SeeDB

Publications

1) Meng-Tsen Ke, Satoshi Fujimoto, Takeshi Imai. “SeeDB: a simple and morphology-preserving optical clearing agent for neuronal circuit reconstruction”. Nature Neuroscience 16, 1154–1161 (2013)  doi: 10.1038/nn.3447

   [PubMed] [PDF] [Supplementary Information] 

   RIKEN press release (English)

   RIKEN press release (Japanese)

   JST press release (Japanese)

   RIKEN RESEARCH

   RIKEN CDB news (Japanese)

2) Meng-Tsen Ke, Takeshi Imai. “Optical clearing of fixed brain samples using SeeDB”. Current Protocols in Neuroscience 66:2.22.1-19 (2014)
[PDF]

3) Meng-Tsen Ke, Satoshi Fujimoto, Takeshi Imai. "Optical clearing using SeeDB". Bio-protocol (2014)

  [HTML version]

TIPs (updated 12-24-2015)

1. Sample preparations

  1-1. sample size: trimming    to allow efficient penetration of SeeDB (up to WD of your objective lens; typically 3-4mm thick)

                                                 Do NOT use very thick samples beyond WD of your objective lens (just useless)

  1-2. agarose embedding    for small and very fragile samples; not recommended for large samples.

  1-3. in utero EP    Do NOT use Fast Green (you can use alexa647 instead).

2. Choice of protocols and conditions

  2-1. neonatal and embryonic brains    SeeDBp

  2-2. young brains (e.g. 3wk)    SeeDB

  2-3. adult brain    SeeDB37 for slice and half brain 

  2-4. adult whole-brain    SeeDB37ht (mild sample expansion and partial quenching may occur) 

3. Microscope and objectives

  3-1. confocal: air vs. water    water-immersion lens is better than air lens; customized lens is not essential.

                                                 oil/glycerol-immersion lenses are also suitable, although WDs are limited.

  3-2. 2P imaging: water vs. Scale vs. customized lens    all are workable, but customized lens for SeeDB works the best.

  3-3. axial calibration for real depth    1.49x for air objective lens; 1.12x (1.49/1.33) for water-immersion lens.

4. Imaging depth possible (adult mouse brain)

  4-1. confocal (water immersion lens)   ~500um at spine resolution; ~1mm at fiber resolution; ~2mm at cellular reslution.

  4-2. 2P (water/scale immersion lens)    ~1mm at spine resolution; 3-4mm at fiber resolution.

5. Sample storage    up to ~1 wk at RT or 37oC (do NOT store in fridge!). We strongly recommend imaging soon after clearing; 

                                prolonged incubation may lead to quenching of dyes and/or accumulation of autofluorescence.

                                For long-term storage, you should restore the sample in PBS.

6. Antibody staining    We could image only up to ~100um depth using standard whole-mount IHC protocol with Triton-X treatment.

                                   Improved antibody staining protocols for SeeDB is also reported (here).

                                   If whole-mount IHC is the primary purpose, you can also chose other clearing agents, such as BABB and TDE.

                                   These clearing agents preserve the fluorescence of Alexa dyes, although fluorescent proteins are quenched.

                                    CLARITY may also be useful for whole-mount immunostaining.

7. Reconstruction software    We are routinely using Neurolucida commercialized from MBF. 

                                             Imaris from Bitplane also supports excellent 3D rendering and neuronal tracing.

                                             Open source software is also avilable (e.g., Vaa3d, neuTube, Simple Neurite Tracer).

                                             We strongly recommend high-spec workstation for reconstruction. 

                                             RAM size should be several times larger than the data volume. Windows 7 Pro supports up up to 192GB.

FAQ (updated 12-24-2015)

Q1. Critical points? Success rate and reproducibility?

    There is no difficult step. However, it is important to ensure efficient penetration of fructose solutions into specimens. Poor clearing occurs when you use too thick samples (such as adult whole brain and rat samples). Because SeeDB is viscous, relatively vigorous shaking on an overhead tube rotator is highly recommended (with ~20ml solution in 50ml conical tube, ~4 rpm). Agarose embedding should be avoided for thick samples. Incubation should be performed at room temperature (25oC or higher; but NOT at 4oC). Because WD of objective lens usually limits the maximum depth (2-3mm), you should not try to use samples much thicker than the WD of your objective lens. Skull, skin, and fibrous tissues are difficult to clear and should be removed beforehand. In general, aged animals are more difficult to clear. For difficult samples, we strongly recommend starting with 1-2 mm thick slices. If preservation of fluorescent protein is not essential in your experiments, you can consider other clearing agents listed here.

Q2. Timing? Shoud I extend incubation time until samples sink in the solution?

    Samples should sink in up to 60% fructose, but may not in higher concentrations. You should use rotator to facilitate penetration of SeeDB into large samples. You can also incubate samples at higher temperatures (e.g., 37oC or 50oC) to facilitate penetration. You should not incubate samples too long (e.g. > 1 days per step), because autofluorescence will accumulate. Prolonged incubation at low conc steps may cause sample expansion.

Q3. Fixatives? 

    We recommend fixation using 4% paraformaldehyde (PFA) in PBS. Transcardial perfusion is recommended if possible. PFA should be freshly prepared. Glutaraldehyde may produce more autofluorescence. Alcohol fixation will quench fluorescent proteins and cause shrinkage of samples.

Q4. Can I put fixed samples directly into the highest conc SeeDB?

    NO. Stepwise incubation (from low conc to high conc) is essential for clearing.

Q5. Temperature?

    We recommend incubation at room temperature (25oC) for standard SeeDB/SeeDBp protocols. SeeDB37 and SeeDB37ht should be performed at 37 and 50oC, respectively.

Q6. Other species? Other organs?

    For insect samples, FocusClear is better. We have tested only mouse brain samples, but SeeDB should be workable for other vertebrate species (e.g., zebrafish) and organs (e.g., liver) to some extent. Fibrous tissues (e.g., heart) are likely more difficult to clear. If you work on large samples, it is important to trim/slice the sample within 3-4 mm thickness (up to WD of your objective lens) to allow for efficient penetration of SeeDB. For example, it is difficult to clear whole rat brains.

Q7. Is a customized lens necessary for imaging?

    Not necessary for confocal microscopy. In the confocal imaging, the maximum imaging depth is 1-2 mm. Within this range, water-immersion lenses provide acceptable resolution.

    In the two-photon microscopy, customized lens is advantageous to image >3 mm depth at high resolution. See Figure 3 of Nature Neuroscience paper for quantitative analysis of resolution.

Q8. Is the customized objective lens available?  new

    The customized objective lenses for SeeDB (and other high index optical clearing agents) have been released from Olympus. XLSLPLN25XGMP (25x, NA 0.9, WD 8mm) covers RI 1.41-1.52.  XLPLN10XSVMP (10x, NA 0.6, WD 8mm) covers RI 1.33-1.52. If you are interested, please contact Olympus.

Q9. Can I image the whole-brain?

    Under the standard SeeDB/SeeDB37 protocol, whole brains can be cleared for up to P21 mice. For older samples, we recommend half brain samples. If the imaging of whole brain is essential for your study, you can use SeeDB37ht protocol, although this may cause mild sample swelling and partial quenching of fluorescent proteins. Also consider the following points.

Q10. Imaging time?

    Imaging time depends on the scanning speed of your microscope and resolution you need. In our imaging setup, it required 6-24 hours to obtain images shown in Supplemental Video 2, 3, 6, 7, 11, 12, 13, for example. Thus, it would require ~1 week to obtain high-resolution images of a whole adult mouse brain (unrealistic at this point).

Q11. Data volume?

    Raw data for Supplemental Video 2, 3, 6, 7, 11, 12, 13 are about 2-20GB. A high-spec workstation is recommended for data analysis. The data volume for a whole adult mouse brain would be 100G-1TB.

Q12. Light-sheet microscopy? 

    We tested Ultramicroscopy from LaVision BioTec. Because light-sheet microscopy does not use a pinhole, it is more sensitive to scattering of emission light. As a result, resolution of the image was not as good as conventional confocal/2P microscopy. If you use Zeiss Light-sheet, samples have to be trimmed to fit the size of a chamber. It should also be noted that Zeiss light-sheet microscopy is not designed for high-index solution such as SeeDB. For example, light sheet optics (e.g., thickness of the light sheet) is suboptimal for imaging SeeDB samples.

Q13. Minimum and maximum thickness of slices?

    You should determine the thickness based on the WD of your objective lens. We recommend slices of up to 3 mm thick. For very thin and fragile slices, you can use agarose embedding.

Q14. Which grade of reagents should I use?  

    We are using >99% D(−)-fructose and >95% α-thioglycerol.

Q15. Can I omit thioglycerol? Can I change the concentration? 

    You can omit thioglycerol for young brain samples. Thioglycerol is recommended for adult samples and SeeDB37/SeeDB37ht protocol, because Maillard reaction is facilitated at higher temperature. 0.5% is recommended for all protocols; higher concentration may cause mild sample expansion. 

Q16. Should I use SeeDB for mounting?

    Yes. You should use SeeDB/SeeDB37 equilibrated with your sample for the sample mounting onto the imaging chamber. Do NOT use freshly prepared SeeDB/SeeDB37 for mounting, because 2P imaging is sensitive to subtle inconsistency in refractive index.

Q17. Can I use SeeDB for immersion of objective lens?

    SeeDB is NOT recommended for immersion because of its viscosity. First, objective lens itself may cause movement artifacts in the viscous solution. Second, imaging of SeeDB samples typically requires long-term imaging. During the imaging, water evaporates from the surface of SeeDB and this causes uneven distribution of refractive index and impair the image resolution. We usually seal the SeeDB samples into a chamber (sandwiched with coverslips) and use water (for water immersion lens), 30% glycerol (for Scale immersion lens), or 80-90% 2,2'-thioglycerol (for SeeDB lens) for immersion. 

Q18. Can SeeDB be combined with EM? 

    Yes, because plasma membrane and ultrastructure remain intact. 

Q19. Minimum fluorescence intensity for single neuron tracing?

    Chemical dyes (e.g., dextran dyes and DiI) are much brighter than fluorescent proteins. For the fluorescent proteins, in utero electroporation and virus (AAV or sindbis) are ideal to express sufficient amount of fluorescent proteins. We could not trace single axons with a knock-in cre reporter line, mT/mG.

Q20. Handling autofluorescence

    Various factors affect autofluorescence. A major cause of autofluorescence is aldehyde-amine compounds. For sample fixation, you should use freshly-prepared PFA/PBS to minimize autofluorescence. Fructose is a reducing sugar and tend to react with various amine, including amino acids (Maillard reaction), producing autofluorescence. Therefore, transcardial perfusion before fixation may reduce the autofluorescence. Autofluorescence may also be produced by the use of additional chemicals (e.g., detergents and salts). FastGreen is commonly used for in utero electroporation, but should NOT be used for SeeDB samples. It remains in the entire brain throughout the life of animals and often produce strong fluorescence after SeeDB treatment. Fructose solutions should not contain additional chemicals, e.g., detergent, because they may also affect autofluorescence. Endogenous autofluorescence may be reduced by pre-treatment with sodium borohydride (NaBH4) or CuSO4. In our experience, two-photon imaging is less sensitive to autofluorescence than one-photon imaging; longer wavelength (e.g., Alexa647) is less sensitive than shorter wavelength. We are developing a new version of SeeDB, which has overcome the autofluorescence issues (unpublished).

Q21. Quenching of fluorescent proteins

    may occur after over-fixation in PFA, prolonged incubation in SeeDB (>several days), or incubation at high temperature. It is best to follow the suggested clearing schedule and perform imaging soon after the clearing (within a few days). Some fluorescent proteins (e.g., tdTomato) are gradually quenched in SeeDB and thus should be imaged immediately after the imaging (within a day). ECFP is not suitable for SeeDB because of quenching. 

Q22. Counterstaining and antibody staining  new

    Permeabilization treatment can improve penetration of nuclear stain and/or antibody (e.g., Triton-X100 or saponin, 0.5-2%). If scattering of DAPI fluorescence is problematic in your sample, NeuroTrace® 640/660 is an alternative choice, because red light scatters much less than blue light and you can expect better image quality. Improved antibody staining protocols for formalin-fixed (iSeeDB) and paraffin-embedded (piSeeDB) tissues are ralso eported. See more information here.

Q23. Combination with CLARITY method  new

    has been published here. In that protocol, SeeDB has been used as an alternative to FocusClear. We, however, do not recommend long-term storage of CLARITY samples in SeeDB.

Questions and feedback are welcome!

Contact: imai.takeshi.457@m.kyushu-u.ac.jp