 s DP spectrum, without scaling. 1D 13C CP spectrum thatpreferentially detects the rigid components was also conducted using 1-ms contacttime. Furthermore, we measured both 13C spin-lattice (T1) relaxation and 1Hrotating-frame spin-lattice relaxation (T1ρ) at 298 K under 10 kHz MAS on aBruker Avance 400-MHz spectrometer. The spin-lock field was 62.5 kHz for themeasurement of 1H-T1ρ. The relaxation data were fit using either a double or singleexponential decay function.

MAS-DNP sample preparation. The stock solution of AMUPol80 was freshlyprepared in the d8-glycerol/D2O/H2O (60/30/10 Vol%) solvent mixture referred asthe DNP matrix, and a final radical concentration of 10 mM. To prepare the DNPsample, 50 µL of the stock solution was added into 5 mg of the 13C,15N-A.fumigatus sample and grinded for 10–15 min to allow the radical solution topenetrate into the cell walls. 3-mg of well-hydrated samples were transferred into a3.2-mm sapphire rotor. A 30-fold enhancement factor of NMR sensitivity withand without microwave irradiation (εon/off) has been achieved. Relatively shortbuildup times of 3–5 s indicate a good mixing of the radicals and biomolecules inthese whole-cell samples. It should be noted that simply mixing the materials withthe DNP matrix failed to give any enhancement regardless of the mixing time.Therefore, grinding the fungi in DNP matrices thoroughly for several minutes isneeded to ensure homogeneous mixing of radicals and biomolecules in thesewhole-cell samples.

MAS-DNP solid-state NMR experiments. All the experiments were performedon a 600 MHz/395 GHz 89 mm-bore MAS-DNP spectrometer equipped with agyrotron microwave source (National High Magnetic Field Laboratory, Tal-lahassee). The cathode currents of the gyrotron were 160 mA. All DNP spectrawere measured using a 3.2 mm probe under 8 kHz MAS frequency. Thetemperature was 103.6 K with the microwave (µW) off and 106 K with theµW on.

To select chitin signals and identify chitin–glucan interactions, 2D 15N–13CTEDOR52 correlation experiments were implemented by a mixing period with 100-ms DARR or 3-s PDSD. Spectral subtraction generates a difference spectrum thatclearly revealed seven intermolecular cross peaks. The total experimental time is 22h. Second, to further improve the spectral resolution, we conducted a 15N,13Cfiltered 2D 13C–13C correlation experiment23, which benefits from the presence oftwo 13C dimensions and provides unambiguous detection of chitin-proximalbiomolecules. It took 4.5 h to measure this experiment on 3 mg 13C, 15N-fungi and25 intermolecular cross peaks have been identified. In addition, a 13C–13C dipolar-INADEQUATE-PDSD spectrum was measured to identify the signals of minorspecies and to further detect long-range correlations. To determine the packing ofdifferent chitin allomorphs, we measured 2D 15N–15N homonuclear correlationspectra using 5 ms and 15 ms PAR mixing46,47. The radiofrequency field strengthsfor PAR were 34 kHz for 15N and 56 kHz for 1H.

Data availability. The data that support the findings of this study are availablefrom the corresponding author upon request.

Received: 1 May 2018 Accepted: 21 June 2018

References1. Brown G. D. et al. Hidden killers: human fungal infections. Sci. Transl. Med. 4,

165rv13 (2012).2. Denning, D. W. Invasive aspergillosis. Clin. Infect. Dis. 26, 781–803 (1998).3. Denning, D. W. Therapeutic outcome in invasive aspergillosis. Clin. Infect.

Dis. 23, 608–615 (1996).4. Latge, J. P. Aspergillus fumigatus and aspergillosis. Clin. Microbiol. Rev. 12,

310–350 (1999).5. Andriole, V. T. Infections with Aspergillus species. Clin. Infect. Dis. 17,

S481–S486 (1993).6. Becksague, C. M. & Jarvis, W. R. Secular trends in the epidemiology of

nosocomial fungal-infections in the United-States, 1980-1990. J. Infect. Dis.167, 1247–1251 (1993).

7. Le, T. et al. A trial of itraconazole or amphotericin B for HIV-associatedtalaromycosis. New. Engl. J. Med 376, 2329–2340 (2017).

8. Ghannoum, M. A. & Rice, L. B. Antifungal agents: mode of action,mechanisms of resistance, and correlation of these mechanisms with bacterialresistance. Clin. Microbiol. Rev. 12, 501–517 (1999).

9. Lin, S. J., Schranz, J. & Teutsch, S. M. Aspergillosis case - fatality rate:systematic review of the literature. Clin. Infect. Dis. 32, 358–366 (2001).

10. Lipke, P. N. & Ovalle, R. Cell wall architecture in yeast: new structure and newchallenges. J. Bacteriol. 180, 3735–3740 (1998).

11. Barreto-Bergter, E. & Figueiredo, R. T. Fungal glycans and the innate immunerecognition. Front. Cell. Infect. Microbiol. 4, 145 (2014).

12. Marco, F., Pfaller, M. A., Messer, S. A. & Jones, R. N. Antifungal activity of anew triazole, voriconazole (UK-109,496), compared with three otherantifungal agents tested against clinical isolates of filamentous fungi. Med.Mycol. 36, 433–436 (1998).

13. Odds, F. C., Brown, A. J. P. & Gow, N. A. R. Antifungal agents: mechanisms ofaction. Trends Microbiol. 11, 272–279 (2003).

14. Perlin, D. S. Current perspectives on echinocandin class drugs. FutureMicrobiol. 6, 441–457 (2011).

15. Bowman, S. M. & Free, S. J. The structure and synthesis of the fungal cell wall.Bioessays 28, 799–808 (2006).

16. Latge, J. P. The cell wall: a carbohydrate armour for the fungal cell. Mol.Microbiol. 66, 279–290 (2007).

17. Kollar, R., Petrakova, E., Ashwell, G., Robbins, P. W. & Cabib, E. Architectureof the yeast-cell wall - the linkage between chitin and beta(1-3)-glucan. J. Biol.Chem. 270, 1170–1178 (1995).

18. Kollar, R. et al. Architecture of the yeast cell wall - beta(1->6)-glucaninterconnects mannoprotein, beta(1-3)-glucan, and chitin. J. Biol. Chem. 272,17762–17775 (1997).

19. Fesel, P. H. & Zuccaro, A. beta-glucan: crucial component of the fungal cellwall and elusive MAMP in plants. Fungal Genet. Biol. 90, 53–60 (2016).

20. Grantham, N. J. et al. An even pattern of xylan substitution is critical forinteraction with cellulose in plant cell walls. Nat. Plants 3, 859–865(2017).

21. Dupree, R. et al. Probing the molecular architecture of arabidopsis thalianasecondary cell walls using two- and three-dimensional 13C solid state nuclearmagnetic resonance spectroscopy. Biochemistry 54, 2335–2345 (2015).

22. Thongsomboon, W. et al. Phosphoethanolamine cellulose: a naturallyproduced chemically modified cellulose. Science 359, 334–338 (2018).

23. Wang, T. et al. Sensitivity-enhanced solid-state NMR detection of expansin’starget in plant cell walls. Proc. Natl. Acad. Sci. USA 110, 16444–16449 (2013).

24. Wang, T., Chen, Y. N., Tabuchi, A., Cosgrove, D. J. & Hong, M. The target ofbeta-Expansin EXPB1 in maize cell walls from binding and solid-state NMRstudies. Plant Physiol. 172, 2107–2119 (2016).

25. Wang, T., Phyo, P. & Hong, M. Multidimensional solid-state NMRspectroscopy of plant cell walls. Solid State Nucl. Magn. Reson. 78, 56–63(2016).

26. Simmons, T. J. et al. Folding of xylan onto cellulose fibrils in plant cell wallsrevealed by solid-state NMR. Nat. Commun. 7, 13902 (2016).

27. Chatterjee, S., Prados-Rosales, R., Itin, B., Casadevall, A. & Stark, R. E. Solid-state NMR reveals the carbon-based molecular architecture of Cryptococcusneoformans fungal eumelanins in the cell wall. J. Biol. Chem. 290,13779–13790 (2015).

28. Zhong, J., Frases, S., Wang, H., Casadevall, A. & Stark, R. E. Following fungalmelanin biosynthesis with solid-state NMR: biopolymer molecular structuresand possible connections to cell-wall polysaccharides. Biochemistry 47,4701–4710 (2008).

29. Ni, Q. Z. et al. High frequency dynamic nuclear polarization. Acc. Chem. Res.46, 1933 (2013).

30. Koers, E. J. et al. NMR-based structural biology enhanced by dynamic nuclearpolarization at high magnetic field. J. Biomol. NMR 60, 157–168 (2014).

31. Rossini, A. J. et al. Dynamic nuclear polarization surface enhanced NMRspectroscopy. Acc. Chem. Res. 46, 1942–1951 (2013).

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

10 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

32. Linden, A. H. et al. Neurotoxin II bound to acetylcholine receptors in nativemembranes studied by dynamic nuclear polarization NMR. J. Am. Chem. Soc.133, 19266–19269 (2011).

33. Takahashi, H., Hediger, S. & De Paepe, G. Matrix-free dynamic nuclearpolarization enables solid-state NMR 13C-13C correlation spectroscopy ofproteins at natural isotopic abundance. Chem. Commun. 49, 9479–9481(2013).

34. Sergeyev, I. V., Itin, B., Rogawski, R., Day, L. A. & McDermott, A. E. Efficientassignment and NMR analysis of an intact virus using sequential side-chaincorrelations and DNP sensitization. Proc. Natl. Acad. Sci. USA 114, 5171–5176(2017).

35. Gupta, R. et al. Dynamic nuclear polarization enhanced MAS NMRspectroscopy for structural analysis of HIV-1 protein assemblies. J. Phys.Chem. B 120, 329–339 (2016).

36. Thurber, K. R. & Tycko, R. Theory for cross effect dynamic nuclearpolarization under magic-angle spinning in solid state nuclear magneticresonance: the importance of level crossings.J. Chem. Phys. 137, 084508(2012).

37. Mentink-Vigier, F., Akbey, U., Oschkinat, H., Vega, S. & Feintuch, A.Theoretical aspects of magic angle spinning - dynamic nuclear polarization. J.Magn. Reson. 258, 102–120 (2015).

38. Hou, G., Yan, S., Trebosc, J., Amoureux, J. P. & Polenova, T. Broadbandhomonuclear correlation spectroscopy driven by combined R2(n)(v)sequences under fast magic angle spinning for NMR structural analysis oforganic and biological solids. J. Magn. Reson. 232, 18–30 (2013).

39. Lu, X., Guo, C., Hou, G. & Polenova, T. Combined zero-quantum and spin-diffusion mixing for efficient homonuclear correlation spectroscopy under fastMAS: broadband recoupling and detection of long-range correlations. J.Biomol. NMR 61, 7–20 (2015).

40. Lesage, A., Auger, C., Caldarelli, S. & Emsley, L. Determination of through-bond carbon-carbon connectivities in solid-state NMR using theINADEQUATE experiment. J. Am. Chem. Soc. 119, 7867–7868 (1997).

41. Cadars, S. et al. The refocused INADEQUATE MAS NMR experiment inmultiple spin-systems: interpreting observed correlation peaks and optimisinglineshapes. J. Magn. Reson. 188, 24–34 (2007).

42. Kasaai, M. R. Determination of the degree of N-acetylation for chitin andchitosan by various NMR spectroscopy techniques: a review. Carbohydr.Polym. 79, 801–810 (2010).

43. Sikorski, P., Hori, R. & Wada, M. Revisit of alpha-chitin crystal structure usinghigh resolution X-ray diffraction data. Biomacromolecules 10, 1100–1105(2009).

44. Yui, T., Taki, N., Sugiyama, J. & Hayashi, S. Exhaustive crystal structure searchand crystal modeling of beta-chitin. Int. J. Biol. Macromol. 40, 336–344 (2007).

45. Kono, H. Two-dimensional magic angle spinning NMR investigation ofnaturally occurring chitins: precise 1H and 13C resonance assignment ofalpha- and beta-chitin. Biopolymers 75, 255–263 (2004).

46. Donovan, K. J., Jain, S. K., Silvers, R., Linse, S. & Griffin, R. G. Proton-assistedrecoupling (PAR) in peptides and proteins. J. Phys. Chem. B 121,10804–10817 (2017).

47. Lewandowski, J. R., De Paepe, G., Eddy, M. T. & Griffin, R. G. 15N-15N protonassisted recoupling in magic angle spinning NMR. J. Am. Chem. Soc. 131,5769–5776 (2009).

48. Wang, T., Yang, H., Kubicki, J. D. & Hong, M. Cellulose structuralpolymorphism in plant primary cell walls investigated by high-field 2D solid-state NMR spectroscopy and density functional theory calculations.Biomacromolecules 17, 2210–2222 (2016).

49. Hall, D. A. et al. Polarization-enhanced NMR spectroscopy of biomolecules infrozen solution. Science 276, 930–932 (1997).

50. Takahashi, H. et al. Rapid natural-abundance 2D 13C-13C correlationspectroscopy using dynamic nuclear polarization enhanced solid-state NMRand matrix-free sample preparation. Angew. Chem. Int. Ed. 51, 11766–11769(2012).

51. Gullion, T., & Schaefer, J. Rotational-echo double-resonance NMR. J. Magn.Reson. 81, 196–200 (1989).

52. Jaroniec, C. P., Filip, C. & Griffin, R. G. 3D TEDOR NMR experiments for thesimultaneous measurement of multiple carbon-nitrogen distances inuniformly 13C,15N-labeled solids. J. Am. Chem. Soc. 124, 10728–10742 (2002).

53. Wang, T., Williams, J. K., Schmidt-Rohr, K. & Hong, M. Relaxation-compensated difference spin diffusion NMR for detecting 13C-13C long-rangecorrelations in proteins and polysaccharides. J. Biomol. NMR 61, 97–107(2015).

54. Wang, T., Jo, H., DeGrado, W. F. & Hong, M. Water distribution, dynamics,and interactions with Alzheimer’s beta-amyloid fibrils investigated by solid-state NMR. J. Am. Chem. Soc. 139, 6242–6252 (2017).

55. White, P. B., Wang, T., Park, Y. B., Cosgrove, D. J. & Hong, M. Water-polysaccharide interactions in the primary cell wall of Arabidopsis thalianafrom polarization transfer solid-state NMR. J. Am. Chem. Soc. 136,10399–10409 (2014).

56. Ader, C. et al. Structural rearrangements of membrane proteins probed bywater-edited solid-state NMR spectroscopy. J. Am. Chem. Soc. 131, 170–176(2009).

57. Rappleye, C. A., Eissenberg, L. G. & Goldman, W. E. Histoplasma capsulatumalpha-(1,3)-glucan blocks innate immune recognition by the beta-glucanreceptor. Proc. Natl. Acad. Sci. USA 104, 1366–1370 (2007).

58. Fontaine, T., Mouyna, I., Hartland, R. P., Paris, S. & Latge, J. P. From thesurface to the inner layer of the fungal cell wall. Biochem. Soc. Trans. 25,194–199 (1997).

59. Aimanianda, V. et al. Surface hydrophobin prevents immune recognition ofairborne fungal spores. Nature 460, 1117–U1179 (2009).

60. Fontaine, T. et al. Molecular organization of the alkali-insoluble fraction ofAspergillus fumigatus cell wall. J. Biol. Chem. 275, 41528–41529 (2000).

61. Ruiz-Herrera J. Fungal Cell Wall: Structure, Synthesis, and Assembly (CRCPress, Boca Raton, FL, 1991).

62. Perez, P. & Ribas, J. C. Cell wall analysis. Methods 33, 245–251 (2004).63. Francois, J. M. A simple method for quantitative determination of

polysaccharides in fungal cell walls. Nat. Protoc. 1, 2995–3000 (2006).64. Lesage, G. & Bussey, H. Cell wall assembly in Saccharomyces cerevisiae.

Microbiol. Mol. Biol. Rev. 70, 317–343 (2006).65. Haido, R. M. et al. Analysis of peptidogalactomannans from the mycelial

surface of Aspergillus fumigatus. Med. Mycol. 36, 313–321 (1998).66. Masuoka, J. Surface glycans of Candida albicans and other pathogenic fungi:

physiological roles, clinical uses, and experimental challenges. Clin. Microbiol.Rev. 17, 281–310 (2004).

67. Cheung, R. C. F., Ng, T. B., Wong, J. H. & Chan, W. Y. Chitosan: an update onpotential biomedical and pharmaceutical applications. Mar. Drugs 13,5156–5186 (2015).

68. Baker, L. G., Specht, C. A. & Lodge, J. K. Cell wall chitosan is necessary forvirulence in the opportunistic pathogen cryptococcus neoformans. Eukaryot.Cell 10, 1264–1268 (2011).

69. Eisenman, H. C. et al. Microstructure of cell wall-associated melanin in thehuman pathogenic fungus cryptococcus neoformans. Biochemistry 44,3683–3693 (2005).

70. Eisenman, H. C. & Casadevall, A. Synthesis and assembly of fungal melanin.Appl. Microbiol. Biot. 93, 931–940 (2012).

71. Heinekamp T. et al. Aspergillus fumigatus melanins: interference with the hostendocytosis pathway and impact on virulence. Front. Microbiol. 3, 440 (2013).

72. Pihet M. et al. Melanin is an essential component for the integrity of the cellwall of Aspergillus fumigatus conidia. BMC Microbiol. 9, 177 (2009).

73. Cosgrove D. J., Jarvis M. C. Comparative structure and biomechanics of plantprimary and secondary cell walls. Front. Plant Sci. 3, 204 (2012).

74. Ciucanu, I. & Kerek, F. A simple and rapid method for the permethylation ofcarbohydrates. Carbohydr. Res. 131, 209–217 (1984).

75. Rienstra, C. M. et al. De novo determination of peptide structure with solid-state magic-angle spinning NMR spectroscopy. Proc. Natl. Acad. Sci. USA 99,10260–10265 (2002).

76. Bennett, A. E. et al. Homonuclear radio frequency-driven recoupling inrotating solids. J. Chem. Phys. 108, 9463–9479 (1998).

77. Baldus, M., Petkova, A. T., Herzfeld, J. & Griffin, R. G. Cross polarization inthe tilted frame: assignment and spectral simplification in heteronuclear spinsystems. Mol. Phys. 95, 1197–1207 (1998).

78. De Paepe G., Lewandowski J. R., Loquet A., Bockmann A., Griffin R. G.Proton assisted recoupling and protein structure determination. J. Chem. Phys.129, 245101(2008).

79. Etzkorn, M. et al. Secondary structure, dynamics, and topology of a seven-helix receptor in native membranes, studied by solid-state NMR spectroscopy.Angew. Chem. Int. Ed. Engl. 46, 459–462 (2007).

80. Sauvee, C. et al. Highly efficient, water-soluble polarizing agents for dynamicnuclear polarization at high frequency. Angew. Chem. Int. Ed. 52,10858–10861 (2013).

AcknowledgementsThis work was supported by Louisiana State University startup funds and LSU Biome-dical Collaborative Research Program. The glycosyl composition and linkage analysiswere supported in part by the Chemical Sciences, Geosciences and Biosciences Division,Office of Basic Energy Sciences, U.S. Department of Energy grant (DE-SC0015662) toCCRC at UGA. P.W. laboratory research was supported by NIH grants AI121451 andAI121460. The National High Magnetic Field Laboratory is supported by National Sci-ence Foundation through DMR-1157490 and the State of Florida. The MAS-DNP systemat NHMFL is funded in part by NIH S10 OD018519 and NSF CHE-1229170. We thankDr. Zhehong Gan and Dr. Ivan Hung for experimental assistance.

Author contributionX.K., A.K., F.M.-V., and T.W. designed and conducted the NMR and MAS-DNPexperiments. X.K., A.K., and A.C. prepared and optimized the DNP samples. A.M. and P.A. conducted the glycosyl composition and linkage analysis. X.K., A.K., M.C.D.W., A.C.,

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0 ARTICLE

NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications 11

P.W., and T.W. analyzed the experimental data. X.K., A.K., P.W., A.M., F.M.-V., andT.W. wrote the manuscript.

Additional informationSupplementary Information accompanies this paper at https://doi.org/10.1038/s41467-018-05199-0.

Competing interests: The authors declare no competing interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,

adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directly fromthe copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2018

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05199-0

12 NATURE COMMUNICATIONS | (2018) 9:2747 | DOI: 10.1038/s41467-018-05199-0 | www.nature.com/naturecommunications

Molecular Architecture of Fungal Cell Walls Revealed

by Solid-State NMR

Kang et al.

Supplementary Figure 1. EI-MS ion fragmentation of partially methylated alditol acetates of 13C

neutral hexoses in fungal cell walls. The proposed glycosyl linkage interpretation of the 13C substituted

(~100%) carbohydrates is detailed.

Supplementary Figure 2. Comparison of A. niger and A. fumigatus cell walls. 2D 13C-13C RFDR spectra

of a, the alkali-insoluble portion of A. niger cell walls, and b, intact cells of A. fumigatus.

Supplementary Figure 3. 13C cross sections of 2D PAR. The 13C cross sections of 2D PAR (black) and

100 ms DARR (yellow) are overlaid for comparison. Dash lines indicate the positions where 13C cross

sections are extracted.

Supplementary Figure 4. DNP Experiments for detecting chitin-glucan interactions. a, 15N-13C

correlation experiment. b, 15N-13C filtered 13C-13C correlation experiment.

Supplementary Figure 5. Parent spectra for generating intermolecular-only 15N-13C correlation

spectrum with improved resolution. a, 15N-13C correlation spectrum with 100 ms mixing times for

intramolecular cross peaks. b, 15N-13C correlation spectrum with 3 s mixing times for both intermolecular

and intramolecular cross peaks. c, Spectral subtraction of b-a results in a difference spectrum with only

long-range intermolecular signals. A Topspin gamma value of -0.65 is used for the subtraction.

Supplementary Figure 6. NMR relaxation curves of polysaccharides. a, 1H-T1ρ and b, 13C-T1 relaxation

curves of polysaccharides in intact A. fumigatus cell walls. The data are collected on a 400 MHz (9.4 Tesla)

spectrometer and best-fits are achieved using single or double exponential equations.

Supplementary Figure 7. NMR analysis of the rigid portion of proteins. a, Comparison of the protein

intensity in 1D 13C CP, DP with 2-s recycle delays and DP with 35-s recycle delays. b, Comparison of

polysaccharide and protein hydration. The normalized intensity is obtained using the intensity ratio between

two water-edited spectra with 4-ms and 49-ms 1H mixing time. The 4-ms spectrum contains signals from

the well hydrated molecules and the 49-ms spectrum reflects the equilibrium condition. The rigid proteins

detected in this CP-based experiment are generally more hydrophobic than the polysaccharides. Error bars

are standard deviations propagated from NMR signal-to-noise ratios.

Supplementary Table 1: Glycosyl composition analysis of the neutral sugars in Aspergillus cell walls.

The intact cell walls of A. fumigatus and the alkali-insoluble portion of A. niger cell walls are reported.

Glycosyl residue of neutral sugars (mole %)

Sample Araa Mana Gala Glca

A. fumigatus 0.3 6.9 14.5 78.3

A. niger nd. 2.9 15.2 81.9

Composition of all sugars (mole %)

Sample Arab Manb Galb Glcb Chitinc Chitosanc

A. fumigatus 0.3 6.2 13.1 70.6 9.0 0.9

a Data on neutral sugars are obtained directly from glycosyl composition analysis.

b Neutral sugar content rescaled by the NMR data of chitin/glucan and chitin/chitosan ratios.

c The amount of nitrogenated sugar in A. fumigatus is calculated using the chitin-to-glucan ratio (1 : 7.8)

obtained by the area of C1-C2 cross peaks in 2D 13C-13C RFDR spectrum (Supplementary Fig. 2) and the

chitin-to-chitosan ratio (9 : 1) obtained using 1D 15N spectra (Fig. 2a).

Supplementary Table 2. 13C and 15N chemical shifts of polysaccharides in A. fumigatus cell walls. Superscripts are used to denote different

allomorphs. Underline denotes the 13C connectivity with ambiguity. Weak signals or minor species are indicated using “w.” Not applicable ( /).

Unidentified (-). Unk: unknown.

C1 C2 C3 C4 C5 C6 CO CH3 N Experimental methods References

β-1,3-glucana 103.6 74.4 86.4 68.7 77.1 61.3 / / / 13C-13C PDSD, 13C CP

J-INADEQUATE Shim et al. 2007

Fairweather et al. 2004

Hazime Saitô et al.

19791-3

β-1,3-glucanb

(w) 104.6 75.2 85.9 69.6 78.6 63.0 / / /

DNP 13C-13C dipolar-

INADEQUATE-PDSD

β-1,3-glucanc

(w) 102.6 72.8 84.4 69.7 72.2 60.5 / / /

DNP 13C CP J-

INADEQUATE

α-1,3-glucana 101.0 71.9 84.6 69.5 71.7 60.5 / / /

13C-13C PDSD, 13C CP

J-INADEQUATE Bhanja et al. 2014 4

Puanglek et al. 2016 5

α-1,3-glucanb 99.9 71.0 80.0 70.4 69.7 60.9 / / /

α-1,3-glucanc 101.2 70.1 84.5 67.7 71.5 60.5 / / /

α-1,3-glucand

(w) 101.5 70.2 86.8 69.3 71.2 62.4 / / /

13C CP J-

INADEQUATE β-1,4-glucan 103.3 69.4 71.7 85.3 74.3 63.4 / / /

β-1,6-glucan 102.6 69.4 79.0 71.6 76.9 67.3 Lowman et al. 2011 6

β-1,4-mannan 101.0 69.4 76.2 83.4 77.1 61.6 / / /

13C DP J-

INADEQUATE

Petkowicz et al. 2001 7

Marchessault et al.

1990 8

arabinana 107.6 81.9 77.1 83.3 69.9 / / / / Renard et al. 19999

arabinanb 108.2 81.7 - - - / / / /

chitina

103.6 55.5 72.9 83.0 75.7 60.9 174.8 22.6 127.9

13C-13C PDSD, 13C CP

J-INADEQUATE, 15N-13C N(CA)CX-DARR Kono et al. 2004 10

Heux et al. 2000 11

Kameda et al. 2004 12

King et al. 2017 13

Tanner 1990 14

103.4 56.0 73.9 - - - - 22.2 127.7 RT 13C CP J-

INADEQUATE, 15N-13C N(CA)CX-DARR

- 55.9 - - - - - 130.8

- 54.7 - - - - - 131.0

- 54.8 - - - - 176.1 129.5

102.0 55.2 - - - - 173.0 129.2

chitinb 102.4 51.0 72.9 81.3 74.7 60.3 177.5/174.6/173.0 24.0 123.6 DNP 15N-13C TEDOR, 15N,13C filtered 2D 13C-

13C PDSD, 15N-13C

N(CA)CX-DARR chitinc 103.2 55.0 73.4 82.2 69.5 - 172.5 22.6 122.6

chitind (w) 103.2 56.3 73.9 86.6 - - - - -

DNP 13C CP J-

INADEQUATE chitine (w) 103.9 57.1 73.3 87.2 - - - - -

chitinf (w) 100.6 52.5 71.0 - - - - - -

unk 102.3 71.0 - - - - - - - 13C CP J-

INADEQUATE

unk 100.9 69.5 - - - - - - -

Supplementary Table 3. The intensity of 65 long-range intermolecular cross peaks. The intensities are

relative ratios of the peak area normalized by the integral of a 13C cross section. For non-DNP experiments,

a peak higher than 2% is categorized as strong restraints (in bold) and 0.8% for intermediate restraints

(underline). For DNP enhanced experiments, given the relatively large values for all cross peaks and the

distribution of peak intensities, the thresholds have been increased to 15% and 8% for strong and

intermediate restraints, respectively. Error bars are standard deviations propagated from NMR sensitivity.

atom 1 atom 2

15 ms CC-

PAR (%)

3s PDSD

(%)

DNP NC-

edited CC

PDSD (%)

DNP NC

with 3s

PDSD (%)

DNP 15

ms NN-

PAR (%)

β-1,3-glucan-

α-1,3-glucan

B5a A3a,c 1.05 9.20

A1a,c B1a/Ch1a 8.11

B5a A4c 7.72

B5a A2a,b 18.54

B5a A6a,b,c 5.57

A6b/Ch6a B6a 0.35

B5a A1a,c 3.91

A1a,c B5a 0.88

A1a,c B6a 0.97

A1a,c B5a 0.39

B1a A2c 15.29

Chitin-

α-1,3-glucan

A5a,c Ch3a,c 2.22

A6bCh6a Ch6c 1.33

Ch3d,e A2c,d 0.15

Ch4a A3a,c 4.63

Ch4a A2c,d 4.62

A1b Ch4b 8.10

A1b Ch6b 3.16

A1a,c Ch4c 0.19

A1a,c Ch2b,c,d 0.36

A3a,c Ch2a,c 0.48

A3a,c Ch1a,c 2.04

A3a,c Ch5a 2.06

Ch2e A3a,c 6.35

A2b ChCH3a,c 4.67

ChNHc A1a,c,d 10.25

ChNHb,c A2b 15.68

ChNHb,c A4b/A2c 9.87

Chitin-

β-1,3-glucan

Ch4a B3b 3.49

Ch3d,e B5c 0.13

Ch2f B6a 0.49

Ch2d,e B6a 11.93

B2a Ch2d 11.54

B2b ChCH3a,c 3.51

B5a ChCH3a,c 2.86

B5a Ch4a 8.92

B5a Ch2d 7.83

Ch4a B6b 7.95

B1b Ch2b 9.53

B1b Ch2d 13.94

ChNHb B4a 10.04

Chitin-Chitin

Ch2d Ch4a 6.50

Ch4a Ch2d 11.47

Ch4a Ch2f 9.85

Ch4a Ch1f 2.41

Chitin-Chitin

Ch4a Ch1e 4.71

ChNHa ChNH

c,d 29.75

ChNHa ChNH

e 24.39

ChNHc ChNH

a 7.88

ChNHa Ch4b 7.85

ChNHc Ch4b 8.34

Chitin-

β-1,6-glucan

ChNHb H4 13.33

H1 Ch1f 1.30

H1 Ch1c,d 0.18

H1 Ch3c 3.65

H1 Ch2f 15.07

H1 Ch2a 12.83

Chitin-

β-1,4-glucan

G6 ChCOa 7.58

Ch4a G4 3.57

β-1,4-glucan-

α-1,3-glucan

A1b G4 1.90

A1a,c G5/B2a 0.21

β-1,6-glucan-

α-1,3-glucan

A1a,c H1 0.93

A1a,c H3 0.98

α-1,3-glucan -

α-1,3-glucan A1b A1a,c 4.92

Supplementary Table 4. The water-edited intensity of polysaccharide cross peaks. The intensity ratios

are obtained by comparing the peak intensity in water-edited and control 2D spectra, with further

normalization by the most hydrated peak, H2-3. Error bars are standard deviations propagated from NMR

signal-to-noise ratios.

Type Cross peaks Intensities Type Cross peaks Intensities

Chitin

Ch1-3a 0.3 ± 0.1

β-1,4-glucan

G1-4 0.6 ± 0.2

Ch1-2a 0.08 ± 0.07 G1-2 0.60 ± 0.08

Ch4-2a 0.3 ± 0.3 G1-6 0.6 ± 0.2

Ch5-2a 0.1 ± 0.3 G5-4 0.6 ± 0.1

Ch3-2a 0.3 ± 0.3 G5-2 0.7 ± 0.1

Ch6-2a 0.02 ± 0.03 G5-6 0.59 ± 0.07

Ch2-1a 0.02 ± 0.03 G2-1 0.58 ± 0.06

Ch2-5a 0.4 ± 0.4 G2-4 0.61 ± 0.08

Ch2-4a 0.1 ± 0.1 G2-5 0.63 ± 0.08

Ch2-3a 0.20 ± 0.1 G1-1/B1-1a 0.57 ± 0.02

Ch2-6a 0.4 ± 0.4 G1-5/B1-2a 0.66 ± 0.03

Ch2-2a 0.37 ± 0.05 G5-1/B2-1a 0.63 ± 0.03

Ch5-3e 0.04 ± 0.01 G5-5 /B2-2a 0.58 ± 0.02

Ch4-3f 0.4 ± 0.1 G2-2/A4-4a/H4-4 0.70 ± 0.01

β-1,3-glucan

B1-3a 0.7 ± 0.1 G2-6 0.7 ± 0.1

B1-5a 0.6 ± 0.1

α-1,3-glucan

A4-1a 0.33 ± 0.02

B1-4a 0.6 ± 0.1 A4-3a 0.26 ± 0.04

B1-6a 0.6 ± 0.1 A4-2a 0.31 ± 0.02

B2-3a 0.7 ± 0.1 A4-6b 0.39 ± 0.06

B2-5a 0.6 ± 0.1 A4-6a 0.24 ± 0.03

B2-4a 0.6 ± 0.1 A6-1a,c 0.23 ± 0.03

B2-6a 0.6 ± 0.1 A6-3a,c 0.20 ± 0.04

B6-1a 0.7 ± 0.2 A6-2b/A6-5c 0.38 ± 0.01

B6-3a 0.8 ± 0.3 A4-6a 0.28 ± 0.02

B6-5a 0.7 ± 0.1 A6-4b 0.25 ± 0.01

B6-2a 0.7 ± 0.1 A6-2/5a 0.39 ± 0.08

B6-4a 0.7 ± 0.1 A5-2b 0.41 ± 0.02

B6-6a/A6-6a 0.6 ± 0.0 β-1,6-glucan

H2-3 1.0 ± 0.3

B4-3a 0.8 ± 0.2 H2-6 0.7 ± 0.3

Supplementary Table 5. The relative intensity of polysaccharide signals in 1D 13C CP and DP

spectra. Error bars are standard deviations propagated from NMR signal-to-noise ratios.

Assignments 13C (ppm) I2s,DP/I35s,DP ICP/I35s,DP

A1a,c 101.1 0.34 ± 0.01 2.06 ± 0.01

A3a,c 84.6 0.25 ± 0.01 2.12 ± 0.02

A4a 69.9 0.38 ± 0.01 1.80 ± 0.01

A6a,c 60.4 0.53 ± 0.01 2.02 ± 0.01

Ch1b 102.4 0.79 ± 0.02 1.48 ± 0.03

Ch4a 83.1 0.57 ± 0.03 1.74 ± 0.07

Ch4b 81.3 0.96 ± 0.05 1.7 ± 0.1

B5a 77.3 0.84 ± 0.02 1.42 ± 0.03

B6a 61.1 0.86 ± 0.03 1.49 ± 0.01

G6 63.7 0.88 ± 0.02 0.46 ± 0.02

G5 74.8 0.93 ± 0.01 1.45 ± 0.05

H3 78.9 0.94 ± 0.04 1.45 ± 0.05

H6 67.5 0.79 ± 0.01 1.00 ± 0.02

Ch4c 82.2 0.85 ± 0.05 1.09 ± 0.06

Ch3c,d,e 73.5 0.88 ± 0.01 0.71 ± 0.01

Ch2e 57.4 0.85 ± 0.04 0.80 ± 0.04

Ch2d 56.9 0.70 ± 0.03 0.95 ± 0.04

Ch2c 54.5 0.95 ± 0.02 0.71 ± 0.02

Ch2f 52.9 0.76 ± 0.03 0.92 ± 0.04

Ch-CH3a,c 22.8 0.85 ± 0.01 1.16 ± 0.02

Supplementary Table 6. 1H-T1ρ relaxation times of polysaccharides in intact A. fumigatus cell walls.

Single and double exponential equations are used to fit the data ( ) 1 ,bt TI t e −

= and

( ) 1 ,a 1 ,bt T t TI t ae be − −

= + , where b=1-a. Error bars are standard deviations of the fit parameters.

Assignment 13C (ppm) a b T1ρ, a(ms) T1ρ, b(ms)

H1 102.8 - 1 - 1.3±0.2

Ch1f 100.6 0.17±0.01 0.83±0.01 0.65±0.04 4.90±0.05

B3a 86.5 - 1 - 1.2±0.1

A3a,c 84.8 0.12±0.02 0.88±0.02 0.7±0.2 5.1±0.1

H3 78.6 - 1 - 0.68±0.04

H5 76.5 - 1 - 0.72±0.04

B2b 75.2 - 1 - 0.92±0.08

Ch3c,d,e 73.7 - 1 - 1.3±0.2

A2b/A5c 71.3 0.29±0.03 0.71±0.02 0.46±0.07 4.7±0.2

A4a/A5b/Ch5c 69.8 0.22±0.03 0.78±0.02 0.42±0.08 4.6±0.2

B6a 61.3 - 1 - 1.6±0.2

A6a,b,c/Ch6a,b 60.7 0.33±0.02 0.67±0.02 0.14±0.02 3.8±0.2

Ch2a,c 54.7 0.14±0.03 0.86±0.02 0.05±0.03 3.2±0.2

Ch2b 51.1 - 1 - 1.7±0.2

Ch-CH3b 24.0 - 1 - 1.9±0.3

Ch-CH3a,c 22.3 - 1 - 2.8±0.2

Supplementary Table 7. 13C-T1 relaxation times of polysaccharides. The data are fit using single and

double exponential equations: ( ) 11 bt TI t e

−= − and ( ) 1 1(1 ) (1 )a bt T t T

I t a e b e− −

= − + − , where a=1-b.

Error bars are standard deviations of the fit parameters.

Assignment 13C (ppm) a b T1a (s) T1b (s)

H1 102.8 - 1 - 1.54±0.15

A1a,c 101.2 - 1 - 2.8±0.9

Ch1f 100.6 - 1 - 3.0±0.9

B3a 86.5 - 1 - 2.5±0.4

A3a,c 84.8 - 1 - 4.3±0.1

H3 78.7 - 1 - 0.4±0.1

Ch5a 76.0 0.26±0.05 0.74±0.05 0.17±0.05 1.3±0.1

B2a/G5 74.3 - 1 - 1.02±0.07

Ch3a,b 72.8 - 1 - 0.95±0.08

A2b/A5c 71.3 - 1 - 1.6±0.1

A4a/A5b/Ch5c 69.8 - 1 - 1.18±0.09

G6 63.2 - 1 - 1.00±0.09

A6a,b,c/Ch6a 60.7 0.38±0.07 0.62±0.07 0.17±0.05 1.6±0.2

Ch2a 55.4 - 1 - 1.14±0.09

Ch2f 52.5 - 1 - 1.0±0.1

Ch-CH3a,c 22.6 - 1 - 1.11±0.09

Supplementary Table 8. The relative intensity of protein signals in 1D 13C CP and DP spectra.

Error bars are standard deviations propagated from NMR signal-to-noise ratios.

13C (ppm) I2s,DP/I35s,DP ICP/I35s,DP

33.7 0.54±0.01 1.58±0.02

31.2 0.92±0.01 1.51±0.01

28.3 0.75±0.01 1.76±0.01

21.6 0.81±0.02 1.42±0.05

20.4 0.76±0.05 2.17±0.01

19.3 0.70±0.01 2.37±0.02

17.6 0.60±0.01 1.85±0.01

16.2 0.72±0.05 1.89±0.02

15.6 0.70±0.02 2.34±0.03

Supplementary Table 9. The water-edited intensity of polysaccharide and protein signals. The ratios

are obtained by comparing the intensity of each 13C peak in 1D water-edited spectra measured with 4-ms

and 49-ms 1H mixing times. Error bars are standard deviations propagated from NMR signal-to-noise ratios.

Type 13C (ppm) intensity Type 13C (ppm) intensity

Polysaccharides

104.0 0.6±0.2

Proteins

32.0 0.3±0.1

101.7 0.31±0.03 31.4 0.3±0.2

99.6 0.9±0.2 30.7 0.3±0.1

87.0 0.58±0.08 30.2 0.32±0.03

85.2 0.3±0.1 29.3 0.2±0.2

83.6 0.50±0.08 28.0 0.28±0.08

81.3 0.73±0.02 26.2 0.32±0.08

79.8 0.71±0.01 25.9 0.3±0.2

78.0 0.68±0.02 23.6 0.4±0.1

74.9 0.59±0.09 20.7 0.21±0.07

74.2 0.56±0.01 19.5 0.19±0.08

72.3 0.41±0.06 19.1 0.20±0.07

70.1 0.49±0.02

68.9 0.6±0.2

68.0 0.77±0.08

67.3 0.90±0.04

63.8 0.6±0.1

61.6 0.5±0.3

Supplementary References

1. Shim J. H., et al. Antitumor effect of soluble beta-1,3-glucan from Agrobacterium sp. R259 KCTC

1019. J. Microbiol. Biotechnol. 17, 1513-1520 (2007).

2. Fairweather J. K., Him J. L. K., Heux L., Driguez H., Bulone V. Structural characterization by 13C-

NMR spectroscopy of products synthesized in vitro by polysaccharide synthases using 13C-enriched

glycosyl donors: application to a UDP-glucose:(1→3)-β-d-glucan synthase from blackberry (Rubus

fruticosus) Glycobiology 14, 775-781 (2009).

3. Saitô H., Ohki T., Sasaki T. A 13C-nuclear magnetic resonance study of polysaccharide gels. Molecular

architecture in the gels consisting of fungal, branched (1 → 3)-β-d-glucans (lentinan and schizophyllan)

as manifested by conformational changes induced by sodium hydroxide. Carbohydr. Res. 74, 227-240

(1979).

4. Bhanja S. K., Rut D., Patra P., Sen I. K., Nandan C. K., Islam S. S. Water-insoluble glucans from the

edible fungus Ramaria botrytis. Bioactive Carbohydrates and Dietary Fibre 3, 52-58 (2014).

5. Puanglek S., et al. In vitro synthesis of linear alpha-1,3-glucan and chemical modification to ester

derivatives exhibiting outstanding thermal properties. Sci. Rep. 6, (2016).

6. Lowman D. W., et al. New Insights into the Structure of (1 -> 3,1 -> 6)-beta-D-Glucan Side Chains in

the Candida glabrata Cell Wall. Plos One 6, (2011).

7. Petkowicz C. L. D., Reicher F., Chanzy H., Taravel F. R., Vuong R. Linear mannan in the endosperm

of Schizolobium amazonicum. Carbohydr. Polym. 44, 107-112 (2001).

8. Marchessault R. H., Taylor M. G. 13C CP/MAS NMR spectra of poly-β-D(1 → 4) mannose: mannan.

Can. J. Chem. 68, 1992-1995 (1990).

9. Renard C. M. G. C., Jarvis M. C. A cross-polarization, magic-angle-spinning, C-13-nuclear-magnetic-

resonance study of polysaccharides in sugar beet cell walls. Plant Physiol. 119, 1315-1322 (1999).

10. Kono H. Two-dimensional magic angle spinning NMR investigation of naturally occurring chitins:

Precise H-1 and C-13 resonance assignment of alpha- and beta-chitin. Biopolymers 75, 255-263 (2004).

11. Heux L., Brugnerotto J., Desbrieres J., Versali M. F., Rinaudo M. Solid state NMR for determination

of degree of acetylation of chitin and chitosan. Biomacromolecules 1, 746-751 (2000).

12. Kameda T., Miyazawa M., Ono H., Yoshida M. Hydrogen bonding structure and stability of alpha-

chitin studied by C-13 solid-state NMR. Macromolecular Bioscience 5, 103-106 (2004).

13. King C., Stein R. S., Shamshina J. L., Rogers R. D. Measuring the Purity of Chitin with a Clean,

Quantitative Solid-State NMR Method. ACS Sustain. Chem. Eng. 5, 8011-8016 (2017).

14. Tanner S. F., Chanzy H., Vincendon M., Roux J. C., Gaill F. High-Resolution Solid-State C-13

Nuclear-Magnetic-Resonance Study of Chitin. Macromolecules 23, 3576-3583 (1990).