Ultrabright and Stable Luminescent Labels for Correlative

Aug 2, 2019 - To identify specific proteins in electron microscopy, immunogold labeling has become the method of choice. However, the distinction of ...
0 downloads 0 Views 672KB Size
Subscriber access provided by BUFFALO STATE

Communication

Ultra-Bright and Stable Luminescent Labels for Correlative Cathodoluminescence Electron Microscopy (CCLEM) Bioimaging Kerda Keevend, Laurits Puust, Karoliine Kurvits, Lukas R.H. Gerken, Fabian H.L. Starsich, Jian-Hao Li, Martin T Matter, Anastasia Spyrogianni, Georgios A. Sotiriou, Michael Stiefel, and Inge K Herrmann Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.9b01819 • Publication Date (Web): 02 Aug 2019 Downloaded from pubs.acs.org on August 6, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

Ultra-Bright and Stable Luminescent Labels for

2

Correlative Cathodoluminescence Electron

3

Microscopy (CCLEM) Bioimaging

4

Kerda Keevend1,2,3, Laurits Puust4, Karoliine Kurvits4, Lukas R.H. Gerken1,3, Fabian H.L.

5

Starsich5, Jian-Hao Li1,3, Martin T. Matter1,3,5, Anastasia Spyrogianni6, Georgios A. Sotiriou7,

6

Michael Stiefel8,º, and Inge K. Herrmann1,3,º,*

7

1

8

Laboratories for Materials Science and Technology (Empa), Lerchenfeldstrasse 5, CH-9014 St.

9

Gallen, Switzerland. 2

10 11 12

Laboratory for Particles Biology Interactions, Department Materials Meet Life, Swiss Federal

Optical Nanomaterial Group, Institute for Quantum Electronics, Department of Physics, ETH Zurich, Auguste-Piccard- Hof 1, CH-8093 Zurich, Switzerland.

3

Laboratory for Nanoparticle Systems Engineering, Institute of Process Engineering, Department

13

of Mechanical and Process Engineering, ETH Zurich, Sonneggstrasse 3, CH-8092 Zurich,

14

Switzerland.

15 16

4

Laboratory of Laser Spectroscopy, Institute of Physics, University of Tartu, W. Ostwaldi St 1, 50411 Tartu, Estonia.

ACS Paragon Plus Environment

1

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

5

Particle Technology Laboratory, Institute of Process Engineering, Department of Mechanical

2 3

and Process Engineering, ETH Zurich, Sonneggstrasse 3, CH-8092 Zurich, Switzerland. 6

Institute of Pharmaceutical Sciences, Department of Chemistry and Applied Biosciences, ETH

4

Zurich, Vladimir-Prelog-Weg 1-5, CH-8093 Zurich, Switzerland. 7

5 6 7

Page 2 of 25

Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, SE-17177 Stockholm, Sweden.

8

Laboratory for Transport at Nanoscale Interfaces, Department Materials Meet Life, Swiss

8

Federal Laboratories for Materials Science and Technology (Empa), Überlandstrasse 129, CH-

9

8600 Dübendorf, Switzerland.

10

ºshared senior author

11

ACS Paragon Plus Environment

2

Page 3 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

ABSTRACT

2

The mechanistic understanding of structure-function relationships in biological systems heavily

3

relies on imaging. While fluorescence microscopy allows the study of specific proteins following

4

their labelling with fluorophores, electron microscopy enables holistic ultrastructural analysis

5

based on differences in electron density. To identify specific proteins in electron microscopy,

6

immunogold labelling has become the method of choice. However, the distinction of

7

immunogold-based protein-labels from naturally occurring electron dense granules, and the

8

identification of several different proteins in the same sample remains challenging. Correlative

9

cathodoluminescence electron microscopy (CCLEM) bioimaging has recently been suggested to

10

provide an attractive alternative based on labels emitting characteristic light. While luminescence

11

excitation by an electron beam enables sub-diffraction imaging, structural damage to the sample

12

by high energy electrons has been identified as potential obstacle.

13

Here, we investigate the feasibility of various commonly used luminescent labels for CCLEM

14

bioimaging. We demonstrate that organic fluorophores and semiconductor quantum dots suffer

15

from a considerable loss of emission intensity, even when using moderate beam voltages (2 kV)

16

and currents (0.4 nA). Rare-earth (RE) element doped nanocrystals, in particular Y2O3:Tb3+ and

17

YVO4:Bi3+,Eu3+ nanoparticles with green and orange-red emission, respectively, feature

18

remarkably high brightness and stability in the CCLEM bioimaging setting. We further illustrate

19

how these nanocrystals can be readily differentiated from morphologically similar naturally

20

occurring dense granules based on optical emission, making them attractive nanoparticle core

21

materials for molecular labelling and (multi)color CCLEM.

ACS Paragon Plus Environment

3

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 25

1

Keywords: Super-resolution microscopy; Immunogold; Molecular labelling; Nanocrystals,

2

Nanoscopy

3

ACS Paragon Plus Environment

4

Page 5 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

Biological systems are enormously complex and our understanding of them relies on the ability

2

to examine relationships between structure and function at various levels of resolution, and

3

within a functionally correlated context.1 In recent years, correlative light and electron

4

microscopy (CLEM)2 has gained increasing importance as it combines imaging based on

5

fluorescent labels and ultrastructural analysis by electron microscopy. However, the resolution

6

mismatch3 between the optical and the electron signal constitutes a major obstacle.4 While recent

7

developments in super-resolution microscopy techniques have enabled fascinating sub-

8

diffraction fluorescence imaging,5-7 direct combination with electron microscopy remains

9

challenging due to differences in sample preparation requirements.8 Recently, the use of

10

(multi)color correlative cathodoluminescence electron microscopy (CCLEM) for the analysis of

11

biological samples has been proposed by us9 and others.10-13 Cathodoluminescence (CL) is

12

generated by an electron beam, which can be focused down to a few nanometers, thereby

13

enabling luminescence data acquisition with deep-subwavelength spatial resolution.14,15 CL

14

spectra can be acquired across a wide spectral range and thus open the possibility of (multi)color

15

imaging, which in turn allows identification of structures directly based on spectral signatures

16

rather than electron density (as in the case of immunogold labelling). Because CL emission from

17

organic materials, including most biological tissues, is usually low, the use of exogenous

18

luminescent labels has been suggested for CCLEM.16 However, the electron beam used for the

19

excitation of the luminescent probes is potentially damaging to the biological samples17 and the

20

labels themselves.18

21

Here, we investigate the feasibility of different luminescent labels for cathodoluminescence

22

bioimaging. We report on the optical properties of a selection of luminescent labels with wide-

23

spread adoption in biological imaging in a photoluminescence (PL, excitation by photons) and a

ACS Paragon Plus Environment

5

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 25

1

cathodoluminescence (excitation by accelerated electrons) bioimaging setting. Additionally, we

2

show how inherent properties of the label (stability, brightness and emission lifetimes) and the

3

electron beam characteristics (acceleration voltage and current) affect CL data quality, and we

4

identify optimal settings for high-resolution CCLEM bioimaging. We then demonstrate that

5

ultra-bright rare-earth (RE) element doped nanocrystals can be used for nanometric CCLEM

6

bioimaging at low electron acceleration voltages (≤ 2 kV). High-resolution CCLEM allows

7

straightforward differentiation between naturally occurring dense granules and nanocrystals,

8

which is pivotal for prospective molecular labelling applications.

9

The labels investigated in the present study include the widely used organic fluorophore 4'-6-

10

diamidino-2-phenylindole

(DAPI),

conjugated

polymer-based

11

semiconductor quantum dots (CdTe and CdSe/CdS), and two types of RE3+ doped nanocrystals

12

(Y2O3:Tb3+ and YVO4:Bi3+,Eu3+), which have previously been employed for photoluminescence

13

bioimaging (Figure 1a,b and see ESI, Figure S1 for structures, Figure S2 for corresponding

14

photo-excitation and emission spectra, Table S1 for PL spectra acquisition parameters, and

15

Figure S3 for RE3+ doped nanocrystals phase identification).19-21 While the organic fluorophores

16

exhibit broad emission, the rare-earth element doped nanocrystals present narrow emission lines

17

with main emission peaks in the green (545 nm, Y2O3:Tb3+) and the orange-red spectral region

18

(620 nm, YVO4:Bi3+,Eu3+) (see ESI, Figure S2). Both organic labels (DAPI and CPN) show

19

rapid bleaching under laser excitation, and the luminescence emission intensity decreased by

20

more than 40% within the first three minutes of illumination (corresponding to a dose of 360

21

J/cm2) (Figure 1c). Such photo bleaching of organic molecules is well-known from fluorescence

22

microscopy and has been attributed to structural damage to the organic molecules.22,23 CdTe and

23

CdSe/CdS semiconductor quantum dots on the other hand show a moderate reduction in

ACS Paragon Plus Environment

quantum

dots

(CPN),

6

Page 7 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

luminescence intensity of ~25% and 10%, respectively. Both RE3+ doped nanocrystals present

2

fluorescence emission stability over time superior to the other labels investigated, in line with

3

previous fluorescence imaging studies.20,21 While YVO4:Bi3+,Eu3+ nanoparticles showed stable

4

luminescence emission over the full duration of the experiment (15 mins), Y2O3:Tb3+

5

nanocrystals presented a minor increase in the luminescence emission intensity, which can be

6

attributed to desorption of surface-bound oxygen species (see ESI, Figure S4).24

7

In addition to brightness and emission stability, short emission lifetimes are crucial for high-

8

resolution imaging in scanning mode. In case the emission lifetime is longer than the dwell time,

9

image quality is strongly compromised by smearing.25 While the organic fluorophores and

10

semiconductor quantum dots have short fluorescence lifetimes (e.g., 2.6 nanoseconds for DAPI,

11

see ESI, Figure S5),26,27 the RE3+ doped nanocrystals have emission lifetimes in the range of

12

milliseconds21,28 (see ESI, Figure S5). Based on the measured PL lifetimes, the dwell times for

13

imaging YVO4:Bi3+,Eu3+and Y2O3:Tb3+ nanocrystals should be no shorter than 1 and 5 ms,

14

respectively.21,28 Although the excitation mechanism for cathodoluminescence is comparable to

15

photoluminescence,14,29 electron beam excitation generally leads to emission by all luminescence

16

mechanisms present in the material, since the electrons act as a supercontinuum source15 and

17

hence lifetimes measured in PL may be underestimating CL lifetimes. Using long dwell times is

18

beneficial in terms of signal-to-noise (S/N) ratio, however, may cause structural damage to the

19

fluorophore as well as specimen heating.30

20

While all of the above labels can be employed for PL bioimaging despite the observed

21

bleaching, exposure to accelerated electrons is typically even more damaging.17 To quantify the

22

label stability in a CL setting, we acquired sequential CL spectra (Figure 1d) using optimized

ACS Paragon Plus Environment

7

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 25

1

beam parameters and monitored the cathodoluminescence intensity over time (see ESI, Table S2

2

for data acquisition parameters).

3 4

Figure 1: a) Transmission electron micrographs of the two synthesized RE3+ doped nanocrystals

5

with optical emission in green and orange-red upon exposure to UV light (λex = 254 nm)

6

(inserts). b) Fluorescence microscopy image of YVO4:Bi3+,Eu3+ nanocrystals (red) in human

7

vascular endothelial cells. The cytoskeleton is stained with Alexa 488 Phalloidin (green) and the

8

nucleus with DRAQ5 (magenta). Scale bar: 25 µm. c) Photoluminescence emission stability over

ACS Paragon Plus Environment

8

Page 9 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

time under laser excitation for the different labels. d) Cathodoluminescence emission spectra, e)

2

brightness, and f) emission stability over time for all the corresponding labels.

3

The beam parameters were optimized in order to retain nanometric resolution but still produce

4

sufficient luminescence emission. The resolution is driven by the highly localized excitation, and

5

is influenced by three factors: (i) the spot size of the electron beam, (ii) the spreading of the

6

electrons of the primary beam in the sample (generation volume), and (iii) the diffusion of the

7

generated secondary carriers in the sample.29,31 The electron beam spot size is mostly determined

8

by the beam current (lower electron beam current leads to smaller spot size).31 At a given

9

current, higher acceleration voltage leads to smaller spot size, however, also increases the

10

electron beam interaction volume (see Monte Carlo simulations for ESI, Figure S6), which in

11

turn increases CL generation volume and therefore reduces spatial resolution. Based on the

12

simulations and experiments at different beam voltages as shown later on Figure 2, 2 kV was

13

found to be optimal in terms of signal/noise ratio. At 2 kV, the brightness was more than 10

14

times higher for the RE3+ doped nanocrystals compared to the other labels (Figure 1e). The

15

YVO4:Bi3+,Eu3+ nanocrystals showed the brightest emissions, followed by the Y2O3:Tb3+

16

nanocrystals. While emission could be detected from all the labels investigated, both organic

17

fluorophores bleached almost immediately under the electron beam excitation, even when using

18

a 2 kV electron beam and the lowest possible current where an emission could still be detected

19

(0.8 nA) (Figure 1f). DAPI and CPN lost around 80% of their initial emission intensity within

20

the first 10 s of beam exposure (equivalent to a dose of 0.04 nC/μm2). A decrease in CL emission

21

intensities over time has been widely reported for organic molecules due to structural damage

22

(degradation of the label) by the electron beam.32 While semiconductor quantum dots were more

23

stable under electron excitation compared to organic materials, CdTe quantum dots showed

ACS Paragon Plus Environment

9

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 25

1

reduced emission (- 40%) within the first 10 seconds and - 80% after 60 seconds (equivalent to a

2

dose of 0.24 nC/μm2). Similarly, CdSe/CdS core/shell quantum dots also exhibited a significant

3

drop (30% reduction) within the first 10 s (equivalent to a dose of 0.08 nC/μm2), however, after

4

60 s (equivalent to a dose of 0.48 nC/μm2) 50% of the initial intensity was retained indicating

5

that the composition and architecture of the semiconductor quantum dots plays an important role

6

in their emission stability. A loss in CL emission intensities of quantum dots has been attributed

7

to electron ionization and trapping in the quantum dot structure.33 Despite the wide-spread

8

adoption of semiconductor quantum dots in fluorescence imaging, the limited brightness and

9

electron beam stability restricts their utility in CL bioimaging settings.14 RE3+ doped

10

nanocrystals on the other hand presented comparatively high electron beam stability. The drop in

11

emission intensity within the first 10 seconds (equivalent to a dose of 0.04 nC/μm2) was 10 -

12

20%, showing total reduction of 40% after a dose of 0.48 nC/μm2 over 2 minutes. The minor loss

13

in emission intensity observed for RE3+ doped nanocrystals may be attributed to the

14

accumulation of active quenchers as indicated by the observed film (e.g., hydrocarbons, see ESI,

15

Figure S7, S8).10,34 No evidence in support of label degradation,23,32 thermal effects,35 CL

16

emission saturation (ESI, Figure S9)34 or charge accumulation for RE3+ doped nanocrystals was

17

found in our setting. Charging effects were ruled out by discharging experiments indicating only

18

incomplete recovery of the CL emission signal after sample grounding (see ESI, Figure S7).

19

In order to further illustrate the feasibility of the nanocrystals for CCLEM bioimaging and

20

especially high-resolution mapping, the different labels were incubated with human vascular

21

endothelial cells, chemically fixed with formaldehyde and osmium tetroxide, embedded in epoxy

22

resin, sectioned with a focused ion beam (FIB), and subsequently imaged with scanning electron

23

microscope (SEM). For semiconductor quantum dot containing cells, no characteristic emission

ACS Paragon Plus Environment

10

Page 11 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

by the quantum dots could be detected (data not shown). Both RE3+ doped nanocrystals could be

2

detected inside the osmium-stained cells based on backscattered electrons (BSE) (Figure 2a, see

3

ESI, Figure S10a,d) and the CL signal (using a photomultiplier tube) (Figure 2b-d, see ESI,

4

Figure 10b,e).

5 6

Figure 2: a) Backscattering image of YVO4:Bi3+,Eu3+ nanocrystal-containing human vascular

7

endothelial cells in a focused ion beam cross-section showing electron dense particles and the

8

cellular ultrastructure. The regions of interest for BSE, SE, and CL co-localization studies have

9

been indicated (low magnification, α-α' and high magnification, β-β'). b-d) Corresponding CL

10

images acquired with 2 kV, 5 kV, and 10 kV acceleration voltages indicating strong voltage

11

dependence of the CL resolution. Scale bars: 1 µm. e) Line profiles of SE and CL signals with

12

different acceleration voltages for low magnification (α-α', A) and high magnification (β-β', B)

13

settings. The profiles obtained with low acceleration voltage show the highest correlation

14

between SE, (BSE) and CL signals.

ACS Paragon Plus Environment

11

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 25

1

The cells containing YVO4:Bi3+,Eu3+ nanocrystals were imaged using different acceleration

2

voltages (1, 2, 5, and 10 kV, Figure 2b-d and ESI, Figure S11, S12). The corresponding

3

secondary electron (SE) micrographs (see ESI, Figure S11) present distinct differences in the

4

interaction volume, resulting in less resolved features in the images acquired with higher

5

acceleration voltages from the region of interest (ROI). The difference in the feature resolution is

6

even more pronounced in CL maps (Figure 2b-d). The images acquired with 2 kV show well-

7

resolved features with good signal-to-noise ratio and high correlation between CL and SE images

8

(Figure 2b,e see ESI, Figure S13), while the images collected with both 5 kV and 10 kV show

9

blurry and non-resolvable features (Figure 2c-e). The acquired CL map from regions containing

10

small nanocrystal clusters (see ESI, Figure S14) indicates that emission can also be observed for

11

the smallest nanoparticle features with a resolution similar to SE images and thus illustrates the

12

potential of YVO4:Bi3+,Eu3+ nanocrystals for protein labelling studies. For the YVO4:Bi3+,Eu3+

13

nanocrystals, it is even possible to obtain CL maps with acceleration voltages as low as 1 kV

14

approaching a critical S/N ratio (see ESI, Figure S12). These experimental results indicate a sub-

15

diffraction resolution of < 100 nm for acceleration voltages of ≤ 2 kV and are in excellent

16

agreement with Monte Carlo (MC) simulations in osmium-containing epoxy resin (see ESI,

17

Figure S6).36 The MC simulations illustrate that spatial resolution is strongly influenced by the

18

interaction of the landing electrons and the sample. The CL generation volume in 7% osmium

19

containing cells was approximated by simulating the electron penetration range as a function of

20

acceleration voltage.37 Increased energy of the incident electrons leads to increased electron

21

penetration depth (20, 60, 225, 700 nm for 1, 2, 5, 10 keV respectively) and maximum lateral

22

spread (15, 40, 300, 800 nm for 1, 2, 5, 10 keV respectively), which results in different electron

23

energy loss profiles in the material (ESI, Figure S6).

37

While low energy electrons (< 2 keV)

ACS Paragon Plus Environment

12

Page 13 of 25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

1

mostly excite nanocrystals proximal to the surface, high energy electrons (5 and 10 keV)

2

primarily interact with nanocrystals in deeper regions.

3

features appear whereas others more proximal to the surface disappear when imaging samples

4

with 10 keV compared to 2 keV electron beam (Figure 2b,e). At the same time, 5 and 10 keV

5

electrons show broadening in the lateral in-plane electron energy depth loss profile with

6

increased depth (see ESI, Figure S6c), explaining the progressive blurring of the CL signal with

7

increased acceleration voltage (Figure 2c,d). These results illustrate necessity to reduce electron

8

acceleration voltages to a minimum in a FIB/SEM setting using full-thickness epoxy embedded

9

biological samples. While low beam energies are generally desirable, they come at the expense

10

of lower signals (less energy transfer), introduction of surface effects (non-radiative decay may

11

dominate) and high injection density, which may lead to saturation. As a consequence, both the

12

acceleration voltage and the current should be reduced to a minimum, which in turn requires

13

highly efficient (bright) CL labels, such as YVO4:Bi3+,Eu3+ nanocrystals which can be imaged at

14

voltages as low as 1 kV and currents of ~0.4 nA.

38,39

As a direct consequence, additional

15

Following optimization of the imaging conditions for nanocrystal-containing epoxy-embedded

16

cells, the cells were imaged with low electron acceleration voltage also in electron backscattering

17

mode, which visualizes the cellular ultrastructure, including mitochondria, vesicular and nuclear

18

membranes and various populations of dense particles (e.g., Figure 3a and see ESI, Figure

19

S10a,d, S15). Simultaneous acquisition of corresponding secondary electron images (Figure 3c

20

and see ESI, Figure S10c,f) and cathodoluminescence maps (Figure 3b and ESI, Figure S10b,e)

21

indicates that not all electron-dense particle-like structures in BSE show emission in CL. In order

22

to identify the underlying reason, corresponding secondary electron image (Figure 3d) and

23

energy-dispersive X-ray (EDX) maps from the same region were recorded (Figure 3e, maps

ACS Paragon Plus Environment

13

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 25

1

shown for the YVO4:Bi3+,Eu3+ nanocrystal-containing cells), which reveal that these dense

2

nanocrystal-like structures are devoid of any rare-earth elements and are endogenous granules.

3

Importantly, while EDX has a slightly smaller generation volume than CL for the same

4

acceleration voltage, the minimum energy required for detection of elements with high atomic

5

number, such as europium, in EDX is significantly higher (typically 5-15 kV) than for CL. 40 The

6

comparison of the SE images recorded at the two respective voltages used for EDX (10 kV,

7

Figure 3d) and CL (2 kV, Figure 3c) illustrates the difference in the probing depth of the EDX

8

and CL mapping, translating into different spatial resolution. Therefore, due to the higher

9

acceleration voltage needed for EDX, additional deep-lying features appear on the EDX map

10

originating from a higher signal generation volume. The lower electron beam acceleration

11

voltages needed for CL mapping and the corresponding smaller signal generation volume further

12

illustrate the potential of CL-based high-resolution imaging compared to other competing

13

techniques, especially in a FIB/SEM setting.

14

Similar non-emitting granules could also be observed in samples containing Y2O3:Tb3+

15

nanocrystals (see ESI, Figure S10). In order to exclude quenching of the emitted photons by

16

osmium-containing epoxy resin covering the nanocrystals, cross section containing Y2O3:Tb3+

17

nanocrystals was polished with the focused ion beam and a thin (