UC Berkeley

Visual observation is a powerful way to examine a biological structure. Since the development of the basic microscope, scientists have attempted to get a closer look at smaller and smaller biological systems and molecules. Until recently, the greatest technical challenge to visualization with visible light has been the diffraction limit. With the advent of photoactivatable light microscopy (PALM), the diffraction limit barrier has been circumvented, allowing for improved resolution by an order of magnitude and a much closer observation of cell organelles and individual proteins.

We set out to elucidate the structural properties of one component of the mammalian mitochondrial fission complex: the Drp1 helical ring. The mitochondria are the powerhouses of the cell, with important roles to play in cellular metabolism and apoptosis. The dynamic balance of mitochondrial network fission and fusion are critical for the health of a cell. The Drp1 helical ring is the main dynamic component of fission and is responsible for membrane scission. In order to visualize the structural properties of Drp1 on mitochondrial tubules in situ, we need a technique to visualize with sub-diffraction resolution. Although the sub-diffraction limited nature of PALM microscopy has enabled the observation of individual proteins, the technique's biological relevance has not yet been recognized. A large part of this lack of recognition is likely due to the limited efficiency of currently published green photoactivatable (or photoswitchable) fluorescent proteins and two-color PALM methods. For a fluorescent imaging technique to truly achieve wide use in biology, at least two useful fluorescent protein colors are required. In this thesis, we accomplish two goals: 1) we develop a novel two color PALM method for imaging in mammalian cells and 2) we explore our original biological question and discern the structural properties of the Drp1 helical ring during fission.

We established that mitochondrial membranes can be distinguished with the available photoactivatable fluorescent protein mEos2. However, we were not able to use any of the published photoactivatable and photoswitchable green fluorescent proteins, predominantly because of an inability to identify individual fluorescent events due to rapidity of the photoswitiching. Based on published crystal structures, we created novel Dronpa variants with increasing steric hindrance around the chromophore, likely partially inhibiting the isomerization. We replaced Val157 with isoleucine, leucine, or phenyalanine. DronpaV157F showed no fluorescence and was discarded. DronpaV157I and DronpaV157L showed photoswitchable green fluorescence, with individual fluorescent events that were more easily discerned. DronpaV157L in particular had bright fluorescent events that were well separated when imaged in mammalian cells at 20 Hz. We named this new variant rsKame. Using PALM we successfully imaged rsKame expressed and localized to the mammalian mitochondrial inner membrane.

In order to characterize rsKame, we purified the protein and performed both bulk spectroscopic and single molecule studies. Purified rsKame shared a similar excitation and emission spectra to Dronpa, but had a reduced quantum yield and brightness. Single molecule studies of the kinetic rates of rsKame and Dronpa uncovered the source of the increased separation of the individual fluorescent events of rsKame, and the cause of the original difficulties with the published photoactivatable or photoswitchable fluorescent proteins. We also proposed and validated a kinetic model for the activation and excitation of Dronpa and rsKame. The time spent in the dark state, T_OFF, of rsKame was significantly increased as compared to Dronpa, indicating that the new variant spends an increased amount of time in the trans (OFF) isomerization of the chromophore. This allows for increased temporal separation of the individual fluorescent events, resulting in a super resolution PALM image. In addition, we observed photoswitching of individual Dronpa and rsKame in the absence of 405 nm activation, demonstrating that Dronpa and rsKame can be both activated and excited by 488 nm. In addition, the cis to trans isomerization of the Dronpa/rsKame chromophore was unaffected by increasing 405 nm and 488 nm laser power densities, demonstrating that the chromophore either relaxes into the trans configuration (dark state) spontaneously or is affected by an unknown variable. We also performed single molecule studies on PAmCherry1 and quantified the photoactivation rate. We noted that under the most intense activation of rsKame, less than 2.5% of PAmCherry1 molecules would be activated.

With the novel photoswitchable fluorescent protein, rsKame, available, we returned to the development of a novel two color PALM method. We chose PAmCherry1 as the partner for rsKame since PAmCherry1 has distinct and well separated excitation/emission spectra from rsKame and is not activated by low 405 nm laser power density. We first imaged rsKame with 405 nm activation at (0.61 mW/mm²) and 488 nm activation/excitation (5.87 W/mm²) to completion. We then imaged PAmCherry1 with increasing 405 nm activation (0.6-6.0 W/mm²) and 561 nm excitation (22 W/mm²). With the novel PALM imaging method, we labeled the inner and outer mitochondrial membranes with large populations of membrane bound rsKame and PAmCherry1 in HeLa and EpH4 cells. We were able to observe and clearly differentiate the two mitochondrial membrane structures and their various morphologies in situ.

With the functional two-color PALM method, we returned to our original investigation of the Drp1 fission ring in situ. In fixed HeLa cells, we continued to label the outer membrane with PAmCherry1 and fused rsKame to the N-terminus of Drp1, separated by a linker. The resultant PALM images allowed for the observation of two previously observed and one hitherto unseen distinct Drp1 morphologies: Constrict, Terminal, and Split. The Constrict morphology was defined as the Drp1 structures that clearly encircle the mitochondrial tubule at various stages of membrane constriction. The Terminal morphology was defined as the Drp1 structures found at the termini of mitochondria, presumably post membrane scission. The Split morphology is a novel morphology and was defined as two Drp1 foci flanking the mitochondrial tubule but not completely encircling it. Quantification of the diameter and length of the Drp1 helical ring structures showed that the mean length of the Drp1 helical rings was consistent between all three morphologies, though a slight decrease was observed for the Terminal morphology, likely due to degradation. We observed a decrease of approximately 40 nm between the Constrict and Terminal mean diameters, consistent with a dynamic change in the Drp1 ring size due to membrane constriction towards membrane scission during mitochondrial fission. The Split morphology had a wide distribution of diameters and warrants further study.

Our earlier difficulties with the large cytoplasmic population Drp1 led to the development of bifurcated complementation PALM imaging (BiFC PALM). We established that PAmCherry1, Dendra2, and rsKame could be bifurcated, and upon complementation, maintain both their fluorescence and photoactivatability or photoswitchability. In ongoing work, we label two subunits of the ATP synthase stator section with the bifurcated halves of PAmCherry1 and attempting to demonstrate that, upon co-expression and complementation, the ATP synthase enzyme complexes along the mitochondrial inner membrane can be PALM imaged. BiFC PALM is a novel technique that still requires significant controls and improvements. However, BiFC PALM has the potential to solve difficulties with determining protein-protein co-localization. Complementation is dependent on the linker length separating the bifurcated half and the protein of interest, and gives relative localization of two proteins to each other. Combining the relative localization of BiFC and the absolute localization of PALM defines the clear interaction of two proteins within the resolution limits of a single PALM fluorescent event. This innovation could make PALM an even more powerful tool for biological investigations.

Over the course of this thesis, we developed and characterized a novel photoswitchable green fluorescent protein, rsKame, and a novel two-color PALM method to accompany it. With the new method, we examined the structural properties of Drp1 helical rings during mitochondrial fission in situ and are developing a method, BiFC PALM, for definitively determining protein-protein co-localization with PALM imaging. Our studies have spanned basic molecular biology, biochemistry, microscopy, and single molecule photophysics.