view release on metacpan or  search on metacpan

t/data/macroman/10542098.macroman  view on Meta::CPAN

   (TRAIL-R4/TRUNDD)^[65]2. The four TRAIL receptors possess two cysteine
   rich repeats, in contrast to TNF-R1 and TNF-R2, which each have four,
   and Fas, which has three. DR4 and DR5 have death domains, while DcR1
   lacks a cytoplasmic domain and DcR2 contains an incomplete
   non-functional death domain. DcR1 and DcR2 are therefore thought to act
   as decoys, protecting cells that coexpress either DR4 or DR5 from
   apoptosis induced by TRAIL. Thus, following activation, peripheral
   blood lymphocytes that are initially insensitive to TRAIL lose
   expression of DcR1 and subsequently become sensitive to TRAIL^[66]4.
   The differential sensitivity of tumors compared to normal cells
   suggested that TRAIL may have greater utility as an anti-tumor agent
   than TNF, which has limited use because of its side effects. In a
   murine model, TRAIL has been shown to be effective against xenogenic
   tumor transplantation without harming the host^[67]5.
   The emerging complexity of the TNF and TNF-R families raises questions
   at the molecular level, such as what is the basis of specificity and
   more particularly for the TRAIL−TRAIL receptor system, what are the
   conserved features that confer the ability to initiate apoptotic
   signaling. We have determined the crystal structure of the TRAIL−DR5
   complex, which throws light on these questions, revealing an unusual
   mechanism for controlling specificity.
   Structure determination
   The complex between soluble forms of recombinant human TRAIL (residues
   91−281, all numbering starts from the initiation codon) and DR5
   (residues 58−184) crystallized with an asymmetric unit consisting of
   one subunit of TRAIL and one copy of DR5, forming the full trimeric
   complex through crystallographic three-fold symmetry. The crystals
   diffracted to 2.2  using synchrotron radiation and the structure was
   phased by a combination of molecular replacement and heavy atom
   derivatives. With the exception of some mobile loop regions, all of
   TRAIL C-terminal to residue 119 and the cysteine rich repeat region of
   DR5 (starting at residue 69) are well ordered in the final crystal
   structure (crystallographic R factor 22.1% and R[free] 27.0%; see
   Methods and [68]Table 1). The overall structure of the TRAIL−DR5
   complex, including representative electron density, is shown in
   [69]Fig. 1.
   [70]Figure 1. The structure of TRAIL, DR5 and the TRAIL−DR5 complex.
   [71]Figure 1 thumbnail

   a, Stereo view of the complex. The three crystallographically
   equivalent copies of the TRAIL subunit (yellow, cyan, pink) and DR5
   (blue, green, red) are depicted schematically and the TRAIL trimer is
   enclosed in a transparent molecular envelope. This orientation defines
   a standard view. b, The complex as depicted in (a) but viewed down the
   three-fold axis. The orientation is such that the cell surface
   presenting DR5 is above the page and that for TRAIL is below the page.
   c, Superposition of TRAIL (pink) with TNF beta (blue). The secondary
   structure elements for TRAIL are also marked on the sequence alignment
   in [72]Fig. 2a. The r.m.s. deviation is 0.9  for 120 structurally
   equivalent C alpha atoms. The major extension of the AA" loop in TRAIL
   is highlighted by yellow stripes. The cell surface position is not to
   scale. d, Comparison of DR5 and TNF-R1. DR5 and TNF-R1 (from the TNF
   beta −TNF-R1 complex) are depicted schematically in the left and right
   panels respectively with equivalent regions in identical colors. The
   central panel is based on superposition of DR5 D1 and TNF-R1 D2. The
   schematic representation of TNF-R1 is shown in gray while that of DR5
   is in green. Disulfide bonds are depicted in yellow as ball-and-stick
   representation. e, Portion of the final 2F[o] - F[c] electron density
   map contoured at 1 sigma , showing a portion of the TRAIL structure in
   the BC loop.
   [73]Full Figure [74]Full Figure and legend (68K)
   [75]Table 1. Crystallographic Statistics
   [76]Table 1 thumbnail
   [77]Full Table [78]Full Table
   The TRAIL trimer
   The TRAIL subunits consist of beta -sandwiches conforming, as expected,
   to the jellyroll topology^[79]6. Each TRAIL monomer contains one
   cysteine at position 230. Crystallization was performed under
   non-reducing conditions and in the crystal a disulfide bond was seen
   between two of the three cysteine residues in each trimer. The
   formation of this disulfide bridge did not affect receptor binding, as
   both TRAIL that had been refolded in iodoacetamide and a TRAIL mutant
   (C230S) immunoprecipitated DR5 in a manner similar to that of wild type
   TRAIL (data not shown). The trimeric interface is extensive with a
   large population of aromatic residues conferring a highly hydrophobic
   character (66% of the 2,250 ^2 buried surface per subunit is apolar
   ([80]Fig. 2a). Comparison with the other known structures for members
   of the TNF superfamily (TNF alpha , TNF beta /LT alpha and
   CD40L)^[81]6, ^[82]7, ^[83]8 highlights a high level of structural
   conservation within the beta -strands and at the trimeric interface
   ([84]Figs 1c, [85]2a), despite the relatively low sequence identity
   between the family members (23−36% for the sequences in [86]Fig. 2a).
   Of particular interest are strands D and E, which have highly divergent
   sequences, yet are almost completely superimposed in structural
   alignments. Structural variation in the family is concentrated in the
   loop regions, with the most dramatic differences seen in the AA", CD
   and EF loops. The TRAIL structure is unique within the TNF family in
   displaying a major extension of the AA" loop that is 16 amino acids
   longer than in TNF beta ([87]Fig. 1c). In the final stages of preparing
   this manuscript the 2.8  resolution crystal structure of uncomplexed
   TRAIL was reported^[88]9. In the absence of coordinate deposition we
   have not been able to make a detailed comparison of the liganded and
   unliganded structures of TRAIL, however inspection of the figures
   suggests that the N-terminal portion of the AA" loop has a markedly
   different conformation in the two structures.
   [89]Figure 2. Comparison of the TRAIL−DR5 and TNF beta −TNF-R1
   complexes.
   [90]Figure 2 thumbnail

   a, Sequence alignments for the TNF and TNF-R1 families. Residues in
   TRAIL, TNF beta , DR5 and TNF-R1 are color coded according to their
   contribution to the interaction patches in (b). Sequence alignments to
   TRAIL are based on pairwise structural superpositions for TNF beta ,
   TNF alpha and CD40L and on manual sequence comparison for FasL.
   Secondary structure assignments and dots at every tenth residue above
   the sequences are for TRAIL. Triangles below the sequences indicate
   TRAIL residues buried in the isolated molecule (less than 20% relative
   side chain solvent accessible area^[91]27), circles indicate surface
   exposed residues mutated in TNF alpha and/or TNF beta (reviewed in ref.
   [92]30). Sequence alignments to DR5 are based on pairwise structural
   superposition for TNF-R1 and on manual sequence comparison for the
   remaining molecules. Dots above the sequences mark every tenth DR5
   residue. Symbols in rows 1 and 2 below the sequences indicate the
   nature of interactions involving DR5 and TNF-R1, respectively. Circles
   in row 3 indicate candidate residues for TRAIL binding specificity in
   DR5, DR4, DcR1 and DcR2. b, Surface representations of TRAIL, DR5, TNF
   beta and TNF-R1. The relative orientation of the two complexes is
   defined by pairwise superposition of the TRAIL and TNF beta subunits.
   The view of the TRAIL and TNF beta surfaces corresponds to that of
   [93]Fig. 1a while the matching DR5 and TNF-R1 surfaces are revealed by
   a 180¡ rotation as in the opening of a book. The solvent accessible
   surfaces (calculated in isolation) of each component are shown in gray
   with areas contributed by residues implicated in the interaction
   surface of the complex highlighted in colored patches. Colors for
   interaction patches are matched between ligand and receptor for each
   complex. Conservation of the color scheme is used to highlight broadly
   equivalent regions of interaction in the two complexes.
   [94]Full Figure [95]Full Figure and legend (83K)
   The DR5 receptor
   Members of the TNF-R family are characterized by extracellular repeats
   containing three disulfide bridges with a cysteine knot topology^[96]7,
   ^[97]10. The number of repeats ranges from six in CD30 through four in
   TNF-R1 and TNF-R2 to three in Fas and only two in the four TRAIL
   receptors. The crystallographically well-ordered portion of the DR5
   molecule includes these two repeats (D1 and D2), which form the ligand
   binding region in the complex. DR5 starts with an N-terminal cap
   containing a partial cysteine knot with a single non-canonical
   disulfide ([98]Fig. 1d). This N-terminal cap corresponds to the
   C-terminal half of the first repeat (D1) in TNF-R1, while DR5 D1 and D2
   correspond to the central two repeats, D2 and D3, of TNF-R1 which form
   the binding interface in the TNF beta −TNF-R1 complex. The sequences of
   DR5 D1 and D2 differ considerably, with only 25% amino acid identity
   compared to 42% between TNF-R1 D2 and D3. This difference is borne out
   at the structural level: the root mean square (r.m.s.) deviation is 1.0
    between 30 equivalent C alpha atoms in DR5 D1 and D2 compared to 0.8
    between 29 equivalent C alpha atoms in TNF-R1 D2 and D3. The two
   ligand binding repeats in these and other TNF-receptor like molecules
   are joined by a CXC motif (CQC in all the TRAIL receptors and CGC in
   TNF-R1) which acts as a flexible articulation point in the isolated
   TNF-R1 structure^[99]10. The complexed structures of DR5 and
   TNF-R1^[100]7 show a clear difference in the relative orientation of
   the two repeats forming their respective ligand binding regions
   ([101]Fig. 1d). Thus, a tilt and rotation totaling 36¡ about the CXC
   pivot point would be required to place DR5 D2 in an orientation similar
   to that of TNF-R1 D3.
   The TRAIL−DR5 complex
   The TRAIL−DR5 complex reflects the trimeric nature of TRAIL, binding
   one copy of DR5 in each of the clefts between subunits ([102]Fig. 1).
   The surface area of 1,420 ^2 buried on TRAIL by each DR5 is relatively
   equally divided between the two subunits. Likewise, the 1,540 ^2 of
   surface buried on the DR5 is shared equally between its two repeats
   ([103]Fig. 2a,b). This contrasts with the TNF beta −TNF-R1
   complex^[104]7, where although the receptor makes approximately equal
   contacts with the two TNF subunits, there is a major difference in the
   contributions of TNF-R1 D2 and D3, with some two thirds of the contact
   area residing on D2.
   Superposition of the two complexes shows the position of DR5 D1 and its
   equivalent in TNF-R1 to be highly conserved, while a change in tilt
   between D1 and D2 of DR5 allows the second domain to more closely
   follow the TRAIL surface, introducing a number of novel contacts.
   A detailed analysis of the receptor−ligand contacts is shown in
   [105]Fig. 2b, where areas of contact are related by color and mapped
   back to the sequence alignments ([106]Fig. 2a). There are four main
   contact patches on TNF beta (apical subunit 1: A1 (blue), basal subunit
   1: B1 (cyan), apical subunit 2: A2 (orange), basal subunit 2: B2 (red);
   [107]Fig. 2). The contacts on TRAIL duplicate these areas while
   containing additional points of contact in the apical portion of
   subunit 1 created by the repositioning of D2 (yellow and green). This
   repositioning also alters which regions of the receptors are involved
   in the conserved patches (A1-blue and A2-orange) on the apical regions
   of the ligands. In the most extreme cases (orange and purple patches)
   the residue usage shifts from one cysteine repeat to the other with the
   A2 (orange) interaction mediated primarily by residues 107−109 in the
   N-terminal binding repeat of TNF-R1 but by residues 153−155 in the C
   terminal repeat of DR5. The more topographically conservative elements
   of the interaction involve DR5 D1, which makes contact with the B1
   (cyan) and B2 (red) patches. Although similarly located, the nature of
   the interacting residues in the B2 (red) patches from the two complexes
   is highly divergent. The hydrophobic interactions of the B1 patch
   (cyan), mediated by the surface exposed Tyr 216 in TRAIL (to Leu 114 in
   DR5) and Tyr 142 in TNF beta (to Leu 100 in TNF-R1), are the only truly
   conserved portions of the interface ([108]Fig. 3a). Mutagenesis of this
   Tyr residue in TNF alpha , TNF beta and FasL abolishes receptor
   binding^[109]11, ^[110]12, ^[111]13. Much of the remainder of the
   interface in both complexes is made up of polar or charged
   interactions. The details of these electrostatic interactions differ
   completely between the two complexes ([112]Fig. 3b).
   [113]Figure 3. Elements of conservation and specificity in
   ligand−receptor binding.
   [114]Figure 3 thumbnail

   a, Conservation in ligand-receptor interactions. Close up of the
   interaction involving the B1 (cyan) surfaces in the TRAIL−DR5 and TNF
   beta −TNF-R1 complexes centered on the key tyrosine residue (Tyr 216 in
   TRAIL). In this and in (c), the polypeptide chains are represented
   schematically as in [115]Fig. 1a and the solvent-accessible surfaces of
   the receptors (calculated in isolation) are shown as semi-transparent
   envelopes. b, Comparison of surface charge between TRAIL, TNF alpha ,
   TNF beta and their receptors. Blue denotes positive, and red negative;
   electrostatic potential is contoured at plusminus 8.0 kT in program
   GRASP^[116]31. The views of ligand and receptor are as in [117]Fig. 2b.
   c, Interaction of Arg 149 in the AA" loop of TRAIL with Glu 147 in DR5.
   d, BIAcore analysis showing binding of DR5 to wild type TRAIL or the
   mutant lacking the AA" loop. e, Immunoprecipitation with DR5-Fc in the
   presence of wild type (W) or the slightly smaller AA" TRAIL mutant (M).
   Lanes 1 and 2 immunoprecipitated material (IP), lanes 3 and 4 material
   left in supernatant (SN) following immunoprecipitation.
   [118]Full Figure [119]Full Figure and legend (83K)
   Features conferring specificity
   Inspection of the four TRAIL binding receptor sequences shows 45−81%
   identity in D1 and 59−80% identity in D2. Most of the residues involved
   in the TRAIL−DR5 interface show conserved characteristics, underscoring
   the role played by these residues in receptor−ligand specificity
   ([120]Fig. 2a). D2 is implicated as a major focus for TRAIL-binding
   specificity, with conservation in this receptor subgroup of residues
   such as Glu 147, Glu 151, Arg 154 and Asp 175 (DR5 numbering) that
   mediate complimentary electrostatic interactions. One of these
   distinctive TRAIL-binding residues (Glu 147) makes a salt bridge to Arg
   149 on the extended AA" loop of TRAIL ([121]Fig. 3c); the shorter AA"
   loop in TNF beta precludes any such interaction in the TNF beta −TNF-R1
   complex. This electrostatic interaction may help establish the correct
   docking orientation for DR5 D2 to TRAIL, stabilizing the appropriate
   tilt between D1 and D2. The importance of this contact was demonstrated
   by mutagenesis where the 19 amino acid sequence in TRAIL
   (^135TLSSPNSKNEKALGRKINS^153) was replaced with three amino acids
   (^76SSL^78) present in the much shorter AA" loop of TNF beta . Binding
   of the mutant TRAIL assessed by BIAcore was reduced by over 95%
   ([122]Fig. 3d). In a separate experiment, DR5-Fc was used to
   immunoprecipitate either the wild type or the AA" loop mutant of TRAIL.
   In agreement with the BIAcore analysis, DR5 binding to the mutant TRAIL
   was completely abolished ([123]Fig. 3e). The modeling study of Cha et
   al.^[124]9 does not appear to replicate the major differences in
   receptor docking observed between the crystal structures of TRAIL−DR5
   and TNF beta −TNF-R1. However, the findings of the accompanying
   mutagenesis study of the AA" loop are in accordance with those reported
   here.
   Discussion
   The TRAIL−DR5 complex highlights the tight interplay between structure
   and function for this receptor−ligand family. Both ligand and receptor
   are constructed from robust structural motifs with signaling mediated
   by the interaction of receptor with the stable trimeric scaffold of the
   ligand. This rigid framework allows free variation of the surface
   residues in both ligand and receptor conferring the biological
   specificity of particular ligand−receptor pairs through numerous
   changes in the interacting pairs of residues. In the ligand, this amino
   acid diversity is perhaps most extreme in the surface-exposed D and E
   strands and flanking loops where very little sequence homology can be
   discerned ([125]Fig. 2a). The surface variation is also clearly
   illustrated in [126]Fig. 3b where it can be seen that although the
   overall surface topology remains intact, the distributions of charge on
   the surfaces of TNF alpha , TNF beta , and TRAIL are very different. In
   the receptors, the orientation selected at the flexible CXC junction
   determines which areas of the C-terminal ligand binding repeat are
   involved in the complex. This switch mechanism may mean that it is
   possible to engineer receptors, with multiple specificities by
   exploiting contact positions unique to individual receptor−ligand
   pairs.
   Members of the TRAIL binding subgroup of receptors have, like the
   chicken death receptor CAR1 (ref. [127]14), been cut down to a minimal
   configuration of only two cysteine-rich repeats, with the second (D2)
   tightly abutting the three-fold axis in the TRAIL−DR5 complex.
   Superposition of the complete, four repeat, TNF-R1 structure^[128]10
   onto the TNF beta −TNF-R1 complex^[129]7 reveals the C-terminus of the
   TNF-R1 cysteine rich region at a distance from the three-fold axis that
   is similar to that reached more directly by the DR5 C-terminus. The
   length of the tether between the cell membrane and second repeat of the
   TRAIL receptors varies from two amino acids in the shorter of two
   alternatively spliced variants of DR5 (ref. [130]15) to 92 in DcR1. The