(A) A stereoview of a ribbon diagram of the structure of TSPN-1. The β strands and α helices are colored in cyan and green, respectively. The irregular proline-rich strand-like segment discussed in the text is labeled as β4′ and is highlighted in purple. The sole disulfide bond (yellow) is drawn in ball-and-stick representation. The jar handle-like motif after the β13 strand is highlighted in yellow. The N and C termini are labeled as N and C, respectively. The molecule is oriented so that its concave β sheet or front sheet faces the reader. This figure was prepared with the program MOLSCRIPT (Kraulis, 1991).
(B) A stereoview of the Cα tracing diagram of the front-sheet, right-side strands of TSPN-1. The diagram displays most of the interactions involved with the irregular strand-like segment β4′ colored in purple, interactions with strand β13 and the following jar handle-like motif colored in yellow, as well as a few interactions from strands β6 and β7 and the loop between them. Three water molecules between β4′ and β13 are colored green and are labeled as W1, W2, and W3. They are conserved in both native and TSPN-1/Arixtra structures and are well defined in electron densities. Another water molecule, W4, potentially involved in stabilizing the jar handle motif conformation is also drawn in green.

(A) A stereoview of TSPN-1 with the major positively (blue) and negatively (red) charged residues labeled. A backbone worm is embedded to show the orientation of the molecule and the positions of the major charged residues. Besides the positively charged patch at the bottom of the domain, there is a negatively charged patch around the left edge and the back of TSPN-1, where fibrinogen reportedly binds.
(B) A view of the positively charged surface patch and the positions of the residues, which may contribute to heparin binding on the bottom of TSPN-1. This site is referred to as the major heparin binding site in the text. The orientation of the molecule shown here is related to the one in (A) by a −60° rotation around the horizontal axis.
These figures were prepared with the program GRASP (Nicholls et al., 1991).

(A) Chemical structure of Arixtra (fondaparinux sodium), C₃₁H₄₃N₃Na₁₀O₄₉S₈.
(B) A simulated annealing omit map (2F_o − F_c at the 1.0σ contour level and colored in green) around the major heparin binding site and the partially disordered Arixtra molecule. All non-protein atoms have been removed from the simulated annealing refinement and map calculation. A SGN unit, which binds to the top of a symmetry-related molecule (see text), is labeled in yellow. The density in the middle on the left side is from residue R23 of another symmetry-related molecule. This residue is on the flexible loop (G20AARKG), which changes conformation in the TSPN-1/Arixtra complex. It seems to be involved in an interaction with Arixtra in the molecular packing of the complex, suggesting that the crystal packing rather than heparin binding may induce the conformational changes in the loop.
(C) The major heparin binding site in unligated native TSPN-1 with the difference electron density map (2F_o − F_c) contoured at 1.0σ and colored in green.
(D) The partially disordered Arixtra molecule and the three arginines from the major heparin binding site. In this drawing, the TSPN-1/Arixtra model was rotated about 90° for comparison of the partial Arixtra model with the chemical structure of Arixtra in (A).

The topology of (A) TSPN-1 is compared with that of (B) CTSP-1 (the C-terminal domain of TSP-1) (Kvansakul et al., 2004) and (C) a typical LG (Laminin A G-domain [LG]/Neurexin/Sex hormone binding globulin [SHBG]) repeat (Rudenko et al., 2001). For the LG structure, only the common positions of the β strands are shown. The disulfide bonds are highlighted in yellow. Only β strands (cyan) are labeled for comparison. The α helices of TSPN-1 and CTSP-1 are shown in green. The two structures can’t be simply superimposed. The best alignment can only be made with the middle strands of the back sheet, β14, β5, β10, and β11 of TSPN-1 and β15, β5, β11, and β12 of CTSP-1. Even in this alignment, the β strands from the two front sheets shift over about one strand and also rotate slightly. The edge strands, especially the N-terminal edge strands, are totally out of alignment. The front concaved sheet of CTSP-1 is distinctly curved, while that of TSPN-1 is relatively flat.

The Sequence Alignment of the N-Terminal Domains of Human TSP-1, TSP-2, TSP-3, and TSP-4; NELL-1 and NELL-2; Collagens α1(XVIII), α1(IX), α1(V), and α1(XI); Xenopus Kielin; and Drosophila TSP
The secondary structures (β strands and α helices) are underlined and are assigned based on the structure of TSPN-1. The irregular, proline-rich strand-like element is marked by a dashed line. The cysteines are colored in yellow and in red shadow. The basic residues contributing to the major heparin binding sites in TSPN-1 are marked in yellow shadow. Those hydrophobic and nonhydrophobic residues that contribute to the core of the TSPN-1 structure or putatively contribute to the cores of other molecule structures are marked in cyan and green shadow, respectively. The RGD triplet in the D-TSP sequence is shadowed in purple. The asparagines that are present within consensus sequences for glycosylation are shown in red. The motif after the last cysteine (e.g., C214 of TSPN-1) in each sequence is colored blue and is assumed to be the linker between a TSPN domain and a helical region that forms either trimeric or pentameric coiled coils. Only part of the linkers of the collagens α1(V), α1(XI), and α1(XVIII) are shown.