BIOPHYSICS

Quantitative mass imaging of singlebiological macromoleculesGavin Young,1* Nikolas Hundt,1* Daniel Cole,1 Adam Fineberg,1 Joanna Andrecka,1

Andrew Tyler,1 Anna Olerinyova,1 Ayla Ansari,1 Erik G. Marklund,2 Miranda P. Collier,1

Shane A. Chandler,1 Olga Tkachenko,1 Joel Allen,3† Max Crispin,3† Neil Billington,4

Yasuharu Takagi,4 James R. Sellers,4 Cédric Eichmann,5 Philipp Selenko,5 Lukas Frey,6

Roland Riek,6,7 Martin R. Galpin,1 Weston B. Struwe,1

Justin L. P. Benesch,1‡ Philipp Kukura1‡

The cellular processes underpinning life are orchestrated by proteins and their interactions.The associated structural and dynamic heterogeneity, despite being key to function, posesa fundamental challenge to existing analytical and structural methodologies.We usedinterferometric scattering microscopy to quantify the mass of single biomolecules insolution with 2% sequence mass accuracy, up to 19-kilodalton resolution, and 1-kilodaltonprecision.We resolved oligomeric distributions at high dynamic range, detectedsmall-molecule binding, and mass-imaged proteins with associated lipids and sugars.These capabilities enabled us to characterize the molecular dynamics of processes as diverseas glycoprotein cross-linking, amyloidogenic protein aggregation, and actin polymerization.Interferometric scattering mass spectrometry allows spatiotemporally resolvedmeasurement of a broad range of biomolecular interactions, one molecule at a time.

Biomolecular interactions and assembliesare central to a wide range of physiolog-ical and pathological processes spanninglength scales from small complexes (1) tothe mesoscale (2, 3). Despite considerable

developments in techniques capable of providinghigh-resolution structural information (4), suchtechniques are typically static, involve averagingover many molecules in the sample, and there-fore often do not fully capture the diversity ofstructures and interactions. Solution-based en-semblemethods enable dynamic studies but lackthe resolution of separation required to distin-guish different species (5–7). Single-moleculemethods offer a means to circumvent heteroge-neity in both structure and dynamics, and prog-ress has been made in terms of characterizinginteractions (8) and mechanisms (9, 10). So far,however, no single-molecule approach has beencapable of quantifying and following the diversity

of interactions of biomolecules with the requiredspatiotemporal accuracy and resolution.Given sufficient sensitivity, light scattering is

an ideal means for detecting and characterizingmolecules in low-scattering in vitro conditionsbecause of its universal applicability. In an in-terferometric detection scheme (Fig. 1A), thescattering signal scales with the polarizability,

which is a function of the refractive index andproportional to the particle volume (11). Combin-ing the approximation that single amino acidseffectively behave like individual nano-objectswith the observation that the specific volumesof amino acids and refractive indices of proteinsvary by only ~1% (fig. S1 and table S1) suggeststhat the number of amino acids in a polypeptide,and thus its mass, are proportional to its scat-tering signal. This close relationship betweenmass and interferometric contrast, which hasbeen predicted (12, 13) and observed (14, 15) tohold coarsely even at the single-molecule level,could thus in principle be used to achieve highmass accuracy.Building on recent advances in the experi-

mental approach (fig. S2) that improved imagingcontrasts for interferometric scattering micros-copy (15, 16), we were able to obtain high-qualityimages of single proteins as they diffused fromsolution to bindnonspecifically to themicroscopecoverslip/solution interface (Fig. 1, B and C, andmovie S1). Reaching signal-to-noise ratios >10,even for small proteins such as bovine serumalbumin (BSA), together with an optimized dataanalysis approach (16), allowed us to extract thescattering contrast for each molecular bindingevent with high precision (Fig. 1D and fig. S3).These data revealed clear signatures of differentoligomeric states of BSA,with relative abundancesfor monomer to tetramer of 88.63, 9.94, 1.18, and0.25% of the detected particles. For nonspecificbinding to an unfunctionalized microscopecoverslip, surface attachment was effectivelyirreversible (12,209 binding versus 372 unbind-ing events). As a result, we could determine

RESEARCH

Young et al., Science 360, 423–427 (2018) 27 April 2018 1 of 5

1Physical and Theoretical Chemistry Laboratory, Departmentof Chemistry, University of Oxford, South Parks Road,Oxford OX1 3QZ, UK. 2Department of Chemistry BiomedicinsktCentrum, Uppsala University, Box 576, 75123 Uppsala, Sweden.3Oxford Glycobiology Institute, Department of Biochemistry,University of Oxford, Oxford OX1 3QU, UK. 4Cell Biology andPhysiology Center, National Heart, Lung and Blood Institute(NHLBI), Bethesda, MD 20892, USA. 5In-Cell NMR Laboratory,Department of NMR-supported Structural Biology, LeibnizInstitute of Molecular Pharmacology (FMP Berlin),Robert-Rössle Straße 10, 13125 Berlin, Germany. 6Laboratoryof Physical Chemistry, Department of Chemistry and AppliedBiosciences, ETH Zürich, 8093 Zürich, Switzerland.7Department of Immunology and Microbiology, The ScrippsResearch Institute, 10550 North Torrey Pines Road, La Jolla,CA 92037, USA.*These authors contributed equally to this work.†Present address: Centre for Biological Sciences and Institute for LifeSciences, University of Southampton, Southampton SO17 1BJ, UK.‡Corresponding author. Email: justin.benesch@chem.ox.ac.uk(J.L.P.B.); philipp.kukura@chem.ox.ac.uk (P.K.)

Fig. 1. Concept of interfero-metric scattering massspectrometry (iSCAMS).(A) Schematic of the experi-mental approach relyingon immobilization of individualmolecules near a refractive indexinterface. Oligomeric statesare colored differently for clarity.(B) Differential interferometricscattering image of BSA. Scalebar, 0.5 mm. (C) Representativeimages of monomers (top left),dimers (bottom left), trimers(top right), and tetramers(bottom right) of BSA. Scalebars, 200 nm. (D) Scatter plotof single-molecule bindingevents and their scatteringcontrasts for 12 nM BSAfrom 14 movies (lower panel).Corresponding histogram(n = 12,209) with a zoomed-inview of the region for largerspecies (upper panel). The reduc-tion in landing rate results froma drop in BSA concentration withtime owing to the large surface-to-volume ratio of our samplecell (supplementary materials).

150

200

100

50

0

0

10

20

30

40

0

1000

2000

3000

0.5 1.0Contrast /10-2

Tim

e /s

Inci

denc

es

1.5 2.0

B

A

C

D

Contrast /10-2-0.8 0.3

Contrast /10-2-1.6 0.8

Incident light Scattered light Reflected light

on May 3, 2020

http://science.sciencem

ag.org/D

ownloaded from

(bulk) binding rate constants, which generallyexhibited only small variations with oligomericstate. These could be accommodated to obtainminor corrections to the recorded mass spectraand yield the solution distribution (fig. S4). Ourresults, including the detection and quantifica-tion of rare complexes such as BSA tetramers,demonstrate the ability of interferometric scat-tering mass spectrometry (iSCAMS) to char-acterize solution distributions of oligomericspecies and molecular complexes at high dy-namic range.The regular spacing in the contrast histogram

of BSA tentatively confirms the expected linearscaling between mass and interferometric con-trast. Repeating these measurements for eightdifferent proteins, spanning 53 to 803 kDa, vali-dates the linear relationship (Fig. 2A and fig. S5A).The deviation between measured and sequencemass was <5 kDa, resulting in an average error of1.9%, and this deviation showed no detectablecorrelation with refractivity in relation to theoverall shape of themolecule (fig. S6A). Even forlarge structural differences, such as those be-tween the extended and folded conformationsof smooth-muscle myosin (530.6 kDa; Fig. 2A

and figs. S5B and S7), we did not find measur-able differences in the apparent molecular massbeyond the increase expected for the additionof glutaraldehyde molecules used to cross-linkmyosin into the folded conformation (extended,528.4 ± 16.2 kDa; folded, 579.4 ± 14.8 kDa; fig. S5B).The resolution, as defined by the full width athalf-maximumof themeasured contrast, reached19 kDa for streptavidin. In all cases, the resolutionwas limited by photon shot noise and influencedby molecular mass, increasing from 19 kDa forstreptavidin to 102 kDa for thyroglobulin (fig. S6,B and C). The <0.5% deviation from sequencemass for species of >100 kDa compareswell withnativemass spectrometry (17) and demonstratesthe intrinsic utility of iSCAMS for the accuratemass measurement of biomolecules with oligo-meric resolution.Moving beyond species composed solely of

amino acids, lipid nanodiscs are an ideal systemfor testing the broad applicability of iSCAMSowing to their flexibility in terms of polypeptideand lipid content (18). For nanodiscs composedof the MSP1D1 belt protein and DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine) lipids,we obtained a mass of 141.0 ± 1.6 kDa, in good

agreement with the range of masses spanning124 to 158 kDa reported from other methods(Fig. 2B and fig. S5D). Replacing MSP1D1 withthe smaller MSP1DH5 reduced the nanodisc di-ameter and the lipid content by ~20%, after ac-counting for the thickness of the protein belt (19).Given the masses of MSP1D1 and MSP1DH5 (47and 42 kDa, respectively), we predicted a massfor the MSP1DH5 nanodisc of 113.6 kDa, in ex-cellent agreementwith ourmeasurement (114.1 ±1.9 kDa). Mass shifts associated with changesin lipid composition, such as those introducedby partially unsaturated lipids and cholesterol,matched those predicted from the assemblyratios (Fig. 2B and tables S2 to S6).To see whether our approach also applies to

solvent-exposed moieties that experience a dif-ferent dielectric environment from those buriedwithin a protein, we selected the HIV envelopeglycoprotein complex (Env), which is a trimerof gp41-gp120 heterodimers. Env is extensivelyN-glycosylated,with the carbohydrates contribut-ing almost half of itsmass (20). For an Env trimermimic expressed in the presence of kifunensine,amannosidase inhibitor that leads predominant-ly to unprocessed Man9GlcNAc2 glycans (Man,

Young et al., Science 360, 423–427 (2018) 27 April 2018 2 of 5

2

0

4

6

0

4

-4

A

D

B

C

0.95

0.55

0.50

0.45

0.40

2.1

1.9

2.3 0 84 72 60 48 362412

0 84726048362412Number of glycans

Env glycosylation

Streptavidin-biotin

Lipid nanodiscs

0.90

0.85

0.80

2.5

-8

+ Kifunensine

Con

tras

t /10

-2

Con

tras

t /10

-2

Mas

s er

ror

/10-2

0 200 400Sequence mass /kDa Mass /kDa

600 800

120110100 130 140 150 160 170

280240 320 6050 70360

SEC

DLS

NMRMS

- Kifunensine

MSP1D1DMPC

MSP1D1DMPC/PC14:1/Chol

MSP1D1PC14:1/Chol

MSP1ΔH5DMPC

+ Biotin•DSG3

+ Biotin + Biotin•ACTH

Exp

ecte

d co

ntra

st c

hang

e

Exp

ecte

d co

ntra

st c

hang

e

Fig. 2. Characterization of iSCAMS accuracy, precision, anddependence on molecular shape and identity. (A) Contrast versusmolecular mass, including for proteins used for mass calibration (black),characterization of shape dependence (yellow), protein-ligand binding(green), lipid nanodisc composition (red), and glycosylation (blue). Masserror (upper panel) is given as a percentage of the sequence mass relativeto the given linear fit. (B) Nanodisc mass measurement for differentlipid compositions and protein belts. Masses obtained by alternativemethodologies for MSP1D1/DMPC are marked and extrapolated to theother compositions. The horizontal bars indicate the expected mass rangeas a function of characterization technique, with the thin bars indicating

the contrast measured and the thick bars representing the measurementuncertainty in terms of the standard error of the mean (SEM) for repeatedexperiments. For each sample, the upper text denotes the membranescaffold protein (MSP) used, and the lower text indicates the lipids in thenanodisc. (C) Recorded differential contrast for Env expressed in thepresence or absence of kifunensine and associated mass ranges expectedfor different glycosylation levels. (D) Mass-sensitive detection of ligandbinding in the biotin-streptavidin system, according to the sequence massof streptavidin and the masses of biotin and two biotinylated peptidesrelative to the calibration obtained from (A). Abbreviations are defined intable S8. In (A) and (D), error bars represent SEM.

RESEARCH | REPORTon M

ay 3, 2020

http://science.sciencemag.org/

Dow

nloaded from

mannose; GlcNAc,N-acetylglucosamine) (fig. S8),we recorded a mass of 350.0 ± 5.7 kDa. Makingthe crude approximation that glycans and aminoacids have similar polarizabilities, this corre-sponds to a glycan occupancy of 74 ± 3 out of 84possible sites (Fig. 2C and fig. S5E), consistentwith recent observations of high occupancy forgp120 expressed with kifunensine (21). For Envexpressed without kifunensine, we recorded alower mass of 315.3 ± 10.5 kDa. The mass differ-ence can be attributed only in part to the loweraverage mass of the processed glycans (fig. S8)and yields a totalN-glycan occupancy of 61 ± 6.Although the exact values for occupancy are be-holden to our calibration (Fig. 2A), the presenceof unoccupied sites is consistent with their ob-servation in proteomics data (22).The high precision of 1.8 ± 0.5% with respect

to the protein mass (Fig. 2A) indicates the po-tential for direct detection of small-moleculebinding. To probe the current limits of iSCAMSin terms of precision, we therefore examined thebiotin-streptavidin system (Fig. 2D and fig. S5C).We measured masses for streptavidin in the ab-sence (55.7 ± 1.1 kDa) andpresence (57.4±0.9 kDa)of biotin, finding a difference of 1.7 ± 1.4 kDa, ingood agreement with the expected 0.98 kDa forcomplete occupancy of the four binding sites.Upon addition of two different biotinylated pep-tides (3705.9 and 4767.4 Da), we found increasesof 16.1 ± 2.8 and 22.0 ± 2.2 kDa (compared withthe expected 14.8 and 19.1 kDa) (Fig. 2D and fig.S5C). These data show that iSCAMS can detectthe association of kilodalton-sized ligands, dem-onstrating its suitability for sensitive ligand-binding studies in solution.After having established the capabilities of

iSCAMS, we sought to test it on more complex

systems that are difficult to assess quantitativelywith existing techniques as a consequence of het-erogeneity andmultistep assemblymechanisms(Fig. 3). In addition, we aimed tomonitor nuclea-tion and polymerization dynamics of mesoscopicstructures down to the single-molecule level,which is challenging because of the simulta-neous requirement for high dynamic range, highimaging speed, and direct correlation betweenthe observed signals and the associated molec-ular events. The biotin-streptavidin system exhib-its nearly covalent binding, raising the questionof whether iSCAMS is capable not only of de-termining mass distributions, but also of quanti-fying weaker equilibria, as are often encounteredfor protein-protein interactions.We therefore investigated the interaction of

Env with the antiviral lectin BanLec, whichneutralizes HIV by binding to surfaceN-glycans(23, 24) through an unknown mechanism. Wewere able tomonitor the interactions and short-lived complexes before aggregation, with the ad-dition of BanLec to Env resulting in a reductionof single Env units coupled to the appearanceof dimers and higher-order assemblies (Fig. 3A).Describing the experimental oligomeric evolu-tion with a simple model (Fig. 3B) enabled usto extract the underlying association constants[KBanLec = 0.12 nM−1; KEnv = 8 nM−1; K′BanLec =0.4 nM−1 (as defined in Fig. 3B)], in good agree-mentwith recent bulk studies (KBanLec =0.19nM

−1),which also found signatures of a secondary bind-ing event (2.85 nM−1) (25). Our ability to followandmodel the evolution of different oligomericspecies allowed us to extract the interactionmechanism and the energetics underlying thelectin-glycoprotein interaction, despite the het-erogeneity of this multicomponent system. As

a result, we could show that binding of Env toBanLec that is already bound to Env (KEnv) ismuch stronger than to free BanLec (KBanLec), akey characteristic of cooperative behavior.More-over, themass resolution of our approach enabledus to quantify the number of BanLecs bound perdimer (one to two), trimer (two to three), andtetramer (three to four) of Env, demonstratingbivalent binding. These results are directly rel-evant to the characterization and optimization ofantiretrovirals, given that multivalency and ag-gregation have been proposed to be linked toneutralization potency (25). We anticipate sim-ilar quantitative insights to be achievable forother therapeutic target proteins and protein-protein interactions in general.An advantage of our imaging-based approach

is its ability to time-resolve mass changes in aposition- and local concentration–sensitiveman-ner. This enables us to examine surface-catalyzednucleation events that may eventually lead toamyloid formation (26). Previous studies usingfluorescence labeling found aggregates of ~0.6 mmin diameter within a minute of addition of theamyloidogenic protein a-synuclein at 10 mMto anappropriately charged bilayer (27). Upon addinga-synuclein to a planar, negatively chargedDOPC(1,2-dioleoyl-sn-glycero-3-phosphocholine)/DOPS(1,2-dioleoyl-sn-glycero-3-phospho-L-serine) (3:1)membrane at physiological pH, we observed theappearance and growth of nanoscopic objectswithin seconds, even at low micromolar concen-trations (Fig. 4A and movie S2). We were unableto determine the sizes of the initial nucleatingspecies or individual assembly steps owing tothe low molecular mass of a-synuclein (14 kDa),but we couldmonitor the nanoscale formation ofassociated structures in the range of hundreds ofkilodaltons and determine the kinetics (Fig. 4B).Growth of these clusters was uniform across thefield of view, with the initial rates following ex-pectations for a first-order process (Fig. 4B andfig. S9A), pointing to a simple growthmechanism.We did not detect such structures on neutral,DOPC-only bilayers, and we found evidence forthioflavin T–positive aggregates after overnightincubation (fig. S9B), suggesting that our assayprobes early stages of amyloid assembly.At the extremes of its current sensitivity,

iSCAMS enables mass imaging of mesoscopicself-assembly, molecule bymolecule. In an actinpolymerization assay, subtraction of the con-stant background revealed the growth of surface-immobilized filaments. In contrast toa-synuclein,where the growth of interest took place withina diffraction-limited spot, in this case we couldquantify length changes of filaments larger thanthe diffraction limit upon the attachment anddetachment of actin subunits (Fig. 4C, fig. S10C,and movie S3). We observed distinct, stepwisechanges in the filament length (Fig. 4D; fig. S10,D to F; andmovie S4), themost frequent forwardand backward step sizes in the traces being 3.0 ±0.8 and 2.7 ± 0.7 nm, respectively—very close tothe expected length increase of 2.7 nm uponbinding of a single actin subunit to a filament(Fig. 4E). Detection of larger step sizes represents

Young et al., Science 360, 423–427 (2018) 27 April 2018 3 of 5

21 3 4

Are

a-no

rmal

ized

pro

babi

lity

dens

ity

Mass /kDa1000 1500500

1.00

0.50

0.75

0.25

0

Mol

e fr

actio

n

[BanLec] /nM

A B KBanLec KEnv KBanLecEnv units

BanLec units0 21 0 2 31 0 2 3 41 1 3 4 52

0.5

5

10

20

40[Ban

Lec]

/nM

’

0.1 1 10

Fig. 3. Single-molecule mass analysis of heterogeneous protein assembly. (A) Mass distributionsfor Env in the presence of 0.5 to 40 nM BanLec monomer, alongside expected positions formultiples of bound BanLec tetramers. Inset, a zoomed-in view of the region for larger species.(B) Oligomeric fractions colored according to (A) versus BanLec concentration, including predictions(curves) using the model shown.

RESEARCH | REPORTon M

ay 3, 2020

http://science.sciencemag.org/

Dow

nloaded from

the addition of multiple actin subunits withinour detection timewindow. The contrast changesassociated with the different step sizes corre-sponded to mass changes of one, two, or threeactinmonomers binding to (andunbinding from)the tip of the growing filaments during acquisi-tion (fig. S10, G and H). Even though we cannotyet distinguish between models invoking mono-mer (28) or oligomer (29) addition to a growingfilament at the current level of spatiotemporalresolution, these results demonstrate the capa-bility of iSCAMS to quantitatively image meso-scopic dynamics and determine how they areinfluenced by associated proteins at the single-molecule level.We anticipate that combining iSCAMS with

established surface modifications (30) will dra-matically expand its capabilities. Passivation de-creases surface binding probabilities and therebyshould provide access to much higher analyteconcentrations (greater than micromolar), andsurface activationwill reducemeasurement timesat low concentrations (less than nanomolar).

Specific functionalization and immobilizationof individual subunits or binding partners couldallow for the determination of on and off rates,in addition to equilibrium constants, and en-able targeted detection in the presence of otheranalytes (14). Although studies within complexthree-dimensional environments such as the cellmay prove to be beyond reach, the advances re-ported here will make iSCAMS a key approachfor dynamic in vitro studies of biomolecular in-teractions, assembly, and structure at the single-molecule level.

REFERENCES AND NOTES

1. S. E. Ahnert, J. A. Marsh, H. Hernández, C. V. Robinson,S. A. Teichmann, Science 350, aaa2245 (2015).

2. K. Rottner, J. Faix, S. Bogdan, S. Linder, E. Kerkhoff, J. Cell Sci.130, 3427–3435 (2017).

3. M. Hemmat, B. T. Castle, D. J. Odde, Curr. Opin. Cell Biol. 50,8–13 (2018).

4. H.-W. Wang, J.-W. Wang, Protein Sci. 26, 32–39 (2017).5. D. M. Kanno, M. Levitus, J. Phys. Chem. B 118, 12404–12415

(2014).6. M. Chakraborty et al., Biophys. J. 103, 949–958 (2012).

7. S. I. A. Cohen, M. Vendruscolo, C. M. Dobson, T. P. J. Knowles,J. Mol. Biol. 421, 160–171 (2012).

8. A. Jain et al., Nature 473, 484–488 (2011).9. B. Schuler, W. A. Eaton, Curr. Opin. Struct. Biol. 18, 16–26 (2008).10. W. J. Greenleaf, M. T. Woodside, S. M. Block, Annu. Rev.

Biophys. Biomol. Struct. 36, 171–190 (2007).11. C. F. Bohren, D. R. Huffman, Absorption and Scattering of Light

by Small Particles (Wiley Interscience, 1983).12. P. Kukura et al., Nat. Methods 6, 923–927 (2009).13. J. Ortega Arroyo et al., Nano Lett. 14, 2065–2070 (2014).14. M. Piliarik, V. Sandoghdar, Nat. Commun. 5, 4495 (2014).15. M. Liebel, J. T. Hugall, N. F. van Hulst, Nano Lett. 17, 1277–1281

(2017).16. D. Cole, G. Young, A. Weigel, A. Sebesta, P. Kukura, ACS

Photonics 4, 211–216 (2017).17. M. van de Waterbeemd et al., Nat. Methods 14, 283–286

(2017).18. I. G. Denisov, S. G. Sligar, Chem. Rev. 117, 4669–4713

(2017).19. S. Bibow et al., Nat. Struct. Mol. Biol. 24, 187–193 (2017).20. L. A. Lasky et al., Science 233, 209–212 (1986).21. W. B. Struwe, A. Stuckmann, A.-J. Behrens, K. Pagel, M. Crispin,

ACS Chem. Biol. 12, 357–361 (2017).22. L. Cao et al., Nat. Commun. 8, 14954 (2017).23. M. D. Swanson, H. C. Winter, I. J. Goldstein, D. M. Markovitz,

J. Biol. Chem. 285, 8646–8655 (2010).24. J. T. S. Hopper et al., Structure 25, 773–782.e5 (2017).

Young et al., Science 360, 423–427 (2018) 27 April 2018 4 of 5

50 100 1500-50-100Mass increment /kDa

0

5 100-5Step size /nm

100

150

50

200

Inci

denc

es

100

10

1

Agg

rega

tion

rate

/MD

a·s-1

1 10[α-synuclein] /µM

Actin filament growth

Con

tras

t /10

-2

-6

3

Con

tras

t /10

-2

-12

8

α-synuclein aggregation

0 s 5 s 10 s0 s 1 s 2 s

A

B

C

D

E

0 0.5 1.0 1.50

5

10

15

Con

tras

t /10

-2

Time /s

1 s

20 n

m

Fig. 4. Mass imaging of mesoscopic dynamics. (A) Schematic andiSCAMS images of a-synuclein (1 mM) aggregation on a negatively chargedbilayer membrane. (B) Initial growth rate versus a-synuclein concentration,shown with the best fit assuming first-order kinetics. Error bars denoteSEM for different particles. Inset, individual (gray) and average (black)growth trajectories for 21 particles from (A). (C) Schematic and iSCAMSimages of actin polymerization. The arrow highlights a growing filament.

(D) Representative traces of actin filament tip position (gray) andcorresponding detected steps (black). (E) Step and mass histogram from1523 steps and 33 filaments, including a fit to a Gaussian mixture model(black) and individual contributions (colored according to fig. S10G). Scalebars, 1 mm. In these experiments, background correction involved removalof the static background before acquisition, rather than continuous differentialimaging as in Figs. 2 and 3 (supplementary materials).

RESEARCH | REPORTon M

ay 3, 2020

http://science.sciencemag.org/

Dow

nloaded from

25. S. Lusvarghi et al., ACS Infect. Dis. 2, 882–891 (2016).26. C. Galvagnion et al., Nat. Chem. Biol. 11, 229–234 (2015).27. A. Iyer, N. Schilderink, M. M. A. E. Claessens, V. Subramaniam,

Biophys. J. 111, 2440–2449 (2016).28. M. Kasai, S. Asakura, F. Oosawa, Biochim. Biophys. Acta 57,

22–31 (1962).29. H. P. Erickson, J. Mol. Biol. 206, 465–474 (1989).30. B. Hua et al., Nat. Methods 11, 1233–1236 (2014).

ACKNOWLEDGMENTS

J.R.S. thanks F. Zhang for technical assistance and the NHLBIelectron microscopy core facility. Funding: G.Y. was supported bya Zvi and Ofra Meitar Magdalen Graduate Scholarship. N.H. wassupported by a DFG (German Research Foundation) researchfellowship (HU 2462/1-1). E.G.M. thanks the Swedish ResearchCouncil and the European Commission for a Marie Skłodowska CurieInternational Career Grant (2015-00559). M.P.C. is a ClarendonScholar supported by the Oxford University Press. S.A.C. issupported by the Biotechnology and Biological Sciences ResearchCouncil and Waters Corp through the iCASE studentshipBB/L017067/1 to J.L.P.B. O.T. acknowledges a Lamb and Flag

Scholarship from St John’s College, University of Oxford, andan Engineering and Physical Sciences Research Council (EPSRC)Studentship. J.Al. and M.C. were supported by the NationalInstitute of Allergy and Infectious Diseases (Center for HIV/AIDSVaccine Immunology and Immunogen Discovery grantUM1AI100663). J.R.S. was supported by NHLBI intramuralprogram HL0001786. C.E. is supported by a Swiss NationalScience Foundation advanced postdoctoral mobility fellowship(P300PA160979). P.S. is funded by a European Research Council(ERC) Consolidator Grant (NeuroInCellNMR, 647474). J.L.P.B.thanks the EPSRC for EP/J01835X/1. P.K. was supported by anERC Starting Investigator Grant (Nanoscope, 337577). Authorcontributions: Conceptualization: W.B.S., J.L.P.B., and P.K.Methodology: G.Y., N.H., D.C., J.An., E.G.M., C.E., P.S., M.R.G.,W.B.S., J.L.P.B., and P.K. Software: G.Y. and N.H. Validation: G.Y.,N.H., J.L.P.B., and P.K. Formal analysis: G.Y., N.H., A.T., A.A.,A.O., J.An., E.G.M., and M.R.G. Investigation: G.Y., N.H., D.C., A.F.,J.An., A.T., A.A., N.B., Y.T., and C.E. Resources: M.P.C., S.A.C., O.T.,J.Al., M.C., N.B., Y.T., J.R.S., C.E., P.S., L.F., R.R., and W.B.S. Writingof original draft: G.Y., J.L.P.B., and P.K. Revision and editing:G.Y., N.H., A.F., A.O., E.G.M., M.P.C., S.A.C., O.T., M.C., J.R.S., C.E.,

P.S., R.R., M.R.G., W.B.S., J.L.P.B., and P.K. Visualization: G.Y.,N.H., J.L.P.B., and P.K. Supervision: P.K. Competing interests:P.K. has filed a patent for the contrast enhancement methodologyand its application to mass measurement of single biomolecules.All other authors declare no competing interests. Data andmaterials availability: All data necessary to support theconclusions are available in the manuscript or supplementarymaterials and are deposited in the University of Oxford ResearchArchive (DOI, 10.5287/bodleian:PmA5Va0a2).

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/360/6387/423/suppl/DC1Materials and MethodsFigs. S1 to S10Tables S1 to S8References (31–51)

Movies S1 to S4

26 November 2017; accepted 26 March 201810.1126/science.aar5839

Young et al., Science 360, 423–427 (2018) 27 April 2018 5 of 5

RESEARCH | REPORTon M

ay 3, 2020

http://science.sciencemag.org/

Dow

nloaded from

Quantitative mass imaging of single biological macromolecules

Justin L. P. Benesch and Philipp KukuraTakagi, James R. Sellers, Cédric Eichmann, Philipp Selenko, Lukas Frey, Roland Riek, Martin R. Galpin, Weston B. Struwe,Erik G. Marklund, Miranda P. Collier, Shane A. Chandler, Olga Tkachenko, Joel Allen, Max Crispin, Neil Billington, Yasuharu Gavin Young, Nikolas Hundt, Daniel Cole, Adam Fineberg, Joanna Andrecka, Andrew Tyler, Anna Olerinyova, Ayla Ansari,

DOI: 10.1126/science.aar5839 (6387), 423-427.360Science 

, this issue p. 423; see also p. 378Scienceassembly.determined in vitro kinetics of fibril and aggregate growth and association constants for a complex protein-glycoprotein biophysical parameters from single molecules and assemblies without labeling. Using this approach, the authorsPerspective by Lee and Klenerman). Movies of protein complex association and dissociation were analyzed to extract

used a light-scattering approach for accurate mass determination of proteins as small as 20 kDa (see theet al.Young Careful measurements of light scattering can provide information on individual macromolecules and complexes.

Watching proteins' weight

ARTICLE TOOLS http://science.sciencemag.org/content/360/6387/423

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/04/25/360.6387.423.DC1

CONTENTRELATED http://science.sciencemag.org/content/sci/360/6387/378.full

REFERENCES

http://science.sciencemag.org/content/360/6387/423#BIBLThis article cites 50 articles, 9 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on May 3, 2020

http://science.sciencem

ag.org/D

ownloaded from