Xiaowei Zhuang develops advanced imaging techniques, in particular single-molecule and superresolution imaging methods, to study problems of biomedical interest. She invented a superresolution imaging method, STORM, which overcomes the diffraction limit using switching and localization of single fluorescent molecules. She continues to push the envelope of supersolution imaging by increasing the spatiotemporal resolution and enabling in vivo imaging. Recently, she invented a multiplexed, error-robust fluorescence in situ hybridization (MERFISH) method for transcriptome imaging. She applies these methods to investigating a variety of biological problems, including the molecular structures inside neurons and neuronal connectivity in circuits, the three-dimensional organization of chromatin and chromosomes in the nucleus, and the regulation of gene expression.
Elucidating the functional mechanisms of cells, and their dysfunction in diseases, requires tools to dissect the intricate molecular processes present in cells. In particular, we need imaging tools with nanometer-scale resolution and dynamic imaging capability to allow direct visualization of these molecular processes, as well as tools that can simultaneously image a large number of genes to probe how different molecules collective give rise to cellular and tissue functions. My lab develops imaging methods with single-molecule sensitivity, nanometer-scale resolution, and genome-scale capacity to meet these challenges, and applies these methods to problems of biomedical interest. Our current research effort focuses on three main directions: (1) developing superresolution fluorescence microscopy techniques to allow imaging of cells and tissues with molecular-scale resolution and applying these methods to study problems in cell biology and neurobiology; (2) developing transcriptome imaging techniques to allow in situ imaging of numerous RNAs in individual cells, and applying these methods to study the spatial organization of transcriptomes inside cells and the spatial organization of transcriptional distinct cell types in complex tissues, and (3) developing and applying various single-molecule imaging methods to study problems related to gene expression and regulation.
Superresolution Fluorescence Microscopy—Technological Development and Biological Applications
Numerous breakthroughs in biology have been enabled by fluorescence microscopy. The live-cell compatibility and molecular specificity make fluorescence microscopy a particularly powerful tool for biological research. However, the spatial resolution of light microscopy is classically limited by the diffraction of light to several hundred nanometers. This resolution limit is nearly 100 times bigger than the sizes of biomolecules, preventing detailed characterizations of most sub-cellular structures.
Zhuang Research Abstract Slideshow
Figure 1: Stochastic optical reconstruction microscopy (STORM).
A: The principle of STORM. Different fluorescent probes are activated at different times, allowing subsets of fluorophores to be imaged without spatial overlap and to be localized with high precision. Iterating the activation and imaging process allows the position of many fluorescent probes to be determined and a superresolution image to be reconstructed from the positions of the localized probes.
B: Conventional immunofluorescence image of microtubules in a fixed mammalian cell.
C: Two-dimensional (2D) STORM image of the same area.
D–G: Conventional (D, F) and STORM (E, G) images corresponding to the boxed regions in panel B.
Panel A adapted from Rust, M.J. et al. 2006 Nature Methods 3:793–795. Panels B–G adapted from Bates, M. et al. 2007 Science 317:1749–1753.
Figure 2: 3D STORM.
A: Conventional immunofluorescence image of clathrin-coated pits in a fixed cell.
B: The xy projection of the 3D STORM image of the same area.
C: The xy cross-section of the 3D STORM image shows the rims of the pits at the plasma membrane.
D: Magnified image of a single coated pit. Upper panel: xy projection. Middle panel: an xy cross-section near the rim of the pit. Lower panel: an xz cross-section through the middle of the pit.
Panels A–C adapted from Huang, B. et al. 2008 Science 319:810–813. Panel D adapted from Wu, M. et al. 2010 Nature Cell Biology 12:902–908.
Figure 3: A novel periodic membrane skeleton in axons revealed by STORM.
A: 3D STORM image of actin in a dendrite of a cultured hippocampal neuron. The z-coordinates are color-coded according to the color scale.
B: 3D STORM image of actin in axons of hippocampal neurons. Inset: yz cross-sections.
C: STORM image of actin (green) and βII-spectrin (magenta) in an axon. βII-spectrin is immunostained against its C-terminal region, which is situated at the center of the spectrin tetramer.
D: STORM image of βII-spectrin (green) and adducin (magenta) in an axon.
E: STORM image of sodium channels (Nav, green) and βIV-spectrin (magenta) in an axon. βIV-spectrin is immunostained against its N-terminal region, which is situated at the two ends of the spectrin tetramer.
F: A model for the membrane skeleton in axons. Short actin filaments, capped by adducin, form ring-like structures wrapping around the circumference of the axon. Spectrin tetramers connect the adjacent actin/adducin rings, creating a lattice structure with a periodicity of ~180–190 nm.
Adapted from Xu, K. et al. 2013 Science 339:452–456.
Figure 4: STORM imaging of synapses in brain tissue.
A, B: Comparison of conventional (A) and STORM (B) images of synapses immunolabeled for presynaptic protein Bassoon (red) and postsynaptic protein Homer1 (green).
C–E: 3D STORM image of a synapse; xz projection (C), yz projection (D), xy projection (E).
F–H: Variation of receptor distribution and composition among synapses. GluR1 subunit of the AMPA receptor (blue) and NR2B subunit of the NMDA receptor (red) are imaged with Homer1 (green). A synapse with GluR1 distributed near the center of the synapses (F). A synapse with GluR1 distributed primarily in the perisynaptic region (G). A synapse with NR2B primarily distributed in the perisynaptic region (H).
Adapted from Dani, A. et al. 2010 Neuron 68:843–856.
Figure 5: A large-volume, super-resolution imaging and analysis platform.
A: Tissues were dissected, fixed for immunohistochemical labeling, postfixed, dehydrated, and embedded in epoxy resin. Ultrathin sections were cut, arrayed on glass coverslips, and etched to expose fluorophores for STORM imaging. Individual serial sections were imaged and aligned to generate 3D reconstructions.
B: Comparison of STORM maximum projection image of a region containing a dendritic branch of a neuron (left) and the corresponding conventional image (right). Neurite is in blue, gephyrin in green, and presynaptic channel in magenta.
C: En face view (top) and side view (bottom) of the STORM maximum intensity projection of a reconstructed On-Off direction-selective ganglian cell (DSGC) (blue) with associated synaptic gephyrin (green) and presynaptic (magenta) clusters.
Adapted from Sigal, Y. M. et al. 2015 Cell 163:493-505.
Figure 6: STORM imaging reveals the t-loop structure and its TRF2 dependence.
A: STORM image of telomeres prepared by a chromatin-spreading procedure shows the t-loop configuration.
B: Comparison of STORM (upper panels) and conventional (lower panels) images of t-loops.
C: Percentage of telomere ends that are in the looped configuration in control cells and in cells with various shelterin proteins deleted. Only a subset of the t-loops are maintained after the chromatin spreading, since a sizable fraction of the loops are not crosslinked before spreading.
Adapted from Doksani, Y. et al. 2013 Cell 155:345–356.
Figure 7: STORM imaging of chromatin domains in different epigenetic state.
A: Comparison of conventional (left) and STORM (right) image of a chromatin domain labeled by fluorescence in situ hybridization with a library of oligonucleotide probes. (leftsynapses in brain tissue.
B: Log-log plot of the median domain volume as a function of domain length for transcriptionally active (red solid circles), inactive (black solid circles) and Polycomb-repressed (light blue solid circles) chromatin domains. The volume versus domain length follows a power-law scalling with distinct scaling exponent.
C: Two-color, 3D-STORM images of pairs of subdomains within transcriptionally active (left), inactive (middle) and Polycomb-repressed (right) domains. Portions of the two subdomains that overlap in 3D are shown in white.
D: Two-color, 3D-STORM images for neighbouring epigenetic domains with repressed::active and inactive::active boundaries.
Adapted from Boettiger, A.N. et al. 2016 Nature 529:418-422.
Figure 8: Multiplexed, error-robust fluorescence in situ hybridization (MERFISH).
Numerous RNA species can be identified, counted, and localized in a single cell using a single-molecule imaging approach that employs combinatorial and sequential labeling with error-robust encoding schemes capable of detection and/or correction of errors. This highly multiplexed measurement of individual RNAs can be used to compute the expression noise, co-variation in expression among genes, and spatial distribution of genes within single cells.
Adapted from Chen, K. H. et al. 2015. Science 348:aaa6090.
To meet this challenge, we invented a superresolution imaging method, stochastic optical reconstruction microscopy (STORM). STORM overcomes the diffraction limit by exploiting photoswitchable probes to separate spatially unresolvable molecules in time, allowing their positions to be determined with high precision and super-resolution images to be reconstructed from these molecular positions. Using STORM, we originally achieved two-dimensional fluorescence imaging of biological structures with ~20-nm resolution (Figure 1). Next, we extended STORM to all three dimensions and achieved 3D superresolution imaging with ~20-nm lateral resolution and ~50-nm axial resolution (Figure 2). We discovered many photoswitchable dyes with different colors, which enabled multiple-color STORM imaging. We continue to improve the spatiotemporal resolution of STORM through novel optical designs or probe development. Using a dual-objective scheme, we obtained ~10-nm lateral and ~20-nm axial resolution. With a nondiffracting self-bending point spread function, we achieved isotropic 3D resolution. We developed ultrabright photoactivatable dyes, which allowed us to push the resolution to several nanometer. We demonstrated live-cell STORM imaging with 1-sec time resolution. We further improved the time resolution by developing imaging analysis algorithms that allow localization of overlapping molecules.
We have applied STORM to a wide variety of biological systems, ranging from cultured cells to complex brain tissues. These studies have led to the discoveries of previously unknown cellular structures and unveiled previous invisible molecular details of biological structures. Below are several examples.
1. A novel periodic membrane skeleton structure in neurons. Using STORM imaging, we discovered a periodic membrane skeleton structure in axons (Figure 3). In this structure, short actin filaments, capped by adducin, form ring-like structures that wrap around the circumference of axons, and these actin rings are periodically spaced along the axon by spectrin tetramers. This sub-membrane lattice structure organizes ion channels into a periodic distribution along the axons using adaptor proteins such as ankyrin. In addition, we observed that this periodic membrane skeleton structure also forms in dendrites and that it forms in dendrites with a lower propensity than in axons. The relatively high local concentration of spectrin in axons is least in part responsible for the high formation propensity of this periodic structure in axons and Ankyrin-B is an important regulator for setting the polarized distribution of spectin in axons versus dendrites.
Recently, we found the ubiquitous existence of this periodic skeleton structure in a wide variety of animal species, ranging from C. elegans and Drosophila to rodents and H. sapiens, and in many neuronal cell types, including excitatory and inhibitory neurons in the central nervous system, as well as, peripheral neurons such as motor and sensory neurons. The prevalence of this structure in diverse neuronal types and animal species suggests that the actin-spectrin-based periodic membrane skeleton is a key functional component of the axon. Indeed, such a periodic structure formed by actin rings connected by flexible spectrin tetramers provides a robust and flexible support for the axonal membrane, which is important for maintaining the integrity of axons under mechanical stress. The ability of this sub-membrane skeleton to organize membrane proteins, such as ion channels, into a periodic distribution may also impact the generation/propagation of action potentials as well as other signaling pathways at the axonal membrane.
2. Molecular architecture of synapses and synaptic field of neurons. In collaboration with Catherine Dulac (HHMI, Harvard University), we have established STORM as an effective tool to analyze the molecular architecture of synapses with nanometer precision in brain tissues (Figure 4). We have determined the spatial organization of many protein components of the presynaptic active zone and the postsynaptic density, and observed a surprisingly large population of immature synapses with activity-dependent plasticity in the adult accessory olfactory bulb.
Recently, we extended super-resolution imaging to the neural-circuit scale and developed a large volume, STORM reconstruction platform, which allows us to map the synaptic connectivity in neural circuits and determine the molecular identities of synapses (excitatory versus inhibitory, neurotransmitter receptor composition, etc.; Figure 5). Using this approach, we mapped the synaptic input fields of several neuron types and revealed novel organization of synapses in the direction-selective circuit in the retina.
3. Chromatin structure in the nucleus. We used STORM to image the structure of functional and dysfunctional mouse telomeres in collaboration with Titia de Lange (The Rockefeller University). We resolved the telomere loop (t-loop) structure in mouse telomeres and found that TRF2, a shelterin protein, is required for t-loop formation (Figure 6). This finding provides a structural basis to explain how DNA damage response through ATM signaling and DNA repair through nonhomologous end joining are inhibited by TRF2 at the ends of linear chromosomes
Increasing evidence suggest that the 3D organization of the genome is critical for gene expression regulation and other genome functions. We recently demonstrated the ability to image the 3D organization of chromatin with nanometer-scale resolution in collaboration with Ting Wu (Harvard Medical School). We studied the 3D structure of chromatin domains in different epigenetic states, including transcriptionally active, inactive, and Polycomb-repressed states. Our results showed substantial differences in the structural organization of chromatin in different epigenetic states with direct implications on gene regulation, as well as molecular mechanisms that are responsible for the different structural organizations (Figure 7).
We have also investigated other cellular structures, for example, the purinosome-mitochondria interaction in collaboration with Stephen Benkovic (Penn State University), the calcium signaling domain in the sperm tail in collaboration with David Clapham (HHMI, Harvard Medical School), and the molecular architecture of bacteria in collaboration with Sunney Xie (Harvard University).
Single-cell Transcriptome Imaging
The ability to measure, at the transcriptome scale, the copy number and location of RNAs within individual cells in their native context will transform our understanding in many areas of biology. Many outstanding questions in topics ranging from the mechanisms of gene regulation, to the development and maintenance of cell fate, and the organization of distinct cell types in complex tissues could be answered directly with such an approach.
Recently, we developed an in situ transcriptome imaging technique that exploits error-robust combinatorial labeling in conjunction with sequential imaging to determine the precise copy numbers and spatial distributions of numerous RNA species in single cells. We termed this method multiplexed, error-robust fluorescence in situ hybridization (MERFISH) (Figure 8). We demonstrated simultaneously imaging of ~1000 RNA species in single cells using MERFISH. We envision that our approach could be further scaled up to detect tens of thousands of genes – the scale of the whole human transcriptome. Our results revealed distinct sub-cellular distributions of RNAs that correlate with the properties of their encoded proteins and predicted functions for nearly 100 previously uncharacterized genes through gene network analysis.
This new transcriptome imaging method allows quantitative in situ single-cell transcriptomic analysis with minimum perturbation, preserved spatial information, and high detection efficiency and accuracy, which should enable the determination of spatial organization of transcriptomes inside cells and spatial organization of transcriptionally distinct cell types in complex tissues.
Single-Molecule Studies of Molecular and Cellular Processes Related to Gene Expression Regulation
1. Chromatin remodeling. The structure of chromatin is in part regulated by ATP-dependent chromatin-remodeling enzymes, which can be classified into several distinct families. Among these, the ISWI family remodelers promote heterochromatin formation and transcriptional silencing, whereas the SWI/SNF-family remodelers are important for transcriptional activation. We used single-molecule FRET to monitor the remodeling of individual nucleosomes in real time and investigated the remodeling mechanism by ISWI-family and SWI/SNF-family enzymes. Our studies elucidated how DNA is translocated across the nucleosome by these enzymes and the translocation step sizes. We have also revealed how the linker DNA length and histone modifications regulate ISWI-mediated nucleosome translocation to generate evenly space nucleosomes in heterochromatin.
2. Translation imaging in live cells. Translation is under tight spatial and temporal controls to ensure protein production in the right time and place, which is important for cellular adaptation to metabolic states or environmental stimuli and for the establishment of polarized subcellular structures. To better understand the spatiotemporal dynamics of translation regulation, we developed a method to image translation on individual mRNA molecules in real time in live cells. This method is based on multivalent fluorescence amplification of nascent polypeptide signal by the SunTag system, which allows translation events to be visualized directly at the translation sites. Using this method, we monitored translation dynamics both in response to environmental stimuli and in different subcellular compartments, and directly captured transient changes in translation in responses to environmental stresses, distinct mobilities of individual polysomes in different subcellular compartments, and 3’UTR-dependent local translation in neuronal dendrites.
These projects are supported in part by the National Institutes of Health.
As of December 2, 2016