672 G. Jones, D. Jones Insect Biochemistry and Molecular Biology 30 2000 671–679
tera and Lepidoptera e.g., Henrich et al., 1990; Oro et al., 1990; Shea et al., 1990; Tzertzinis et al., 1994; Kapit-
skaya et al., 1996; Jindra et al., 1997; Perera et al., 1998; Vogtli et al., 1999. The sequence identity between these
USPs versus the vertebrate and invertebrate RXRs is in the range of 40–50, while that between invertebrate
RXRs and vertebrate RXRs is around 70. However, these statistics in and of themselves do not provide a
basis for inferring whether USPs bind ligands and what those ligands might be. For example, vertebrate RAR
and RXR both bind the same ligand 9-cis RA, yet the percent identity between them is only around 27
Mangelsdorf et al., 1990. The relevant indicators are whether the residues that have been retained identically
or with conservative substitution are those residues necessary to preserve the secondary structures for basic
architecture of the ligand-binding domain, and those necessary to preserve a functional ligand-binding pocket
within the ligand-binding domain. Thus, in the analyses below, we do not confer importance to a residue in USP
merely because it is conserved in USPs, but rather also because of its location in USP relative to the functional
location of ligand- or co-activator-associated residues conserved in other nuclear receptors.
2. Secondary and tertiary structure
The crystal structure for unliganded human RXR α
hRXR α
has been reported, including the residues part- icipating in the various secondary structures Bourguet
et al., 1995. It is possible to gauge the accuracy of algorithms for prediction of secondary structure by com-
paring the consensus predictions of several methods for hRXR
α against the secondary structure arrangement
actually observed in crystallized hRXR α
. Using the vari- ous secondary structure prediction alogithms available at
the http:pbil.ibcp.frNPSA web site, we performed such an analysis. Fig. 1 shows the consensus locations of pre-
dicted secondary structures for hRXR α
vs those actually observed in the crystal structure, for the regions of the
ligand-binding domain that contain corresponding resi- dues known to line the ligand-binding pocket of hRAR
γ .
In general, the analyses reasonably predict for hRXR α
the locations of α
-helices. We then applied those analyti- cal methods to Drosophila melanogaster USP dmUSP.
The predictions of α
-helical secondary structure for dmUSP generally parallel the predictions for hRXR
α Fig. 1. These data on the predicted arrangement of sec-
ondary structures in dmUSP support the hypothesis that the general tertiary architecture of the ligand-binding
domain LBD of dmUSP is similar to that of the LBD for hRXR
α . The remaining analyses below are rested on
this hypothesis that the general tertiary architecture of secondary structures of the LBD of dmUSP is similar to
that of the LBD for hRXR α
.
3. Receptor residues contacting oxygen-containing moieties
The crystal structure of the ligand-binding domain has been reported for apo-hRXR
α , holo-hRAR
γ , holo-rat
thyroid hormone receptor rTR β
1, holo-human estrogen receptor hER, and both apo- and holo-human pro-
gesterone receptor hPPAR γ
Bourguet et al., 1995; Renaud et al., 1995; Wagner et al., 1995; Brzozowski et
al., 1997; Nolte et al., 1998; Shiau et al., 1998; Tanen- baum et al., 1998; Uppenberg et al., 1998; Williams and
Sigler, 1998; Xu et al., 1999. The ligand-binding pock- ets of these LBDs are lined primarily with hydrophobic
amino acid side chains Fig. 1. Those residues in the pocket that are charged are in most cases those residues
that hydrogen bond with hydroxyl, keto or carboxyl groups of the respective ligands Fig. 2. In every case
examined thus far of a crystallized or modelled, for hRXR
α nuclear receptor complexed with an endogen-
ous ligand containing a ‘terminal’ hydroxyl, keto or car- boxyl group at its ‘leading end’, that group is hydrogen
bonded to two residues which each are placed on differ- ent secondary structures; either one on
α -helix 3 and the
other on α
-helix 5, or one on α
-helix 5 and the other on a
β -turn S1–S2 Figs. 1 and 2. Thus far, the hydrogen
bonds of the hydroxyl-, keto- and carboxyl-containing moieties always include at least a basic residue arginine
or histidine on either α
-helix 3 or 5, and a nonbasic, but hydrogen-bonding residue glutamine, glutamic acid
or serine on either the other helix or on the β
-turn Fig. 2. hPPAR
γ , which has some additional secondary struc-
tures and a tertiary arrangement most different from the other nuclear receptors, also binds the carboxyl group of
its ligand through histidine on α
-helix 5, but is excep- tional in that a histidine on
α -helix 11 also hydrogen
bonds with that same carboxylate group. On the basis of the above pattern of the nature and locations of hydro-
gen-bonding residues in a variety of nuclear receptors, we make the following hypothesis: if USP binds to an
endogenous ligand containing an oxygen-substituted end-group, that binding will be by hydrogen-bond of that
end-group through at least arginine or histidine contacts at one of either
α -helices 3 or 5, and perhaps through a
second contact on the other respective α
-helix 3 or 5 or on the
β -turn.
In hRXR α
, the apparent closest relative of dmUSP that has been crystallized, the modelled structure of 9-
cis RA in the RXR ligand-binding pocket places I268 and Q275 on
α -helix 3 as straddling on each side of
the ligand, near the backbone double bond positioned between the
α and
β carbons respective to the carboxyl
group C14 and C15, Fig. 2. Those two residues are conserved in higher insect USPs as an invariant L and
the correspondng invariant Q, respectively. Closer examination of the the residues on
α -helices 3
and 5 and the β
-turn that contact the oxygen-substituted
673 G. Jones, D. Jones Insect Biochemistry and Molecular Biology 30 2000 671–679
674 G. Jones, D. Jones Insect Biochemistry and Molecular Biology 30 2000 671–679
Fig. 1. Conservation of primary, secondary and tertiary structure of
portions of ligand-binding domain of vertebrate nuclear receptors in comparison with primary and inferred secondary and tertiary structure
of portions of ligand-binding domain of ultraspiracle USP proteins from higher insects. For the vertebrate receptors, the
α -helices barrels
or β
-sheet block arrows contain the primary sequence shown by crys- tallography to constitute that secondary structure in the respective
receptor. Open rectangles around hRXR α
and around the sequence for dmUSP show the consensus of secondary structure predictions. Resi-
dues coded in orange are present in the ligand-binding pocket of the crystallized receptor. Residues coded in dark brown touch oxygen-con-
taining moieties on the ligand in the ligand-receptor cocrystal. Blue bars connecting residues in two receptors are positions in which differ-
ences in hydrophobic side chains discriminate between ligands of the two receptors. Insect USP and dmE75A sequences are shown in align-
ment below the vertebrate receptors. Light brown coded residues denote highly conserved insect sequences at selected positions. Purple
coded dmE75 residues denote positions in which the dmE75 residue is a non-conservative change from the conserved series at the respect-
ive USP position. Dark brown underlining for
α -helices 3, 1011 and
the β
-turn indicate narrower regions in vertebrate receptors that contact oxygen-containing moieties on the respective ligands. Primary
sequences and alignments for vertebrate receptors are from Wurtz et al. 1996. Sources for secondary and tertiary structure designations
and receptor acronyms are as follows: hPR =
human progesterone recep- tor Williams and Sigler, 1998; hER
= human estrogen receptor
Brzozowski et al., 1997; hRAR γ=
human retinoic acid receptor γ
Renaud et al., 1995; hRXR α=
human retinoid X receptor α
Bourguet et al., 1995; rTR
β 1
= rat thryoid hormone receptor
β 1; dm
= Drosophila
melanogaster Oro et al., 1990; aa =
Aedes aegypti Kapitskaya et al., 1996; ct
= Chironomus tetans Vogtli et al., 1999; bm
= Bombyx mori
Tzertzinis et al., 1994; cf =
Choristoneura fumiferana Perera et al., 1998; ms
= Manduca sexta Jindra et al., 1997.
leading end of ligands of a variety of crystalized ligand- receptor complexes leads to additional hypotheses on
candidate residues in dmUSP that may also contact an oxygen-substituted endogenous ligand. The residues on
α -helix 3 reported to hydrogen-bond with the hydroxyl-,
keto-, or carboxyl moiety of the respective ligands are all in positions within 5 amino acids of each other on
helix 3 Fig. 1. In this immediate area on hRXR α
are also the I268 and Q275 that are modelled as just distal
to the leading carboxyl group. The only invariant resi- dues in that region for the USPs are those in the motif
LXXXXXKQ, where L and Q correspond to I268 and Q275 of hRXR
α . Further, on
α -helix 5, hPR, hER,
hRAR γ
and hRXR α
all possess a conserved arginine that participates in hydrogen-bonding to the hydroxl-, keto
or carboxyl- moiety of their ligand. However, in several of the USPs, this residue is neither arginine nor another
basic residue. Based on the nature and locations of con- served residues in USP
α -helix 3 relative to the corre-
sponding placement of ligand-associated residues in other crystallized receptors, we hypothesize that should
higher insect USPs bind an endogenous ligand with ter- minal oxygen-substitution, then either the conserved K
or Q on helix 3 participates in hydrogen-bonding to that moiety.
More problemmatic is conjecturing which residues are candidates for a second hydrogen bond with the oxygen-
containing moiety of a putative USP ligand. The short region between S1 and S2 of hRAR
γ and rTR
β 1 contains
in those receptors the serine residue making the second hydrogen bond to the leading-end oxygen-containing
moiety. It is likely of significance that this short region has been substituted for a larger and more variable
region in the higher insect USPs Fig. 1. This region in each USP apparently contains an invariant aspartic acid,
in addition to at least one serine.
Also relevant to this analysis is comparison of the status of these hypothesized ligand-associated residues
of USP with the corresponding residues in dmE75A. This latter regulator has also been hypothesized to be
an orphan receptor, or, at least, there have been no reports of it binding to juvenile hormones or JH-like
molecules Thummel, 1995; Buszczak and Segraves, 1998. dmE75A does not conserve the invariant Q on
α -helix 3 of USP which is instead V417, the invariant
K on α
-helix 3 is instead H408, and the invariant IL on
α -helix 3 is instead F410. In addition, two other resi-
dues on that α
-helix 3 that substantively participate in creating the hydrophobic environment of the ligand-
binding pocket in each of the crystallized vertebrate receptors have been changed to charged E and R resi-
dues at the corresponding positions in α
-helix 3 of dmE75A E407 and R413. On
α -helix 5, dmE75A has
also replaced an otherwise invariantly hydrophobic USP residue IL with a charged R452. Likewise, a hydro-
phobic residue otherwise invariantly conserved on helix 7 in both RXRs and USP, and which contributes to the
hydrophobicity of the pocket in several receptors, is instead a polar threonine residue T491 in dmE75A.
One current hypotheses is that E75A and USP are orphan receptors that do not have a ligand. Another is
that either may have a ligand that is as yet unknown. There has been discussion in the literature of whether
the differences that exist in identity of residues between the USPs and RXRs in the ligand-binding domain
reflect that USP does not have a ligand-binding func- tion, i.e. there would then be a loss of selection for
maintenance in USP of those RXR-particular residues Kapitskaya et al., 1996; Hayward et al., 1999. Under
the hypothesis that USP and E75A are orphan receptors with no ligand, neither E75A nor USP would be under
selection pressure to maintain a ligand-binding pocket architecture that is favorable for binding hydrophobic,
end-oxygenated ligands. However, the analyses here on comparison of the differences between dmE75A and
RXRsUSPs for conservation of ligand-associated resi- dues in RXR
α -helices 3 and 5 raises the important con-
sideration of why USP does not show similar encroach- ment of charged residues in its ligand-binding pocket
as does dmE75A, if USP is not under selection pressure to maintain a hydrophobic ligand-binding pocket to bind
hydrophobic ligands? This consideration does not
675 G. Jones, D. Jones Insect Biochemistry and Molecular Biology 30 2000 671–679
Fig. 2. Summary of specific hydrogen bonding contacts between specific ligand-binding pocket residues and oxygen-containing motifs of the
respective ligands, as shown by co-crystallography. The residue of a particular color moiety of the same color for each receptor-ligand combination shown. Also provided are the hydrogen-bonding distances from the indicated residue to the respective site on the ligand, or to a water molecule
that in turn h-bonds with the oxygen-containing moiety. EPA =
eicosapentaenoic acid; AT-RA =
all trans retinoic acid; DIMIT =
3,5-dimethyl-3 9-
isopropylthryonine. Each specific oxygen-containing moiety shown is a part of the indicated ligand and not an extra moiety added to it. All other abbreviations as in Fig. 1.
imply that E75A does not have a ligand, but rather inquires as to why the USPs, but not E75A, have been
selected to conserve residues that in RXR provide a hydrophobic ligand-binding environment.
The end of the ligand-binding pocket that is near the ligand entrance point shows more divergence among
receptors. However, a narrow region on α
-helix 11 is the location of residues that in some crystallized recep-
tors make additional hydrogen-bonds, again with other hydroxyl or keto moieties that are also present on the
opposite end of the respective ligand e.g., progesterone and thryoid hormone, Figs. 1 and 2. In both hER and
rTRb1, it is a basic histidine that makes the respective hydrogen-bond Fig. 2. In the same narrow respective
region of dmUSP is an invariant K472 and an invariant H476, neither of which is conserved in dmE75A
although the change from histidine in dmE75A by K578 is conservative. Based on these available data, we
hypothesize that if USP does bind an endogenous ter- penoid ligand through a hydrogen bond to an oxygen-
containing moiety on the ligand’s distal end, then either the conserved lysine or conserved histidine participates
in that bond. We also note that the insect juvenile hor- mones possess an oxygen-substitution epoxide group
at its end opposite from its carboxyl ester group, and that we did not observe methyl farnesoate, that is miss-
ing the epoxide, to bind to dmUSP Jones and Sharp, 1997.
676 G. Jones, D. Jones Insect Biochemistry and Molecular Biology 30 2000 671–679
4. Conservation of residues participating in ligand- induced coactivator-binding