Remarkably, some of the residues in mutated human transthyretin that weaken or abolish the interaction with retinol-binding protein are present in piscine transthyretin, consistent with
Trang 1interactions between transthyretin and retinol-binding
protein
Giuseppe Zanotti1,2, Claudia Folli3, Laura Cendron1,2, Beatrice Alfieri3, Sonia K Nishida4,
Francesca Gliubich1,*, Nicola Pasquato1, Alessandro Negro5and Rodolfo Berni3
1 Department of Chemical Sciences and Institute of Biomolecular Chemistry-CNR, University of Padua, Italy
2 Venetian Institute of Molecular Medicine, Padua, Italy
3 Department of Biochemistry and Molecular Biology, University of Parma, Italy
4 Department of Medicine, Federal University of Sa˜o Paulo, Brazil
5 Department of Experimental Veterinary Sciences and CRIBI, University of Padua, Italy
Keywords
Fab; mutational analysis; protein–protein
interactions; retinol-binding protein;
transthyretin
Correspondence
G Zanotti, Department of Chemical
Sciences, University of Padua, Via Marzolo
1, 35131 Padua, Italy
Fax: +39 049 8275239
Tel: +39 049 8275245
E-mail: giuseppe.zanotti@unipd.it
R Berni, Department of Biochemistry and
Molecular Biology, University of Parma, V.le
G.P Usberti 23 ⁄ A, 43100 Parma, Italy
Fax: +39 0521 905151
Tel: +39 0521 905645
E-mail: rodolfo.berni@unipr.it
*Present address
KCL Enterprises Ltd, London, UK
Database
Atomic coordinates and structure factors
have been deposited at the Protein Data
Bank (PDB) (http://www.rcsb.org) for
imme-diate release: PDB 3BSZ for the TTR–RBP–
Fab complex and PDB 3BT0 and 3CXF
for the V20S and Y114H TTR variants,
respectively
(Received 23 July 2008, revised
22 September 2008, accepted 25
September 2008)
doi:10.1111/j.1742-4658.2008.06705.x
Transthyretin is a tetrameric binding protein involved in the transport of thyroid hormones and in the cotransport of retinol by forming a complex
in plasma with retinol-binding protein In the present study, we report the crystal structure of a macromolecular complex, in which human transthy-retin, human holo-retinol-binding protein and a murine anti-retinol-binding protein Fab are assembled according to a 1 : 2 : 2 stoichiometry The main interactions, both polar and apolar, between retinol-binding protein and transthyretin involve the retinol hydroxyl group and a limited number of solvent exposed residues The relevance of transthyretin residues in com-plex formation with retinol-binding protein has been examined by muta-tional analysis, and the structural consequences of some transthyretin point mutations affecting protein–protein recognition have been investigated Despite a few exceptions, in general, the substitution of a hydrophilic for a hydrophobic side chain in contact regions results in a decrease or even a loss of binding affinity, thus revealing the importance of interfacial hydro-phobic interactions and a high degree of complementarity between retinol-binding protein and transthyretin The effect is particularly evident when the mutation affects an interacting residue present in two distinct subunits
of transthyretin participating simultaneously in two interactions with a reti-nol-binding protein molecule This is the case of the amyloidogenic I84S replacement, which abolishes the interaction with retinol-binding protein and is associated with an altered retinol-binding protein plasma transport
in carriers of this mutation Remarkably, some of the residues in mutated human transthyretin that weaken or abolish the interaction with retinol-binding protein are present in piscine transthyretin, consistent with the lack
of interaction between retinol-binding protein and transthyretin in fish
Abbreviations
PDB, Protein Data Bank; RBP, retinol-binding protein; TTR, transthyretin.
Trang 2Transthyretin (TTR), a homotetramer of
approxi-mately 55 kDa, is a thyroid hormone-binding protein
present in the extracellular fluids of vertebrates,
where it participates, together with other binding
proteins, in the distribution of thyroid hormones
(thyroxine and triiodothyronine) [1] It was generated
during early vertebrate evolution as a result of a
duplication event in the gene encoding
5-hydroxyiso-urate hydrolase, an enzyme distributed in several
prokaryotes and in several eukaryotic lineages and
involved in purine metabolism [2–6] The
extracellu-lar transport of retinol (vitamin A alcohol) is
specifi-cally mediated by retinol-binding protein (RBP, also
known as RBP 4), a monomeric protein of 21 kDa
that delivers the vitamin molecule to the target cells
[7,8], where a membrane RBP receptor represents a
major mediator of cellular vitamin A uptake [9,10]
TTR and RBP are synthesized primarily by the
hepatocytes and are secreted into the circulation,
where RBP is found bound to TTR The association
of RBP with TTR increases the stability of the
reti-nol–RBP complex [11,12] and, according to various
lines of evidence [13–15], is believed to reduce the
glomerular filtration of the relatively small RBP
mol-ecule In turn, the stability of the RBP–TTR
com-plex is strongly affected by the presence of retinol
bound to RBP within the complex, a feature that is
believed to be of physiological significance The
affinity of holoRBP for TTR is significantly higher
than that of apoRBP [16,17], which is consistent
with holoRBP being retained in the circulation as
the protein–protein complex and with the
uncom-plexed apoRBP molecule, resulting from the delivery
of retinol, being selectively cleared from the
circula-tion by glomerular filtracircula-tion TTR has been
associ-ated with human diseases It is one of several
proteins that can produce the extracellular
accumula-tion in tissues of protein aggregates, in the form of
fibrils, which are responsible for degenerative diseases
known as amyloidoses; to date, more than 100 point
mutations have been described for TTR, most of
which are involved in familial amyloidoses [18,19]
Moreover, a protective role of TTR in Alzheimer’s
disease has recently been proposed [20,21] RBP is
an adipocyte-derived ‘signal’ that may contribute to
the pathogenesis of insulin resistance [22]
Well-refined crystal structures of TTR from
differ-ent vertebrate species, including mammals [23–25],
chicken [26] and fish (sea bream) [27,28], and of
RBP from mammals [29–33] and chicken [34], have
been described The crystal structures of heterologous
(human TTR–chicken RBP) [35] and homologous
(human TTR–human RBP) [36] TTR–holoRBP
com-plexes have also been determined, both characterized
by a 1 : 2 TTR : RBP stoichiometry in which each TTR-bound RBP molecule interacts simultaneously with three TTR subunits [35,36] TTR is a tetrameric protein formed by the assembly of four identical sub-units Each monomer is composed of eight anti-par-allel b-strands (A–H), arranged in a topology similar
to the Greek key b-barrel, with a short a-helix located at the end of b-strand E In the tetramer, the four monomers are organized as a dimer of dimers Specifically, two monomers are held together
to form a stable dimer through a net of H-bond interactions involving the two edge b-strands H and
F To form the tetramer, two dimers associate back
to back, mainly through hydrophobic contacts between residues of the loops formed by b-strands A and B and b-strands G and H One of the two-fold symmetry axes of the tetramer is coincident with a long channel that transverses the entire molecule and harbors two binding sites for thyroid hormones RBP is a single domain protein, made up of an N-terminal coil, eight anti-parallel b-strands (A–H) and a short a-helix close to the C-terminus The core of the protein is the internal cavity of an eight-stranded up-and-down b-barrel The vitamin mole-cule is accommodated within the cavity of the barrel: the b-ionone ring is innermost, the polyene chain is fully extended and the hydroxyl end group is almost solvent exposed, in the region of the loops that con-nect b-strands A and B, C and D and E and F and surround the entrance of the b-barrel at the open end of the cavity As a result of evolutionary restraints imposed by the multiple interactions estab-lished by both RBP and TTR, a high degree of structural similarity appears to be preserved for these two proteins from phylogenetically distant vertebrate species It should be noted, however, that piscine TTR and RBP lack the ability to form a protein– protein complex [27] The molecular basis of the evolution of the two piscine proteins into proteins that possess the ability to interact with each other in terrestrial vertebrates remains to be clarified
In the present study, we report on the structure of
a complex formed by the association of human TTR, human holoRBP and a murine anti-RBP Fab, and
on a mutational analysis of the RBP-binding determi-nants present in the human TTR molecule The data provide insight into the molecular basis of the altered plasma transport of RBP in carriers of a relevant amyloidogenic TTR mutation (I84S) and of the changes in the TTR molecule that have affected its ability to interact with RBP during the course of vertebrate evolution
Trang 3Structure of a complex between human TTR,
human holoRBP and an anti-RBP Fab
The crystal structure of the human TTR–RBP
com-plex bound to an anti-RBP Fab could be determined
at a resolution of 3.36 A˚, similar to that obtained for
the crystal structures of TTR–RBP complexes [35,36]
The molecular model, obtained by molecular
replace-ment starting from the available high resolution
crys-tal structures of RBP and TTR, is of reasonable
quality The anti-RBP Fab is bound to an RBP
epi-tope that is well separated from the TTR-binding
determinants present in the RBP molecule It
inter-acts with RBP on the side opposite to that involved
in the binding to TTR (Fig 1A), so that the
interac-tions between RBP and TTR are not affected by the
RBP-bound Fab The entire complex, composed of
one TTR tetramer and two holoRBP molecules in
complex with Fab, is present in the asymmetric unit
(Fig 1A) The two RBP–Fab sub-complexes are
arranged symmetrically around the two-fold axis of
TTR running through the central channel that
trans-verses the TTR molecule RBP and TTR
substan-tially maintain the structure they have in the
uncomplexed state, with only minor changes in
con-tact regions
The region of entrance of retinol into the b-barrel
cavity of RBP (i.e loops 32–36, 63–67 and 92–98),
and retinol itself, participate in the recognition of
TTR, involving residues that are essentially located in
loops connecting b-strands of the tetrameric protein,
in such a way that one RBP molecule interacts
simul-taneously with three TTR subunits (Table 1 and
Fig 1A,B) The RBP–TTR interactions are both
polar or apolar (Table 1) The contact surface of
TTR is characterized by a prevalence of hydrophobic
residues in the case of the subunits B and C and of
hydrophilic residues in the case of the subunit A
Val20, Trp79, Leu82, Ile84, Pro113 and Tyr114 from
both subunits B and C of TTR form a hydrophobic
patch, which is in contact in the protein–protein
com-plex with the hydrophobic patch formed by the
resi-dues Trp67, Leu63, Leu64, Val69, Phe96 and Leu97
of RBP Instead, the interacting hydrophilic residues
from the subunit A of TTR are closer to the border
of the contact surface of TTR Four H-bonds, mainly
between RBP and the subunit B of TTR, are also
present (Table 1) The area buried on the two
pro-teins upon complex formation is 1443 A˚2, out of a
total area of 9307 and 21502 A˚2 for the two
sepa-rated RBP and TTR molecules, respectively
The Fab–RBP interaction involves the region pre-ceding the a-helix and the C-terminal b-strand of RBP and the hyper-variable regions of Fab: loops 53–56 and 100–103 and the short helix 28–32 of chain H and loops 31–36 and 53–56 of chain L The interactions, which are mainly polar and comprise several H-bonds, are summarized in Table 2 The number of residues involved in the formation of the Fab–RBP complex is smaller than that of interacting residues in the RBP– TTR complex (Table 2) but the surface buried upon complex formation is comparable to that for the RBP– TTR complex (1588 A˚2) The contacts between inter-acting surfaces of RBP and Fab are shown in Fig 1C
Mutational analysis of the RBP-binding determinants of TTR
The amino acid sequences of RBP and TTR from dif-ferent vertebrate species showing the residues of human RBP and TTR that are mainly involved in interactions, according to our structure of the TTR– RBP–Fab complex, are shown in Fig 2 The func-tional and structural consequences of several TTR point mutations on the RBP–TTR interactions have been investigated The values of the dissociation con-stants for several complexes between human TTR vari-ants and human RBP, as determined by means of fluorescence anisotropy titrations (Fig 3A), are reported in Table 3 For some TTR mutations that abolish or weaken significantly the RBP–TTR interac-tions, the crystal structures of the variants have been determined or already available structures have been examined to provide details of the interference with protein–protein recognition by mutations
V20S TTR variant Two Val20 residues present in two distinct TTR sub-units (B and C, according to our designation of TTR subunits) are located at the center of a large hydropho-bic patch in the contact area between TTR and RBP,
so that their replacement by a hydrophilic residue is expected to significantly impair protein–protein inter-actions Accordingly, the V20S mutation has been found to almost abolish the binding affinity between RBP and TTR (Table 3 and Fig 3A) To investigate the structural consequences of the V20S replacement
on the RBP–TTR recognition, we have determined the 1.6 A˚ resolution crystal structure of the V20S TTR variant The overall structure is very similar to that of the wild-type protein (PDB: 1F41) [25]: a superposition
of equivalent Ca atoms gives an rmsd of 0.40 and 0.68 A˚ for subunits A and B, respectively However,
Trang 4B
C
Fig 1 Structure of the TTR–RBP–Fab complex (A) Representation of the overall structure of the TTR–RBP–Fab complex TTR: chain A, red; chain B, green; chain C, magenta; chain D, orange RBP: chains E and F, yellow Fab: heavy chains H and N, cyan; light chains L and M, blue (B) Detail of the contact between the TTR subunits A, B and C and the RBP molecule E [left side, color codes as in (A)] Center and right drawings show the interacting surfaces of RBP (center) and of the TTR subunits A, B and C (right) rotated by approxi-mately 90 counterclockwise and clockwise, respectively, compared to the previous view It is possible to appreciate how a protuberance formed by loops 63–67 and 92–98 on the RBP surface fits into a crevice formed by the arrangement of three TTR subunits (C) Detail of the contact between the RBP molecule F and the Fab chains H and L [left side, color codes as in (A)] Center and right drawings show the interacting surfaces of RBP (center) and Fab (right), rotated through a vertical axis by approximately 90 counterclockwise and clock-wise, respectively, compared to the previous view Residues 163–169 and the region preceding the a-helix contribute to the formation of the RBP epitope.
Trang 5significant differences are observed for loop region
98–103, which connects b-strands G to F and is
flexi-ble in most TTR structures, and for loop 19–22, which
connects b-strands A and B and hosts the mutation
The movement of the latter loop is not large (the Ca
atom of Val20 is displaced by 1.6 A˚ from its position
in the wild-type structure) but the entire loop 19–22 is
displaced by more than 1.5 A˚ from its original position
(Fig 3B), so that the change in the positions of Val20
and Arg21 of two distinct TTR subunits (Table 1) may
interfere simultaneously with two interactions
estab-lished with RBP
I84S and I84A TTR variants
A drastic effect on the recognition between TTR and RBP is caused by mutations at position 84 The in vitro binding of the I84S and I84A TTR variants to RBP is abolished or becomes almost negligible, respectively [37] (Table 3 and Fig 3A) Accordingly, the lack of interaction between RBP and the amyloidogenic I84S TTR is known to lead to a markedly lowered plasma concentrations of RBP in individuals carrying the mutation due to the impaired transport function of the TTR variant [38] The explanation for this effect is not straightforward because the structures of the I84S (PDB: 2G4G) [39] and I84A (PDB: 2G4E) [39] TTR variants at neutral pH are very similar to that of the wild-type protein In both cases, the amino acid replacements are not associated with significant local conformational changes Ile84 occupies a central posi-tion in a hydrophobic patch of TTR involved in hydro-phobic interactions with RBP; its replacement by a serine can perturb the polarity of the microenvironment
at the interface, thereby impairing protein–protein rec-ognition A reduction of the steric hindrance of the side chain of Ile84 as a consequence of the I84A mutation can instead perturb the interaction It is concluded that the residue at position 84 is particularly relevant for protein–protein recognition In this respect, it should
be noted that, as in the case of Val20 and Arg21, two lle84 residues of two subunits present in two different dimers of the TTR tetramer are involved at the same time in the interactions with one RBP molecule (Table 1) Therefore, an amino acid substitution at position 84 in TTR can lead to the loss of two relevant contacts between two residues of an RBP molecule and both dimers of the TTR counterpart Moreover, we have demonstrated that, at variance with wild-type TTR, at pH 4.6, both I84S and I84A mutations induce
a remarkable conformational change in the region com-prising residue 84, with the disruption of the a-helix itself [39] It is tempting to speculate that the two amino acid replacements cause a destabilization of the TTR region hosting the mutation, as revealed by the structural alteration at acidic pH, which in turn may affect the interaction with RBP
S85A TTR variant Ser85 appears to be relevant for the TTR–RBP inter-action on the basis of the structure of the complex because it participates in several contacts with RBP residues: two are H-bond interactions, one between the Ser85 -OH group and the amide nitrogen of Lys99 of RBP and the other between the amide nitrogen of
Table 1 Contacts between amino acid residues in the RBP–TTR–
Fab complex characterized by interatomic distances within 4.0 A ˚ :
interactions between RBP and TTR Interactions were analyzed
using the program CONTACT of the CCP4 package [54].
RBP chain E TTR chain A TTR chain B TTR chain C
a
Denotes the participation of TTR residues in at least one polar
contact b An H-bond distance is present between at least two
atoms of interacting residues of RBP and TTR.
Table 2 Contacts between amino acid residues in the RBP–TTR–
Fab complex characterized by interatomic distances within 4.0 A ˚ :
interactions between RBP and Fab Interactions were analyzed
using the program CONTACT of the CCP4 package [54].
a An H-bond distance is present between at least two atoms of
interacting residues of RBP and Fab.bDenotes the participation of
Fab residues in at least one polar contact.
Trang 6Ser85 and the carbonyl oxygen of Phe96 of RBP
(Table 1) Although the latter can be preserved in the
S85A variant, the former is lost, possibly contributing
to an approximately five-fold decrease in binding affin-ity caused by the mutation (Table 3) It can be specu-lated that the loss of interactions caused by the
A
B
Fig 2 Multiple sequence alignments for RBP (A) and TTR (B) from different vertebrate species The residues that are identical in all the sequences considered for each alignment are shaded in red; the residues that are identical or chemically similar in at least five sequences for each alignment are denoted by red characters (similarity groups are: HKR, DE, STNQ, AVLIM, FYW, PG, C) The amino acid residues in the human TTR and RBP sequences more directly involved in RBP–TTR interactions (Table 1) are denoted by arrowheads Numbering and secondary structure elements are based on the structures of human RBP (PDB: 1JYD) and TTR (PDB: 1F41) GenBank or SwissProt acces-sion numbers are: human TTR, PO2766; rat TTR, NP_036813; chicken TTR, CAA43000; zebrafish TTR, AAH81488; sea bream TTR, AF059193; trout TTR, CB497711 (EST sequence); human RBP, P02753; rat RBP, NM_013162; chicken RBP, P41263; zebrafish RBP, EF373650; sea bream RBP, AAF79021; trout RBP, P24774 Sequences containing the signal peptide have been reported when the N-termini
of the mature proteins are not known Sequence alignments were constructed by CLUSTALW [58] and rendered with ESPRIPT [59].
Trang 7mutation may be partially compensated by a novel
hydrophobic interaction in a hydrophobic patch
involving Ala85
D99A and S100E TTR variants These are important mutations because each of them causes an approximately 20-fold decrease in binding affinity of TTR for RBP (Table 3 and Fig 3A) Both replaced residues interact with Lys99 of RBP and, moreover, Ser100 is close to Trp91 (Table 1) Conse-quently, it is conceivable that the replacements of Asp99 by a hydrophobic residue and of Ser100 by a charged residue significantly perturb the interactions However, it is likely that the effects of the D99A and S100E mutations are less drastic compared to the I84S and V20S replacements because the interactions involv-ing residues at positions 99 and 100 are present at the periphery of a contact area between RBP and TTR Moreover, the new electrostatic situation for the TTR variants could result in new interactions of the mutated residues with the solvent
Y114F and Y114H TTR variants
A special case is represented by Tyr114: its substitution for a phenylalanine lowers the dissociation constant of the protein–protein complex by an approximate factor
of two, whereas the dissociation constant is three-fold higher when Tyr114 is replaced by a histidine (Table 3 and Fig 3A) We have also determined the crystal structure of the amyloidogenic Y114H variant: no sig-nificant conformational changes have been observed, and, in particular, the side chain of His114 maintains the same position and orientation compared to Tyr114 The same holds for the position and orienta-tion of the Phe114 residue present in the structure of both chicken TTR [26], which interacts well with RBP [17,40], and piscine TTR [27] The -OH group of Tyr114 forms an H-bond with the -OH group of Ser95
of RBP Moreover, Tyr114 is in a hydrophobic patch
of TTR, despite its proximity to the protein surface Its replacement by a phenylalanine is not drastic in terms of modification of the surface potential but it leads to the loss of an H-bond interaction It must be assumed that the loss of a H-bond interaction for the Y114F variant is compensated by some conforma-tional rearrangements that result in stronger
hydro-A
B
Fig 3 Human TTR mutations affecting TTR–RBP interactions.
(A) Typical fluorescence anisotropy titrations of human holoRBP
(3 l M ) with human TTR: wild-type, black; V20S TTR, gray; I84S
TTR, red; D99A TTR, green; Y114H TTR, blue; Y114F TTR, orange.
Fluorescence anisotropy values are plotted as a function of human
TTR molar concentration Lines represent theoretical binding curves
(for details, see Experimental procedures) corresponding to
dissoci-ation constants of 0.34 l M for wild-type TTR, 5.99 l M for D99A
TTR, 1.04 l M for Y114H TTR and 0.17 l M for Y114F TTR (B)
Ste-reo view showing the superposition of the Ca chain traces of
wild-type TTR (PDB: 1F41, red) and the V20S TTR variant (green) in the
area around residue 20 The side chains of residues Ser20 or Val20
and Arg21 are shown.
Table 3 Dissociation constants of complexes between human holoRBP and human TTR variants as determined by means of fluorescence anisotropy titrations Data represent the average of at least three independent measurements.
TTR Wild-type V20S a V30M a,b L55P a,b L58H b T60A a,b I84S a,b,c I84A a S85A D99A c S100E c Y114H a,b Y114F c
a The structure for the TTR variant is available: V20S and Y114H [present study]; V30M [55]; L55P, [56]; T60A, [57]; I84S and I84A, [39].
b Amyloidogenic TTR variant c Position affected by significant amino acid replacements in piscine TTR compared to human TTR (Fig 2B).
d Almost negligible interaction e Lack of interaction.
Trang 8phobic interactions, possibly explaining the higher
affinity between RBP and this TTR variant The
replacement of Tyr114 by a potentially charged
histi-dine could have a marked effect on the surface
poten-tial; the finding that the Y114H mutation does not
drastically impair the RBP–TTR interaction suggests
that His114 is not protonated in the protein–protein
complex
Modeling of the interactions of piscine TTR and
RBP with protein counterparts within the
TTR–RBP–Fab complex
The observation that human TTR and RBP are bound
in the TTR–RBP–Fab complex without undergoing
sig-nificant conformational changes compared to the
un-complexed proteins prompted us to assess the ability of
the two proteins from fish, which are unable to interact
with each other [27], to fit within the structure of the
TTR–RBP–Fab complex by replacing the
correspond-ing human proteins present in the complex The
struc-ture of sea bream (Sparus aurata) TTR is known [27],
whereas no structure for a fish RBP is available to date
Therefore, only a theoretical model for the structure of
sea bream RBP could be obtained with the
Swiss-Model server [41] The structure of sea bream TTR was
superimposed on that of human TTR present in the
TTR–RBP–Fab complex, giving rise to a hypothetical
model of the mixed piscine TTR–human RBP complex
(Fig 4) The most relevant difference can be observed
for loop 98–102 of all the subunits of TTR, especially
for subunits A and C, due to some remarkable
muta-tions affecting interacting residues (Table 3 and
Fig 2B) The conformation of loop 80–85 of piscine
TTR also changes slightly (Fig 4), possibly due to
relevant mutations in this area (Table 3 and Fig 2B)
The same holds for a model in which the human RBP
structure is also replaced by a theoretical sea bream RBP structure, giving rise to a hypothetical piscine RBP–TTR complex (Fig 4) Conformational differ-ences for TTR along with point mutations that have no effect on TTR structure (Fig 2B) may account for the experimentally determined lack of binding affinity between piscine TTR and RBP, as well as between piscine TTR and human RBP [27] Only limited confor-mational differences between piscine and human RBP are found Accordingly, the degree of conservation of the putative interacting residues is remarkably higher in piscine RBP than in piscine TTR; the only significant amino acid difference in piscine RBP compared to human RBP is K99T (Fig 2A) These features may explain the existence of an affinity, albeit weak, between piscine RBP and human TTR [27,42]
Discussion
The human holoRBP–TTR complex is relatively weak, being characterized by a dissociation constant of approximately 0.35 lm (Table 3) It has been suggested that this feature may be correlated with the need for the presence in plasma of a small but significant amount of uncomplexed holoRBP, which can thus leave the circulation more easily to deliver the retinol
to the target tissues [7] A limited number of residues and retinol itself are mainly responsible for the rela-tively weak RBP–TTR interaction An important role played by the retinol hydroxyl end group in the interaction is consistent with the low binding affinity
of apoRBP compared to that of holoRBP for TTR [16,17] and with the drastic interference with the interaction between the two proteins by RBP-bound fenretinide, a retinoid that bears a bulky end group in place of the retinol hydroxyl group [43,44] Moreover, the conformational change affecting one of the loops
Fig 4 Modeling of piscine TTR and RBP within the TTR–RBP–Fab complex Stereo view showing interacting regions between RBP (magenta) and TTR (orange) in the human RBP–TTR complex bound to Fab, with a superimposed model of the piscine RBP–TTR complex (green) based on the structure of sea bream TTR (PDB: 1OO2) and on a hypothetical model of the piscine RBP structure The two regions of TTR that differ significantly in the structures (98–102 and 80–85) are labeled.
Trang 9surrounding the opening of the b-barrel (in particular,
residues Leu35 and Phe36) in apo-RBP compared to
holoRBP [30,31] is likely to contribute to the
weaken-ing of the interaction of apoRBP with TTR due to the
involvement of such a loop in RBP–TTR recognition
Despite a few exceptions, the substitutions of
hydro-philic for hydrophobic side chains in TTR contact
regions generally have a rather pronounced
dissociat-ing effect on the RBP–TTR complex, consistent with
the important role played by interfacial apolar
inter-actions In general, the changes in the TTR molecule
induced by amyloidogenic mutations do not interfere
with the interactions between RBP and TTR, unless
the mutations are located in contact areas The
amyloi-dogenic mutations V30M, L55P, L58H and T60A have
a limited effect on the binding affinity between RBP
and TTR (Table 3), in accordance with the lack of the
replaced residues in contact areas; moreover, it can be
inferred that such amyloidogenic mutations do not
cause large conformational changes in the TTR
mole-cule that might affect indirectly the RBP–TTR
recogni-tion Conversely, the amyloidogenic I84S mutation,
which affects a residue that is crucial for
protein–pro-tein interactions, causes the lack of recognition
between RBP and TTR and an altered plasma
trans-port of RBP by TTR [37,38] It might be hypothesized
that the ability of RBP to interact well with relevant
amyloidogenic TTR variants, such as V30M, L55P,
L58H and T60A, can protect them from amyloid
aggregation However, it should be noted that the
plasma concentration of TTR is significantly higher
than that of RBP [7], so that a protective effect of
RBP on TTR can only be limited
Despite the high symmetry of TTR, which is a
homo-tetramer with virtually four identical binding sites for
RBP, a 1 : 1 TTR : RBP complex is believed to be
present in plasma due to the excess of TTR over RBP
[7] Binding data obtained in solution [17,37,45] and
structural data [35,36] have shown that a maximum of
two RBP molecules can be bound by one TTR
tetra-mer The binding of two RBP molecules to an
uncom-plexed TTR tetramer partially hinders the potential
binding of two nearby RBP molecules, thereby limiting
the possible interactions with tetrameric TTR to two
RBP molecules [35,36] However, two distinct
macro-molecular organizations, both accounting for the 1 : 2
TTR : RBP stoichiometry, have been described for
the heterologous and the homologous RBP–TTR
complexes [35,36] The crystal structure of the TTR
tetramer is characterized by 222 symmetry One of the
three orthogonal two-fold axes runs through the
central channel harboring the two thyroid hormone
binding sites In the heterologous chicken RBP–human
TTR complex, the two TTR-bound RBP molecules are related by one of the two available two-fold axes that are perpendicular to the central channel of TTR [35] Instead, in the homologous human RBP–human TTR complex, as well as in the case of our structure of the TTR–RBP–Fab complex, the two TTR-bound RBP molecules are related by the two-fold axis running through the central channel of TTR [36] (Fig 1A) Because the two situations are chemically equivalent, a possible explanation for the observed different assem-bly is that, in solution, both modes of assemassem-bly can be present and that the crystallization process selects one
of them according to the best packing
By comparing the structure of the human TTR– human RBP complex bound to Fab with those of the heterologous human TTR–chicken RBP complex [35] and of the homologous human TTR–human RBP complex [36], a good correspondence between these structures with regard to interacting surfaces of TTR and RBP has been found In the case of the human TTR–human RBP complex, however, it should be noted that one of the two TTR-bound RBP molecules has been reported to participate in the interaction with the last C-terminal amino acid residues (especially Leu182 and Leu183), thereby generating an asymmetry within the complex [36] At variance with this observa-tion, our TTR–RBP–Fab structure does not reveal the presence of interactions between the carboxy terminus
of RBP and TTR On the other hand, it should be noted that chicken RBP, in which eight C-terminal res-idues are missing compared to human RBP (Fig 2A), binds to human TTR with an affinity similar to that exhibited by human RBP [40; C Folli and R Berni, unpublished data], which suggests that the carboxy terminus of human RBP is not so crucial for the inter-action with TTR
The RBP–TTR complex is normally isolated from the serum of terrestrial vertebrates, such as mammals and birds [7] Moreover, purified human and chicken RBP and TTR have been found to cross-interact [40]
By contrast, RBP could be isolated from the serum of different fish species only as uncomplexed protein [27,42,46], suggesting that, in fish, it is present in the circulation as an uncomplexed protein without affinity for TTR In accordance with this observation, the lack
of binding affinity between purified piscine RBP and TTR has been established [27] The comparison of the amino acid sequences of piscine RBPs and TTRs with those of the same proteins from terrestrial vertebrates reveals the presence of remarkable differences in regions involved in protein–protein interactions for TTR, whereas only limited differences are present in the case of RBP (Fig 2A,B) In particular, the amino
Trang 10acid replacements at positions 82, 84, 99 and 100 in
piscine TTR compared to the human and chicken
pro-teins are drastic and are present at positions critical
for the interaction between RBP and TTR (Table 1
and Fig 2B), and some of them (I84S, D99A and
S100E) are shown in the present study to impair or
abolish protein–protein recognition (Table 3 and
Fig 3A) The results obtained are consistent with the
notion that evolutionary changes affecting a limited
number of surface-exposed residues led to the
appear-ance in terrestrial vertebrates of the TTR function of
cotransport of retinol through the interaction with
holoRBP in plasma, in addition to that of the
distribu-tion of thyroid hormones in the extracellular fluids
Experimental procedures
Materials
HoloRBP was purified from human plasma as reported
previously [17] Recombinant wild-type human TTR and
TTR variants I84S and I84A were prepared and quantified
as described previously [39] All chemicals were of analytical
grade
Site-directed mutagenesis, bacterial expression
and purification of human TTR variants
The recombinant human TTR variants V20S, L55P, L58H,
T60A, S85A, D99A, S100E, Y114F and Y114H were
pre-pared by PCR using the plasmid pET11b-human TTR [39]
as template, a high-fidelity thermostable DNA polymerase
(Pfu Ultra II Fusion HS DNA polymerase; Stratagene, La
Jolla, CA, USA) and mutagenic primers complementary to
opposite strands For each mutation, the product of
reac-tion was treated with DpnI (New England Biolabs, Beverly,
MA, USA) to digest the parental DNA template This
pro-cedure allowed us to select the newly synthesized and
potentially mutated plasmids The products of each
diges-tion were used to transform Escherichia coli XL1 Blue cells
Single clones were then sequenced to confirm the
occur-rence of the desired mutation Finally, mutant plasmids
were electroporated into E coli BL21 (DE3) cells The
expression of TTR variants was induced by 1 mm isopropyl
cells were disrupted by sonication TTR variants were
puri-fied as described for wild-type TTR [39]
Determination of the amino acid sequences of
anti-RBP Fab variable domains
Total RNA obtained from the cell line producing the
anti-RBP murine monoclonal antibody A8P3 [47] was subjected
to retrotranscription into cDNA employing the Superscript
Preamplification System (Gibco, Gaithersburg, MD, USA) The cDNA sequences encoding for the variable domains of the H and L chains of the monoclonal antibody A8P3 were PCR amplified using a mixture of 18 5¢ primer VKBACK mix and a mixture of five 3¢ primer VKFOR for the VK gene and 20 5¢ primer VHBACK mix and a mixture of five 3¢ primer VHFOR mix for the VH gene [48] Each domain was cloned using the TA cloning kit (Invitrogen, Carlsbad,
CA, USA) and sequenced using an automated model 377 sequenator (Applied Biosystems, Foster City, CA, USA) The amino acid sequences of the variable domains of L and
H chains of the antibody A8P3 are provided in Fig S1
Binding assay for the interaction between holoRBP and TTR variants
To study the in vitro interaction of holoRBP with TTR variants, the highly fluorescent RBP-bound retinol provides
an intense signal which is suitable for fluorescence polariza-tion measurements [40] The intensities of the vertical (I ||) and horizontal (I^) components of the fluorescence of RBP-bound retinol (excitation at 330 nm and emission at
460 nm) were recorded at an angle of 90 to the vertically polarized excitation beam A correction factor, G, equal to
perpendicular direction) was used to correct for the unequal transmission of differently polarized light Fluorescence anisotropy (A) was determined according to the equation:
3.0 lm) in 0.05 m sodium phosphate (pH 7.2) and 0.15 m
concen-trated solutions of human TTR (wild-type or mutant forms) to the RBP-containing cuvette and the increase in fluorescence anisotropy of the RBP-bound retinol upon complex formation was monitored The fraction of RBP bound by TTR (a) was calculated for every point of the
value of RBP-bound retinol for a certain molar
anisotropy values (i.e in the presence of an excess saturat-ing TTR and in the absence of TTR, respectively) Bindsaturat-ing data were analyzed as described [12] Fluorescence anisot-ropy measurements were carried out with a LS-50B spectro-fluorometer (Perkin-Elmer, Waltham, MA, USA)
Crystallization, data collection, structure determination and refinement for the TTR–RBP–Fab complex and the V20S and Y114 TTR variants
Crystallization and preliminary X-ray data for the macro-molecular complex formed by human transthyretin, human holoRBP and a murine anti-RBP Fab have been reported