Encode-Guess-Educated
view release on metacpan or search on metacpan
t/data/cp1252/10676968.cp1252 view on Meta::CPAN
Guanylate-binding proteins (GBP1 and 2) were originally identified as
proteins from an extract of human fibroblasts treated with interferons,
gamma -interferon being the most effective, that bind to agarose-bound
GMP, GDP and GTP^[32]1, ^[33]2. Smaller guanylate-binding proteins of
relative molecular mass 47,000 (M[r] = 47K)^[34]6 are also induced by
gamma -interferon, whereas alpha - and beta -interferon induce
antiviral GTP-binding Mx proteins^[35]7. Human (h)GBP1 is expressed to
mediate an antiviral effect against vesicular stomatitis virus and
encephalomyocarditis virus^[36]4. The biochemical properties of GBPs
are clearly different from those of Ras-like and heterotrimeric
GTP-binding proteins. They bind guanine nucleotides with low affinity
(micromolar range), are stable in their absence and have a high
turnover GTPase^[37]3, ^[38]8. In addition to binding GDP/GTP, they
have the unique ability to bind GMP with equal affinity and hydrolyse
GTP not only to GDP but also to GMP^[39]3. As a first step towards
understanding the biochemistry and biology of GBPs, we have determined
the three-dimensional structure of hGBP1. Furthermore, we show
nucleotide-dependent oligomerization of hGBP1 and concentration
dependence of its GTPase reaction rate.
We determined the structure of full-length, histidine(His[6])-tagged
hGBP1 in the absence of nucleotide to 1.8 Å. The final model comprises
residues 6–583 and 341 water molecules ([40]Fig. 1a, [41]b), with some
poorly defined, probably mobile loops (dashed lines). The
crystallographic data are summarized in [42]Table 1. The structure can
be divided into a compact, globular alpha , beta -domain (6–278), which
we term the LG (Large G) domain, and an elongated, purely alpha
-helical domain. The domains are connected by a short intermediate
region consisting of one alpha -helix and a short two-stranded beta
-sheet. The connecting region is not an independent domain; it is
packed, via helix alpha 6, against the beta 1/ alpha 1 region of the LG
domain, away from the presumed nucleotide-binding site (see below). It
could be involved in stabilizing the relative location of the two
domains against each other. The helical domain is composed of seven
helices, which extend 90 Å away from the LG domain.
§5§ [43] Figure 1: Primary, secondary and tertiary structure of hGBP1. §5§
[44]Figure 1 : Primary, secondary and tertiary structure of hGBP1.
Unfortunately we are unable to provide accessible alternative text for
this. If you require assistance to access this image, or to obtain a
text description, please contact npg@nature.com
a, A model of the tertiary structure of hGBP1 presented as a ribbon,
where the LG domain is in purple, the connecting region in green, the
helical domain in yellow and alpha 12/ alpha 13 in cyan. Insertions,
marked I2–I5 in b, are in violet. Dashed lines indicate disordered
regions in the molecule. The tentative nucleotide-binding area,
identified by a Ras–GBP overlay, is indicated by a sphere with radius
7 Å. The topology is shown schematically using the same colour code. b,
Sequence alignment of hGBP1 (Swissprot accession no. P32455) with Mag-2
(EMBL acccession no. M81128) from mouse and chicken GBP1 (EMBL
accession no. X92112), with the secondary structure assignment as
determined using the programme DSSP^[45]29, with the same colour code
as in a, and aligned with the secondary structure of Ras dot
GppNHp^[46]30 (light blue lines). Contacts between helix alpha 12 and
the rest of the protein are indicated: asterisk, direct; circle,
water-mediated polar interactions; hash sign for both. For brevity,
sequences with lowest homology have been chosen.
[47]High resolution image and legend (233K)
§5§ [48] Table 1: Crystallographic data statistics §5§
[49]Table 1 - Crystallographic data statistics
[50]Full table
The LG domain of GBPs contains the conserved sequence elements of
GTP-binding proteins with modifications. Originally the N/TKxD motif
was believed to be absent in the GBPs^[51]2. An Asp-Asn mutation can
produce a change in specificity from guanine to xanthine nucleotides in
many GTP-binding proteins such as EF-Tu^[52]9. As an Asp-Asn mutation
in the ^181TLRD^184 motif of hGBP1 behaves similarly, it is postulated
that Asp 184 should bind the guanine base through a bidentate hydrogen
bond^[53]8. It was thus expected that GBP1 would contain the Ras G
domain or a variation thereof. The structure shows that the 278-residue
globular domain largely resembles the canonical architecture of Ras (
approx 170 residues), allowing for additions and insertions (labelled
I). Superimposing the structures of hGBP1 and Ras-GDP ([54]Fig. 2a)
gives a root mean square (r.m.s) deviation of 1.1 Å for 112 common C
alpha atoms. The LG domain consists of an eight-stranded beta -sheet
with six parallel and two anti-parallel strands surrounded by nine
helices, whereas Ras contains six beta -strands and five helices
([55]Fig. 1b). Using the secondary structure elements of Ras as the
basis for the comparison and retaining the corresponding numbering, the
additional elements are beta 0, alpha 0 and beta -1 on the N-terminal
side of the sheet, and helices alpha 3' (I3) and alpha 4' (I4)
([56]Fig. 1b). Apart from a short insertion (I1) in switch I, there are
comparatively long loop insertions between the ^97DxxG^100 motif and
alpha 2 (I2), and between beta 6 and alpha 5 (I5) of the canonical G
domain, respectively ([57]Figs 1 and [58]2). Whereas I1, I2 and I5
could be involved in nucleotide binding, I3 and I4, on the opposite
side of the protein, apparently mediate contact with the C-terminal
helix.
§5§ [59] Figure 2: Comparison of hGBP1 and Ras structures. §5§
[60]Figure 2 : Comparison of hGBP1 and Ras structures. Unfortunately we
are unable to provide accessible alternative text for this. If you
require assistance to access this image, or to obtain a text
description, please contact npg@nature.com
a, Superimposition of the LG domain of hGBP1 with the G domain of Ras
in complex with GDP(PDB accession no. 1Q21) as a stereo view.
N-terminal residues 1–36 of hGBP1 up to beta 1 have been omitted for
clarity. The colour code is as in [61]Fig. 1; Ras is in cyan. b,
Putative location of nucleotide-binding site in hGBP1. The regions of
hGBP1 potentially involved in binding the guanine nucleotide are shown
as obtained from a structural superimposition of RasGDP (in cyan) with
the corresponding regions in hGBP1 (purple), highlighting functionally
important residues necessary for binding and conformational change as
balls or in ball-and-stick. Whereas Gly 60^ras overlays very well with
Gly 100^hGBP1, residues D119/D184 and T35/T75 do not.
[62]High resolution image and legend (79K)
As GBP is stable in the absence of nucleotide, whereas Ras-like and G
alpha GTP-binding proteins are not, it was of interest to investigate
the effect of the absence of nucleotide on the structure. As all
P-loop-containing proteins^[63]10 bind the beta / gamma -phosphate of
the nucleotide in a similar manner, and as the role of Asp 184 in
binding the guanine base is similar to that of the Asp of the canonical
N/TKxD motif, we can locate the nucleotide-binding site of hGBP1 using
the RasGDP–hGBP1 overlay ([64]Fig. 2b). From this comparison we can
also see that, although part of the binding site is more accessible to
the solvent than in Ras–nucleotide complexes, part of the polypeptide
chain is in a position that interferes with nucleotide binding. Perhaps
owing to the absence of nucleotide, the polypeptide chain around the
binding site is mobile, as no electron density is visible for residues
69–72 (I1) in the region analogous to switch I, residues 190–193 close
to the ^181TLRD^184 motif and residues 244–257 in I5, close to the
SAK/L motif, which is conserved only in the Ras family and is absent in
GBPs.
The (phosphate-binding) P loop^[65]10, residues 45–52, adopts a
structure different from that of the Ras–nucleotide complexes. The
invariant lysine residue of the P loop does not interact with the
main-chain carbonyls for stabilization. Instead, in hGBP1 the loop is
stabilized by interactions with the region analogous to switch II,
involving hydrogen bonds between Tyr 47 (backbone N) and Asp 103,
Lys 51 and Thr 98. Furthermore, the structure is not suited for
nucleotide binding as the phosphates would clash with Tyr 47. The
region corresponding to switch I in Ras is disordered in hGBP1. Thr 75
appears to be analogous to Thr 35 in Ras, but is 5 Å away in the
overlay ([66]Fig. 2b). The ^97DxxG^100 motif of hGBP1 superimposes well
with that of switch II in Ras. D184 is 6 Å away from the corresponding
D119 of the canonical N/TKxD motif and would have to move accordingly
to occupy a similar position in the nucleotide-bound form. In general,
it appears that the guanine nucleotide-binding site is partly open
([67]Fig. 2a, [68]b) such that the incoming nucleotide would enter the
binding site base first and would then, after a corresponding
conformational change, bind into the phosphate-binding area, as
suggested for Ras by the structure of the Ras–Sos complex^[69]11.
Ras proteins have an intrinsic GTPase reaction rate in the order of
0-001–0.1 min^-1, whereas hGBP1 (ref. [70]8) has a rate of up to
80 min^-1 (see below). The catalytic mechanism for Ras and G alpha
proteins involves a glutamine (Gln 61 in Ras) that stabilizes the
transition state and is itself stabilized by GAP in the GAP-catalysed
mechanism^[71]12. As Gly 60^ras and Gly 100^hgbp1 are very close to
each other ([72]Fig. 2b), we would predict that Gln 61 of Ras is
structurally homologous to Leu 101 in hGBP1. Considering the importance
of Gln 61 in the intrinsic and GAP-catalysed GTPase reaction of Ras, we
conclude that the chemical mechanism of hydrolysis of GTP by hGBP1 is
different from that of Ras. A similar hydrophobic residue corresponding
to Gln 61 of Ras has also been identified in dynamins and Mx proteins.
Discounting the influence of oligomerization on the rate of the GTPase
reaction, only insertion I2 following Leu 101 appears to be close
enough to the phosphate-binding site to be involved in catalysis. This
^103DxEKGD^108 motif is conserved in GBPs and contains charged residues
that could have a catalytic role, either similar to that of the Arg in
G alpha proteins or GAPs^[73]12 or as a catalytic base.
The helical domain of hGBP1 can be considered to consist of two
three-helix bundles, formed by helices alpha 7, alpha 8, alpha 9
(residues 311–403) and helices alpha 9, alpha 10, alpha 11 (404–482),
where helix alpha 9 is common to both ([74]Fig. 1a). Adjacent to and
covering these two bundles is a very long helix alpha 12 that reaches
back to the LG domain. This helix, which is predicted from the sequence
to be a coiled-coil structure, is 78 residues long and stretches over
118Å. At its C-terminal end there is a helical turn leading into
another short helix, alpha 13, which makes a coiled-coil type of
interaction with alpha 12. The last eight residues, including the
prenylation-recognition Caax motif, not modified owing to recombinant
expression in E. coli, are not visible in the structure.
The two three-helix bundles give the molecule an elongated shape. The
core of the helical domains is formed by hydrophobic residues, whereas
the charged amino acids are exposed towards but interact only weakly
with the residues from the long helix alpha 12. Thus, alpha 12 has
almost no contact with the second (more remote) helical bundle (a
single hydrogen bond between Glu 490 and Tyr 433, [75]Fig. 1b) whereas
it forms seven direct side-chain and nine water-mediated contacts with
the first. It is stabilized further by four direct and eight
water-mediated contacts with insertions I3 and I4 ([76]Fig. 1) of the
LG domain. The electrostatic surface potential ([77]Fig. 3) indicates
that alpha 12 may mask some of the charged residues exposed by the two
helical bundles. In conclusion, the helical domain appears to actually
consist of two subdomains.
§5§ [78] Figure 3: Interaction of the C-terminal helix motif alpha 12/13
with the helical and the LG domains. §5§
[79]Figure 3 : Interaction of the C-terminal helix motif |[alpha]|12/13
with the helical and the LG domains. Unfortunately we are unable to
provide accessible alternative text for this. If you require assistance
to access this image, or to obtain a text description, please contact
npg@nature.com
The electrostatic surface potential shows that the highly charged
regions of the helical and LG domains are masked by an alpha 12/13
motif, as indicated in the lower panel by showing alpha 12/ alpha 13 in
worm representation.
[80]High resolution image and legend (40K)
Mx and dynamins are commonly grouped into a separate class of large
GTP-binding proteins^[81]5. Proteins of this family have several
variations in their domain structure, but they all possess at least a
'GTPase' domain of approx 300 residues, a 'middle' or 'assembly' domain
(150–200 residues) and a 'GED' domain (100 residues). These values are
close to the sizes of the LG, helical bundle and alpha 12/13 domains of
hGBP1, respectively. Biochemically, they share common properties such
as relatively low affinity for nucleotides (K[m] 10–500 micro M), the
ability to form oligomers and a high intrinsic rate of GTP hydrolysis
(k[cat] 2–100 min^-1)^[82]5, ^[83]7. It was thus interesting to find
out whether hGBP1 has to undergo multimerization to hydrolyse GTP
rapidly, as observed for dynamin. [84]Figure 4a shows that the specific
GTPase activity does increase at least eight-fold with increasing
concentrations under the conditions employed. This clearly indicates a
cooperative mechanism of GTP hydrolysis by hGBP1, similar to that
observed for dynamin^[85]13. Note, however, that for technical reasons
the data do not allow us to estimate the GTPase reaction rate at lower
concentrations.
§5§ [86] Figure 4: Cooperative GTP hydrolysis and nucleotide-dependent
multimerization of hGBP1. §5§
[87]Figure 4 : Cooperative GTP hydrolysis and nucleotide-dependent
multimerization of hGBP1. Unfortunately we are unable to provide
accessible alternative text for this. If you require assistance to
access this image, or to obtain a text description, please contact
npg@nature.com
a, Concentration dependence of the specific GTPase activity of hGBP1.
The data were fitted to a model involving dimer formation yielding 0.6
micro M for the apparent dimerization constant of hGBP1. b,
Size-exclusion chromatography of GDP dot AlF[x]-, GppNHp-, GDP-bound
and nucleotide-free hGBP1.
[88]High resolution image and legend (20K)
The homology to the dynamin family is further corroborated by
demonstrating nucleotide-dependent oligomerization ([89]Fig. 4b). The
elution profiles from standardized analytical size-exclusion
chromatography show that hGBP1 is monomeric in the absence and in the
presence of GDP, whereas it shows higher molecular mass when bound to
the non-hydrolysable GTP analogue GppNHp. The molecular mass is higher
again for hGBP1 in complex with GDP and aluminium fluoride which
together are believed to mimic the transition state of GTP
hydrolysis^[90]8, ^[91]12. It seems that, for efficient GTP hydrolysis
to occur, higher order structures have to be formed, as suggested for
dynamin^[92]13, ^[93]14, ^[94]15. The dependence of nucleotide
hydrolysis on oligomerization has been proposed to be important for
dynamin's ability to assemble around the necks of clathrin-coated
vesicles and regulate receptor-mediated endocytosis. A role for the
oligomerization-dependent hydrolysis of GTP by hGBP1 in
interferon-mediated cellular functions remains to be established.
In addition to the biochemical experiments, structural considerations
lead us to propose that hGBP1 belongs to the dynamin family of large
GTP-binding proteins. Secondary structure predictions suggest a helical
and/or coiled-coil structure for the middle and GTPase effector (GED)
domains of Mx and dynamin, just as the secondary structure prediction
for alpha 7– alpha 12 of hGBP1 is in tune with the current structure.
Limited proteolysis experiments^[95]16 and yeast two-hybrid
studies^[96]17 have shown that the C-terminal end of Mx, which is
required for fast GTP hydrolysis, contacts the nucleotide-binding
domain and the middle domain just as helix alpha 12 folds back onto the
LG domain and the helical bundles in the structure of hGBP1. A similar
pattern of interactions has also been observed for dynamin where the
binding of the GED domain to both the middle and the LG domain has been
shown using various techniques^[97]18, ^[98]19. Similarly, the internal
GAP-related GED domain of dynamin is implicated not only in GTP
hydrolysis but also in the regulation of assembly: two processes that
are intimately coupled^[99]14, ^[100]20. With the structure of hGBP1
now available, we can design experiments to test the hypothesis of a
common structural make-up for these proteins. We would also investigate
the mechanism of GTP hydrolysis in GBPs, the unique nature of the
resulting products (GMP and GDP) and whether an internal helical GED
domain with properties similar to that of dynamin exists in GBPs.
Furthermore, the role of the C-terminal end in the biological function
of GBPs in the antiviral defence should be investigated, as the
functionally related Mx protein requires its C-terminal end to interact
with thogoto virus nucleocapsids^[101]21.
[102]Top of page
§3§ Methods §3§
§4§ Protein preparation §4§
hGBP1 with an N-terminal His[6] tag was expressed from a pQE9 vector in
E. coli strain BL21(DE3) as described^[103]8. The GTPase activity of
hGBP1 was measured at 37 °C in a buffer containing 50 mM Tris/HCl,
pH 8.0, 5 mM MgCl[2] and 300 micro M GTP. Aliquots were taken at
adequate time intervals and the progress of GTP hydrolysis was analysed
by reversed phase chromatography (C-18, Hypersil)^[104]8. Rates were
derived from a linear fit to the initial reaction (<30% GTP
hydrolysed).
We analysed the states of multimerization of hGBP1 by size-exclusion
( run in 1.932 second using v1.01-cache-2.11-cpan-8f98c5d2c55 )