Skip header information|
Wadsworth Center Home - Science in the Pursuit of Health|
NY.gov Portal State Agency Listing
Wadsworth Center, Department of Health, New York StateD Wadsworth Center New York State Department of Health
Main Body

Investigators and Program Directors

image of Haixin Sui

Haixin Sui

Research Scientist,
Wadsworth Center, Cellular and Molecular Basis of Diseases

Assistant Professor,
School of Public Health, Biomedical Sciences, Structural and Cell Biology

Ph.D., Dalian University of Technology, China (1996)
Postdoctoral training, Lawrence Berkeley National Laboratory

Email: hsui@wadsworth.org


Research Interests Continued

Cryo-EM study of primary cilia

Primary cilia are membrane-enveloped microtubule-based projections on the apical surface of cells. They extend to extra-cellular environment when cells exit cell cycles, and resorb as cells re-enter cell cycles. Research results in the past decade have concluded that primary cilia are "antenna" collecting environmental signals for control of cell proliferation. In some human organs, primary cilia on the epithelial cells bend reversibly and serve as the mechanosensors to liquid flow. The elastic bending is believed to modulate Ca2+ influx across the ciliary membrane, thereby regulates epithelial cell proliferation and maintains normal architecture of mature tissues. Defects related to this mechano-signaling pathway are the primary cause for cystic diseases in the kidney, liver, and pancreas. Currently, fundamental knowledge about primary cilia is still missing despite significant progress relating them to many physiological processes. In order to determine the major component complexes responsible for structural architecture, assembly and mechanical properties of primary cilia, we have been developing methods to characterize primary cilia by coordinated studies using light microscopy and electron microscopy.

Primary cilia of LLC-PK1 (Pig Kidney Epithelial Cells): (a) Morphology of primary cilia of LLC-PK1 cells with GFP-fusion SSTR3 under DIC, and (b) florescence imaging. (c) Electron micrograph of a vitreously frozen primary cilium.
( + enlarge image)
 Primary cilia of LLC-PK1 (Pig Kidney Epithelial Cells): (a) Morphology of primary cilia of LLC-PK1 cells with GFP-fusion SSTR3 under DIC, and (b) florescence imaging. (c) Electron micrograph of a vitreously frozen primary cilium.


Back to Top

Structures of various types of microtubules

Microtubules are quite stiff in mechanical property measurements but are often observed to bend with high curvature in cells. Using cryo-electron microscopy, we investigated the molecular basis of their mechanical properties that are well adapted to their various roles in the cell. The resultant high quality maps provided structural insight into the lateral deformation and bending property of microtubules.

Density maps of different structural types of microtubules containing protofilament numbers of 11 (pink), 12 (orange), 13 (yellow), 14 (green), 15 (blue), 16 (purple). These density maps were reconstructed from cryo-electron micrographs of microtubule mixtures obtained by polymerization of purified tubulin. They clearly revealed secondary structural features. The background image represents the typical morphology of the frozen-hydrated microtubule mixture with microtubules of various diameters.
( + enlarge image)
 Density maps of different structural types of microtubules containing protofilament numbers of 11 (pink), 12 (orange), 13 (yellow), 14 (green), 15 (blue), 16 (purple). These density maps were reconstructed from cryo-electron micrographs of microtubule mixtures obtained by polymerization of purified tubulin. They clearly revealed secondary structural features. The background image represents the typical morphology of the frozen-hydrated microtubule mixture with microtubules of various diameters.


Inter-protofilament interactions are conserved in microtubules with various diameters. Flexible loop-loop interactions enable microtubule lateral deformation and bending.
( + enlarge image)
 Inter-protofilament interactions are conserved in microtubules with various diameters. Flexible loop-loop interactions enable microtubule lateral deformation and bending.


Movie: Sequential morphing of the structural models for MT 11-3, MT 12-3, MT 13-3, MT 14-3, MT 15-4, and MT 16-4
( + watch movie clip)
Windows Media Format
Movie: Sequential morphing of the structural models for MT 11-3, MT 12-3, MT 13-3, MT 14-3, MT 15-4, and MT 16-4 .


These density maps clearly resolved secondary structural features and enabled docking of the tubulin crystal structure to produce pseudo-atomic models. Analysis of the models suggests the structure surrounding the connecting loops contributes to microtubule rigidity, and seams have little or no effect on the mechanical properties of microtubules.
( + enlarge image)
 These density maps clearly resolved secondary structural features and enabled docking of the tubulin crystal structure to produce pseudo-atomic models. Analysis of the models suggests the structure surrounding the connecting loops contributes to microtubule rigidity, and seams have little or no effect on the mechanical properties of microtubules.


Back to Top

Molecular architecture of axonemal microtubule doublets

Microtubule doublets are structural components of axonemes in eukaryotic cilia. In motile cilia, they usually found in arrays of nine doublets arranged around two singlet microtubules. Coordinated sliding of adjacent doublets, which involves a host of other proteins in the axoneme, produces periodic beating movements of the axoneme. Both singlet and doublet microtubules are composed mainly of tubulin, but the doublets also incorporate a number of other structural protein components. Knowing how the doublets are constructed is essential to understanding the mechanical properties and interactions with other proteins that lead to movement.

We have obtained a 3D density map of intact microtubule doublets using cryo-electron tomography and image averaging. Our map, with a resolution of about 3 nm, provides insights into locations of particular proteins within the doublets and the structural features of the doublets that define their mechanical properties.

The receding gray-scale image is one of the images in one tomographic tilt series. The image just below is a vertical section through the 3-D reconstruction generated from this series. The density in yellow is the resultant 3D maps of microtubule doublets with tubulin structural model fitting into it. The density also revealed 8% lateral distortion to the component microtubules
( + enlarge image)
 The receding gray-scale image is one of the images in one tomographic tilt series. The image just below is a vertical section through the 3-D reconstruction generated from this series. The density in yellow is the resultant 3D maps of microtubule doublets with tubulin structural model fitting into it. The density also revealed 8% lateral distortion to the component microtubules.


Based on the density map of microtubule doublet, 8% lateral deformation is introduced to the structural model of microtubule, with distortions limited at the loop-loop connection. The blue and red arrow point to the different packing environment of the neighboring protofilaments. The distortion model was validated by the high-resolution structural study of various types of microtubules (Sui et al 2010).
( + enlarge image)
 Based on the density map of microtubule doublet, 8% lateral deformation is introduced to the structural model of microtubule, with distortions limited at the loop-loop connection. The blue and red arrow point to the different packing environment of the neighboring protofilaments. The distortion model was validated by the high-resolution structural study of various types of microtubules (Sui et al 2010).


End-on view of the difference density map highlighting microtubule associated proteins in the microtubule
( + enlarge image)
 End-on view of the difference density map highlighting microtubule associated proteins (in yellow).

Proteins densities other than tubulins are clearly revealed in the microtubule doublets. Inside the A-tubule, intraluminal microtubule associated proteins are visualized on the intraluminal wall of microtubules, where inter-protofilament interactions locate for lateral integration of the microtubule.
( + enlarge image)
 Proteins densities other than tubulins are clearly revealed in the microtubule doublets. Inside the A-tubule, intraluminal microtubule associated proteins are visualized on the intraluminal wall of microtubules, where inter-protofilament interactions locate for lateral integration of the microtubule.


Density of nexin links containing associated doublets. (a) Projection view of a 42nm-thick cross-section of a denoised cryo-tomogram. Densities can be seen linking several pairs of the doublets (arrows). (b) Projection images of seven 21nm-thick cross-sections, extracted from the tomogram shown in (a), were averaged to enhance the linkage density that is interpreted as nexin. Red circles indicate PF positions and indicate that the linker connects PF B2 to the region of A9-A11. Several of the PFs are labeled for reference.
( + enlarge image)
 Density of nexin links containing associated doublets. (a) Projection view of a 42nm-thick cross-section of a denoised cryo-tomogram. Densities can be seen linking several pairs of the doublets (arrows). (b) Projection images of seven 21nm-thick cross-sections, extracted from the tomogram shown in (a), were averaged to enhance the linkage density that is interpreted as nexin. Red circles indicate PF positions and indicate that the linker connects PF B2 to the region of A9-A11. Several of the PFs are labeled for reference.


Schematic of proteins associated with microtubule doublets. Locations of tektins, Sp77, Sp83 and nexin are from our studies, locations and shapes of dyneins, the dynein regulatory complex and spokes31, 32 are based on publications of others.
( + enlarge image)
 Schematic of proteins associated with microtubule doublets. Locations of tektins, Sp77, Sp83 and nexin are from our studies, locations and shapes of dyneins, the dynein regulatory complex and spokes are based on publications of others.


Back to Top

Structural basis of water specific transport through Aquaporin 1

Water channels facilitate the rapid transport of water across cell membranes in response to osmotic gradients. These channels are believed to be involved in many physiological processes that include renal water conservation, neuro-homeostasis, digestion, regulation of body temperature and reproduction. Members of the water channel superfamily have been found in a range of cell types from bacteria to human. In mammals, there are currently 10 families of water channels, referred to as aquaporins (AQP): AQP0 - AQP9. We determined the structure of the water specific channel aquaporin 1 (AQP1) to 2.2Å resolution. The channel consists of three topological elements, an extracellular and a cytoplasmic vestibule connected by an extended narrow pore or selectivity filter. Within the selectivity filter, four bound waters are localized along three hydrophilic nodes. This structure offered a great deal of structural insight into the mechanism of water specificity for water channel proteins.

AQP1water channel (left: PDB#1J4N, Sui et al. 2001) consists 6 transmembrane helices and two loop-helix motifs as high-lighted. It is a membrane channel protein only permeable to water molecules. The water specificity is achieved due to the structural feature of the restriction region that is close to the upper mouth and the partially polar environment in the channel pore along the loops of the loop-helix motifs. A NPA-NPA interaction between the two motifs at the center of the channel pore stabilizes the architecture and prevent protons to leak through. In fact, the loop-helix motifs are present in a number of membrane channel proteins. A different configuration of loop-helix motifs in KcsA potassium channel (right: PDB#1BL8, Doyle et al 1998) forms the channel pore in the middle of four loops and enables its ion
( + enlarge image)

( + watch Windows Media movie clip)

AQP1water channel (left: PDB#1J4N, Sui et al. 2001) consists 6 transmembrane helices and two loop-helix motifs as high-lighted. It is a membrane channel protein only permeable to water molecules. The water specificity is achieved due to the structural feature of the restriction region that is close to the upper mouth and the partially polar environment in the channel pore along the loops of the loop-helix motifs. A NPA-NPA interaction between the two motifs at the center of the channel pore stabilizes the architecture and prevent protons to leak through. In fact, the loop-helix motifs are present in a number of membrane channel proteins. A different configuration of loop-helix motifs in KcsA potassium channel (right: PDB#1BL8, Doyle et al 1998) forms the channel pore in the middle of four loops and enables its ion


Back to Top

Method of single particle reconstruction for multiple structural types of microtubules

The image processing method followed a single-particle strategy. The programs are developed using customized scripts written in the SPIDER package (Frank et al., 1996), in combination with PYTHON. This program package has produced high-quality density maps of various types of microtubules (Sui and Downing, Structure, 2010)<br/><br/> Image boxes, each containing about 10 dimer repeats, were extracted along microtubules in the digitized micrographs. Boxes from each microtubule were treated as a group and processed as described previously (Li et al., 2002), with some modifications as detailed in the main text. The power spectra of selected image boxes should show the layer-lines. As described below, all extracted images were aligned parallel to a box edge, using the Radon transform of power spectra described by Li et al.in 2000 (shown in the right panel). The rotationally aligned image boxes are then subjected to supervised classification (shown in the left panel). For each structural type, the iterative helical real space reconstruction strategy (IHRSR), was applied to refine the symmetry parameters and reconstruct density maps (Egelman, 2007).
( + enlarge image)

The image processing method followed a single-particle strategy. The programs are developed using customized scripts written in the SPIDER package (Frank et al., 1996), in combination with PYTHON. This program package has produced high-quality density maps of various types of microtubules (Sui and Downing, Structure, 2010)

Image boxes, each containing about 10 dimer repeats, were extracted along microtubules in the digitized micrographs. Boxes from each microtubule were treated as a group and processed as described previously (Li et al., 2002), with some modifications as detailed in the main text. The power spectra of selected image boxes should show the layer-lines. As described below, all extracted images were aligned parallel to a box edge, using the Radon transform of power spectra described by Li et al.in 2000 (shown in the right panel). The rotationally aligned image boxes are then subjected to supervised classification (shown in the left panel). For each structural type, the iterative helical real space reconstruction strategy (IHRSR), was applied to refine the symmetry parameters and reconstruct density maps (Egelman, 2007).



Back to Top

Method for sub-volume averaging of motile-cilium microtubule doublets

Sub-volumes along microtubule doublets (MtDs) are extracted from the original tomograms. The extracted sub-volumes were roughly orientated with the axis of the doublet roughly parallel to the z axis and the missing wedge roughly 45 degrees away from the x and y axes.<br /> Each individual sub-volume was then projected onto a 2D image along the x axis. The protofilament structure contributes strongly to the equator in the power spectrum of this projection, and a Radon transform of the power spectrum shows a strong peak indicating the rotational offset required to align the protofilament precisely parallel to the z axis in the projection. The volume was then projected along the y axis for determination of the other rotation angle. Combining the above two rotation angles, the doublet 3D volumes were aligned parallel to the z axis. <br /> The projection of the aligned 3D volume along the z axis show a distinctive protofilament distribution and was used to calculate the rotational alignment around the z axis. After an orientation alignment by combining the above three rotation angles, cross-correlation was used to find the translational shifts. The parameters were used to re-extract 3D images along the doublet microtubule tomogram. 3D volumes were then averaged for each doublet microtubule. <br /> This alignment method is developed using libraries of the SPIDER package and the BSOFT package. It efficiently improved the quality of the 3D density map.
( + enlarge image)

Sub-volumes along microtubule doublets (MtDs) are extracted from the original tomograms. The extracted sub-volumes were roughly orientated with the axis of the doublet roughly parallel to the z axis and the missing wedge roughly 45 degrees away from the x and y axes.

Each individual sub-volume was then projected onto a 2D image along the x axis. The protofilament structure contributes strongly to the equator in the power spectrum of this projection, and a Radon transform of the power spectrum shows a strong peak indicating the rotational offset required to align the protofilament precisely parallel to the z axis in the projection. The volume was then projected along the y axis for determination of the other rotation angle. Combining the above two rotation angles, the doublet 3D volumes were aligned parallel to the z axis.

The projection of the aligned 3D volume along the z axis show a distinctive protofilament distribution and was used to calculate the rotational alignment around the z axis. After an orientation alignment by combining the above three rotation angles, cross-correlation was used to find the translational shifts. The parameters were used to re-extract 3D images along the doublet microtubule tomogram. 3D volumes were then averaged for each doublet microtubule.

This alignment method is developed using libraries of the SPIDER package and the BSOFT package. It efficiently improved the quality of the 3D density map.



Results of sub-volume averaging: (a) Power spectrum of the projection of one doublet from a tomographic reconstruction. The doublet image was computationally straightened before computing the Fourier transform. The layer line at 4 nm is the most prominent. (b) Power spectrum of the averaged tomogram from nine doublets, showing prominent layer lines at multiples of 1/(16 nm) and extending to a resolution better than 3 nm.
( + enlarge image)

Results of sub-volume averaging: (a) Power spectrum of the projection of one doublet from a tomographic reconstruction. The doublet image was computationally straightened before computing the Fourier transform. The layer line at 4 nm is the most prominent. (b) Power spectrum of the averaged tomogram from nine doublets, showing prominent layer lines at multiples of 1/(16 nm) and extending to a resolution better than 3 nm.



Back to Top