CD1e molecules interacted with lipid surfaces enriched in anionic lipids, such as bismonoacylglycerophosphate, a late endoso-mal⁄ lysosomal lipid, especially at acidic pH, or when the pr
Trang 1CD1e at endosomal pH
Natalia Bushmarina1,2, Sylvie Tourne1,2, Gae¨lle Giacometti1,2, Franc¸ois Signorino-Gelo1,2,
Luis F Garcia-Alles3,4, Jean-Pierre Cazenave2,5, Daniel Hanau1,2and Henri de la Salle1,2
1 INSERM, UMR-S725, INSERM-Universite´ de Strasbourg, France
2 Etablissement Franc¸ais du Sang-Alsace, Strasbourg, France
3 CNRS, IPBS (Institut de Pharmacologie et de Biologie Structurale), Toulouse, France
4 Universite´ de Toulouse, UPS, IPBS, France
5 INSERM, UMR-S949, Etablissement Franc¸ais du Sang-Alsace, Strasbourg, France
Introduction
CD1 molecules are nonclassic major histocompatibility
complex class I molecules that are mainly expressed by
dendritic cells, the professional antigen-presenting cells
of the immune system CD1 proteins are heterodimers composed of a poorly polymorphic membrane-anchored a-chain and b2-microglobulin (b2m) In humans, four
Keywords
CD1e; conformational changes; lipid binding;
structure; surface plasmon resonance
Correspondence
H de la Salle or N Bushmarina,
Etablissement Franc¸ais du Sang-Alsace,
10 rue Spielmann, 67065 Strasbourg, France
Fax: +33 388 212 544
Tel: +33 388 212 525
E-mail: henri.delasalle@efs-alsace.fr;
natalia.bushmarina@gmail.com
(Received 25 January 2011, revised 30
March 2011, accepted 4 April 2011)
doi:10.1111/j.1742-4658.2011.08118.x
The plasma membrane proteins CD1a, CD1b and CD1c are expressed by human dendritic cells, the professional antigen-presenting cells of the immune system, and present lipid antigens to T lymphocytes CD1e belongs to the same family of molecules, but accumulates as a membrane-associated form in the Golgi compartments of immature dendritic cells and
as a soluble cleaved form in the lysosomes of mature dendritic cells In lysosomes, the N-terminal propeptide of CD1e is also cleaved, but the functional consequences of this step are unknown Here, we investigated how the pH changes encountered during transport to lysosomes affect the structure of CD1e and its ligand-binding properties Circular dichroism studies demonstrated that the secondary and tertiary structures of recombi-nant CD1e were barely altered by pH changes Nevertheless, at acidic pH, guanidium chloride-induced unfolding of CD1e molecules required lower concentrations of denaturing agent The nonfunctional L194P allelic vari-ant was found to be structurally less stable at acidic pH than the functional forms, providing an explanation for the lack of its detection in lysosomes The number of water-exposed hydrophobic patches that bind 8-anilino-naphthalene-1-sulfonate was higher in acidic conditions, especially for the L194P variant CD1e molecules interacted with lipid surfaces enriched in anionic lipids, such as bis(monoacylglycero)phosphate, a late endoso-mal⁄ lysosomal lipid, especially at acidic pH, or when the propeptide was present Altogether, these data indicate that, in the late endosomes⁄ lyso-somes of DCs, the acid pH promotes the binding of lipid antigens to CD1e through increased hydrophobic and ionic interactions
Abbreviations
ANS, 8-anilinonaphthalene-1-sulfonate; bis-ANS, 4,4¢-bis(1-anilinonaphthalene-8-sulfonate); b2m, b 2 -microglobulin; BMP,
bis(monoacylglycero)phosphate; ER, endoplasmic reticulum; NBD, nitrobenzoxadiazole; PtdCho, phosphatidylcholine; PtdSer,
phosphatidylserine; PtdIns, phosphatidylinositol; PtdInsM6, hexamannosylated phosphatidylinositol; rsCD1e, recombinant soluble CD1e; rsCD1b, recombinant soluble CD1b; rsCD1e2, recombinant soluble CD1e2; rsCD1e4, recombinant soluble CD1e4; sCD1e, soluble CD1e; TMP, transition midpoint.
Trang 2forms (CD1a–CD1d) are directly involved in the
presen-tation of lipid antigens to T cells These proteins are
internalized from the plasma membrane, and then traffic
through the endocytic pathway, where they capture
anti-genic ligands, before returning to the plasma membrane
to stimulate antigen-specific T cells In contrast, newly
assembled CD1e molecules are transported from the
endoplasmic reticulum (ER) to the endocytic pathway
without passing through the plasma membrane [1] In
late endosomal compartments, CD1e undergoes double
processing and becomes functional; it is cleaved into a
soluble form [soluble CD1e (sCD1e)], and a nonpolar
12-residue N-terminal propeptide (APQALQSYHLAA)
is removed [2,3] This propeptide is unique to CD1e
among classic and nonclassic human
leukocyte-associ-ated class I molecules It facilitates the assembly of
a-chain–b2m complexes in the ER, but has no other
ascribed function [3] Soluble lysosomal CD1e molecules
participate in the processing of antigenic glycolipids
pre-sented by CD1b [4,5] Thus, coexpression of both CD1b
and CD1e by antigen-presenting cells is indispensable
for the activation of specific T-cell clones by
hexaman-nosylated phosphatidylinositol (PtdInsM6), a
structur-ally complex mycobacterial glycolipid The use of
antigen-presenting cells deficient in lysosomal
a-man-nosidase and of recombinant sCD1e (rsCD1e) produced
in Drosophila cells has allowed us to demonstrate that
sCD1e molecules bind PtdInsM6and facilitate the
com-plete processing of its four a-mannoses by lysosomal
a-mannosidase into dimannosylated PtdIns, a
CD1e-independent CD1b-restricted antigen [5] Additional
investigations showed that, among the six natural
vari-ants of CD1e, only one is unable to sustain PtdInsM6
presentation This molecule, CD1e4, is characterized by
the replacement of Leu194 by proline, as compared with
the common CD1e1 variant In human cells, only small
amounts of CD1e4 reach late endosomal compartments,
and soluble lysosomal forms are not detected
Never-theless, recombinant soluble CD1e4 (rsCD1e4) can be
normally expressed in insect cells and, like other natural
variants, assists in vitro digestion of PtdInsM6 by
a-mannosidase [6]
CD1b, CD1c, CD1d and CD1e transit through
acidic endosomal compartments and CD1b and CD1d,
at least, are subject to pH-dependent conformational
changes Acidification of late endosomal compartments
is required for the presentation of several
CD1b-restricted and CD1d-CD1b-restricted antigens, for at least
two reasons First, as shown for CD1b and CD1d,
lipid loading appears to be mediated by lysosomal
lipid transfer proteins [7–9], which are optimally
func-tional at acidic pH [10] Second, acid-induced
confor-mational changes allow CD1b and CD1d to adopt a
conformation with partially unfolded a-helices, thereby facilitating the access of hydrophobic ligands to their antigen-binding pockets [11–13]
The aim of this study was to determine how acidifi-cation modifies different structural features of CD1e and affects its interaction with lipid membranes and ligands The role played by the propeptide in these processes was also investigated Finally, we looked at how these properties are modified in the immunologi-cally nonfunctional CD1e4 variant
Results
The secondary structure of rsCD1e is stable at physiological pH
In this work, recombinant soluble CD1e2 (rsCD1e2) molecules including or not including the propeptide (rsCD1e2+ or rsCD1e2)) were produced in Drosoph-ila melanogasterS2 cells First, we investigated how
pH variations similar to those occurring during transport from neutral Golgi to acidic lysosomal com-partments influence the secondary structure of the active lysosomal form, rsCD1e2) The alteration of the secondary structure was followed with far-UV (190–240 nm) CD spectroscopy Similar experiments were performed with rsCD1e2+, in order to determine the impact of the propeptide on the stability of CD1e
As shown in Fig 1, the circular dichroism spectra of rsCD1e2) and rsCD1e2+ did not differ significantly, and remained unaltered over pH values ranging from
4 to 7 (Fig 1), being affected only at pH < 3.5 (data not shown) The pronounced minimum at 219 nm and the maximum at 195–196 nm are characteristic features
of a⁄ b class proteins with a major b-sheet content The percentages of different secondary structures calculated from these spectra, namely 16 ± 1% a-helices,
37 ± 1% b-strands, and 47 ± 1% other structures [14,15], are in full agreement with the content deduced from a homology model derived from the crystal struc-ture of recombinant soluble CD1b (rsCD1b) [5]
Physiological pH changes induce minor perturbations in the tertiary structure of CD1e
We next examined the changes in the tertiary structure
of rsCD1e2) and rsCD1e2+ induced by acidification,
by near-UV (250–320 nm) circular dichroism spectros-copy This method allows characterization of the envi-ronment of the aromatic amino acid side chains in proteins, and thus gives information about the com-pactness of the tertiary structure It is widely used to study the conformational changes caused by
Trang 3physico-chemical perturbations [16] At pH 7, the near-UV
spectra of rsCD1e proteins displayed several
pro-nounced peaks characteristic of native proteins with a
compact tertiary structure, with no significant
differ-ences between the two types of rsCD1e2 molecule
(Fig 2A,B) Shifting the pH to 4.8 only slightly
decreased the ellipticity, revealing a subtle increase in
the flexibility of the tertiary structure at lysosomal pH
(Fig 2A,B)
Enhanced interaction between rsCD1e2 molecules
and hydrophobic probes at endosomal pH
As the overall structure of rsCD1e2 molecules
under-went no major pH-induced alterations, we decided to
investigate whether pH variations influence solvent
accessibility to the hydrophobic interfaces of CD1e,
which include the lipid-binding groove Hence, we
mea-sured the fluorescence resulting from binding to CD1e
of 8-anilinonaphthalene-1-sulfonate (ANS) and
4,4¢-bis(1-anilinonaphthalene-8-sulfonate) (bis-ANS) These
probes are widely used to study the solvent-accessible hydrophobic surfaces of proteins [17] ANS binds strongly to hydrophobic clusters associated with loose tertiary contacts or hydrophobic binding sites Bis-ANS
is a superior molecular probe for nonpolar cavities in proteins, and allows the determination of saturation curves
At a given pH and probe concentration, rsCD1e2+ and rsCD1e2) bound similar amounts of ANS A shift from neutral to acidic pH caused a significant increase in fluorescence intensity in the two forms of CD1e, with values that were a function of the ANS concentration (Fig 3A) Experiments with bis-ANS demonstrated a similar dependence on bis-ANS con-centration for both rsCD1e2+ and rsCD1e2) How-ever, bis-ANS binding saturation could only be attained
at pH 4.5 and not at neutral pH, regardless of the CD1e2 molecule studied It is also noteworthy that a nearly two-fold higher maximal fluorescence was obtained for rsCD1e2+ than for rsCD1e2) (Fig 3B)
At pH 4.5 and substoichiometric bis-ANS⁄ CD1e
Fig 1 pH dependence of the secondary structure of rsCD1e molecules rsCD1e2 +
or rsCD1e2) were diluted to 4 l M in 5 m M monoso-dium ⁄ disodium phosphate (pH 7) or 5 m M sodium phosphate ⁄ citrate (pH 4) buffer containing 150 m M sodium sulfate The solutions were incubated overnight, and the far-UV circular dichroism spectra were recorded in 1-mm cuvettes MRW, mean residue weight; res, residue.
Fig 2 pH dependence of the tertiary structure of rsCD1e molecules rsCD1e proteins were diluted to 20 l M in 5 m M sodium phosphate buffer (pH 7 or pH 4.8) containing 150 m M sodium sulfate (A, B) The near-UV circular dichroism spectra of CD1e2)(A) and CD1e2 + (B) were recorded at pH 7 and pH 4.8 (C) Comparison of the near-UV circular dichroism spectra of rsCD1e2)and rsCD1e4)at pH 7 MRW, mean residue weight; res, residue.
Trang 4ratios, the fluorescence was proportional to the
probe concentration (Fig 3C) This allowed us to draw
a Scatchard plot of the data obtained in three
indepen-dent experiments (one representative experiment is
shown in Fig 3C), and to deduce that rsCD1e2) and
rsCD1e2+bound, respectively, 12 ± 0.3 and 24 ± 0.7
bis-ANS molecules, with an apparent Kd of
16 ± 0.2 lm for the two proteins In conclusion, experi-ments with ANS and bis-ANS confirmed a beneficial effect of lysosomal pH on the binding of hydro-phobic ligands to both rsCD1e2+ and rsCD1e2) In addition, bis-ANS fluorescence data indicated that the CD1e propeptide influences the number of binding sites
A
B
C
D
Fig 3 Binding of ANS and bis-ANS to
rsCD1e molecules rsCD1e proteins were
diluted to 2 l M in pH 7 and pH 4.5 buffers
as described in Fig 1, and incubated in the
presence of different concentrations of ANS
(A) or bis-ANS (B, C, D) for 30 min The
fluo-rescence intensity of the solutions was then
measured [k ex = 370 nm and k em = 480 nm
(ANS); kex= 390 nm and kem= 490 nm
(bis-ANS)] in a FlexStation automate Each
condition in each row (B and D) was tested
in triplicate on a same day However each
of these rows corresponds to a set of
experiments performed on an different day.
Mean values of triplicate analyses with their
respective standard deviations are shown.
The binding of bis-ANS to CD1e proteins
was studied at low [bis-ANS] ⁄ [CD1e] ratios
(< 1), or in the presence of an excess of
bis-ANS (up to 25-fold molar excess) The
left panel shows the bis-ANS fluorescence
as a function of [bis-ANS], for [bis-ANS] ⁄
[CD1e2] < 1 Scatchard representations of
the binding of bis-ANS to rsCD1e2)and
rsCD1e2 + in one representative experiment
at pH 4.5 are shown in the middle and right
panels (D) Comparison of the binding of
bis-ANS to rsCD1e2 and rsCD1e4, with and
without the propeptide.
Trang 5Unfolding of rsCD1e is facilitated at acidic pH
To determine the influence of pH on the structural
robustness of rsCD1e2 molecules, we examined the
effect of the denaturing agent guanidium chloride at
pH 7 and pH 4.8 First, the stability of the secondary
structure of the molecules was analyzed by far-UV
cir-cular dichroism spectroscopy The values of the
normalized ellipticity at 219 nm (h219 nm) for rsCD1e2)
and rsCD1e2+ are plotted in Fig 4 as a function of
guanidium chloride concentration The differences
between rsCD1e2)and rsCD1e2+proved to be rather
subtle, as the transition midpoint (TMP) of the
unfold-ing curve of the secondary structure of rsCD1e2+
(Table 1) was slightly higher than that of rsCD1e2)at
pH 7 (4 m versus 3.7 m guanidium chloride,
respec-tively), whereas the two TMPs were equal at pH 4.8
(3.8 m)
The resistance of tertiary contacts in rsCD1e2
mole-cules to guanidium chloride-induced denaturation was
then explored by measuring the intrinsic fluorescence
of the proteins and the extent of ANS binding A
com-parison of the intrinsic fluorescence of rsCD1e2+ and
rsCD1e2), at different guanidium chloride
concentra-tions and pH values, revealed that the two CD1e
mole-cules were more susceptible to guanidium
chloride-induced denaturation at acidic pH (Fig 5A and
Table 1) Although the two forms of rsCD1e2 bound
equivalent amounts of ANS, the concentration of
gua-nidium chloride required to induce maximal ANS binding decreased from 3–3.2 m at neutral pH to 2.3 m
at acidic pH (Fig 5C); these values are close to the TMPs of the denaturation curves of the tertiary struc-tures (Table 1)
Altogether, these data suggest that, upon arrival in the lysosomes, CD1e could become less stable through exposure of hydrophobic surfaces, including, presum-ably, the lipid-binding pocket
The propeptide moderately influences lipid binding to rsCD1e molecules
The binding of lipids to CD1e proteins was investi-gated by using phosphatidylserine (PtdSer) with a fluo-rescent nitrobenzoxadiazole (NBD) moiety linked to the terminus of one of the fatty acid chains The bind-ing of PtdSer–NBD to rsCD1e was analyzed by mea-suring the increase in the fluorescence of the NBD group upon its insertion into the hydrophobic lipid-binding pocket of CD1e The kinetics of lipid-binding to the different rsCD1e molecules were compared at pH 7 and pH 4.5 Three independent experiments were per-formed, resulting in similar profiles for the different protein and experimental conditions The reproducibil-ity of the experiments was validated by determining the fluorescence intensities at the end of the assays, and the times required to reach half these values The standard deviation of these parameters deduced from the three experiments fell between 2% and 5% (data not shown); representative curves for each condition are shown in Fig 6 Regardless of the pH, the binding
of PtdSer–NBD to CD1e reached a plateau for all pro-teins, except for rsCD1e2), at pH 7 For a given rsCD1e molecule, the maximal fluorescence intensity was barely affected by the pH The time required to reach half the maximal lipid binding was significantly less at acidic pH for rsCD1e2+ but not for rsCD1e2) (data not shown) These results indicate that the two CD1e2 proteins, with and without the propeptide, are fairly equivalent in terms of PtdSer binding
Enhanced interaction of rsCD1e with anionic lipids at endosomal pH
The interaction of rsCD1e molecules with liposomes immobilized on sensor chips was studied by surface plasmon resonance The liposomes were composed
of mixtures of neutral phosphatidylcholine (PtdCho) (70 molÆ%) and either PtdSer, phosphatidylinositol (PtdIns) or sulfatide (30 molÆ%), or PtdSer plus bis(monoacylglycero)phosphate (BMP) (15 molÆ% of each) for BMP-enriched liposomes RsCD1e2+ and
Fig 4 Guanidium chloride-induced unfolding of the secondary
struc-ture of rsCD1e molecules rsCD1e2 + (h, j), rsCD1e2)(4,m) and
rsCD1e4) (), ¤) were diluted to 20 l M in 10 m M monosodium ⁄
disodium phosphate buffer containing 150 m M sodium sulfate at
pH 7 (open symbols) or pH 4.8 (closed symbols) The ellipticity at
219 nm at a given molar concentration of guanidium chloride
(h[219; M ]) was normalized with the following formula: (h[219; M ] –
h[219;6]) ⁄ h[219;0].
Trang 6rsCD1e2) were injected over the lipid surfaces for
300 s, after which the surfaces were rinsed with buffer
for the same period of time For comparison, rsCD1b
was also analyzed
Irrespective of the pH, in terms of resonance units,
the interaction of rsCD1e2) with liposomes was
con-siderably higher than that of rsCD1b molecules
(Fig 7) Moreover, at acidic pH, the interaction of
rsCD1e2) with liposomes increased significantly, the
intensity of the signal being three times higher The
protein–liposome interactions were also dependent on
the liposome composition Thus, the order of preference
was sulfatide > BMP + PtdSer > PtdIns@ PtdSer for rsCD1e2)at pH 7 or pH 4.8, whereas for rsCD1b it was sulfatide@ PtdSer > PtdIns @ BMP + PtdSer at
pH 7, and sulfatide > PtdSer@ PtdIns @ BMP + Ptd-Ser at pH 4.8 The behavior of rsCD1e2+ in the presence of liposomes was qualitatively comparable to that of rsCD1e2), although the interaction appeared
to be stronger, with a two-fold higher resonance signal (Fig 7, right panel), demonstrating that the ability of CD1e to interact with membranes at neutral
or acidic pH does not rely on cleavage of the propep-tide
Table 1 TMPs of guanidium chloride-induced denaturation transitions in the secondary and tertiary structures of rsCD1e proteins Equations fitting the data in Fig 4 or in Fig 5A,B were used to deduce the guanidium chloride-induced denaturation transitions in the secondary and tertiary structures of rsCD1e proteins, respectively The form of the equations was Y = 1 ⁄ (1 + 10^((logEC50 – [guanidium chloride])*HillS-lope)), proposed by GRAPHPAD PRISM , where Y is the normalized fluorescence intensity at 340 nm or normalized ellipticity at 219 nm, logEC50
is the guanidium chloride concentration at which the ellipticity is half its initial value in the absence of guanidium chloride (TMP), [guanidium chloride] is the molar concentration of guanidium chloride, and HillSlope is the Hill constant or slope factor defining the steepness of the curve ss, secondary structure; ts, tertiary structure; ANS, [guanidium chloride] inducing the maximal ANS fluorescence.
CD1e2 + , pH 7 CD1e2), pH 7 CD1e4, pH 7 CD1e2 + , pH 4.8 CD1e2), pH 4.8 CD1e4), pH 4.8
Fig 5 Guanidium chloride-induced changes in the fluorescence of rsCD1e proteins rsCD1e2 +
or rsCD1e2)and rsCD1e4)were diluted to
2 l M in 10 m M monosodium ⁄ disodium phosphate buffer containing 150 m M sodium sulfate at pH 7 (open symbols) or pH 4.8 (closed sym-bols) Symbols are the same as in Fig 4 (A, B) Normalized tryptophan fluorescence at 340 nm (C, D) Normalized ANS fluorescence at
490 nm (molar ANS ⁄ protein ratio is 40 : 1) The fluorescence intensities were normalized by dividing by the maximal intensity of the spec-trum of the native protein For each condition, the curves represent the means of three independent experiments Because of their values, the bars corresponding to the standard deviations are smaller than the symbols, and are therefore not represented (A, C) Comparison of rsCD1e2)and rsCD1e2+ (B, D) Comparison of rsCD1e2)and rsCD1e4) Normalized I, normalized intensity.
Trang 7Compromised stability of the CD1e4 variant
We previously reported that the CD1e4 allelic variant
does not facilitate the presentation of PtdInsM6
anti-gens to CD1b-restricted T cells, and that this was
attributable to inefficient transport of this variant to
CD1b+compartments and the absence of a detectable
soluble lysosomal form Here, we compared the
con-formational properties of rsCD1e4) with those of
rsCD1e2), from which it differs by the replacement of
Leu194 with a potentially helix-destructuring proline
At pH 7, the far-UV (data not shown) and near-UV
(Fig 2C) circular dichroism spectra showed similar
secondary and tertiary structures of rsCD1e2) and
rsCD1e4) Shifting the pH to 4.8 resulted in only
minor changes in the tertiary structure of rsCD1e4)as
compared with rsCD1e2) (data not shown) At pH 7
and guanidium chloride concentrations below 3 m, the
normalized ellipticity of rsCD1e4)was nevertheless
sig-nificantly weaker than that of rsCD1e2) In addition,
at least 1 m lower concentrations of guanidium
chlo-ride were required for rsCD1e4) than for rsCD1e2)
to induce a similar decrease in ellipticity, at acidic or
neutral pH (Fig 4), to reduce the intrinsic protein flu-orescence (Fig 5B), or to reach maximal ANS binding (Fig 5D) Thus, although the L194P substitution seems to only weakly affect the secondary structure of CD1e, it appears to have a profound effect on the sta-bility of its tertiary structure
Under nondenaturing conditions, rsCD1e4) and rsCD1e2)bound comparable amounts of ANS at neu-tral pH Conversely, incubation at pH 4.5 resulted in a dramatic increase in the binding of ANS to rsCD1e4), but not to rsCD1e2) (Fig 3A) In experiments with bis-ANS, rsCD1e4) and rsCD1e2) behaved compara-bly at pH 4.5 when [bis-ANS]⁄ [rsCD1e] ratios were below 20 At higher stoichiometries, bis-ANS binding
to rsCD1e4)appeared to be unsaturable (Fig 3D) At acidic pH, with rsCD1e4+ the fluorescence of bound bis-ANS reached a plateau, although at a higher value
of fluorescence intensity than for CD1e2+ At low [bis-ANS]⁄ [rsCD1e4+ or rsCD1e4)] ratios (i.e < 1), the fluorescence was proportional to the concentration
of bis-ANS Calculations indicated that the apparent number of bis-ANS-binding sites on CD1e4+ was 12.4 ± 0.3, i.e half that on CD1e2+
Fig 6 Interaction of rsCD1e molecules with PtdSer–NBD The fluorescence of NBD in reaction mixtures containing rsCD1e2 + , rsCD1e2), rsCD1e4+or rsCD1e4)and PtdSer–NBD (both 2 l M ) at pH 7 (left) or pH 4.5 (right) was recorded for 6000 s Curve C represents the fluores-cence of PtdSer–NBD alone in the buffer.
Fig 7 Interaction of rsCD1e2 and rsCD1b molecules with liposomes Liposomes containing 70 molÆ% PtdCho and 30 molÆ% PtdSer, PtdIns
or sulfatide, or 70 molÆ% PtdCho, 15 molÆ% PtsSer and 15 molÆ% BMP, were adsorbed onto L1 chips in a Biacore 3000 system RsCD1e2) and rsCD1b proteins (0.2 l M ) in 10 m M monosodium ⁄ disodium phosphate buffer containing 150 m M NaCl at pH 7 (left) or pH 4.8 (middle) were injected over the chips for 300 s, after which the surfaces were rinsed with the same buffer for the same period of time The interac-tions of rsCD1e2 + with liposomes at pH 7 and pH 4.8 were compared (right) in a similar manner.
Trang 8At neutral pH, the fluorescence of PtdSer–NBD
bound to CD1e4+ reached a plateau with a higher
value than for CD1e2 molecules at acidic pH (Fig 6)
These observations suggest that, at neutral pH,
CD1e4+ molecules are more receptive to lipids, or
form more stable complexes with PtdSer–NBD
CD1e4)behavior appears to be the opposite of that of
CD1e4+ Indeed, at acidic pH, CD1e4+and CD1e2+
displayed similar PtdSer–NBD binding curves In
con-trast, rsCD1e4) bound almost two times less PtdSer–
NBD than rsCD1e2)
Discussion
CD1 molecules and human leukocyte antigen class I
and class II molecules are structurally related proteins
In particular, their antigen-binding pockets are formed
of a-helices lying on b-sheets, and include hydrophobic
residues pointing to the groove, in an optimal
architec-ture for lipid binding [18] The 3D strucarchitec-ture of CD1e
is not known, and no rigorous structural study of this
protein has been reported to date With the intention
of filling this gap and gaining a more precise picture of
the role played by the CD1e propeptide sequence,
we performed structural investigations on different
forms of CD1e, using complementary biophysical
approaches
Prediction of the secondary structure on the basis of
far-UV circular dichroism spectra indicated that the
a-helix and b-sheet contents of rsCD1e2+or rsCD1e2)
and rsCD1e4)were almost identical to those of CD1b
and major histocompatibility complex class I and
clas-s II moleculeclas-s [11,19] We alclas-so analyzed rclas-sCD1b
pro-duced in S2 cells by circular dichroism spectroscopy,
and the results obtained were in excellent agreement
with the structural contents deduced from
crystallo-graphic CD1b structures (data not shown), which
vali-dates our experimental approach In subsequent
studies, the secondary structure of rsCD1e was found
to remain unaltered when the pH was shifted from
neutral to acidic (Fig 1A), and structural changes only
became evident under nonphysiological conditions
(pH < 3.5, data not shown) A comparison with
liter-ature data suggested that the a1-helix and a2-helix of
CD1e are less sensitive to pH changes than those of
CD1b or CD1d [11,12]
In contrast, several lines of evidence presented in
this work indicate that CD1e could gain flexibility in
its tertiary structure while transiting from the ER to
acidic CD1b+ late endosomes⁄ lysosomes First, the
exposure of hydrophobic surfaces to ANS increased
considerably at acidic pH in both rsCD1e2) and
rsCD1e2+ (Fig 3A) On the other hand, circular
dichroism studies (Fig 4) and analyses of intrinsic flu-orescence (Fig 5A) and ANS binding (Fig 5C) in the presence of various concentrations of guanidium chlo-ride revealed increased structural instability at acidic
pH These data suggest that acidification could gener-ate stable conformational intermedigener-ates with loose ter-tiary contacts and improved access to the lipid-binding groove An enhancement of ANS binding at lysosomal
pH has been reported for CD1b and CD1d molecules, and found to correlate with an accompanying increased capacity to bind lipids [11,12]
A second important point addressed in this study was whether the CD1e propeptide, which is present on the membrane-associated but not on the soluble lyso-somal form of the molecule, influences protein struc-ture and stability Circular dichroism data (Figs 1 and 2), ANS binding experiments (Fig 3A) and unfolding experiments with guanidium chloride (Figs 4 and 5) failed to reveal any significant differences between rsCD1e2+and rsCD1e2) In contrast, rsCD1e2+ was found to bind twice as many bis-ANS molecules as rsCD1e2) (Fig 3B,C) This intriguing observation is difficult to explain, as the estimated stoichiometries of interaction (24 and 12 bis-ANS molecules for one rsCD1e2+ and one rsCD1e2) molecule, respectively) are largely in excess of the number of bis-ANS mole-cules that could be expected to interact directly with the CD1e lipid-binding groove or the 12-residue pro-peptide Nonetheless, our observation is supported by the fact that the propeptide also triggered an almost two-fold higher response in surface plasmon resonance experiments on the interaction of CD1e with surfaces containing anionic lipids (Fig 7C) These interactions may be partly driven by electrostatic forces Indeed, theoretical estimations with a 3D homology model of CD1e and the propka algorithm [20] indicate that the overall charge of CD1e could shift from ) 3 to + 18
as the pH drops from 7 to 4.5, whereas these parame-ters are) 7 and + 2 for CD1b Such a strong positive charge on CD1e proteins could partially explain why saturation was only attained at acidic pH with high bis-ANS binding stoichiometries Similarly, this prop-erty would explain why CD1e interacted better with surfaces enriched in anionic lipids when the pH was acidic (Fig 7), and would strongly suggest that this effect could control its interaction with lysosomal sur-faces rich in negatively charged lipids such as BMP [10] It nevertheless remains challenging to elucidate why the presence of the propeptide leads to an almost two-fold increase in bis-ANS binding stoichiometry and in surface plasmon resonance signals
This study was also intended to shed light on the structural behavior of the natural variant CD1e4,
Trang 9which is inefficiently transported from the ER to
lyso-somes, with the result that a soluble lysosomal form
cannot be detected [6] Although rsCD1e4 molecules
can be efficiently expressed in S2 cells, and the circular
dichroism data presented here demonstrate that the
protein is folded adequately, there is cumulative
evi-dence for lower intrinsic stability and higher structural
sensitivity in acidic environments than for rsCD1e2
Thus, the ANS-binding data revealed a remarkable
exposure of hydrophobic patches at pH 4.5 (Fig 3A)
Moreover, significantly lower guanidium chloride
con-centrations were required to cause a similar
destabili-zation or exposure to ANS as in rsCD1e2 molecules
(Figs 4 and 5) The instability of rsCD1e4) at acidic
pH might explain why CD1e4 could be
immunolocal-ized in CD63+ compartments, whereas no soluble
form could be detected Our data strongly suggest that,
once they are in acidic late endosomal⁄ lysosomal
com-partments, CD1e4 molecules probably adopt a more
water-exposed conformation, which would render the
protein susceptible to proteolysis, thus preventing its
accumulation
Overall, the results presented in this study suggest
that the conformational behavior of CD1e is optimized
to facilitate its interaction with and binding of lipids in
a pH-regulated manner Thus, the arrival of CD1e in
acidic late endosomal and lysosomal compartments
would be synchronized with proteolytic events
permit-ting release of the soluble domain from the
mem-branes, cleavage of the propeptide, and interaction
with anionic lipids, possibly the anionic lipid domains
present in CD1b+ compartments The propeptide
would have no influence on the conformational
struc-ture of CD1e or the properties of the lipid-binding
groove This short N-terminal oligopeptide might, on
the contrary, interact with key residues to prevent the
occurrence of intermolecular or intramolecular
con-tacts How these properties combine to optimize the
repertoire of CD1e ligands and their subsequent
pre-sentation by CD1b molecules in dendritic cells remains
to be investigated
Experimental procedures
Reagents and recombinant proteins
Recombinant soluble sCD1 (rsCD1) molecules were
expressed in transfected Drosophila melanogaster S2 cells
and purified as previously described [5,6] Briefly, rsCD1e
molecules with (CD1e2+ and CD1e4+) or without
(CD1e2) and CD1e4)) the propeptide, i.e amino acids 20–
305 or 32–305 of the pre-a-chain of the corresponding
vari-ant, were expressed fused to the signal peptide of
heat-shock 70-KD protein 5 (HSPA5) and a C-terminal tag including a tandem WSHPQFEK(streptag II)-His8 peptide tag In the case of CD1b, amino acids 17–300 of the pre-a-chain were expressed by use of the same vector The a-chains were coexpressed in S2 cells with human b2m Recombinant proteins were purified by metal chelate chro-matography followed by affinity purification on immobi-lized Strep-Tactin (Qiagen, Courtaboeuf, France) Eluted proteins were concentrated to 15 mgÆmL)1 (0.32 mm) in
10 mm sodium phosphate buffer containing 150 mm NaCl (NaCl⁄ Pi) Protein purity was checked with the Experion system (BioRad, Marne la Coquette, France), and found to
be 95–99%, depending on the CD1 preparation Purified rsCD1e preparations were biologically active in vitro in PtdInsM6digestion assays [5] All CD1e variants and CD1b were able to bind lipids in vitro [15] (this article and data not shown) The concentrations of purified rsCD1 molecules were determined by measurement of the absorbance at
280 nm, using extinction coefficients of 1.8 and 1.7 mgÆ
mL)1Æcm)1 for rsCD1e and rsCD1b, respectively (protpa-ram; http://www.expasy.ch/tools/protparam.html) Lipids were purchased from Sigma (Saint Quentin Fallavier, France) and Avanti Polar Lipids (Alabaster, AL, USA) When necessary, stock protein solutions were diluted directly in buffer at the desired pH, which was checked with a microelectrode (Inlab 423; Mettler-Toledo GmbH, Giessen, Germany) For reversibility measurements, small quantities (1–10% of the volume) of disodium phosphate (pH 9.9) or 100 mm NaOH were added to the solution to obtain a pH of 7 or 4.8, and the results were corrected for protein dilution
Liposome preparation Lipids solubilized in chloroform or chloroform⁄ methanol (2 : 1) were dried under a gentle stream of nitrogen, and then placed under vacuum for at least 2 h to remove any traces of the organic solvents The thin lipid film was resus-pended in 10 mm NaCl⁄ Pi (pH 7) by vortex mixing, and hydrated for 1 h at room temperature (10 mgÆmL)1) Large unilamellar vesicles were prepared with a Mini-extruder (Avanti Polar Lipids), according to the manufacturer’s instructions Briefly, the lipid suspension was subjected to seven freeze–thaw–vortex cycles, consisting of freezing on dry ice for 10 min, immersion in a water bath at 37C for
10 min, and thorough vortexing of the sample The suspen-sion was then passed through a 100-nm-pore membrane at room temperature, stored at 4C, and used within 1 week Liposomes of different composition were prepared: neu-tral liposomes containing 100% PtdCho; negatively charged liposomes containing 70% PtdCho and 30% Ptd-Ser, PtdIns or sulfatide (w⁄ w); and BMP-enriched lipo-somes containing 70% PtdCho, 15% PtdSer, and 15% BMP The vesicle size was checked by electron microscopy
Trang 10(negative staining) and dynamic light scattering, and found
to be 100 ± 10 nm
Fluorescence measurements (ANS and bis-ANS)
For measurement of the fluorescence of ANS or bis-ANS
(Invitrogen, Cergy-Pontoise, France) in the presence or
absence of protein, the concentration of ANS or bis-ANS
was determined from the absorbance at 350 or 390 nm,
with an extinction coefficient of 5000 or 16 790 m)1Æcm)1,
respectively Measurements were performed with 200 lL of
reagents in 96-well polystyrene plates (Becton-Dickinson,
Meylan, France), using a FlexStation3 (Molecular Devices,
Saint Gre´goire, France) As this robot provides relative
data that cannot be directly compared from one day to
another, each comparison of conditions corresponds to a
set of experiments performed in triplicate on a same day
The excitation and emission wavelengths were kex=
370 nm and kem= 480 nm for ANS, and kex= 390 nm
and kem= 490 nm for ANS In experiments with
bis-ANS, the fluorescence was corrected for the inner filter
effect, with the equation Fcorr= Fobsantilog[(Aex=
Aem)⁄ 2], where Fcorr and Fobs are the corrected and
observed fluorescence intensities at the emission and
excita-tion wavelengths, respectively [21] Data were analyzed as
described previously [22] Briefly, the measured fluorescence
of bis-ANS (F) was confirmed to be proportional to
[bis-ANS], when [bis-ANS]⁄ [CD1e] < 1 (F = B · [bis-ANS])
At higher [bis-ANS]⁄ [CD1e] ratios, the number of bis-ANS
molecules bound to rsCD1e molecules was calculated to be
F⁄ B This allowed calculatation of r= [bis-ANS
bound]⁄ [rsCD1e] The Scatchard representation of r versus
r⁄ [bis-ANS] = n ⁄ Kd) r ⁄ Kd, where n is the number of
binding sites and Kd the apparent dissociation constant,
then enables determination of n and Kd
Fluorescence measurements (PtdSer–NBD)
To quantify the binding of PtdSer–NBD to CD1e molecules,
100 lg of acyl-labeled PtdSer–NBD
(1-oleoyl-2-{12-[(7-
nitro-2-1,3-benzoxadiazole-4-yl)amino]lauroyl}-sn-glycero-3-phosphoserine) (Avanti Polar Lipids) was dissolved in
100 lL of ethanol⁄ binding buffer (150 mm NaCl, 10 mm
monosodium⁄ disodium phosphate, pH 7 or 4.5) (50 : 50,
v⁄ v), diluted in 1 mL of buffer, and sonicated for 10 min
(final PtdSer–NBD concentration of 200 lm) A 2-lL
ali-quot of diluted PtdSer–NBD was then added to 200 lL of
2 lm CD1e in binding buffer To standardize the beginning
of the experiments, all measurements were started 12 s
after the addition of PtdSer–NBD to the protein solution
The fluorescence was recorded in a PTI spectrofluorimeter
(PTI, Birmingham, NJ, USA), using a 1-cm path length
and 100-lL minimal volume quartz cuvettes, with slit
widths of 3 nm for excitation (kex= 460 nm) and 4 nm for
emission (kem= 525 nm) The kinetic curves were fitted by
use of a two-phase association fitting curve (graphpad prism) The fluorescence at 6000 s and the time required to reach half this fluorescence intensity were deduced from these curves
Circular dichroism spectroscopy Protein samples were diluted to 2 or 21 lm in the appropri-ate buffer containing 0.15 m sodium sulfappropri-ate and the indi-cated concentration of guanidium chloride Sodium phosphate⁄ citrate buffers (5 mm) and monosodium ⁄
disodi-um phosphate buffers (5 mm) were used for the pH ranges 2.5–4.8 and 5–9, respectively Samples were incubated over-night at room temperature to permit the denaturation reac-tion to reach equilibrium before measurement of the circular dichroism In the case of kinetic and reversibility experiments, the incubation time is indicated Spectra were acquired with a Jasco J-810 spectropolarimeter (Jasco, Tokyo, Japan) with a PTC-423S temperature controller and
a Peltier cell holder The far-UV spectra were recorded with
a 0.02-cm path length rectangular ‘sandwich’ cuvette (closed) or a 0.1-cm cuvette The spectra were acquired with
a scan rate of 50–100 nmÆmin)1, a response time of 2–4 s, and a step and band width of 1 nm, and were averaged over three to five acquisitions The near-UV spectra were recorded at 20C with a 1-cm cuvette and a protein con-centration of 21 lm The spectra were acquired with a scan rate of 50 nmÆmin)1, a response time of 1 s, a step of 0.2 nm, and a band width of 1 nm, and were averaged over
10 acquisitions Buffer spectra were subtracted from the sample spectra The ellipticity was converted to the mean residue weight ellipticity, using the path length of the cuv-ette, the protein concentration, and mean residue weights
of 113 for rsCD1e and 112.6 for rsCD1b The secondary structure content was calculated with the program cdsstr
of the DichroWeb Site (http://dichroweb.cryst.bbk.ac.uk/ html/home.shtml) [23], reference set SP175 [24]
Surface plasmon resonance Surface plasmon resonance experiments were performed in
a Biacore 3000 system (GE Healthcare Biacore AB, Uppsala, Sweden) An L1 chip was used for liposome immobilization, and the running buffer was 25 mm mono-sodium⁄ disodium phosphate at pH 7 or 4.8, containing
150 mm NaCl The chip was first washed with three injections of isopropanol⁄ NaOH solution (2 : 3, v ⁄ v) at
30 lLÆmin)1, and then rinsed thoroughly with buffer Dif-ferent liposome mixtures (2 lm) were injected separately at
1 lLÆmin)1 through each of the four L1 chip flow cells, until the surfaces were saturated (generally 20–30 min) A resonance level of 7000–10 000 RU was usually achieved The chips covered with liposomes were then washed with three 1-min pulses of 100 mm NaOH at 30 lLÆmin)1 to remove loosely bound lipids and stabilize the surfaces The