Novel Strategies For Super-resolution Technique

Recently developed super-resolution imaging techniques, such as photoactivated localization microscopy (PALM), fluorescence PALM (FPALM) and stochastic optical reconstruction microscopy (STORM), collectively referred to here as (F)PALM/STORM, are revolutionizing the study of cellular ultrastructures at unprecedented resolutions. After several years of development, this new technique has been applied to every corner of biology. Despite these breakthroughs that increase the resolution to10nanometers,(F)PALM/STORM also faces many techniquechallenges. This thesis introduces several novel strategies in super-resolution techniques in four aspects:1. Developed a series of photoswitchable fluorescent proteins (mGeos-M and mGeos-ES). The green fluorescent protein Dronpa is currently the PSFP (photoswitchable fluorescent protein) of choice for most cell biology applications. However, the poor localization precision potential of Dronpa (26nm) limits its wide application in (F)PALM/STORM experiments. The mEos2, a homologue of Dronpa but an irreversible green-to-red photoconverter, was chosen as an initial template. By site directed random mutation, series of mutations in His62position of mEos2wre performed and found that many of them yielded green species with no conversion ability. Those series of monomer green Eos were renamed as mGeos. The surprising thing is that mGeos could be turned off by488-nm laser light and be reactivated by405-nm laser which is really similar to Dronpa. In (F)PALM/STORM experiments, mGeos-M exhibits the highest number of photons collected per switching cycle and hence the highest localization precision potential of all the green PSFPs. This results reflect that mGeos-M is quite suitable for single-molecule based (F)PALM/STORM technique. On the other hand, mGeos-ES in this family could switch on and off many cycles without bleaching, which is suitable for RESOLFT microscopy like reEGFP.(see chapter3)2. Rational design of true monomeric fluorescent proteins, mEos3.1and mEos3.2. Monomeric (m)Eos2is an engineered photoconvertable fluorescent protein widely used for super-resolution microscopy. We found that mEos2forms oligomers at high concentrations and forms aggregates when labeling membrane proteins The ultracentrifuge analysis also gave us the dissociation constant which was20.21μM for mEos2. To generate a true monomeric mEos2, we solved its crystal structureand rationally designed improved versions. We finally identified two variants, mEos3.1and mEos3.2, with substantially improved brightness with monomeric properties. By fusing different PAFPs with Lifeact, a short peptide that binds actin, we compared their label densities in HeLa cells. mEos3.2gave the highest label density and was approximately threefold higher of that of mEos2, which gave a1.7-fold increase in Nyquist resolution. In conclusion, mEos3.2was a truly monomeric, brighter, faster in maturation, and exhibit higher photon budget and label density in PALM imaging, which will definitely make it popular in future super-resolution imaging,(see chapter4)3. Determination of suitable pairs for two color (F)PALM/STORM imaging. Determination of the relative positions of proteins within cells using multicolor superresolution microscopy is essential for understanding their biological function. We tested many two color pairs and found two excellent ones:mGeos-M/PA-Cherry1and PS-CFP2/mEos3.2. In thus, if you want to get a two-color (F)PALM/STORM imaging simultaneously, we have demonstrated that mGeos-M/PA-mCherry1pair is the best choice now. Both of them could give similar localization precision potential (approximately10nm) with minimal leak through of mGeos into the red channel and virtually no overlap between mGeos-M and PAmCherryl when they were attached to twodifferently localized proteins. If you want to get a two color image with higher spatial resolution, PS-CFP2/mEos3.2(mEos3.2is a green to red FP, so you have to image mEos3.2first and then image the green channel PS-CFP2) pair is the best choice for you.(see chapter3and4)4. Ultrafast, accurate, and robust localization strategy based on artificial neural network. The effective spatial resolution of (F)PALM/STORM depends on localization precision of single molecules. Hence, to improve the resolving ability of (F)PALM/STORM, large amount of efforts had been made to develop localization algorithms of the point spread function (PSF) of single molecular fluorophore. However, the commonly used Gaussian function is not the true PSF, which will produce systematic errors depending on the dipole orientation and the amount of defocus. In thus, we developed an algorithm based on an artificial neural network for detecting the center of PSFs using true PSF model with varies advantages. First, it is the fast fitting algorithm reported till now. Second, compared with Gaussian fitting (such as MLEG and NLLSG) it increased the localization precision of single molecules up to20%in PALM imaging. Third, it could be used to monitor the rotation of single molecules. Fourth, it could expand to3D imaging.(see chapter5)...