Optimized two-color super resolution imaging of Drp1during mitochondrial fission with a slow-switchingDronpa variantAlyssa B. Rosenblooma,b,c,1, Sang-Hyuk Leeb,c,d,1, Milton Toa, Antony Leec,e, Jae Yen Shinb,c,d,and Carlos Bustamantea,b,c,d,e,2

aDepartment of Molecular and Cell Biology, bCalifornia Institute for Quantitative Biosciences, cJason L. Choy Laboratory of Single Molecule Biophysics,dHoward Hughes Medical Institute, and eDepartment of Physics, University of California, Berkeley, CA 94720

Edited by Jennifer Lippincott-Schwartz, National Institutes of Health, Bethesda, MD, and approved July 24, 2014 (received for review October 24, 2013)

We studied the single-molecule photo-switching properties ofDronpa, a green photo-switchable fluorescent protein and a pop-ular marker for photoactivated localization microscopy. We foundthe excitation light photoactivates as well as deactivates Dronpasingle molecules, hindering temporal separation and limiting superresolution. To resolve this limitation, we have developed a slow-switching Dronpa variant, rsKame, featuring a V157L amino acidsubstitution proximal to the chromophore. The increased sterichindrance generated by the substitution reduced the excitationlight-induced photoactivation from the dark to fluorescent state.To demonstrate applicability, we paired rsKame with PAmCherry1in a two-color photoactivated localizationmicroscopy imagingmethodtoobserve the innerandoutermitochondrialmembranestructures andselectively labeled dynamin related protein 1 (Drp1), responsible formembrane scission during mitochondrial fission. We determined thediameter and length of Drp1 helical rings encircling mitochondria dur-ing fission and showed that,whereas their lengths alongmitochondriawere not significantly changed, their diameters decreased signifi-cantly. These results suggest support for the twistase model ofDrp1 constriction,with potential loss of subunits at the helical ends.

photo-physics | PALM | suborganelle structures

Photoactivated localization microscopy (PALM) allows forsubdiffraction optical resolution via the stochastic temporal

separation of individual photoactivatable or reversibly photo-switchable fluorescent proteins (PA-FPs or PS-FPs) and their sub-sequent localization in space. Successful PALM imaging requirestwo conditions: (i) temporally separated stochastic activation ofa population of PA-FPs or PS-FPs, and (ii) accurate localization ofeach molecule in space (1, 2).For PALM imaging to be more biologically applicable, robust

two-color PALM is necessary. Several published protocols fortwo-color PALM exist, one featuring EosFP and Dronpa, a GFP-like photo-switchable protein (3–5). Initially dark, upon photo-activation by 405-nm light, Dronpa becomes fluorescent with anexcitation maximum at 503 nm and an emission maximum at 515nm. When excited by 488 nm, activated Dronpa emits greenfluorescence (“ON” state) until it is photo-induced back into thedark state (“OFF” state) and can be reactivated multiple timesbefore photobleaching (Fig. 1A). In addition, Dronpa was ob-served to spontaneously recover from the OFF to the ON state intens of seconds even under 488-nm illumination alone, whichhypothetically has been ascribed to thermal activation (4, 6, 7).As a result, large overlapping populations of Dronpa moleculescan be excited simultaneously, especially in densely labeledsamples, hindering single-molecule identification and localiza-tion. rsFastLime, a Dronpa variant, has the amino acid mutationV157G (8). The valine, not directly part of the chromophore, isinvolved in the photoactivated isomerization of the chromo-phore. The V157G mutation is thought to reduce the sterichindrance around the chromophore, allowing rapid switchingbetween cis and trans conformers (8). This observation led us todesign a Dronpa variant (DronpaV157L), termed “rsKame.” We

theorized that the increased steric hindrance associated with thebulkier leucine residue would retard both undesired 488 nm- andtypical 405 nm-driven photoactivation. Subsequent single-moleculephoto-physical analysis of rsKame confirmed our hypothesis.To demonstrate the efficacy of rsKame, we paired it with the

dark-to-red PA-FP PAmCherry1 in a two-color PALM imagingapplication, to simultaneously visualize the inner and outermitochondrial membranes (IMM and OMM) (9). This studyrevealed diverse spatial organizations of the membranes inrelation to each other at various morphologically distinct statesof the organelle network.Mitochondrial fission maintains the network by facilitating

dynamic exchange of matrix and membrane contents (10–12).Dynamin related protein 1 (Drp1), a largely cytoplasmic proteinresponsible for OMM scission in a mammalian cell, is recruitedto its receptor, mitochondrial fission factor (Mff), on the OMMand self-assembles into a helical ring upon GTP binding (12–17).On the basis of EM in vitro studies of the Saccharomyces cerevisiaehomolog dynamin related-1 (Dnm1), the current mitochondrialscission model suggests that the assembled subunits form a heli-cal ring around the OMM and slide past each other upon GTPhydrolysis, causing dynamic constriction of the ring diameterby ∼60 nm (18–20). However, the structural properties of theDrp1/Dnm1 helical ring on mitochondria have not been well

Significance

Optimal performance of super resolution fluorescence localiza-tion microscopy relies on a clear understanding of the photo-physical properties of photoactivatable (photo-switchable)fluorescent proteins [PA(PS)-FPs] at the single-molecule level.Our comparative study of Dronpa and a novel variant, rsKame,demonstrates the crucial role of photo-switching kinetics in su-per resolution imaging. rsKame, with its superior properties,significantly broadens the green PA(PS)-FP palette. We demon-strate the efficacy of rsKame and our two-color super resolutionimaging method (paired with PAmCherry1) by visualizing theinner and outer mitochondrial membranes and in situ structuralparameters of dynamin related protein 1 helical rings duringmitochondrial fission. Our two-color super resolution imagingmethod presented here is a reliable and user-friendly techniquewithout complicated sample preparation.

Author contributions: A.B.R., S.-H.L., J.Y.S., and C.B. designed research; A.B.R., S.-H.L., M.T.,and A.L. performed research; A.B.R. and S.-H.L. contributed new reagents/analytic tools;A.B.R., S.-H.L., and A.L. analyzed data; and A.B.R., S.-H.L., A.L., and C.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1A.B.R. and S.-H.L. contributed equally to this work.2To whom correspondence should be addressed. Email: carlos@alice.berkeley.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1320044111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1320044111 PNAS | September 9, 2014 | vol. 111 | no. 36 | 13093–13098

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0

characterized in vivo owing to the resolution limit of conven-tional optical microscopy.Here we used rsKame, together with PAmCherry1, in a se-

quential two-color PALM imaging method to structurally char-acterize the Drp1 helical rings in situ. We observed a ∼60-nmdecrease in ring diameter during fission and no significant changein length, suggesting support for a twistase scission model withpotential subunit loss at the helical termini (12).

ResultsThe Photo-Physical Limitations of Dronpa. When we initially attemp-ted to examine suborganelle structures with two-color PALM, weencountered the photoactivation/emission of dense populations ofDronpa molecules per frame (50 ms) even under low or zero ac-tivation laser power (Movie S1). Unlike other red PA-FPs, such asthe Eos-FP family and PAmCherry1, in which photoactivation istightly controlled by 405-nm light (21), this spurious “basal” or“spontaneous” photoactivation rate of Dronpa prevents the iden-tification of single-molecule events and causes a high backgroundfluorescence level even in the absence of 405-nm laser, partic-ularly in densely labeled samples. This effect fundamentally limitsthe discrimination required for single-molecule identification andlocalization achievable with Dronpa. When we attempted to in-activate Dronpa molecules with strong illumination by 488-nmlaser power before imaging, as recommended in the publishedmethod (3), the molecules could no longer be reactivated or ex-cited and were presumed to be photobleached. Although theseproblems likely have prevented Dronpa from wider use in superresolution microscopy, systematic studies on these issues havebeen lacking. We hypothesized that the excitation light (488 nm)was also capable of photoactivation and the source of the practicalproblems with Dronpa (Fig. 1A). To validate this idea, we exam-ined in vitro the photo-physics of individual Dronpa molecules inTris buffer by a total internal reflection microscope (SI MaterialsandMethods) (22). Similar to previous work, the emission trace ofa single Dronpa molecule, under 488-nm excitation only, featuresseveral bursts separated by a few seconds (Fig. 1B) (4, 6). How-ever, in contrast to the previously published observations, wefound that the interval between bursts becomes significantlyshortened with increasing 488-nm powers. For the complete char-acterization of Dronpa at the single-molecule level, three quantitieswere measured and statistically analyzed for various 488-nm and

405-nm power: the number of times singlemolecules photo-switch(NBlink), the time spent in the fluorescent state (TON), and the timespent in the dark state (TOFF). NBlink and TON are related to thekinetics of leaving the fluorescent state “F,” whereas TOFF is re-lated to the kinetics of entering the fluorescent state (Fig. 1A).NBlink and TON are decreased by the increase of 488-nm ex-

citation laser power, although not affected by 405-nm activationlaser power (Fig. S1 A and B). Thus, the Fluorescent (F)→Dark(D) and Fluorescent (F) → Bleached (B) kinetics depend onlyon 488-nm light (Fig. 1A). In contrast, TOFF is altered by 405 nmand 488 nm laser power (Fig. S1C). As revealed by the cumu-lative probability distribution function (CDF) of TOFF (Fig. 1C),TOFF increases as 405-nm laser power decreases, confirmingthat the photoactivation kinetics (D → F) is proportional to the405-nm activation laser power. Noticeably, even at zero 405-nmlaser power, 50% of the Dronpa molecules are photoactivatedwithin 3.4 s (Fig. 1C). This basal photoactivation rate increases asthe 488-nm excitation laser power increases, indicating that thephotoactivation kinetics (D → F) is also proportional to the 488-nm activation laser power (Fig. 1D). At high 488-nm laser powerthe basal photoactivation rate dominates, such that the 405-nm laserpower contribution becomes almost negligible (Fig. 1D). There-fore, a more realistic model of Dronpa includes 488-nm, as wellas 405-nm, light in the photoactivation kinetics (Fig. 1A). To miti-gate the issue, we developed a Dronpa variant with an approxi-mately twofold lower basal 488 nm-induced photoactivation.

Design of a Slow-Switching Dronpa Variant. The V157G mutationof rsFastLime significantly decreases the TOFF by fourfold inbulk measurements compared with Dronpa, likely through thereduction of steric hindrance to the chromophore’s cis/transisomerizations (8, 23). When we attempted to use rsFastLime ata 20-Hz frame rate and simultaneous illumination with 405-nmand 488-nm light, fluorescent events occurred with such rapiditythat no individual events could be distinguished per frame.Therefore, we hypothesized that increasing the steric hindranceat V157 would inhibit the cis/trans isomerization of the Dronpachromophore, slowing the 488 nm-induced basal photoactivation

BA

0 5 10 15 200

0.2

0.4

0.6

0.8

1

Toff

(sec)

Cu

mu

lati

ve D

ist.

Fu

nc.

(C

DF

) Dronpa (488nm Laser = 0.5 W/mm2)

405nm Laser (mW/mm2)

0.0 0.6 1.6 3.5 7.3

0 5 10 15 200

0.2

0.4

0.6

0.8

1

Toff

(sec)

Cu

mu

lati

ve D

ist.

Fu

nc.

(C

DF

) Dronpa (488nm Laser = 5.9 W/mm2)

405nm Laser (mW/mm2)

0.0 0.6 1.6 3.5 7.3

DC

1.1 3.4

D F BD FF488 nm

488 nm405 nm

(trans)

(cis) 0

15

0 10 20 30 40 500

100

Flu

ore

scen

ce (

AU

) 488 nm = 0.5 W/mm2

488 nm = 5.9 W/mm2

Time (s)

Toff

Fig. 1. Limits of Dronpa. (A) Trans to cis isomerization is activated by both 488-nm and 405-nm wavelengths. Cis to trans isomerization and photobleaching isdriven by 488-nm illumination. (B) Photo-switching of Dronpa for two different488-nm excitation powers in the absence of 405-nm light. Red arrows indicatethe time at which the molecule is photobleached. Camera integration time: 50ms. (C and D) TOFF cumulative probability distributions of Dronpa under lowand high-excitation (488 nm) laser power over a range of photoactivation (405nm) laser power. The fit curves were obtained using triple exponential func-tions (Fig. S1C).

C DA

0 5 10 15 200

0.2

0.4

0.6

0.8

1405nm Laser = 0

Toff

(sec)

Cu

mu

lati

ve D

ist.

Fu

nc.

(C

DF

)

488nm Laser (W/mm2)

0.5 (Dronpa) 5.9 (Dronpa) 0.5 (rsKame) 5.9 (rsKame)6.23.4

Time

Fluo

resc

ence

Time

Fluo

resc

ence

Time

Fluo

resc

ence

Frame

Frame

Frame

Dronpa

rsFastLime (Dronpa V157G)

rsKame (Dronpa V157L)

Frame

Frame

Frame

Flu

ore

scen

ceF

luo

resc

ence

Flu

ore

scen

ce

Time

rsKame(Dronpa V157L)

rsFastLime(Dronpa V157G)

Dronpa

0 100 200 3000

0.5

1

Photoactivation Time (s)

No

rmal

ized

Eve

nts

EDronpa : 48 srsKame: 76 s

Decay time

0 20 40 600

0.1

0.2

Localization Uncertainty (nm)

Pro

bab

ility

DronparsKameDronparsKameDronparsKameF

DronparsKame

B

Dronpa (IMM) rsKame (IMM)

Fig. 2. Rational design and photo-physical characterization of rsKame. Rationaldesign of rsKame was based on the V157G mutation of rsFastLime. (A) rsKamehas a V157L mutation to slow the trans to cis transition by increasing sterichindrance around the chromophore and reduce the number of overlappingmolecules at a given time. (B) TOFF cumulative probability distributions ofrsKame and Dronpa under high- and low-excitation (488 nm) laser powers inthe absence of 405-nm activation. The fit curves were obtained using tripleexponential functions (Fig. S1C). (C and D) Super resolution images of IMMlabeled with Dronpa (C) or rsKame (D) in HeLa cells. (E) The number ofphotoactivation events for both Dronpa and rsKame decrease exponentiallywith time. The rsKame decay rate is 1.6 slower than Dronpa, demonstratingslowed basal photoactivation of rsKame in vivo. (F) The average localizationuncertainty of rsKame is smaller than Dronpa, increasing the resolution ofrsKame compared with Dronpa. (Scale bar, 200 nm for C and D, Inset.)

13094 | www.pnas.org/cgi/doi/10.1073/pnas.1320044111 Rosenbloom et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0

and improving temporal separation between single molecules(Fig. 2A). We chose to replace Dronpa V157 with the nonpolaramino acid leucine or phenylalanine. Both mutants and wild typewere expressed in Escherichia coli and tested for fluorescenceand photo-switchability. DronpaV157L maintained fluorescenceand UV induced photo-switching. DronpaV157F lacked visiblefluorescence and was discarded for the remainder of the studies.

In Vitro Bulk Characterization of DronpaV157L. We determined thebulk spectroscopic properties of purified in vitro DronpaV157L andDronpa. The excitation and emission maxima of DronpaV157L didnot differ significantly from Dronpa; however, DronpaV157L has∼25% reduced brightness (Fig. S2 A and B). Comparison of theUV-induced photo-switching properties of Dronpa and Dron-paV157L were determined by an in vivo photo-switching assayof the fluorescent proteins expressed in E. coli (Fig. S2 C and D).DronpaV157L showed a slightly slower photo-decay time (∼0.7 s)than Dronpa (∼0.5 s), but the fluorescence growth rate was notaffected (Fig. S2 E and F). In addition, DronpaV157L photo-bleached more slowly (2.5 vs. 1.7 min) than Dronpa (Fig. S2 C andD). We renamed DronpaV157L “rsKame.” Dronpa was named for“dron,” a Japanese ninja word for vanishing, and PA for photo-activation (24). rsKame is a combination of reversibly switchable(rs) and “kame,” the Japanese word for turtle. Although the slowerphoto-switching of rsKame could be inferred by the ensemblemeasurement, we chose to more accurately understand its kineticsat the single-molecule level.

Single-Molecule Characterization of rsKame. To better understand theeffects of increased steric hindrance proximal to the Dronpa chro-mophore, we examined rsKame as single molecules using the samephoto-physical single-molecule characterization as was described forthe characterization of Dronpa (SI Materials and Methods) (22).For moderate 488-nm laser power (0.5 W/mm2), the NBlink of

rsKame (2.4) was slightly less than that of Dronpa (3.1), whereasTON of rsKame (44 vs. 42 ms) was slightly longer (Fig. S3 A andB). As expected, these are consistent with slowed 488-nm de-activation [F (cis) to D (trans)] kinetics by the increased sterichindrance posed by the leucine side chain of rsKame at position157. However, the effect is minor and becomes almost negligiblewhen the photobleaching becomes dominant at strong 488-nmlaser power (5.9 W/mm2) (Fig. S3 A and B).

The TOFF of rsKame in the presence of low 405 nm and 488 nmis ∼2 times as long as Dronpa (Fig. S3 C and D), demonstratingthat the increased steric hindrance affects the photoactivation[trans (D) to cis (F)] kinetics more than the deactivation kinetics.However, the difference in TOFF between rsKame and Dronpa isreduced and eventually becomes negligible as 405-nm laser powerincreases. Data sets from a third laser power (0.2 W/mm2), whichdemonstrated similar photo-physical trends, further clarified theeffect of 488-nm light on the photo-switching properties ofrsKame and Dronpa (Fig. S3).For the super resolution imaging application, the suppression

of basal level 488 nm-induced activation at zero 405-nm power isespecially important. The CDF of TOFF shows that the basal ac-tivation of rsKame is reduced by twofold compared with Dronpa(Fig. 2B). Fully analyzed super resolution images of the OMMlabeled with rsKame in EpH4 cells or the IMM inHeLa cells, withonly 488-nm illumination (Movie S1), show clearly delineatedmembranes with increased resolution compared with those labeledwith Dronpa (Fig. 2 C and D and Fig. S3 E and F). Moreover, the

I

−400−200 0 200 400

1

Distance (nm)

Inte

nsi

ty (

AU

) LK

JHG

BA C

−400−200 0 200 400

1

Distance (nm)

Inte

nsi

ty (

AU

) IMMOMME F

D

OMM IMM

Fig. 3. Two-color PALM distinguishes inner and outer mitochondrial membranes in mammalian cells. EpH4 cells were cotransfected with mitochondrialmembrane-specific PA(PS)-FPs. (A, B, G, H) OMMs were labeled with PAmCherry1-Lk-BclXl201-233. IMMs were labeled with BCS1L1-160-Lk-rsKame. (C–E)Elongated and (I–K) punctate mitochondria are observed with distinctly defined membranes (∼20-nm resolution; Fig. S5 C and D) in comparison with thecorresponding (Insets, 0.5× zoom) diffraction-limited images. (F and L) Intensity profiles across individual mitochondrion demonstrate differentiation be-tween the IMM and OMM. The residual red fluorescence observed toward the center of the mitochondria is due to the out-of-focus contribution from the top orthe bottom of the mitochondrion. (Scale bars, 5 μm for A, B, G, and H, 500 nm for C–E and I–K.)

A B C D

OMM

IMM

Merge

Fig. 4. Super resolution morphologies of IMM and OMM. Observed IMMand OMM morphologies include (A and B) single or double branched donut.Observed IMM structures within the OMM include (C) disorganized, (B and D)structured, and (C and D) partially filled. Some regions of the IMM (A, zoom)show linear and transverse fluorescence patterns. (Scale bar, 500 nm for A–D.)

Rosenbloom et al. PNAS | September 9, 2014 | vol. 111 | no. 36 | 13095

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0

number of rsKame photoactivation events in the presence of 488nm alone decays 1.6 times slower than for Dronpa, further dem-onstrating a slower basal photoactivation rate in vivo for rsKame(Fig. 2E). Because fewer molecules are photoactivated at any giventime, single-molecule events can be better separated in time forrsKame. In addition, a reduction in the background fluorescencefrom molecules out of focus improves the contrast of molecules infocus and thus their localization accuracy, although rsKame andDronpa have similar photon budgets (Fig. 2F, Fig. S3G andH, andMovie S1). Therefore, the increased steric hindrance of leucine atposition 157 in rsKame mitigates the 488 nm-induced photo-activation of the molecule and renders it more suitable for PALMimaging than Dronpa.We also compared rsKame with another green PS-FP, mGeos-M,

that has similar photo-physical properties (25). However,mGeos-M exhibits more significant basal photoactivation by 488nm than rsKame (Fig. S4 and Movie S2).

Two-Color PALM. We chose to pair rsKame with PAmCherry1, adark to red PA-FP. Because 405-nm light activates both rsKameand PAmCherry1, to develop our sequential two-color PALMimaging protocol, we measured the photoactivation rate ofPAmCherry1 for varying 405-nm laser powers, using the pre-viously described single-molecule photo-physical characteriza-tion method. The photoactivation rate of PAmCherry1 was morethan 1,000-fold slower than rsKame for any given 405-nm laserpower (Fig. S5A). We estimated that over a 5-min experiment(the average time for imaging rsKame), less than 2.5% ofPAmCherry1 molecules would be activated. Therefore, sequentialtwo-color PALM by imaging first rsKame and then PAmCherry1is possible (Fig. S5B; Materials and Methods).To demonstrate the efficacy of our two-color PALM method,

we chose to visualize two major substructures of the mitochon-dria, the repeatedly and deeply invaginated IMM and the smoothOMM. Nuclear encoded mitochondrial proteins are directed tomitochondria by localization sequences (MLS), often found at theN or C terminus (26). These labels and additional transmembraneor localization sequences direct mitochondrial proteins to specificcompartments and/or membranes (26). We chose two proteinswhose MLS and specific membrane integration had been char-acterized by truncations fused to reporter proteins.To coat the OMM with PAmCherry1, we fused the C-terminal

MLS and transmembrane region of murine B-cell lymphoma-extra

large (BclXl) to the C terminus of PAmCherry1 with a 25-amino-acid linker (PAmCherry1-Lk-BclXl201-233). BclXl is a protein in-volved in apoptosis and anchored to theOMMwith themain bodyof the protein in the cytoplasm (27, 28). To coat the IMM, wefused the N-terminal MLS and transmembrane region of humanubiquinol-cytochrome c reductase synthesis-like (BCS1L) tothe N terminus of rsKame (BCS1L1-160-Lk-rsKame). BCS1L is amitochondrial translocase protein and anchored to the IMM withthe main body of the protein in the matrix (29, 30).EpH4 cells, cotransfected with PAmCherry1-Lk-BclXl201-233

and BCS1L1-160-Lk-rsKame, were imaged by our two-colorPALM method (Materials and Methods). Two-color PALMimages allowed us to elucidate cells containing predominantlyelongated mitochondrial networks (Fig. 3 A and B) from cellscontaining predominantly punctate mitochondria (Fig. 3 G andH). From the clear OMM boundary in PALM images, we couldmeasure the width of elongated mitochondria (200-250 nm) andpunctate round mitochondria (300-350 nm) (Fig. 3 C and I). Incontrast, the width of IMM was shorter than OMM, appearingcompletely enclosed by the OMM (Fig. 3 D, E, J and K). Thedistinction is further demonstrated in intensity profiles made bydrawing a line transversely across an individual mitochondrionand measuring PALM localization intensity along the line (Fig. 3F and L). In elongated IMM, we could observe structurescomposed of a series of transverse and somewhat linearly ori-ented fluorescent molecules, presumably outlining the highlyfolded structure of the cristae (Figs. 3D and 4A and Fig. S3I).Two-color PALM images of the IMM and OMM revealed

diverse mitochondrial morphologies (Fig. 4). We observed “donut”-shaped mitochondria, a recently described morphology withpathophysiological significance under stress conditions (Fig. 4A and B) (31, 32). In previous studies the typical diameter was∼1.3 μm by conventional fluorescence microscopy. However,we measured donuts as small as ∼500 nm (Fig. 4 A and B).Previously observed donut mitochondria with small diameterswere likely misclassified as a separate “blob” morphology be-cause of diffraction limited resolution. Some observed IMMsonly partially filled the OMM compartments (Fig. 4 C and D).Such heterogeneous IMM structures may represent mitochon-dria undergoing IMM remodeling (33, 34).

Drp1. We applied our two-color PALM method to examine thestructural characteristics of mammalian Drp1 oligomerization

EB C D

Initial

Drp1

Linker

rsKame

Uncertainty

Intermediate

Constricted Scissioned Terminal

12

3

4

1

3

4

2

A

HG

F

W W1

prd Ldrp1

2Δ

L

WL

OMMDrp1

Initial Intermediate Terminal

N 69 48 121

W 183 ± 56 149 ± 43 121 ± 32

Wdrp1 139 ± 60 105 ± 48 77 ± 39

L 112 ± 34 112 ± 46 92 ± 34

Ldrp1 68 ± 40 68 ± 51 48 ± 38

TerminalInitialIntermediate

Fig. 5. Quantification of Drp1 helical ringdiameter and length obtained with two-color PALM. (A–E) Super resolution char-acterization of Drp1 helical ring structuredimensions at various fission stages.(Insets) Diffraction-limited images forOMM (red) (0.5× zoom) and Drp1 (green)(1× zoom). (B) Elongated or directlytransverse Drp1 clusters on unconstrictedor marginally constricted mitochondrialtubules (Initial). (C and D) Drp1 clusterson constricted and almost fully dividedmitochondrial tubules (Constricted andScissioned, respectively). (E) Drp1 clustersat the termini of mitochondrial tubules(Terminal). (F) Overestimation of di-ameter and length results from meanuncertainty of fluorophore localization(Δ). Drp1 helical ring diameter (Wdrp1) orlength (Ldrp1) is determined by subtract-ing 2Δ from the measured diameter (W)or length (L). (G and H) Quantification ofthe raw measured diameter (W) andlength (L) of the Drp1 helical ring struc-tures pre constriction through post scis-sion.NCell= 36. (Scale bars, 1 μmforA, 200nm for B–E.)

13096 | www.pnas.org/cgi/doi/10.1073/pnas.1320044111 Rosenbloom et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0

during mitochondrial fission in situ. Drp1 helical rings wereimaged by labeling the OMM with PAmCherry1-Lk-BclXl201-233and fusing Drp1 to rsKame. Previous studies with Drp1 subunitsfused N-terminally to fluorescent proteins incorporated intofission rings during assembly showed normal fission progression(15, 18, 35, 36). We fused rsKame to the N terminus of humanDrp1 separated by a 14-amino-acid linker (rsKame-Lk-Drp1).HeLa cells were cotransfected with PAmCherry1-Lk-BclXl201-233and rsKame-Lk-Drp1 and imaged by PALM as previously described(Materials and Methods). We determined the ratio of rsKame-Lk-Drp1 to endogenous Drp1 to be 0.62 to 0.38, which, assuming thata low percentage of rsKame will misfold, indicates at least half ofthe Drp1 molecules in the helical rings are labeled (Fig. S6A).The high cytoplasmic population of rsKame-Lk-Drp1 could

induce errors in identifying colocalization between Drp1 ringsand mitochondria. We reduced this potential error by using acustom built program to analyze Drp1 clusters, filtering outclusters of n < 8 events (Fig. S7). The majority of the large Drp1clusters were observed to colocalize with the OMM (Fig. 5A andFig. S7). Distinct OMMmorphologies can be used as a referenceto determine fission sites; the diameter of the OMM at the sitedecreases as fission progresses (Fig. S8). Therefore, on the basisof the observed OMM morphology and the incorporation ofDrp1 clusters, we identified four different fission states. Thefission state in which an elongated or two distinct Drp1 clustersare found on unconstricted or slightly constricted mitochondriawas termed Initial (Fig. 5B and Fig. S8 A–D). This is presumed tocorrespond to the 2D section of a Drp1 ring in its relaxed formbefore the GTP-hydrolysis-driven conformational change. Con-stricted is used to describe the fission state in which the Drp1cluster is localized to a clearly constricted outer membrane (Fig.5C and Fig. S8 E–H). Presumably this state corresponds toa Drp1 ring that is partially constricted and undergoing GTPhydrolysis. We termed the fission state in which membranesappear to be very tightly constricted beneath Drp1 clustersScissioned (Fig. 5D and Fig. S8 I–L). The fission state Terminal isused to describe Drp1 clusters observed at one end of mito-chondria (Fig. 5E and Fig. S8 M–P). This is likely a partiallydegraded ring, which is found predominantly at only one of themitochondrial termini resulting from fission (Fig. 6D). Because itwas difficult to determine the precise degree of membraneconstriction between the Constricted and Scissioned states ofthe OMM, we grouped these two states together for analysisunder Intermediate (Fig. 5 C and D). We also noted numerousindividual Drp1 clusters found at shallow invaginations in the OMM(Fig. 6 A–C). These invaginations and clusters were usually ob-served only on one side of the mitochondria and may reflect nascentfission sites or early binding of Drp1 to local clusters of Mff (16, 37).To measure the diameter and the length of individual Drp1

rings, we first drew a rectangular box around each ring (Fig. 5 Band F). The box is reduced to the minimum rectangle thatincludes all of the Drp1 molecules. The widths and the lengths ofthese minimal rectangles (Fig. 5 G and H) overestimate thedimensions of the Drp1 ring by twice the mean localization un-certainty of rsKame (22 nm; Fig. S5C) in the PALM images (Fig.5F). After this correction, the Drp1 ring diameters were as fol-lows: Initial = 139 ± 60 nm, Intermediate = 105 ± 48 nm, andTerminal = 77 ± 39 nm (Fig. 5G). The corresponding ringlengths were as follows: Initial = 68 ± 40 nm, Intermediate = 68 ±51 nm, and Terminal = 48 ± 38 nm (Fig. 5H). Significantly,although the Drp1 ring diameter decreases by ∼60 nm, accom-panying its transition between the Initial and Terminal fissionstates, no significant change in ring length was observed. We alsoquantified the diameters of the OMM within the Initial andIntermediate fissions states, which correlated with the Drp1diameters within error and showed a distinct decrease from theInitial to Intermediate state (Fig. S8 Q and R).To determine the sensitivity of our analysis method to the

number of incorporated labeled Drp1 molecules, we performedsimulations on Dnm1 helical rings, based on cryo-EM structuraldata (20), varying the amounts of fluorescently labeledmonomers.

The resulting values were consistent at coverage greater than 0.3(Fig. S6 B and C), significantly less than the experimental cover-age (≥0.5) (Fig. S6A). Compared with our measured Drp1 width(W) (Fig. 5G), our values were within the error of the Dnm1simulations, for both non- and constricted rings; whereas thelengths (L) were much shorter than predicted for a two-turn helix(Fig. S6 B and C).

DiscussionWe have shown that rsKame, when paired with PAmCherry1,can be used to perform two-color super resolution PALM im-aging. The increased steric hindrance introduced by the V157Lmutation of rsKame leads to reduced basal photoactivation bythe 488-nm excitation light, compared with Dronpa. These effectsimprove the temporal separation of single-molecule fluorescenceevents and reduce the background fluorescence, increasing theaccuracy ofmolecular localization. The large difference in 405-nmphotoactivation rates between PAmCherry1 and rsKame makethem an excellent pair for two-color PALM. Two-color superresolution images of OMMs and IMMs showed distinct morpho-logical differences, including much smaller donut-shaped mito-chondria than were previously reported with use of conventionalfluorescence microscopy. More strikingly, we observed mitochon-dria with the inner membrane clustered to a small area within theboundaries of the outer membrane. Application of our two-colorPALM method to investigate IMM structure using proteins withdiffering submembrane localizations could yield important in-formation about the organization of the inner membrane.Here we used rsKame and PAmCherry1 to obtain 2D super

resolution (∼20 nm) images of Drp1 rings during mitochondrialfission and quantified the structural properties of three distinctfission states. The diameters of the Drp1 rings measured withthis method (139 ± 60 nm for the pre mitochondrial membraneconstriction and 77 ± 39 nm for the post membrane scission)(Fig. 5G) agree well within error with the published data for invitro Drp1/Dnm1 helical rings (Pre: 121 ± 25 nm; Post: 71 ±23 nm) (11, 18). The in situ observations presented here confirmthat the GTP-induced constriction of the Drp1 helical ring in-deed plays a crucial role in mitochondrial membrane fission.The Drp1 helical ring length (∼68 nm) between the Initial and

Intermediate states exhibits negligible change and is only slightlyreduced in the Terminal state, despite the significant (∼60 nm)change in overall diameter of the helical ring. Although unknownfor Drp1, the helical pitch of Dnm1 has been measured as ∼30 nm,which does not change upon constriction (18). Therefore, in vivoDrp1 helical rings may have a mean number of ∼1−4 helical turns,similar to dynamin, as opposed to the extended Drp1 and Dnm1helical structures observed in vitro over liposomes (14, 19, 38, 39).Interestingly, for a two-turn helix, we expect a minimal 50-nm in-crease in Drp1 helical length during fission if Drp1 constrictionfollows the current twistase scission model and its helical pitchremains unchanged (Fig. S6C). However, we observed no increase,indicating potential degradation of the helical termini during fission.We also observed numerous Drp1 clusters localized on small

invaginations on one side of the OMM. A current model of

A B C DOMMDrp1

Fig. 6. Drp1 clusters form at shallow unilateral invaginations on the mito-chondrial outer membrane. (A–C) Small Drp1 clusters not involved in de-monstrable fission states are often found on the OMM at small invaginations,primarily on one side of unconstricted mitochondria. (D) Drp1 clusters ob-served at the Terminal state appear to associate primarily with a singularresulting mitochondrial terminus. (Scale bars, 500 nm.)

Rosenbloom et al. PNAS | September 9, 2014 | vol. 111 | no. 36 | 13097

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0

mitochondrial fission describes a process by which the endo-plasmic reticulum (ER) independently and physically constrictsthe OMM to ∼130 nm (18, 38). Drp1 forms helical rings on thesesites and proceeds through fission (12, 18, 19). However, wehypothesize that Drp1 may form clusters at ER–mitochondriacontact sites that are in the initial stages of membrane con-striction (37). Highly localized subpopulations of Mff at the ERcontact sites could facilitate Drp1 oligomerization into saidclusters (18, 40). Super resolution imaging of the interactionsbetween the ER membrane, the OMM, Mff, and Drp1 couldlead to improved understanding of mitochondrial fission.Super resolution fluorescence microscopy techniques, like

PALM, can potentially revolutionize biology by visualizingsubcellular structures in situ that previously could not be ob-served owing to the diffraction limit. However, the applicability ofmulticolor PALM has been hindered by the limited palette ofwell-characterized PA(PS)-FPs. We have added a novel PS-FP,rsKame, to the green palette and demonstrated its effectiveness,along with PAmCherry1, in super resolution two-color imaging bythe visualization of two suborganelle structures: the outer andinner mitochondrial membranes and the Drp1 fission ring. Thedevelopment of rsKame and its successful use in two-color PALMimaging opens the door to a wealth of information to be gainedfrom closer inspection of proteins and subcellular structures withmulticolor super resolution microscopy.

Materials and MethodsTwo-Color Sequential PALM Imaging. To first image rsKame, the sample wascoilluminated with 488-nm excitation/photoactivation (5.9 W/mm2) and405-nm activation at 0.6 mW/mm2. Fluorescent data were collected at 20 Hz

with the 525-nm emission filter until all molecules were photobleached. Toimage PAmCherry1, the sample was coilluminated with the 561-nm excita-tion (22.0 W/mm2) and 405-nm photoactivation begun at 0.1 W/mm2 andslowly increased to 1.4 W/mm2 over time. Fluorescence was collected until allmolecules were photobleached. Both PALM datasets were analyzed sepa-rately and their super resolution images were overlaid using our customMATLAB-based analysis software package (22).

Drp1 and Mitochondrial Fission-Cell Preparation. rsKame-Lk-Drp1 was placeddownstream of a TetOn promoter and cotransfected with PAmCherry1-Lk-BclXl201-233 into the T-Rex HeLa cells (Invitrogen TetOn system). Expression ofrsKame-Lk-Drp1 was induced by doxycycline. Twelve hours before fixationand imaging, cells were fed 40 nm Au fiducial markers coated in FBS (60,000particles/mL). Four hours before imaging, doxycycline (Invitrogen) and z-vad-fmk (Promega) were added to the total growth media to final concentrationsof 0.75 μg/mL and 20 μM, respectively. HeLa cells were then incubated for110 min at 37 °C. Staurosporine (Cell Signaling Technologies) was added toa final concentration of 1 μM, and cells were incubated for another 100 minat 37 °C. Just before fixation, mitochondria were labeled with MitoTrackerDeep Red (Life Technologies/Invitrogen) at a final concentration of 30 nM.Treated cells were fixed in 1% formalin (Sigma-Aldrich) in 1× PHEM (60 mMPipes, 25 mM Hepes, 10 mM EGTA, and 2.0 mM MgCl2, pH 7.0) for 10 min atroom temperature and imaged in 1× PHEM.

ACKNOWLEDGMENTS. We thank Jodi Nunnari for excellent advice andcritical comments throughout these studies; the Nunnari laboratory atUniversity of California, Davis, for their generous gift of Drp1; the JanLiphardt laboratory at University of California, Berkeley for their gener-ous gift of PamCherry1; Eric Betzig of Janelia Farm Research Campus forhis advice regarding Dronpa imaging; and Young-Woo Seo of the KoreaBasic Sciences Institutes for his advice on mitochondrial biology.

1. Betzig E, et al. (2006) Imaging intracellular fluorescent proteins at nanometer reso-lution. Science 313(5793):1642–1645.

2. Rust MJ, Bates M, Zhuang X (2006) Sub-diffraction-limit imaging by stochastic opticalreconstruction microscopy (STORM). Nat Methods 3(10):793–795.

3. Shroff H, et al. (2007) Dual-color superresolution imaging of genetically expressedprobes within individual adhesion complexes. Proc Natl Acad Sci USA 104(51):20308–20313.

4. Habuchi S, et al. (2005) Reversible single-molecule photoswitching in the GFP-likefluorescent protein Dronpa. Proc Natl Acad Sci USA 102(27):9511–9516.

5. Subach FV, Patterson GH, Renz M, Lippincott-Schwartz J, Verkhusha VV (2010) Brightmonomeric photoactivatable red fluorescent protein for two-color super-resolutionsptPALM of live cells. J Am Chem Soc 132(18):6481–6491.

6. Flors C, et al. (2007) A stroboscopic approach for fast photoactivation-localizationmicroscopy with Dronpa mutants. J Am Chem Soc 129(45):13970–13977.

7. Ando R, Flors C, Mizuno H; Hofkens J, Miyawaki A (2007) Highlighted generation offluorescence signals using simultaneous two-color irradiation on Dronpa mutants.Biophys J 92:L97–L99.

8. Andresen M, et al. (2008) Photoswitchable fluorescent proteins enable mono-chromatic multilabel imaging and dual color fluorescence nanoscopy. Nat Biotechnol26(9):1035–1040.

9. Subach FV, et al. (2009) Photoactivatable mCherry for high-resolution two-colorfluorescence microscopy. Nat Methods 6(2):153–159.

10. Chan DC (2006) Mitochondrial fusion and fission in mammals. Annu Rev Cell Dev Biol22:79–99.

11. Berman SB, Pineda FJ, Hardwick JM (2008) Mitochondrial fission and fusion dynamics:the long and short of it. Cell Death Differ 15(7):1147–1152.

12. Hoppins S, Lackner L, Nunnari J (2007) The machines that divide and fuse mito-chondria. Annu Rev Biochem 76:751–780.

13. Pitts KR, Yoon Y, Krueger EW, McNiven MA (1999) The dynamin-like protein DLP1 isessential for normal distribution and morphology of the endoplasmic reticulum andmitochondria in mammalian cells. Mol Biol Cell 10(12):4403–4417.

14. Yoon Y, Pitts KR, McNiven MA (2001) Mammalian dynamin-like protein DLP1 tubu-lates membranes. Mol Biol Cell 12(9):2894–2905.

15. Smirnova E, Griparic L, Shurland DL, van der Bliek AM (2001) Dynamin-related proteinDrp1 is required for mitochondrial division in mammalian cells. Mol Biol Cell 12(8):2245–2256.

16. Otera H, Mihara K (2011) Discovery of the membrane receptor for mitochondrialfission GTPase Drp1. Small GTPases 2(3):167–172.

17. Otera H, et al. (2010) Mff is an essential factor for mitochondrial recruitment of Drp1during mitochondrial fission in mammalian cells. J Cell Biol 191(6):1141–1158.

18. Friedman JR, et al. (2011) ER tubules mark sites of mitochondrial division. Science334(6054):358–362.

19. Mears JA, et al. (2011) Conformational changes in Dnm1 support a contractilemechanism for mitochondrial fission. Nat Struct Mol Biol 18(1):20–26.

20. Ingerman E, et al. (2005) Dnm1 forms spirals that are structurally tailored to fit mi-tochondria. J Cell Biol 170(7):1021–1027.

21. Chudakov DM, Matz MV, Lukyanov S, Lukyanov KA (2010) Fluorescent proteins andtheir applications in imaging living cells and tissues. Physiol Rev 90(3):1103–1163.

22. Lee SH, Shin JY, Lee A, Bustamante C (2012) Counting single photoactivatable fluo-

rescent molecules by photoactivated localization microscopy (PALM). Proc Natl Acad

Sci USA 109(43):17436–17441.23. Stiel AC, et al. (2007) 1.8 A bright-state structure of the reversibly switchable fluo-

rescent protein Dronpa guides the generation of fast switching variants. Biochem J

402(1):35–42.24. Ando R, Mizuno H, Miyawaki A (2004) Regulated fast nucleocytoplasmic shuttling

observed by reversible protein highlighting. Science 306(5700):1370–1373.25. Chang H, et al. (2012) A unique series of reversibly switchable fluorescent proteins with

beneficial properties for various applications. Proc Natl Acad Sci USA 109(12):4455–4460.26. Stojanovski D, Bohnert M, Pfanner N, van der Laan M (2012) Mechanisms of protein

sorting in mitochondria. Cold Spring Harb Perspect Biol 4(10):1–18.27. Horie C, Suzuki H, Sakaguchi M, Mihara K (2002) Characterization of signal that di-

rects C-tail-anchored proteins to mammalian mitochondrial outer membrane. Mol

Biol Cell 13(5):1615–1625.28. Kaufmann T, et al. (2003) Characterization of the signal that directs Bcl-x(L), but not

Bcl-2, to the mitochondrial outer membrane. J Cell Biol 160(1):53–64.29. Fölsch H, Guiard B, Neupert W, Stuart RA (1996) Internal targeting signal of the BCS1

protein: A novel mechanism of import into mitochondria. EMBO J 15(3):479–487.30. Wagener N, Neupert W (2012) Bcs1, a AAA protein of the mitochondria with a role in

the biogenesis of the respiratory chain. J Struct Biol 179(2):121–125.31. Liu X, Hajnóczky G (2011) Altered fusion dynamics underlie unique morphological

changes in mitochondria during hypoxia-reoxygenation stress. Cell Death Differ

18(10):1561–1572.32. Ahmad T, et al. (2013) Computational classification of mitochondrial shapes reflects

stress and redox state. Cell Death Dis 4:e461.33. Mannella CA (2006) Structure and dynamics of the mitochondrial inner membrane

cristae. Biochim Biophys Acta 1763(5-6):542–548.34. Frey TG, Sun MG (2008) Correlated light and electron microscopy illuminates the role

of mitochondrial inner membrane remodeling during apoptosis. Biochim BiophysActa 1777(7-8):847–852.

35. Labrousse AM, Zappaterra MD, Rube DA, van der Bliek AM (1999) C. elegans dyna-

min-related protein DRP-1 controls severing of the mitochondrial outer membrane.

Mol Cell 4(5):815–826.36. Sesaki H, Jensen RE (1999) Division versus fusion: Dnm1p and Fzo1p antagonistically

regulate mitochondrial shape. J Cell Biol 147(4):699–706.37. Kornmann B (2013) The molecular hug between the ER and the mitochondria. Curr

Opin Cell Biol 25(4):443–448.38. Morlot S, Lenz M, Prost J, Joanny JF, Roux A (2010) Deformation of dynamin helices

damped by membrane friction. Biophys J 99(11):3580–3588.39. Ferguson SM, De Camilli P (2012) Dynamin, a membrane-remodelling GTPase. Nat Rev

Mol Cell Biol 13(2):75–88.40. Gandre-Babbe S, van der Bliek AM (2008) The novel tail-anchored membrane protein

Mff controls mitochondrial and peroxisomal fission in mammalian cells. Mol Biol Cell

19(6):2402–2412.

13098 | www.pnas.org/cgi/doi/10.1073/pnas.1320044111 Rosenbloom et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 12

, 202

0