Live cell imaging and confocal microscopy Laurent Gelman Facility for Advanced Imaging and Microscopy Friedrich Miescher Institut, Basel

Acknowledgments Preparation of the lecture -

Arne Seitz (EPFL)

-

Stefan Terjung (EMBL Heidelberg)

-

Jens Rietdorf (FMI / Zeiss, Munich)

Data -

Rico Kunzmann (group A. Peters, FMI)

-

Vincent Dion (group S. Gasser, FMI)

-

Julia Kleylein-Sohn (Novartis)

-

Flavio Donato (group P. Caroni, FMI)

-

Gerald Moncayo (group B. Hemmigs)

-

Jérôme Feige (UNIL / Novartis)

Overview When performing live cell microscopy, cell viability must be at the forefront of any measurement to ensure that the biological processes that are under investigation are not altered in any way. Growing cells under a microscope… – Physical integrity – Attachment, immobilization of the sample – Incubation systems (temperature, gases, humidity, pH control) Imaging living cells – Autofluorescence – Photodamage – Microscope settings (speed, laser power, pinhole, voxel size…) – Microscope setups: spinning-disk versus single-beam LSM and wide-field microscope.

Mounting: Physical Integrity

Attachment, Immobilization Cells usually adhere better on plastic (treated) but can be also plated on glass. Coating the coverglass / Substrate Poly-L-Lysine Fibronectin Concanavalin A Other…

0.5 - 1% LMP agarose CyGel 1.5 - 3% methyl cellulose Silicon oil, Halocarbon oil Transparent films Some dishes are made of a special plastic with low autofluorescence where cells adhere well and allowing the use of immersion objectives.

(www.ibidi.de) (www.willcowells.com)

Temperature Small heating stages + Objective heater

Temperature control (on microscope stage) Temperature controller for slides (www.embl-em.de)

Multiwell stage-top incubator (OKOlab, www.oko-lab.com)

+ Relatively cheap - System temperature not constant (focus drift possible!)

Focus Drift Constant temperature of the microscope is also important for time-lapse imaging:

Focus drift due to material extension/shrinkage due to temperature changes

Microscope incubators (www.lis.ch) (www.oko-lab.com) (www.embl-em.de)

Advantages • Constant conditions (10 – 40 °C) • Stabilized focus • Atmosphere controllable (CO2 …) Disadvantages • Expensive • Microscope access impaired

Gases, pH, Humidity, Osmolarity Physiological pH during imaging is required! Without CO2 control (short-term): • HEPES buffered medium instead of carbonate e.g. 30mM HEPES, 0.5g/l Carbonate instead of 2.2g/l Carbonate. • Seal the sample chamber, e.g. with silicon (no gas exchange) With CO2 control (long-term): • Use perfusion chambers • Use incubators

Gases, pH, Humidity, Osmolarity Humidity control during long-term experiments: • Osmolarity of the medium stays constant • sample doesn´t run dry Possible solutions: • Seal the sample chamber, e.g. silicon (gas exchange ?) • Use perfusion chambers • Use incubators with humidity control • Use silicon oil

HEPES Buffered Media

Advantages • Open system, easy to manipulate. • Easy to handle and control.

Disadvantages • Usable for ~ 1 hour. • Toxicity of HEPES with some cells. • Evaporation.

Sealed chamber

Advantages

Disadvantages

• Easy to handle and control.

•

• No evaporation.

• •

• Cheap (reusable).

Usable for max 3 hours depending on volume. No manipulation. Mounting of the chamber might be tedious.

Gas Permeable Chamber (www.ibidi.de)

Advantages • High optical quality (DIC) • Gas permeable • Water impermeable • Perfusable • Usable for days. Disadvantages • No direct access to cells

Perfusion chambers

Advantages

Disadvantages

• Constant conditions.

• •

• Manipulation of media. • Usable for days. • No evaporation problems.

Hard to assemble and control. Shear stress possible.

“scratch” or “would healing assay” Mammalian cells. Wide-field automated inverted microscope. 2.5x objective Time-lapse: 50 points, 1 stack every 20min (16.5 hours total).

Courtesy of Gerald Moncayo (Hemmings lab, FMI)

Overview Growing cells under a microscope… – Physical integrity – Holders, attachment – Incubation systems (Temperature, Gases, Humidity, pH…) Imaging living cells – Autofluorescence – Photodamage – Microscope settings: speed, laser power, pinhole, voxel size… – Microscope setups: Spinning-disk versus single-beam LSM and wide-field microscope.

Choice of the imaging setup Which microscope is the most appropriate for a live cell experiment?  Go more for sensitivity than resolution!

Autofluorescence Specific sources of autofluorescence – Aromatic amino acid residues (UV) – Reduced pyridine nucleotides (UV) – Flavins (UV, blue) – Chitin (broad) – Yolk – Chlorophyll (blue, green) – Phenol red indicator  use ‘Imaging medium’ ! General sources of autofluorescence – Dead cells (broad) Cures – Longer wavelengths – Avoid stress – Use imaging medium (no phenol red) – Choose excitation and detection windows properly

Photodamage and photobleaching With live-cell microscopy, there must be a compromise between acquiring beautiful images and collecting data that provide a high enough signal-to-noise ratio (S/N) to make meaningful quantitative measurements of a living specimen. A lot of light, not absorbed by the fluorophore, can also potentially trigger chemical reactions in the sample by bringing directly energy to molecules or heating up locally the sample. Fluorescent molecules used for the labeling of the sample can themselves cause deleterious effects.

Triplett state: • chemical reactive • radicals • bleaching / phototoxicity

Example: Be very careful when using DNA intercalating agents, such as DAPI or Hoechst!

Recognize damaged cells - Cells detach and round up - Bulges form (“Blebbing”) - Mitochondria swell and/or get isolated - Large vacuoles appear - Cells do not make it through mitosis - Any sign of necrosis or apoptosis

Monitoring of cellular metabolic activity can be monitored for example with AlamarBlue (Invitrogen).

Avoid Photodamage - Minimize illumination power, tolerate more noise and less S/N ratio - Optimise detection: Filtersets Detectors Resolution (x, y, z, t, intensity value, channels) - Use decent dyes (high quantum efficiency). - Add antioxidants (Trolox, ascorbic acid 2mg/ml)

Is max resolution needed for my live cell experiment? The brightness of an image varies directly as the fourth power of the NA but inversely as the square of the magnification. Relative Brightness of objective lenses NA4/M2

Detector

40x/1.3

20x/0.8 10x/0.45

63x/1.4

25x/0.8 100x/1.3

5x/0.15

Sample Magnification (M)

Increasing the number of pixels or increasing magnification decreases the intensity in each pixel  Resolution is increased only as long as the signal-to-noise ratio is good

Resolution, pixel size, binning Increasing the number of pixels decreases the intensity in each pixel, so resolution is increased only as long as the signal-to-noise ratio is good (and optical resolution not limiting). N.B.: Decreasing pixel size by two decreases signal-to-noise ratio by 2 also! (see also later section on noise). Binning pixels increases intensity but decreases resolution. Exposure time is also decreased (as well as file size)  Interesting for live cell experiments. No binning

Binning 3x

Binning, contrast and resolution What are my object size and shape? I can’t be sure… Unless I do binning ! NB: “binning” is achieved on a CLSM by reducing the number of pixels while keeping the field of view constant (“Frame size” 512X512  256x256).

No binning

Binning 2x2

Zeiss Z1, 100x/1.4

Binning, contrast, speed and photodamage Mouse brain section, fixed section, Alexa488 staining, Zeiss Z1, 20x/0.8 Binning 2x2, exposure time 10ms

No binning, exposure time 320ms

Playing on pixel size (binning) and contrast (do not use full dynamic range/well capacity) allowed here to reduce by 32 the exposure time  more time for Z-stacks and less photodamage. Courtesy of Flavio Donato (group P. Caroni, FMI)

Is max resolution needed for my live cell experiment? Ex: FRAP experiment GFP-labeled protein in the paternal pronucleus of a mouse zygote. LSM710 Zeiss microscope, 40x/1.3 objective. Time-lapse: 250 points, 1 frame every 134msec. Pinhole ～1.5 AU

Region bleached for FRAP

Courtesy of Rico Kunzmann (Group A. Peters, FMI)

Refraction around the coverslip

NA and immersion medium The higher the refractive index of the immersion medium, the higher the NA…

air, n = 1

oil, n = 1.51

The higher the NA, the brighter the sample and the higher the resolution…

Immersion medium and mounting medium refractive indexes The refractive index of the immersion medium must fit that of the mounting medium when objects are located far away from the coverslip.

oil, n = 1.51 Thick tissue slice n < 1.51

Objective numerical aperture - definition Moss in water

Apochromat 63x/1.3 glycerol immersion

Apochromat 63x/1.2 water immersion

(Courtesy of Nathalie Garin, Leica)

Working distance The working distance is defined as the distance (in millimeters) from the front lens element of the objective to the closest surface of the coverslip when the specimen is in sharp focus. In most instances, the working distance of an objective decreases as magnification increases.

Principle of Confocal Laser Scanning Microscopy (CLSM) PMT Pinhole Emission Filter Dichroic Laser

Objective

Specimen focal plane

Laser beam shape and photobleaching

Laser beam shape

confocal microscopy

multi-photon microscopy

focal plane

Photobleaching pattern

coverslide focal plane

Scanned area

Scanned area Reduced phototoxicity but heating problems and higher photobleaching may occur due to high laser power.

Principle of Confocal Laser Scanning Microscopy (CLSM) PMT Pinhole Emission Filter Dichroic Laser

Objective

Specimen focal plane

Principle of Confocal Laser Scanning Microscopy (CLSM) The acquisition speed depends on - pixel dwell-time (time spent per pixel) - number of pixels

PMT Pinhole

tone frame > number of pixel * pixel dwell time

Emission Filter Dichroic Laser

Objective

Specimen focal plane

Example 1: x*y format = 512 * 512 pixel dwell time = 2.16 μs t > 512 * 512 * 2.16 = 0.57 s Example 2: with x*y format = 1024*1024 pixel dwell time = 3.5 μs average = 4 t > 1024 * 1024 * 305 * 4 = 14.7 s  Scanning is a slow process !!  To go faster, parallel acquisition of several pixels is required (spinning-disk microscope)

Live Cell Confocal Settings • Avoid averaging, but if it is necessary then use line average instead frame average .

Line average

Frame average

• Use line-by-line sequential instead of frame-by-frame sequential • Try to acquire channels in parallel and not sequentially

Line sequential

Frame sequential

• Minimize laser power, compensate by: – Increased sensitivity (higher gain)  more noise, process the image after acquisition (see later) – Opening pinhole  increased optical thickness – Less pixels  larger voxel size (x, y, z) or scan every second line.

Parallel versus single point acquisition

Nipkow Disk

Optical Path Configuration in CSU22

Comparison CLSM, SD and wide-field systems

.

… PMT

Camera chip

Laser

Pinhole

Camera chip

Laser

Hg Bulb

Higher sensitivity in SD/WF due to (EM-)CCD cameras

- NB: Confocal microscopes become more and more sensitive though, owing to a new generation of detectors: the GaAsP detectors (“BIG” for Zeiss, “HyD” for Leica).

- Read-out noise can be lowered when using lower read speeds of the camera.

Signal to noise ratio (S/N) S/N =

Signal √ (read noise)2 + (dark noise)2 + (Shot noise)2 noise dominated by photonic component

S/N

noise dominated by read and dark components

Number of photons

Andor catalogue (2006)

Duplication of supernumerary centrosomes Single-beam scan

mitotic DdCP224-GFP/GFP-histone2B cells

Multi-beam scan

Comparison of signal to noise ratio between LSM and SD Single-beam LSM scan

Multi-beam SD scan

SD have weak illumination powers

Imaging limitations: available signal (Laser) excitation is not the only limiting factor, fluorophore saturation decreases overall signal 1

Saturation

0.8

Extended Jablonski diagram

0.6

0.4

0.2

0 0.00

0.05

0.10

0.15

0.20

0.25

Excitation photons x1026

0.30

0.35

Saturation and triplet state in single- and multi-beam scanning 1mW 488nm, 40x 1.25 NA

Multi-beam scan

Single-beam scan Triplet population at steady state

Fluorophore saturation

1.0

Multi-beam scan Single-beam scan

0.8 0.6 0.4 0.2 1E20 1E21 1E22 1E23 1E24 1E25

Intensity (photons cm-2s-1)

1.0 0.8 0.6 0.4 0.2 1E20 1E21 1E22 1E23 1E24 1E25

Intensity (photons cm-2s-1)

Intensity levels in single and multi-beam scanning

Beams Illumination Emission rate Fluorophore saturation Detection efficiency Overall efficiency

1 1 mW 1.26 x 108 photons/sec 63% 0.25 1

Beams Illumination Emission rate Fluorophore saturation Detection efficiency Overall efficiency

> 1000 0.4-0.6 µW 1.72 x 105 photons/sec 0.09% 0.9 4

Model calculations based on scanning fluorescein at 1 mW 488 nm excitation with a 40x/1.25NA objective, from Live Cell Spinning Disk Microscopy, Ralph Gräf , Jens Rietdorf & Timo Zimmermann, Adv Biochem Engin/Biotechnol (2005) 95: 57–75

Light-dependent photobleaching? “Photochemical damage is one of the most important yet least understood aspects of the use of fluorescence in biology; in this discussion we can do little more than define our ignorance.” Roger Tsien (Nobel Prize for Chemistry 2008)

Triplett state: • chemical reactive • radicals • bleaching / phototoxicity

Consequences of a not adjustable pinhole size in a SD N.A. MAG λ Resolution

→ → ↘ ↗

↘ → → ↘~

→ ↘ → ↘

Comparison LSM vs. Spinning-Disk confocal Property

Single-beam scanning confocals

Multi-beam scanning confocals

Wide-field setup

Acquisition

point by point slow

Numerous points fast

All frame at a time very fast

Detection

Photomultiplier Tube low sensitivity

CCD/EM-CCD Camera high sensitivity

CCD/EM-CCD Camera high sensitivity

Multichannel imaging

Parallel/Sequential = fast

Sequential/(Parallel) slow (time x nb of channels)

Sequential/(Parallel) slow (time x nb of channels)

Region bleaching

Integrated

Add-on

Add-on

Optical sectioning

Adjustable pinholes Flexible a lot of light is lost No crosstalk between scan beams

Fixed pinhole size inflexible/suboptimal a lot of light is lost Crosstalk between scan beams

No sectioning all light collected Possibility to do TIRF

Image processing in live cell imaging Very often, images are noisy, as low laser powers are used, and/or short exposure times, and/or fluorescent proteins with low quantum yields. Deconvolution or denoising can often improves dramatically image quality.

Image processing in live cell imaging: PureDenoise plugin PureDenoise is a plugin in ImageJ, Florian Luisier at the Biomedical Imaging Group (BIG), EPFL,

Switzerland, http://bigwww.epfl.ch/algorithms/denoise/

Mammalian cells, GFP-labeled microtubule-organizing centre (MTOC) SD-microscope, 40x/1.3 objective, exposure time: 100ms per frame. Z-stack: 81 planes, dist=500nm. 17’010 frames within 7 hours Time-lapse: 210 points, 1 stack every 2 min. After acquisition, images are denoised with PureDenoise over time and MIPs over Z-dimension are generated. After denoising

Before denoising

Courtesy of Julia Kleylein-Sohn (Novartis)

Image processing in live cell imaging: Deconvolution YFP-labeled locus in Yeast nucleus. SD-microscope, 100x/1.45 objective, exposure time: 30ms per frame. Z-stack: 14 planes, dist=200nm. 2’814 frames within 5 minutes Time-lapse: 201 points, 1 stack every 1.5 sec. After acquisition, images are deconvolved with Huygens professional (www.svi.nl) and MIPs over Zdimension are generated. After deconvolution

Before deconvolution

Courtesy of Vincent Dion (Gasser lab, FMI)

Working with fluorescent fusion proteins: be careful! - Check expression level sand degradation A-ECFP ng of vector transfected 10 20 30 40

- Check the functionality of the fusion protein… Adjusting expression levels to those of the endogenous protein!

B-EYFP EYFP-B 10 20 10 20

Working with fluorescent fusion proteins: be careful! - Look for artificial clustering of the over-expressed proteins YFP-PPARα

End