Cell, Vol.
68, 465-477, February 7, 1992, Copyright 0 1992 by Cell Press
The Human Class II MHC Protein HLA-DRI
Assembles as Empty ap Heterodimers
in the Absence of Antigenic Peptide
Lawrence J. Stern’ and Don C. Wiley”t
*Department of Biochemistry and Molecular Biology
rHoward Hughes Medical Institute
Harvard University
Cambridge, Massachusetts 02138
Summary
We have produced the human class II histocompatibil-
ity protein, HLA-DRl, as a soluble, secreted glycopro-
tein in insect cells infected with baculovirusescarrying
truncated a and p subunit genes. The peptide-binding
site is empty, and the empty molecules are fully com-
petent to bind antigenic peptide. We used the empty
molecules to measure an intrinsic rate for peptide as-
sociation, and to investigate the role of peptide in sta-
bilizing the class II structure. Peptide binding kinetics
for the empty molecule are only lo-fold faster than for
peptide exchange into an occupied site, suggesting
that a conformational change may accompany peptide
binding. The native up heterodimer assembles in the
absence of antigenic peptide, but peptide binding sta-
bilizes the empty heterodimer against aggregation and
against SDS-induced denaturation.
Introduction
Major histocompatibility complex (MHC) proteins are
highly polymorphic cell surface glycoproteins that bind an-
tigenic peptides and display them at the cell surface (re-
viewed in Rothbard and Gefter, 1991). T lymphocytes initi-
ate immune responses by recognizing a specific peptide
bound to an MHC protein. Class I MHC proteins are be-
lieved to bind endogenous peptides in the endoplasmic
reticulum (Nuchtern et al., 1989; Yewdell and Bennink,
1990), while class II MHC proteins generally bind exoge-
nously derived peptides in a specialized post-Golgi com-
partment (Guagliardi et al., 1990; Neefjes et al., 1990;
Harding et al., 1990, Davidson et al., 1991; Germain and
Hendrix, 1991). Both class I and class II MHC proteins
must bind peptides tightly to prevent peptide exchange at
the cell surface and an inappropriate immune response.
The peptide-binding sites of class I molecules are usu-
ally occupied with a mixture of peptides (Bjorkman et al.,
1987; Jardetzky et al., 1991; Falk et al., 1991), and class
I molecules as isolated do not easily exchange or bind
peptides in vitro (Chen and Parham, 1989). Studies using
mutant cell lines that do not load peptides onto class I
molecules havesuggested that peptide binding is required
for assembly of the class I heterodimer, and for stable cell
surface expression (Townsend et al., 1989, 1990; Ljung-
gren et al., 1990; Ortiz-Navarrete and Hlmmerling, 1991).
Class II MHC proteins are also isolated from lymphoid
cells as very stable complexes with antigenic peptides
(Buus et al., 1988; Rudensky et al., 1991). Studies from a
number of laboratories have shown that less than 20% of
class II molecules will bind antigenic peptide added in vitro
(Watts and McConnell, 1988; Buus et al., 1987; Jardetzky
et al., 1990; O’Sullivan et al., 1990; Roche and Cresswell,
1990), or in vivo (Ceppellini et al., 1989; Busch and Roth-
bard, 1990). The peptide-binding sites on the remainder
of the proteins are presumably occupied with tightly bound
peptides, which interfere with the binding of exogenous
peptide and complicate the interpretation of binding stud-
ies (Tampe and McConnell, 1991).
In this report we describe the expression of a native
human class II MHC protein, HLA-DRl, in insect cellscoin-
fected with baculoviruses carrying genes for the a and 5
subunits. The antigen-binding site of soluble DRl pro-
duced in insect cells appears to be empty. The “empty”
DRl from insect cells binds significantly more antigenic
peptide than DRl isolated from human cells, with a lo-fold
faster association rate. The properties of “empty” DRl sug-
gest that a conformational change may accompany pep-
tide binding. In contrast to the heavy and light chains of
class I MHC proteins, which appear to require peptide
for stable association, the subunits of DRl assemble a8
heterodimers in the absence of antigenic peptide. Al-
though not required for heterodimer assembly, peptide
binding stabilized DRl against aggregation and against
sodium dodecyl sulfate (SDS)-induced denaturation.
Results
Recombinant Baculoviruses Direct the Synthesis
and Secretion of DRa9 Heterodimers
in Coinfected Sf9 Cells
Recombinant baculoviruses carrying full-length genes for
the a and 8 subunits of human DRl (BV-DRa and
BV-DR8), or carrying truncated genes (BV-DRasol and
BV-DRPsol), were generated by homologous recombina-
tion in the insect ovarian cell line Sf9 (fall armyworm,
Spodoptera frugiperda). The truncated genes code for pro-
teins of 192 (a) and 198 (5) residues, which terminate just
before the beginning of the predicted transmembrane
spans (Figure 1, bottom panels). Insect cells infected with
BV-DRa or with BV-DR8 expressed polypeptides of the
expected apparent molecular weight, which reacted with
antisera specific for the appropriate subunit of DRl (Figure
1, lanes 5). No reactivity was observed in the extracellular
medium (Figure 1, lanes 8), nor in insect cells infected with
a control baculovirus, BV$gal (Figure 1, lanes 3 and 4).
Insect cells infected with BV-DRasol or BV-DRPsol,
which carry the truncated genes, expressed polypeptides
that exhibited somewhat faster mobility on SDS-PAGE
(Figure 1, lanes 7) than the full-length forms, as expected
for the removal of the transmembrane and cytoplasmic
domains. The truncated constructs were expressed at a
significantly greater level than the full-length proteins. A
fraction of the protein produced in the singly-infected cells
was secreted into the extracellular medium (Figure 1,
lanes 8). The protein retained within the cells exhibited
multiple bands per subunit by SDS-PAGE, probably due
Cell
466

A B
anti-DRa anti-DRf3
I I
LG2
“GIIIBsd DRa DRa.SOL DRa3OL
+DRaSOL LG2 “& INal DR6 DR6.SOL
III
DRaSOL
+DRPSOL
c cmcmcmcm Mr c c m cmcmcm
-66-

-43-

- 29 -

-lS-

1 2345676910 12 3 4 5 67 8 910
1 765 1 601
DW 4 ss 1 61 Iv CP TM CYT
I
I
167 ...PmGluThr~ffiluAsnV~V~C~laLeu... tw 193 ...GluSerAlaGlnSe~~lysMetleuSelGlyVal... ax
1 654 1 664
DRasol ss ( al 1 a2 1 CP

187 ...PmGluThrThrGluAsn-COOH WE? 193 ...GluSerAlaGlnSerlys-COOH 1%

Figure 1. Expression of Full-Length and Truncated DRa and DRb Polypeptides in Baculovirus-Infected Sf9 Insect Cells
Upper panels: Western blots for DRa and DRf3. Sf9 cells were harvested 72 hr postinfection, and aliquots of cell lysate (c) and extracellular medium
(m) were analyzed by 12.5% acrylamide SDS-PAGE and Western blotting with specific antisera directed against DRa (A) or DR6 (B). Lane 1, human
LG2 cell lysate; lane 2, affinity-purified, papain-solubilized DRI from LG2 cells (DRlpap); lanes 3 and 4, cell lysate and extracellular medium from
Sf9 insect cells infected with control baculovirus BV-6gal; lanes 5 and 6, cell lysate and extracellular medium from insect cells infected with either
BV-DRa or BV-DR6; lanes 7 and 6, cell lysate and extracellular medium from insect cells infected with either BV-DRasol or BV-DR6sol; lanes 9
and 10, cell lysate and extracellular medium from insect cells coinfected with both BV-DRasol and BVDRf3sol. Samples of lysates and extracellular
medium represented 1 x IO5 cells in (A) and 2.5 x lo4 cells in (S). Papain-solubilized DRl from human cells (DRlpap) was used at 100 ng (A)
and 25 ng (B) per lane.
Lower panels: HtA-DRI genes used to construct recombinant baculoviruses. Nucleotide numbering, beginning at the initiation codon, is indicated
above the boxes. Amino acid numbering begins after the signal sequence at the N-terminus of the mature polypeptide. Portions of the amino acid
sequence near the C-terminal end of the extracellular domain, along with amino acid residue numbers, are indicated below the boxes. DRa and
DR6 contain the entire coding sequences of the parent cDNAs. DRasol and DR6sol have been truncated just before the transmembrane domain
as indicated. Open boxes indicate coding regions; SS, signal sequence; al, ~2, 61, 62, HLA extracellular domains; CP, connecting peptide; TM,
transmembrane domain; CMO. cytoplasmic domain.

to incomplete signal sequence cleavage and partial glyco-
sylation, but the secreted protein exhibited predominantly
one band per subunit. In cells coinfected with both BV-
DRasol and BV-DRBsol, secretion into the extracellular
medium was much more efficient (Figure 1, lanes 9 and 10).
The differences in mobility on SDS-PAGE between the
subunits of DRl expressed in insect cells and those of
full-length DRl (Figure 1, lanes 1) or papain-solubilized
DRI (Figure 1, lanes 2) produced by human cells are due
to differential glycosylation in the insect cell and human

Table 1. Glycosylation of DRI Produced in Human and Insect Cells

DRI Reactivity

Insect Human

Assay Specificity a P a P

Glycosidase sensitivity
Endoglycosidase H High mannose or hybrid + + +I-
Glycopeptidase F Most N-linked + + + +
Lectin reactivity
GNA (Galanthus nivalis agglutinin) ManaMan + + -
SNA (Sambucus nigra agglutinin) SAa(P6)Gal - + +
MAA (Maackia amurensis agglutinin) SAa(2-3)Gal
DSA (Datura straminium agglutinin) Galfi(l4)GlcNAc +
PNA (peanut agglutinin) Galp(l-3)GalNAc

For glycosidase analysis, purified DRl samples were denatured, digested with the appropriate glycosidase, and analyzed by SDS-PAGE. A
difference in mobility in the glycosidase-treated samples relative to mock-digested samples was scored as positive. For lectin analysis, purified DRI
samples were analyzed by SDS-PAGE and Western blotting using labeled lectins. In both assays, full-length and soluble DRl behaved identically.
The expected oligosaccharide specificity is shown beside the name of each glycosidase or lectin (Man, mannose; SA, sialic acid; Gal, galactose;
GlcNAc, N-acetylgluoosamine; GalNAc, N-acetylgalactosamine).
Empty HLA-DRl from Insect Cells
467

1.5
:A DRa

upSol medium

apsol cell

a6 medium
0.0 singly infected-
0 20 40 60 80 100
Time after infection (hours)
Red Fluorescence

Figure 2. Cell Surface Expression of DRl in Infected Sf9 Cells Figure 3. Time Course of Expression of Soluble and Membrane-
Bound HLA-DRl in Insect Cells
Baculovirus-infected Sf9 insect cells along with LG2 human lympho-
blastoid cells were analyzed by flow cytometry at 46 hr postinfection. Sf9 cells (IO6 cells per ml) were coinfected with BV-DRa+BV-DRf3
Surface expression of DRI was detected using phycoerythrin- (squares) or with BV-DRasol+BV-DRpsol (circles), or were singly in-
conjugated anti-DRl monoclonal L243 (shaded). Background fluores- fected with either BV-DRa or BV-DRf3 alone (triangles). DRI concentra-
cence was estimated with nonspecific phycoerythrin-conjugated tion in the extracellular medium (open symbols) or in cell lysates
mouse antibody (open). (A) Sf9 cells infected with BV-DRa alone. (B) (closed symbols) was determined by ELISA, using the conformation-
Sf9 cells infected with BV-DRj3 alone. (C) Sf9 cells coinfected with ally sensitive monoclonal antibody L243 as the capture antibody. De-
BV-DRa+BV-DR6. (D) LG2 cells. terminationswith monoclonal antibody LB3.1 produced similar results.
The dashed line indicates cell viability estimated by trypan blue
exclusion.

cell lines. Both the a and 8 chains of the full-length and

truncated forms of DRl from insect cells were sensitive to
endoglycosidase-H and glycopeptidase-F (Table 1). Both
a and 8 chains bound GNA lectin but not SNA, MAA, DSA,
or PNA lectins, indicating that both chains contain high-
mannose, N-linked polysaccharides. In contrast, DRl
isolated from human cells carries a complex, sialated poly-
saccharide on each chain, along with a second, nonsia-
lated polysaccharide on the a chain (Table 1 and Shackel-
ford and Strominger, 1983). After deglycosylation, the
subunits of full-length and truncated DRl from insect cells
exhibited the same mobility as the corresponding deglyco-
sylated subunits of intact or papain-solubilized DRl from
human cells (data not shown).
Sf9 insect cells coinfected with the full-length con-
structs, BV-DRa+BV-DR8, express DRl on the cell sur-
face as detected by flow cytometry using monoclonal anti-
body L243 (Figure 2C). This antibody recognizes
conformational determinant on the correctly folded DRl
heterodimer (Lampson and Levy, 1980; Gorga et al.,
1987). No reactivity was observed with Sf9 cells singly
infected with BV-DRa alone (Figure 2A) or with BV-DRf3
alone (Figure 28). The surface expression of DRl on the
coinfected SF9 insect cell surface was weaker and more
heterogeneous than that exhibited by LG2, a human
lymphoblastoid cell line (Figure 2D).
The time course of DRl expression in insect cells was
monitored by enzyme-linked immunosorbent assay (ELISA)
using the anti-native DRl monoclonal antibody L243 (Fig-
ure 3). DRl expression in BV-DRa+BV-DRP-coinfected
cells increased from 24 to 48 hr postinfection, then re-
mained relatively constant (closed squares). DRl could
not be detected in the extracellular medium (open
squares). No L243 reactivity was observed in lysates of
singly-infected cells (shaded triangles), indicating that this
antibody does not recognize DRa or DR8 monomers, or
any a2 or p2 homodimers that may be produced by the
singly infected cells. Similar results were obtained with
LB3.1 (Gorgaet al., 1988), another conformation-sensitive
monoclonal antibody that recognizes the DRaf3 hetero-
dimer. Insect cells coinfected with the truncated con-
structs, BV-DRasol and BV-DRPsol, produced heterodi-
merit DRa8 complex that was detected in cell lysates
(closed circles) and also in the extracellular medium (open
circles). Secretion of DRl to the extracellular medium sig-
nificantly lagged behind expression within the cell and con-
tinued to increase very late in infection. The overall expres-
sion level of soluble DRl (cells plus medium) remained
fairly constant after 48 hr postinfection, at approximately
a 2 mg per liter of culture medium, more than six times the
expression level of the full-length, membrane-bound form.
Purification of DRl from Insect Cells
Insect cell cultures were harvested for protein purification
at 72 hr postinfection. Soluble DRl (l-2 mg per ml of
culture) was isolated from the extracellular medium of
coinfected cells in 800/o-90% yield by immunoaffinity chro-
matography using monoclonal antibodies that recognize
the native DRaf3 heterodimer (LB3.1 or L243). No DRa or
DR8 subunits could be detected on Western blots of the
material that did not bind to the affinity column, indicating
that all of the secreted DRa and DRf3 was present as a8
heterodimer. The immunoaffinity-purified soluble DRI ex-
hibited predominantly two bands (DRa and DR8) by SDS-
PAGE, along with a significant but variable amount of a
second DRa band (Figure 4, lane 1). The three bands were
Cell
466

DRl : Insect Human and the a and 8 chains migrate as a heterodimer on SDS-
/
Peptide:
--II
-I - + - Mr PAGE if the samples are not boiled prior to loading (Figure
Boil: +-+-+-+- 4, lane 8; Gorga et al., 1987). After boiling in SDS, the a
- 68 and 8 subunits disassociate (Figure 4, lane 7). In contrast,
the soluble DRl secreted from coinfected insect cells was
- 43 sensitive to dissociation by SDS at room temperature, and
migrated mostly as monomeric a and 8 chains, along with
several faint bands near the position expected for the het-
erodimer (Figure 4, lane 4). After boiling, soluble DRl from
- 29 insect cells migrated as the expected a and 8 monomers
(Figure 4, lane 3). Preincubation with an antigenic peptide
from influenza hemagglutinin, HA(306-318), caused the
soluble DRl from insect cells to become resistant to SDS-
induced dissociation. Soluble DRl from insect cells, incu-
1 2345678 bated with HA(306-318) peptide and subsequently treated
Figure 4. Peptide Binding Increases the Stability of Soluble HLA-DRl
with SDS at room temperature (Figure 4, lane 2) migrates
from Insect Cells against Denaturation by SDS as a strong band that corresponds to the a8 heterodimer
SDS-PAGE analysis of soluble DRI from insect and human cells. seen with DRl from human cells After boiling, the sub-
Soluble DRI (60 uM) from insect cells (lanes l-4) or papain-solubilized units dissociate (Figure 4, lane 1). Incubation of DRl from
DRl from human cells (lanes 5-6) was incubated in the presence human lymphoblastoid cells with peptide had no effect on
(lanes 1, 2, 5, and 6) or absence (3, 4, 7, and 8) of 360 uM HA(306-
the stability to SDS-induced dissociation (Figure 4, lanes
318) peptide, for 100 hr at 37OC. After incubation, samples were mixed
with SDS-PAGE loading buffer (final [SDS] = 1%). One half of each 5 and 6).
sample was boiled for 3 min before loading (odd lanes); the other half While the DRl isolated by immunoaffinity purification
was loaded without boiling (even lanes). Samples were analyzed by appeared to be substantially free of contaminating pro-
SDS-PAGE on 12.5% polyacrylamide with Coomassie brilliant blue teins by SDS-PAGE, it eluted from a gel permeation col-
R250 detection. Positions of molecular weight markers BSA (SSOOO),
ovalbumin (43000) carbonic anhydrase (29000) and 6-lactoglobulin
umn in anumberof peakswith apparent molecularweights
(18400) are indicated at right. of 50,000 and greater (Figure 5A). Each of these peaks
contained material that reacted with anti-DRaf3 antibodies.
After incubation for 72 hr at 37% with antigenic peptide
HA(306-318), most of the protein eluted in a single peak
subjected to N-terminal sequencing. Both DRa band8 had corresponding to 50,000 daltons (Figure 5B), as expected
the sequence NH*-IKEEH..., and the DR8 band had the for the DRaf3 heterodimer and as seen with DRl isolated
sequence NH2-GDTRP... These are the N-termini ex- from human lymphoblastoid cells (Figure 5C). Incubation
pected for the mature subunits, indicating that the native of DRl from insect cells without the addition of peptide
DRl signal sequences were efficiently removed by the had no effect on its aggregation behavior. The aggregation
insect cell. The purified DRl was tested against 13 mono- was not a result of the isolation procedure, as whole condi-
clonal antibodies that recognize native DRl from human tioned medium also exhibited multiple, DRl-containing
cells. Each of the antibodies tested, DA2(81specific), peaks (data not shown). [‘*51]HA(306-318) peptide in-
DA6.147(a), DA6.231(81), IVA12(81), L227(81), SG171(81), cluded with soluble DRl from insect cells in the incubation
TAL8.1(81), TAL14.1(81), Tii36(82), Ti.i39(81), and TU43 mixture comigrated with the strong DRl a8 peak (Figure
(up), as well as the antibodies used for affinity purification 5D, open bars). Radiolabeled peptide binding could be
L243(a) and LB3.l(a), bound to the soluble, insect-cell- competed with an excessof unlabeled peptide (solid bars).
derived DRl. The effect of added peptide in converting the heteroge-
Approximately half of the total soluble DRl produced neous DRl isolated from insect cells (Figure 5A) to a
by BV-DRa+BV-DR8-coinfected cells was retained within mostly homogeneous species (Figure 5B) thus occurs
the cells. This material could be isolated from a lysate through peptide binding to the DRl molecule.
of coinfected cells by ion exchange and immunoaffinity Gel filtration HPLC was used to isolate the complex of
chromatographies. Soluble DRl isolated from cell lysates soluble, insect-cell-derived DRl with HA(306-318) pep-
behaved similarly to soluble DRl isolated from the extra- tide. The purified DRl-peptide complex retained binding
cellular medium. Full-length DRl (0.1 ug per ml of cul- to all of the anti-DRl monoclonal antibodies described
ture) could be isolated in detergent solution from lysates above. DRl -HA(306-318) peptide complexes were crys-
of BV-DRa+BV-DR8-coinfected cells, by including 1 O/o tallized by vapor diffusion from polyethylene glycol, under
3-[(3-cholamidopropyl)dimethylammonio]-l-propanesuIfo- conditions previously developed for crystallization of DRl
nate (CHAPS) in all solutions throughout the purification from human cells (Gorga et al., 1991). These crystals were
procedure. morphologically similar to those produced from papain-
solubilized DRl isolated from human lymphoblastoid
Purified, Soluble DRl Is Stabilized cells.
by Antlgenic Peptide The SDS-PAGE (Figure 4) and HPLC gel filtration (Fig-
DRl isolated from human lymphoblastoid cells is substan- ure 5) result8 indicate that DRl isolated from insect cells
tially resistant to dissociation by SDS at room temperature, was less stable to denaturation and aggregation than DRl
Empty HLA-DRl from Insect Cells
469

Figure 5. Peptide Binding Increases the Sta-

B Insect + bility of Soluble HLA-DRl from Insect Cells
peptide against Aggregation
HPLC gel filtration analysis of soluble DRl from
insect and human cells. Soluble DRl (80 uM)
from insect cells was incubated in the absence
(A) or the presence (6) of 500 uM HA(306-318)
peptide for 86 hr at 37% before HPLC analy-
h--Jvv, 3 sis. The elution profile of papain-solubilized
DRI from human cells (C) was unaltered by
incubation with HA(306-318) peptide (data not
shown). In a separate experiment(D), 0.3 pM
soluble DRI from insect cells was incubated
with 1 pM [1251]HA(306-318) peptide (open
bars) or with labeled peptide and a 50-fold
excess of unlabeled HA(306-318) peptide
(shaded bars), and was analyzed by gel filtra-
tion HPLC. Fractions (0.5 ml) were collected,
and the amount of radioactivity in each fraction
was determined by gamma counting. The inset
Elution Volume (ml) Elution Volume (ml) to (A) indicates the elution position of molecular
weight standards blue dextran (void volume
Vo), thyroglobulin (670.000), immunoglobulin
G (158,000) ovalbumin (44,000) myoglobin
(17.000), and vitamin 812 (1,300).

isolated from human lymphoblastoid cells. In both assays, panel, squares) and by human lymphoblastoid cells (cir-
preincubation of the insect-cell-derived DRl with antigenic cles). The initial rate of peptide binding to insect-cell-
peptide caused it to behave similarly to DRl from human derived DRl was 0.11 mol peptide per mole DRl per hour,
cells. In contrast, incubation of human-cell-derived DRl significantly faster than the 0.0093 mol peptide per mole
with peptide had no effect on its behavior in HPLC gel DRl per hour observed for human-cell-derived DRl.
filtration or SDS-PAGE, presumably because the protein These initial rates correspond to pseudo-first-order rate
as isolated is already saturated with tightly bound pep- constants of 12 M-W for DRl from insect cells and 1 .O
tides. These results prompted us to investigate the peptide M-W for DRl from human cells. The extent of radiola-
binding activity of the insect-cell-derived DRl. beled HA(306-318) peptide binding to DRl from insect
and human cells was determined from the data in Figure
Binding of Antigenic Peptide to Soluble DRl 6 (left panel). At times after 24 hr, the amount of peptide
from Insect Ceils bound to the insect-cell-produced DRl was 1 .O k 0.3 mol
The kinetics of radioiodinated HA(306-318) peptide bind- peptide per mole DRl. For human-cell-produced DRl the
ing to DRl were measured at pH 7.2 and 37X, for soluble (extrapolated) maximum extent of binding was 0.2 f 0.1
DRl produced by coinfected insect cells (Figure 6, left mol peptide per mole DRl

..-_l -1.1 Figure 6. Association and Dissociation Kinet-

t
ics of Antigenic Peptide Binding to HLA-DRl
Insect from Human and Insect Cells
, ,q _ _ _. - - - - -. i Left panel: association kinetics. Soluble DRl
I’ isolated from insect cells (0.14 uM), or pro-
I duced by papain cleavage of DRI from human
cells (0.21 pM), was incubated with 2.5 uM
[‘*“l]HA(306-318) peptide at 37%. At the indi-
cated times the binding reaction was stopped,
and the amount of bound peptide was deter-
i
mined by immunoabsorption. Squares, soluble
Human
DRl from coinfected insect cells; circles,
papain-solubilized DRl from human lympho-
blastoid cells. Closed symbols, DRl + 9-
0.1 A 1 labeled HA peptide, open symbols, DRl + 1*51-
0 20 40 60 80 0 100 200 300 labeled HA peptide + 20-fold excess cold HA
Trme (hrs) Time (hrs) peptide. Solid lines indicate the initial rate of
peptide binding; dashed lines indicate the best
fit single exponential equations, with r = 7.8 hr and a maximum of 12,000 cpm for insect-cell-produced DRl, and with r = 81 hr and an extrapolated
maximum of 4,800 cpm for human-cell-produced DRl.
Right panel: dissociation kinetics, DRI-peptide complexes were formed as described above, isolated by spin ultrafiltration, and diluted to 25 nM
DRl in binding buffer containing 0.25 mM unlabeled peptide. At the indicated times DRl-peptide complexes were again isolated, and the amount
of radiolabeled peptide remaining bound to DRl was determined by gamma counting. Dashed lines indicate single exponential fits with r = 81 hr
for DRI from insect cells and ‘I = 52 hr for DRl from human cells.
Cell
470

8x105, Table 2. Stoichiometry of Peptide Binding to HLA-DRl from Human

and Insect Cells

Peptide Released
[‘251]HA(306-3 18) by Acid Treatment
Peptide Bound (mol amino acid
DRI Source (mol peptide/mol DRl) residuelmol DRl)

Human 0.17 + 0.07 14

Insect 0.90 f. 0.1 ND
Insect
(preloaded) - 13
0
The extent of [‘251]HA(305-318) peptide binding to soluble DRl pro-
4 5 6 7 8
duced in insect cells and in human cells was determined for the experi-
PH ments shown in Figures 5,6, and 7 (pH 7 values only) and in four other
Human Insect trials. In each experiment, DRl samples from insect and human cells
n t2%peptide q l2%peptide were treated in parallel. Occupancy values for ‘251-labeled HA peptide
are given as the ratio of moles peptide bound per mole DRl , deter-
0 125bePide 0 1251-peptide
+ unlabeled mined using the measured specific activity of the [‘Z51]HA(306-318)
+ unlabeled
preparation and the concentration of DRl determined by ELISA or
Figure 7. pli Dependence of Antigenic Peptide Binding to HLA-DRl absorbance at 280 nm. Values are the average of seven trials with
from Human and insect Cells the observed standard deviation. The amount of endogenous peptide
Soluble DRl, produced by insect cells (0.2 PM, shaded bars) or pre- bound to soluble DRl was determined for papain-solubilized human
pared by papain digestion of DRl purified from human cells (0.35 uM, DRl and secreted insect cell DRl , and also for insect cell DRl pre-
solid bars), was incubated with 1.8 uM [‘251]HA(30~318) peptide at loaded with HA(306-318) peptide. Bound peptides were released by
37% in 0.1 M sodium citrate-phosphate buffer at the indicated pH. acid treatment and isolated by spin ultracentrifugation (see Experimen-
tal Procedures). Occupancy values are given as the ratio of moles
After 96 hr. the amount of radioactive peptide bound to DRl was deter-
mined by spin ultrafiltration. Radiolabeled peptide binding in the pres- amino acid residue in the peptide fraction, determined by amino acid
ence of 25 uM unlabeled peptide is indicated by open bars for DRI analysis, per mole DRI, determined by absorbance at 280 nm. ND,
from human cells and by lightly hatched bars for DRI from insect cells. none detected above the background reactivity observed for a buffer
blank. Detection limit was approximately 5 amino acid residues per
mole DRl.

For measurement of dissociation kinetics(Figure6, right

panel), DRl samples were equilibrated with excess
[1251]HA(306-318) peptide for 73 hr at 37%. After this time, Soluble DRl produced in insect cells reproducibly bound
DRl-peptide complexes were separated from free pep- nearly a stoichiometric amount of peptide (0.90 + 0.15
tide, diluted into buffer containing excess unlabeled pep- mol per mole DRl), while DRl from human cells bound
tide, and returned to 37%. Samples were removed at the 5-fold less peptide (0.17 + 0.07 mol peptide per mole
indicated times, and the amount of peptide remaining DRl). The low binding capacity, slow association kinetics,
bound to DRl was measured. The kinetics of peptide dis- and pH dependence of peptide binding for class II mole-
sociation were extremely slow for DRl from either source, cules isolated from mammalian cells are all believed to be
and no significant difference in dissociation rate was ob- due to the presence of tightly bound peptides occupying
served over 300 hr. The dissociation data for DRl from the antigen-binding site, which must dissociate before ex-
both sources are consistent with a first-order dissociation ogenously added peptide will bind (Buus et al., 1986;
constant of about 4 x 10Y6 s-l. Roche and Cresswell, 1990; Tampe and McConnell,
The pH dependence of peptide binding to DRl from 1991). Taken together, the increased peptide binding ca-
human and insect cells was also determined. Binding of pacity, increased binding rate, and decreased pH depen-
excess [1251]HA(306-318) peptide to DRl was measured dence of peptide binding for DRl produced in insect cells
after 72 hr incubation at 37%, for soluble DRl from human indicate that, as isolated, the antigen-binding site is largely
cells (Figure 7, solid bars) and from insect cells (shaded empty.
bars). Open and hatched bars show binding in the pres- To confirm this result, we directly measured the amount
ence of excess unlabeled peptide. Peptide binding to hu- of endogenous peptide bound to DRl, using a procedure
man lymphoblastoidcell-derived DRl increased at lower previously used to characterize peptides bound to class I
pH. In contrast, peptide binding to insect-cell-derived DRl and class II molecules (Van Bleek and Nathenson, 1990;
was relatively independent of pH. At every pH tested, in- Falk et al., 1991; Rudensky et al., 1991). A pool of bound
sect cell DRl bound more peptide than DRl from human peptides was released from the DRl-binding site by acid
lymphoblastoid cells. The extent of peptide binding ob- denaturation, isolated by spin ultrafiltration, and finally
served for the insect-cell-derived DRl corresponds to quantitated by amino acid analysis. Papain-solubilized
1 .l it 0.2 mol peptide per mole protein. For the DRl iso- DRl isolated from human cells carried the equivalent of
lated from human cells, the extent varied from 0.06 (pH 8) 14 amino acid residues per mole (Table2). Full-length DRl
to 0.3 (pH 4). from human cells gave essentially the same result. This
The measurements of peptide binding capacity were corresponds to approximately 95% occupancy, with en-
repeated using different preparations of [1251]HA(306-31 8) dogenous peptides having an average length of 15 resi-
peptide and DRl from insect and human cells (Table 2). dues (L. J. S. and D. C. W., unpublished data). As acontrol,
Empty HLA-DRl from Insect Cells
471
soluble DRl from insect cells was analyzed after loading
with HA(306-318) peptide. The isolated DRl-peptide
complexes carried 13 amino acid residues per mole DRl ,
consistent with the length of the HA(306-318) peptide and
a 1: 1 molar ratio of bound peptide to DRl . In contrast, no
amino acid residues were detected in the pool from soluble
DRl from insect cells, above the reactivity observed in a
buffer blank.
Discussion
We have expressed the human class II major histocompat-
ibilityglycoprotein HLA-DRl in baculovirus-infected insect
cells in a membrane-bound and a soluble, secreted form.
In both cases the a and p subunits associate to form the
native heterodimer. By several criteria, DRl purified from
insect cells has the native protein fold: it reactswith several
conformation-sensitive monoclonal antibodies directed
against the native human protein; it forms a long-lived com-
plex with HA(306-318), an antigenic peptide from influ-
enza virus; and it crystallizes under the same conditions
that crystallize DRl from human cells. However, the anti-
genie peptide-binding site of DRl produced in insect cells
is not occupied by tightly bound endogenous peptides, as
it is for DRl and other human and mouse class II proteins
isolated from lymphoblastoid cells. Instead, the peptide-
binding site appears to be empty. The empty DRl mole-
cules are fully competent to bind a stoichiometric amount
of added peptide. The availability of empty DRl allowed
us to investigate the mechanism of peptide binding, and
to determine the effect of peptide in stabilizing the DRl
structure.
Using the empty DRl expressed in insect cells, we were
able to measure directly an intrinsic rate of peptide associ-
ation. In earlier studies the rate of exchange was mea-
sured between an added peptide and the collection of en-
dogenous peptides already bound into the antigen-binding
site, and only 5%-20% of the class II molecules could bind
the added peptide (Watts and McConnell, 1986; Buus et
al., 1987, 1988; Ceppellini et al., 1989; Jardetzky et al.,
1990; O’Sullivan et al., 1990; Roche and Cresswell, 1990;
Busch and Rothbard, 1990; Tamp&and McConnell, 1991).
In those experiments, slow peptide binding could not be
clearly distinguished from the slow dissociation of endoge-
nous peptides (Tamp& and McConnell, 1991). For DRl
and HA(306-318) peptide, we found that at pH 7.0 the kinet-
ics of peptide association were 1 O-fold faster than for ex-
change into an occupied site (Figure 6). This association
rate is still exceedingly slow compared with similar pro-
cesses and suggests that a protein conformational change
occurs during formation of the DRl-peptide complex. For
comparison, the reaction of penicillopepsin with peptides,
which involves binding up to six amino acids of a peptide
and a change in the conformation of a flap on the enzyme,
occurs with a forward rate constant of 1 04-1 O6 M-Is-l (Hof-
man et al., 1988), many thousand-fold faster than the rate
of peptide binding to class II histocompatibility antigens
measured in vitro. While peptide binding to class II proteins
in vivo may involve factors not present in the in vitro sys-
tems, studies of peptide binding have found that peptide
association occurs in vivo on the same slow time scale
measured in vitro (Busch and Rothbard, 1990; Nygard et
al., 1991). The slow kinetics for peptide association even
when the antigen-binding site is unoccupied argue that a
slow conformational rearrangement is required to accom-
modate a peptide within the antigen-binding site.
The tendency of the empty DRl from insect cells to
aggregate, and the striking disaggregation after peptide
loading, also suggest an altered conformation in the empty
molecule. Although much of the native structure must be
preserved in the empty state, as all antibodies tested still
bound to the empty molecule, a conformational alteration
in one or both subunits may not have been detected by the
antibodies used. Alternate conformational states of class
II MHC proteins have been previously proposed, based
on a “floppy” intermediate observed during denaturation
(Dornmair et al., 1989) and refolding (Dornmair and
McConnell, 1990), on the kinetics (Sadegh-Nasseri and
McConnell, 1989) and pH dependence (Jensen, 1991;
Sadegh-Nasseri and Germain, 1991; Wettstein et al.,
1991) of peptide binding, and on differential antibody reac-
tivity (Germain and Hendrix, 1991).
We were also able to investigate the role of peptide in
stabilizing the class II up heterodimer against SDS-
induced dissociation. Mouse and human class II proteins
are unusually resistant to SDS and migrate as up com-
plexes on SDS-PAGE if the samples are not heated prior
to loading. However, under certain circumstances class II
molecules have been observed that are readily dissoci-
ated by SDS: in a mutagenized lymphoblastoid line that is
deficient in antigen processing (Mellins et al., 1990), for
newly synthesized class II before dissociation of invariant
chain (Germain and Hendrix, 1991) and for purified protein
treated in vitro at high pH (Gorga et al., 1986) or low pH
(Sadegh-Nasseri and Germain, 1991). In the last instance,
resistance to SDS was partially recovered after exposure
to antigenic peptides before neutralization, and it was sug-
gested that peptide contributes to the resistant structure
(Sadegh-Nasseri and Germain, 1991). In the present case,
the recovery of SDS resistance after stoichiometric pep-
tide binding clearly demonstrates that resistance to SDS-
induced chain dissociation is directly due to peptide bind-
ing, and not to a pH or other effect.
In vivo, nascent class II molecules are believed to en-
counter antigenic peptide in an endosomal compartment
after transit through the endoplasmic reticulum (Guagli-
ardi et al., 1990; Neefjes et al., 1990; Davidson et al.,
1991). Low endosomal pH has been suggested to be im-
portant in generating a conformation of class II that is able
to bind peptide (Jensen, 1990; Sadegh-Nasseri and Ger-
main, 1991; Wettstein et al., 1991). We find that DRl pro-
duced in insect cells is fully competent to bind antigenic
peptide without low pH treatment. While the low endoso-
mal pH may be required for efficient generation of peptides
by proteolysis of endocytosed antigens, or may help to
dissociate class II from bound invariant chain molecules,
it does not appear to be required for peptide binding. We
have not, however, investigated any effect of pH on the
rate of peptide association to the empty molecules (Jen-
sen, 1991).
Cell
472
Similarly, our in vitro results indicate that the class II-
associated invariant chain, which appears to bind to na-
scent Class II molecules in vivo (Kvist et al., 1982; Cress-
well et al., 1987) and escort them to the peptide binding
compartment (Bakke and Dobberstein, 1990; Lotteau et
al., 1990), is not required for stoichiometric binding of pep-
tide to DRl Invariant chain association in vivo may serve
to block peptide binding until the nascent DRl reaches the
proper compartment (Teyton et al., 1990; Germain and
Hendrix, 1991; Roche and Cresswell, 1991). The invariant
chain may also act as a chaperonin by maintaining empty
class II molecules in a disaggregated state. In mammalian
cells, aggregated protein may be targeted for degradation
(Hurtley and Helenius, 1989; Klausner and Sitia, 1990),
and most empty, aggregated class II molecules may not
survive to bind antigenic peptides (Germain and Hendrix,
1991). It seems possible that the fate of class II molecules
recycling from the cell surface into a low pH compartment,
where they would lose peptide, would also be to aggregate
and then be discarded, in the absence of binding the invari-
ant chain or another chaperonin. This could provide a con-
trol point to assure that most class II molecules present
peptides encountered in the compartment designated by
the trafficking signals of the invariant chain. We and others
(Traunecker et al., 1989; R. Germain, personal communi-
cation) have had difficulty producing soluble class II mole-
cules in nonlymphoid mammalian cells. Insect cells may
lack a retention/degradation pathway for aggregated pro-
teins, or it may become disabled during baculovirus infec-
tion (Jarvis and Summers, 1989; Davidson and Castellino,
1991), perhaps accounting for the general success of the
baculovirus-insect cell system in producing and secreting
recombinant proteins and its utility in producing empty
class II molecules.
Empty DRl molecules from insect cells form stable ctp
heterodimers in the absence of antigenic peptide. The
empty heterodimers remain associated throughout immu-
noaffinity isolation and week-long incubation at 37°C. This
stability is in marked contrast to class I histocompatibility
antigens, where the subunits do not stably assemble in the
absence of peptide. The heavy and light chains of class I
proteins transcribed in vitro (Kvist and Hamann, 1990) and
in mutant cell lines defective in peptide processing (Town-
send et al., 1990; Ljunggren et al., 1990; Hosken and Be-
van, 1990; Ortiz-Navarrete and HBmmerling, 1991) do not
stably assemble in the absence of antigenic peptide. The
empty class I heterodimer produced in the absence of
peptide quickly loses conformational epitopes and dissoci-
ates, unless stabilized by exogenously added peptide.
The difference in stability between empty class I and
class II can be understood in terms of the roles of class I
and class II proteins in the antigen-presenting cell. T cell
recognition of a class I-peptide complex on the cell sur-
face targets the cell for lysis. Inappropriate T cell-medi-
ated lysis is prevented by tight regulation of the peptides
that are presented: peptide loading on class I molecules
is restricted to the endoplasmic reticulum, where peptides
are apparently provided by MHC-linked cytoplasmic prote-
ases (Brown et al., 1991; Glynne et al., 1991; Ortiz-
Navarrete et al., 1991; Martinez and Monaco, 1991) and
trans-endoplasmic reticulum peptide transporters (Dever-
son et al., 1991; Trowsdaleet al., 1991; Spieset al., 1991;
Spies and DeMars, 1991). Class I molecules that do not
acquire a peptide are not stable and for the most part are
not transported to the cell surface (Ljunggren et al., 1990;
Townsend et al., 1990; Ortiz-Navarrete and HBmmerling,
1991). By contrast, class II molecules generally present
peptides derived from exogenous antigens, which are pro-
teolytically processed in the endosomal compartments of
specialized antigen-presenting cells. Class II molecules
must normally transit the endoplasmic reticulum in the
empty state and arrive at the endosomes competent to
bind peptide. Appropriately, class II molecules can form
stable up heterodimers in the absence of peptide, in vitro
and presumably in the endoplasmic reticulum; and peptide
binding is a separate step that occurs when the class II
molecules arrive in the proper cellular compartment.
Transport of class II molecules to this compartment and
prevention of peptide binding in theendoplasmic reticulum
both seem to be functions of the invariant chain. The ten-
dency of the empty class II molecules to aggregate as
observed here could also argue that the invariant chain
may possess a chaperonin-like activity that can stabilize
empty class II molecules against aggregation.
In conclusion, expression of the human class II histo-
compatibility antigen HLA-DRl in insect cells has provided
a source of native, soluble up heterodimers that have an
empty peptide-binding site. Heterodimer formation is not
dependent on peptide binding, as it is for class I histocom-
patibility antigens. Peptide binding does, however, stabi-
lize the molecule against aggregation and increases the
resistance to SDS-induced denaturation. Low endosomal
pH is not needed for peptide binding, although it may be
required for antigen processing, or invariant chain dissoci-
ation, or it may influence the rate of the binding reaction.
Peptide binding at pH 7.0 is an intrinsically slow process,
even for the empty molecule, indicating that an unusual
conformational change is required to accommodate (or
trap) a peptide within the binding site.
Experimental Procedures
Reagents and Materials
Oligonucleotides were synthesized with a Milligen model 3700 DNA
synthesizer using Pcyanoethyl phosphoroamidite chemistry, and were
purified by denaturing acrylamide gel electrophoresis and reverse-
phase chromatography on Seppack (Millipore) cartridges. Baculovi-
rus transfer plasmids pVL1393 and pAC360-Bgal and the wild-type
baculovirusACMNPV-EPwereprovided by M. Summers(TexasA&M).
Restriction enzymes and DNA modifying enzymeswere obtained from
New England Biolabs, Boehringer Mannheim, US Biochemicals, and
Promega.
Hybridoma cells secreting anti-DR monoclonal antibody L243
(IgG*d were obtained from the American Type Culture Collection
(ATCC #HB55) and were maintained in Dulbecco’s modified Eagle’s
medium (DMEM; Sigma) plus 10% fetal bovine serum (FBS). LB3.1-
secreting (lgGZD)cells were obtained from J. Strominger (Harvard Uni-
versity) and were maintained in RPM 1640 (Sigma) plus 10% FBS.
For antibody production, cells were grown with immunoglobulin G-free
FBS (Gibco) in roller bottle culture or in serum-free WRC935 me-
dium (Amicon) in a mini-FlowPath bioreactor (Amicon). Antibodies
were purified from clarified tissue culture medium by ammonium sul-
fate fractionation followed by affinity chromatography on protein A-
agarose (Repligen) of protein G-Sepharose Fast Flow (Pharmacia).
Empty HLA-DRl from Insect cells
473

Phycoerythrin-conjugated L243 and control mouse immunoglobulin annealing (55OC, 1 min), and extension (72OC, 3 min). The reaction
G were obtained from Becton-Dickinson. Rabbit antiserum against product was isolated, cut with BamHl and Xbal, and inserted into the
papain-solubilized DRl was produced by Hazelton. Anti-DRI mono- corresponding restriction sites of pUCl9. One of three clones se-
clonal antibodies lVAl2@1), TAL14.1@1), Tii36@2), Tii39@1), quenced had no unexpected substitutions, and the DRBsol gene was
Tii43(ab), and SG171(61) and biotinylated monoclonal antibodies excised from this clone and inserted between the BamHl and Xbal
DAP(Pl), DA6.147(01), DA6.321(a), TALB.l(Pl), and L227(61) were sites of pVL1393.
obtained from D. Vignali (Harvard University). The specificity of each
Construction of Recombinant Baculovlrus Clones
antibody for DR domains is indicated in parentheses (D. Vignali, un-
Recombinant baculoviruses BV-Pgal, BV-DRa, and BV-DRfi were pro-
published data). Rabbit antisera specific for the a and p chains of DRl
duced by homologous recombination following cotransfection of 2 x
were a generous gift of H. Ploegh (The Netherlands Cancer Institute).
106 cells with 5 pg of plasmid and 1 wg of viral (wild-type ACMNPV-EP)
Goat anti-rabbit or anti-mouse secondary antibodies were obtained
DNAs, as described (Summers and Smith, 1966). Recombination effi-
from Boehringer Mannheim (horseradish peroxidase-labeled) and
ciencies varied from 0.1% to 1%. Viral clones were isolated by limiting
Promega (alkaline phosphatase-labeled). Streptavidin-alkaline phos-
dilution in 96-well tissue culture plates. Recombinant viruses were
phatase was from Biorad.
identified by dot-blot DNA hybridization of alkali-lysed cells (Summers
Immunoaffinity-purified DRl, isolated from the human lymphoblas-
and Smith, 1966), using a “P-labeled probe carrying both DRa and
toid cell line LG2, and soluble DRI, produced by limited papain diges-
DRB sequences. Three or four rounds of dilution and screening were
tion of immunoaffinity-purified DRl from LG2, were generous gifts of
required to obtain single isolates free of wild-type virus. Recombinant
J. Gorga and J. Strominger (Harvard University). Glycosidases,
baculoviruses BV-DRasol and BV-DRBsol were similarly produced and
digoxygenin-labeled lectins, and detergents were from Boehringer
isolated, except that BV-6gal viral DNA was used instead of wild-type
Mannheim. HA(306-316) peptide (NH2-PKYVKQNTLKLAT-COOH)
ACMNPV-EP. This simplified the identification of nonrecombinant vi-
was synthesized with an ABI model 431 peptide synthesizer using
ruses, which were easily observed by including 5-bromo-4-chloro-3-
Fmoc chemistry, and was purified by reverse-phase high-pressure
indole-fl-o-galactoside (0.2 mg/ml) in the culture medium.
liquid chromatography(HPLC) on &ProPep (Vydac) in 0.1% trifluoro-
acetic acid using a 40%-60% acetonitrile gradient. The purified pep
tide was characterized by amino acid analysis (Harvard Microsequenc- Sf9 Growth and Infection
ing Facility) and by mass spectrometry (Harvard Spectrometry Lab) Spodoptera frugiperda (Sf9) were obtained from the American Type
and shown to be homogeneous. Peptide concentration was deter- Culture Collection (ATCC# CRL1711) and were maintained at 27°C in
mined by ultraviolet absorbance, using ssW = 1600 M-‘cm-‘. TNM-FH medium (Gibco) plus 10% FBS. Viral stocks were produced
by infection at low multiplicity and were stored at 4°C. Viral titerg
v&e usually greater than 108 plaque-forming units per ml. For protein
Construction of Transfer Plasmids Carrying DRa, DRP, production, cells were grown in spinner flasks (100 ml or larger) in
Truncated DRa, and Truncated DRB Genes serum-free media SF900 (Gibco) or Excel1410 (JRH Scientific). Cells
cDNA clones for the a and p subunits of HLA-DRl (DRA and were infected at 1 x 106 cells per ml with a multiplicity of infection of
DRB7 ‘0707, GENBANK identifiers: Hummhdram.pr and Hum- 20 for each virus, using the procedures described in Summers and
mhldrl b.pr) were obtained from Zhi Yang (Harvard University). Trans- Smith (1966).
fer plasmids carrying DRa and DRP genes were constructed by isola-
tion of the genes as BamHl fragments from the appropriate cDNA SDS-PAGE and Western Slotting
clones and insertion of these genes into the unique BamHl site of Cell lysates for SDS-PAGE analysis were prepared by mixing washed
the baculovirus transfer plasmid pVL1393. In this vector the inserted cells with 1110 culture volume phosphate-buffered saline (PBS: 20
genes are under transcription control of the strong late polyhedrin mM phosphate, 130 mM NaCl [pH 7.21) containing 1% CHAPS and a
promoter. The initiation codon of the polyhedrin gene has been altered mixture of protease inhibitors (1 mM EDTA, 1 mM phenylmethylsulfo-
to ATT (Luckow and Summers, 1969), so that translation is initiated at nyl fluoride (PMSF), 0.1 mM iodoacetamide, 0.3 NM aprotinin, 1 PM
the first ATG in the inserted gene. Clones carrying DRa or DR!3 inserts pepstatin, 1 PM leupeptin). The mixture was stored at 4OC for 1 hr,
in the proper orientation were isolated, and the expected sequences and nuclei and cell debris were removed by low-speed centrifugation.
were confirmed throughout the entire coding regions. Samples of extracellular medium for SDS-PAGE were prepared by
DRasol (Figure 1) was constructed by using a synthetic oligonucleo- acetone or trichloroacetic acid precipitation. Samples for SDS-PAGE
tide duplex that codes for DRa sequence from the unique Pstl site at were mixed with SDS-PAGE sample buffer (Laemmli, 1970) con-
nucleotide 566 to the Asn-192 codon ending at nucleotide 651, fol- taining 1% SDS and 100 mM dithiothreitol (Dll) (final concentrations)
lowed by the termination codon TAA, and Notl and Kpnl cloning sites. and boiled for 3 min before application to 12.5% acrylamide SDS-
The sequences of the constituent oligonucleotides were: PAGE slab gels (7 x 7 x 0.075 cm). After electrophoresis, gels were
transferred to polyvinylidene fluoride (PVDF) membranes (Immobi-
Ion-P, Millipore). Membranes were blocked in 3% bovine serum albu-
5”GGGTGGAGCACTGGGGCTTGGATGAGCCTCTTCTCAAGCATT
min (BSA) in PBS. DRa and DR6 polypeptides were detected using
GGAATTCGATGCTCCAAGCCCTCTCCCAGAGACTACAGAGAACTA
appropriate antisera, followed by alkaline-phosphatase conjugated
AGCGGCCGCGGTAC-3’
anti-rabbit serum, and nitro-blue tetrazolium and bromochloroindole
5”CGCGGCCGCTTAGTTCTCTGTAGTCTCTGGGAGAGGGCTTGGA
phosphate as described (Blake et al., 1964).
GCATCGAATTCCAATGCTTGAGAAGAGGCTCATCCAAGCCCCAGT
GCTCCACCCTGCA-3
Flow Cytometry
Altered sequences relative to the DRa gene are italicized; the first Baculovirus-infected cells were analyzed by flow cytometry at 46 hr
two substitutions are silent changes to introduce a unique EcoRl site. postinfection, before significant virus-induced cell lysis. In order to
The syntheticduplex was inserted into pVLl393-DRa between the Pstl avoid the strong green autofluorescence intrinsic to Sf9 insect cells,
site in the DRa gene and a Kpnl site downstream in the disabled long-wavelength fluorophore R-phycoerythrin (PE) was used. At 46
polyhedrin gene. Oneclonecarrying theinsertwassequenced through hr postinfection, 106 cells were pelleted, gently resuspended in 1110
the altered region and shown to have the expected sequence. culture volume Grace’s medium (Gibco), 2% FBS, 0.01% NaNJ, and
DR6sol (Figure 1) was constructed by polymerase chain reaction- incubated for 1 hr on ice with PE-conjugated L243 or PE-conjugated
mediated amplification of the DR6 gene. The”forward”oligonucleotide nonspecific control mouse immunoglobulin G. The cells were washed
primer, complementary to the coding strand, 5”GACTTGGATCCTA- three times with Grace’s medium and were finally resuspended at
TAAATATGGTGTGCTGAAGCTCCT-3’, introduces a BamHl site up 1110 the initial culture volume in PBS and fixed with 2% paraformalde-
stream of the initiation ATG codon, and the reverse primer, 5’-ACAGCT- hyde. Red fluorescence was measured with a Becton-Dickinson FACS-
CTAGATTAClTGCTCTGTGCAGAlTCAGA-3’, introduces a termi- can flow cytometer.
nation TAA codon starting at nucleotide 662, followed by an Xbal clon-
ing site. Sequences not present in the DRB gene are italicized. The Enzyme-Linked lmmunosorbent Assay (ELISA)
truncated gene was amplified by 10 cycles of melting (94OC, 1 min), The ELISA used to measure DRl concentration was a sandwich type,
Cell
474

with solid phase L243 or LB3.1 monoclonal antibodies used to capture containing fractions were pooled and concentrated into PBS using a
native DRl, and rabbit anti-DRl and alkaline-phosphatase-labeled spin ultrafiltration device (Centricon-30, Amicon). For purification of
goat anti-rabbit antibodies used to detect bound DRl Ninety-six-well the full-length protein from BV-DRa+BV-DR8-coinfected cells, all solu-
microtiter plates (Maxisorp, Nunc) were coated with 200 ng of purified tions contained 1% CHAPS. The concentration of purified DRl was
L243 or LB3.1 monoclonal antibody in 100 mM sodium carbonate (pH determined by ultraviolet absorbance at 280 nm using an extinction
9.6) blocked with 3% BSA in PBS, and stored at 4OC. All subsequent coefficient of 77,000 Mm’cm-‘.
incubations were for 30 min or 1 hr at room temperature using 0.1 ml N-terminal sequence analysis of purified soluble DRl from insect
per well and were followed by three washes with 0.05% Triton X-100 cells was performed after separation of the subunits by SDS-PAGE
in PBS (PBST). Dilutions of samples and DRI standards (0.1-100 ng and transfer to PVDF, by automated Edman degradation, as described
per well) were prepared in PBST plus 0.3% BSA and applied to the (Matsudaira, 1987).
plate. After binding, DRl was detected using rabbit anti-DRl serum
(1:50,000 in PBST plus 0.3% BSA) followed by horseradish peroxi-
dase-coupled goat anti-rabbit antibody (15 pglml in PBST plus 0.3% Glycosylation Analysis
BSA). The plate was developed with the peroxidase substrate For glycosidase analysis, purified DRl samples were denatured by
2,2’azino-di[3-ethyl]benzthiazoline sulfonate (ABTS, Boehringer boiling in 1% SDS plus 1% 8-mercaptoethanol, then cooled and diluted
Mannheim) in perborate-citrate-phosphate buffer. After 5-15 min, the lo-fold into PBS containing protease inhibitors and 1% dodecylmalto-
reaction was stopped with 0.2% NaN,, and the absorbance at 405 nm side. Endoglycosidase-H (EC 3.2.1.96, 0.005 U per mg of DRl), or
was measured. For quantitation of DRl, triplicate sample dilutions glycopeptidase-F (EC 3.2.2.18, 1 U per mg of DRl), or an equivalent
were compared to a standard curve produced using purified, papain- volume of PBS, was added, and the mixtures were incubated at 37OC
solubilized DRl from human lymphocytes. The four-parameter binding for 12 hr. The reactions were stopped by again boiling in SDS and the
equation (Tijssen, 1965) reaction products were analyzed by 12.5% acrylamide SDS-PAGE.
For lectin analysis, purified DRl samples were analyzed as described
A = (A,, - A,:,) I (1 + (C/C,,$) + A,,,
above for Western blotting. Parallel blots were incubated with each of
where A is the absorbance caused by a sample of concentration C, the digoxigenin-labeled lectins, and then with alkaline phosphatase-
and A,,, A,,., &, and b are adjustable parameters, was fit to the labeled anti-digoxigenin, and were developed as described above.
standard curve by a nonlinear least squares algorithm, and was used DRa and DR8 bands were identified by comparison with parallel blots
to convert sample absorbances to DRl concentrations. analyzed with rabbit anti-DRu and anti-DR8 sera.
For determination of the reactivity of DRl from insect or human cells
with a panel of anti-DRI antibodies, a direct-binding ELISA was used.
Microtiter plates were coated with 200 ng per well of DRl and blocked Peptide Binding to Purified DRl
as above. Serial dilutions of monoclonal antibodies or biotinylated Immunoaffinity-purified, soluble DRl (0.05-l .O FM) from insect or hu-
monoclonal antibodies were added to the plate, and bound antibodies man cells was used in binding reactions, with a 2- to lo-fold molar
were detected by alkaline-phosphatase goat anti-mouse antibodies, excess HA(306-318) peptide. Standard binding conditions were 37OC
or with streptavidin-alkaline phosphatase, and pnitrophenylphos- for >72 hr in PBS (pH 7.2) with 1 mM EDTA, 1 mM PMSF, 0.1 mM
phate (Harlow and Lane, 1966). The antibody dilution that produced iodoacetamide, and 3 mM NaN3. Incubation time, pH, and buffer were
one-half the maximal absorbance was used to compare the affinity of varied in someexperiments(seefigure legends). SDS-PAGEanalyses
each antibody for human- and insect-cell-derived DRl. were performed as described above, except that larger gels (14 x 14
x 0.15 cm) were used, and some samples were not boiled prior to
laolatlon of DRl from Coinfected Insect Cells loading, as noted in the figure legend. After electrophoresis, proteins
The procedures used to purify DRl from insect ceils were based on were detected with Coomassie brilliant blue R-250. HPLC analyses
those developed for isolation of DRl from human lymphoblastoid cells were performed using a 7.8 x 300 mm Waters Protein-Pak SW300 gel
(Gorga et al., 1967). Soluble DRI was isolated from the conditioned filtration column, equipped with a Waters l-125 guard column, and
culture medium of Sf9 insect cells coinfected with DRasol and DRf3sol. variable wavelength absorbance detector. PBS was used as the mobile
At 72 hr postinfection, cells were removed by centrlfugation and the phase, with a flow rate of 0.5 mllmin.
mixture of protease inhibitors was added. The culture medium was For quantitation of peptide binding, [12SI]HA(306-318) peptide was
concentrated approximately 1 O-fold using a spiral membrane cartridge used. Peptide (10 ug) was radiolabeled with 1 mCi of Na[?] and 50
(Amicon SIYIO) and used for immunoaffinity purification. pg of chloramine-T in phosphate buffer in a total volume of 50 ul for
Soluble DRl was also isolated from lysates of cells coinfected with 2 min at room temperature, the reaction was stopped by the addition
DRasol and DRpsol. Washed cells were lysed in 10 mM Tris-Cl (pH of excess Na2S205, and the peptide was isolated by gel filtration over
8.0) containing the protease inhibitor mixture, by repeated passage Sephadex G-15 (Pharmacia) in PBS. Peptide concentration in the la-
through a 23 gauge needle. The lysate was centrifuged (200,000 x g, beled preparations was determined using a bicinchoninic acid assay
30 min, 4OC) and the clarified lysate was enriched for DRl by ion by comparison with dilutions of an unlabeled peptide standard. Spe-
exchange chromatography on Q-Sepharose (Pharmacia) in 10 mM cific activities of the labeled peptide were 30,000-i 60,000 cpmlpmol
imidazole-HCI (pH 6.4) using a 50-250 mM NaCl gradient. DRl- in different preparations. Peptide bound to DRl was separated from
containing fractions were pooled and used for immunoaffinity purifi- free peptide by HPLC gel filtration (as above), immunoabsorption, or
cation. spin ultrafiltration. Bound ‘251-labeled peptide was measured by
Full-length DRl was isolated in detergent solution from BV- gamma counting.
DRa+BV-DR8-coinfected Sf9 insect cells. Washed cells were lysed For immunoabsorption, polystyrene microtiter wells (RIA/EIA 8-well
with 1% CHAPS in PBS containing the protease inhibitor mixture. strips, Costar) were coated overnight with 2 pg of purified L243 in 50
Nuclei and insoluble materials were removed by low-speed centrifuga- mM sodium carbonate (pH 9.8) and blocked with 5% nonfat dry milk.
tion (2,500 x g, 5 min, 4OC). The supernatant was cleared by ultracen- Milk was used to reduce nonspecific absorption, rather than BSA as
trifugation (200,000x g. 30 min, 4OC) and used for immunoaffinity in the ELISA assay, since radiolabeled HA(308-316) showed some
purification. binding to BSA. The DRl binding capacity of these plates was deter-
lmmunoaffinity matrices were prepared using anti-native DRl mined to be 50 ng per well, and they were always used with subsaturat-
monoclonal antibodies LB3.1 or t-243. Purified antibodies were cou- ing DRl concentrations. Peptide binding mixtures (in triplicate) were
pled at 5 mglml to protein A-agarose (Repligen) or to protein G-Sepha- added to an equal volume of blocking solution in the antibody-coated
ross Fast Flow using dimethyl pimelimidate as described (Harlow and wells and were allowed to bind for 1 hr at room temperature. The wells
Lane, 1988). Samples for immunoaffinity purification were passed were washed five times with PEST before gamma counting. For spin
through uncoupled protein A or protein G columns before application ultrafiltration, DRl and DRl-peptide complexes were separated from
to the immunoaffinity column. lmmunoaffinity columns were washed free peptide by five cycles of concentration and 25-fold dilution into
with PBS, and DRl was eluted with 50 mM sodium cycloheylaminepro- PBS, using Centricon- ultrafiltration devices (Amicon). Before use,
panesulfonate (CAPS) buffer (pli 11.5). Eluted fractions were immedi- the Centricon- devices were blocked with 5% nonfat dry milk and
ately neutralized with 100 mM sodium phosphate (pH 6.0). Protein- washed with PBS.
Empty HLA-DRl from insect Cells
475

Isolation and Crystallization of Soluble serological similarity of MHC-linked LMP and proteasome (multicata-
DRl -Peptlde Complexes lytic proteinase) complexes. Nature 353, 355-357.
Immunoaffinity-purified soluble DRI (0.5-l mM) was incubated with Busch, R., and Rothbard, J. 8. (1990). Detection of peptide-MHC
2- to 5fold molar excess HA(306-316) peptide and with 1 mM EDTA, complexes on the surface of intact cells. J. Immunol. Meth. 734, 1-22.
1 mM PMSF, 0.1 mM iodoacetamide, and 3 mM NaN3 at 37°C for
Buus, S., Sette, A., Colbn, S. M., Jenis, D. M., and Grey, H. M. (1986).
>72 hr. DRl-peptide complexes were separated from free peptide,
Isolation and characterization of antigen-la complexes involved in T
aggregated DRl, and residual contaminating proteins by gel filtration
cell recognition. Cell 47, 1071-1077.
HPLC. The sharp peak corresponding to a molecular weight of about
Buus, S., Sette, A., and Grey, H. M. (1987). The interaction between
50,000 was collected and concentrated by spin ultrafiltration. For crys-
protein-derived immunogenic peptides and la. Immunol. Rev. 98,115-
tallization, DRl-peptide complexes (5 mg/ml) were transferred to 10
141.
mM Tris-Cl, 0.01% NaN3 (pH 6.0). Crystals were obtained by vapor
diffusion against 14-17% PEG 6000, 100 mM glycine (pli 3.5) using Buus, S., Sette, A., Colon, S. M., and Grey, H. M. (1986). Autologous
hanging drops on silanized microscope cover slips over precipitant peptides constitutively occupy the antigen binding site on la. Science
solution in 24-well tissue culture plates. 242, 1045-I 047.
Ceppellini, R., Frumento, G., Ferrara, G. B., Tosi, R., Chersi, A., and
Elutlon and Measurement of Bound Peptldes Pernis, 8. (1989). Binding of labelled influenza matrix peptide to HLA
A procedure similar to that published for elution of peptides from class DR in living B lymphoid cells. Nature 339, 392-394.
I MHC (Van Bleek and Nathenson, 1990; Falk et al., 1991; Jardetzky Chen, B. P., and Parham, P. (1989). Direct binding of influenza pep
et al., 1991) was used to elute DRI-associated peptides. DRI samples tides to class I HtA molecules. Nature 337, 743-745.
(50 ug) were separated from low-molecular-weight material by gel filtra- Cresswell, P., Blum, J. S., Kelner, D. N., and Marks, M. S. (1987).
tion HPLC as above, except that 170 mM aqueous ammonium acetate Biosynthesis and processing of class II histocompatibility antigens.
was used as the mobile phase. DRl containing fractions from each CRC Crit. Rev. Immunol. 7, 31-53.
sample was pooled, and any residual low-molecular-weight material
Davidson, D. J., and Castellino, F. J. (1991). Asparagine-linked oligo-
was removed by three cycles of 25fold concentration and dilution into
saccharide processing in lepidopteran insect cells. Temporal depen-
170 mM aqueous ammonium acetate using a Centricon- ultrafiltra-
dence of the nature of the oligosaccharides assembled on asparagine-
tion device as above. Bound peptides were eluted from the final con-
289 or recombinant human plasminogen produced in baculovirus
centrate by25-fold dilution into 10% aceticacid and incubation at 70°C
vector-infected Spodoptera frugiperda (IPLB-SF-21 AE) cells. Bio-
for 15 min. The samples were cooled and concentrated once again.
chemistry 30, 6167-6174.
The final filtrate provided the pool of peptides eluted by acid denatur-
ation. Filtrates were concentrated to 100 pl by vacuum centrifugation, Davidson, H. W., Reid, P. A., Lanzavecchia, A., and Watts, C. (1991).
and a portion was used for amino acid analysis on an ABI 420AI130A Processed antigen binds to newly synthesized MHC class II molecules
derivatizer/HPLC after hydrolysis with 6N HCI for 24 hr in vacua. A in antigen-specific B lymphocytes. Cell 67, 105-t 16.
sample of 170 mM ammonium acetate was processed in parallel Deverson, E. V., Gow, I. R., Coadwell, W. J., Monaco, J. J., Butcher,
through the HPLC, washing, elution, and analysis steps, to control for G. W., and Howard, J. C. (1991). MHC class II region encoding proteins
background and nonpeptidic reactivity. related to the multidrug resistance family of transmembrane transport-
ers. Nature 348, 736-741.
Acknowledgments Dornmair, K., and McConnell, H. M. (1990). Refolding and reassembly
of separate a and 8 chains of class II molecules of the major histocom-
We would like to thank Mia Frayser for excellent technical assistance patibility complex leads to increased peptide-binding capacity. Proc.
throughout the course of this work. The authors also wish to thank Joan Natl. Acad. Sci. USA 87, 4134-4138.
Gorga, Polina Klimovistsky, and Jack Stromingerfor DRl isolated from Dornmair, K., Rothenhtiusler, B., and McConnell, H. M. (1989). Struc-
human cells; Leslie Berg and Penelope Robbins for assistance with tural intermediates in the reactions of antigenic peptides with MHC
collection of fluorescence data; William Lane and Andrew Tyler for pep molecules. Cold Spring Harbor Symp. &ant. Biol. 54, 409-416.
tide characterization; Renee Robinson and William Lane for amino
Falk, K., Rotzschke, O., Stevanovic, S., Jung, G., and Rammensee,
acid analysis; Nine1 Sinitskaya for peptide and oligonucleotide synthe-
H.-G. (1991). Allele-specific motifs revealed by sequencing of self-
sis; M. Summers for insect cell materials and assistance; Hsiao Lin peptides eluted from MHC molecules. Nature 351, 290-296.
for assistance with cell culture; Ted Jardetrky, David Garboczi, and
Germain, R. N., and Hendrix, L. Ft. (1991). MHC class II structure,
Edward Collins for helpful discussions; and Ronald Germain fordiscus-
sions during revision of this manuscript. L. J. S. is supported by a occupancy and surface expression determined by post-endoplasmic
Damon Runyon-Walter Winchell Cancer Fund Fellowship (DRG1041). reticulum antigen binding. Nature 353, 134-139.
This research was supported by the Howard Hughes Medical Institute. Glynne, R., Powis, S. H., Beck, S., Kelly, A., Kerr, L.-A., andTrowsdale,
The costs of publication of this article were defrayed in part by J. (1991). A proteasome-related gene between thetwo ABC transporter
the payment of page charges. This article must therefore be hereby loci in the class II region of the human MHC. Nature 353, 357-360.
marked “advertisemenl in accordance with 18 USC Section 1734 Gorga, J. C., Knudsen, P. J., Foran, J. A., Strominger, J. L., and
solely to indicate this fact. Burakoff, S. J. (1986). lmmunochemically purified DR antigens in lipo-
Somes stimulate xenogenic cytolytic T cell in secondary in vitro cul-
Received August 27, 1991; revised November 12, 1991. tures. Cell. Immunol. 703, 160-173.
Gorga, J. C., Horejsi, V., Johnson, D. R., Raghupathy, R., and Strom-
References inger, J. L. (1987). Purification and characterization of class II histo-
compatibility antigens from a homozygous human B cell line. J. Biol.
Bakke, 0.. and Dobberstein, B. (1990). MHC class II-associated invari- Chem. 262, 16087-16094.
ant chain contains a sorting signal for endosomal compartments, Cell Gorga, J. C., Brown, J. H., Jardetzky, T. J., Wiley, D. C., and Strom-
63, 707-716. inger, J. L. (1991). Crystallization of HLA-DRl antigens. Res. Immu-
Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennett, W. S., Strom- not 742, 401-407.
inger, J. L., and Wiley, D. C. (1987). Structure of the human class I Guagtlardi, L. E., Koppelman, B., Blum, J. S., Marks, M. S., Cresswell,
histocompatibility antigen, HLAA2. Nature 329, 506-512. P., and Brodsky, F. M. (1990). Co-localization of molecules involved in
Blake, M. S., Johnston, K. H., RusselJones, G. J., and Gotschlich, antigen processing and presentation in an early endocytic compart-
E. C. (1984). A rapid, sensitive method for the detection of alkaline ment. Nature 343, 133-139.
phosphatase-conjugated anti-antibody on Western blots. Anal. Bio- Harding, C. V., Unanue, E. R., Slot, J. W., Schwartz, A. L., and Geuze,
them. 736, 175-179. H. J. (1990). Functional and ultrastructural evidence for intracellular
Brown, M. G., Driscoll, J.. and Monaco, J. J. (1991). Structural and formation of major histocompatibility complex class II-peptide com-
plexes during antigen processing. Proc. Natl. Acad. Sci. USA 67, of an immunogenic influenza hemagglutinin-human leukocyteantigen
5553-5557. class II complex. J. Exp. Med. 7 74, 243-251.
Harlow, E., and Lane, D. (1988). Antibodies: A Laboratory Manual Ortiz-Navarrete, V., and Hammerling, G. J. (1991). Surface appear-
(Cold Spring Harbor, New York: Cold Spring Harbor Laboratory). ance and instability of empty H-2 class I molecules under physiological
Hofman, T., Allen, B., Bendiner, M., Blum, M., and Cunningham, A. conditions. Proc. Natl. Acad. Sci. USA 88, 3594-3597.
(1988). Effect of secondary substrate binding in penicillopepsin: contri- Ortiz-Navarrete, V., Seelig, A., Gernold, M., Frentzel, S., Kloetzel,
butions of subsites SI and Sp’ to be,, Biochemistry 27, 1140-l 148. P. M., and Hammerling, G. J. (1991). Subunit of the”20S” proteasome
Hosken, N. A., and Bevan, M. J. (1990). Defective presentation of (mUltiCatalytiC proteinase) encoded by the major histocompatibility
endogenous antigen by a cell line expressing class I molecules, Sci- complex. Nature 353, 882-884.
ence 246387-370. O’Sullivan, D.. Sidney, J., Appella, E., Walker, L., Phillips, L., Col6n,
Hurtley, S. M., and Helenius, A. (1989). Protein oligomerization in the S. M., Miles, C., Chesnut, R. W., and Sette, A. (1990). Characterization
endoplasmic reticulum. Annu. Rev. Cell. Biol. 5, 277-307. of the specificity of peptide binding to four DR haplotypes. J. Immunol.
Jardetzky, T. S., Gorga, J. C., Busch, R., Rothbard, J., Strominger, 145, 1799-1808.
J. L., and Wiley, D. C. (1990). Peptide binding to HLA-DRI: a peptide Roche, P. A., and Cresswell, P. (1990). High-affinity binding of an
with most residues substituted to alanine retains MHC binding. EMBO influenza hemaggtutinin-derived peptide to purified HLA-DR. J. Immu-
J. 9, 1797-1803. not 744, 1849-1858.
Jardetzky, T. S., Lane, W. S., Robinson, R. A., Madden, D. R., and Roche, P. A., and Cresswell, P. (1991). Proteolysis of the class II-
Wiley, D. C. (1991). Identification of self peptides bound to purified associated invariant chain generates a peptide binding site in intracel-
HLA-B27. Nature 353, 328-329. lular HLA-DR molecules. Proc. Natl. Acad. Sci. USA 88, 3150-3154.
Jarvis, D. L.. and Summers, M. D. (1989). Glycosylation and secretion Rothbard, J. B., and Gefter, M. L. (1991). Interactions between immu-
of human tissue plasminogen activator in recombinant baculovirus- nogenic peptides and MHC proteins. Annu. Rev. Immunol. 9, 527-
infected insect cells. Mol. Cell. Biol. 9, 214-223. 585
Jensen, P. E. (1990). Regulation of antigen presentation by acidic pH. Rudensky, A. Y., Preston-Hurlburt, P., Hong, S-C., Barlow, A., and
J. Exp. Med. 171, 1779-1784. Janeway, C. A. (1991). Sequence analysis of peptides bound to MHC
Jensen, P. E. (1991). Enhanced binding of peptide antigen to purified class II molecules. Nature 353, 822-827.
class II major histocompatibility glycoproteins at acidic pH. J. Exp. Sadegh-Nasseri, S., and Germain, R. N. (1991). A role for peptide in
Med. 774, 1111-1120. determining MHC class II structure. Nature 353, 187-170.
Klausner, R. D., and Sitia, R. (1990). Protein degradation in the endo- Sadegh-Nasseri, S., and McConnell, H. M. (1989). A kinetic intermedi-
plasmic reticulum. Cell 82, 811-814. ate in the reaction of an antigenic peptide and I-EX. Nature 337, 274-
278.
Kvist, S., and Hamann, U. (1990). A nucleoprotein peptide of influenza
A virus stimulates assembly of HlA-827 class I heavy chains and Shackelford, D. A., and Strominger, J. L. (1983). Analysis of the oligo-
8,-microglobulin translated in vitro. Nature 348, 448-448. saccharideson the HLA-DRand DC1 Bcell antigens. J. Immunol. 730,
274-282.
Kvist, S., Wiman, K., Claesson, L., Peterson, P. A., and Dobberstein.
B. (1982). Membrane insertion and oligomeric assembly of HLA-DR Spies, T., and DeMars, R. (1991). Restored expression of major histo-
histocompatibility antigens. Cell 29, 81-89. compatibility molecules by gene transfer of a putative peptide trans-
porter. Nature 351, 323-324.
Laemmli, U. K. (1970). Cleavage of structural proteins during the as-
sembly of the head of bacteriophage T4. Nature 227, 880-885. Spies, T., Bresnahan, M., Bahram, S., Arnold, D., Blanck, G., Mellins,
E., Pious, D., and DeMars, R. (1991). A gene in the human major
Lampson, L. A., and Levy, R. (1980). Two populations of la-like mole-
histocompatibility complex class II region controlling the class I antigen
cules on a human B cell line. J. Immunol. 725, 293.
presentation pathway. Nature 348, 744-747.
Ljunggren, H.-G., Stam, N. J., Chlen, C., Neefjes, J. J., Hbglund, P.,
Summers, M. D., and Smith, G. E. (1988). A Manual of Methods for
Heemels, M.-T., Bastin, J., Schumacher, T. N. M., Townsend, A.,
Baculovirus Vectors and Insect Cell Culture Procedures: Texas Ag-
Karre, K., and Ploegh, H. L. (1990). Empty MHC class I molecules
ricultural Experiment Station Bulletin No. 1555 (College Station,
come out in the cold. Nature 346, 478-480.
Texas).
Lotteau, V., Teyton, L., Peleraux, A., Nilsson, T., Karlsson, L., Schmid,
Tamp& R., and McConnell, H. M. (1991). Kineticsof antigenic peptide
S. L., Quaranta, B., and Peterson, P. A. (1990). Intracellular transport
binding to the class II major histocompatibility molecule I-Ad. Proc.
of class II MHC molecules directed by invariant chain. Nature 348,
Natl. Acad. Sci. USA 88, 4881-4885.
800-805.
Teyton, L., O’Sullivan, D., Dickson, P. W., Lotteau, V., Sette, A., Fink,
Luckow, V. A., and Summers, M. D. (1989). High level expression of
P., and Peterson, P. A. (1990). Invariant chain distinguishes between
nonfused foreign genes with Autographa californica nuclear polyhedro-
the exogenous and endogenous antigen presentation pathways. Na-
sis virus expression vectors. Virology 7 70, 31-39.
ture 348, 39-44.
Martinez, C. K., and Monaco, J. J. (1991). Homology of proteasome
Tijssen, P. (1985). Practice and Theory of Enzyme Immunoassays:
subunits to a major histocompatibility complex-linked LMP gene. Na-
Laboratory Techniques in Biochemistry and Molecular Biology 75
ture 353, 884-887.
(New York: Elsevier Science Publishers).
Matsudaira, P. (1987). Sequence from picomole quantities of proteins
Townsend, A., Chlen, C., Bastin, J., Ljunggren, H.-G., Foster, L., and
electroblotted onto polyvinlylidene difluoride membranes. J. Biol.
Karre, K. (1989). Association of class I major histocompatibility heavy
Chem. 262, 10035-l 0038.
and light chains induced by viral peptides. Nature 340, 443-448.
Mellins, E., Smith L., Arp, B., Cotner, T., Celis, E., and Pious, D. (1990).
Townsend, A., Elliot, T., Cerundolo, V., Foster, L., Barber, B., and Tse,
Defective processing and presentation of exogenous antigens in mu-
A. (1990). Assembly of MHC class I molecules analyzed in vitro. Cell
tants with normal HLA class II genes. Nature 343, 71-74.
62, 285-298.
Neefjes. J. J., Stollorz, V., Peters, P. J., Geuze, H. J., and Ploegh,
Traunecker, A., Dolder, B., Oliveri, F., and Karjalainen, K. (1989). Solu-
H. L. (1990). The biosynthetic pathway of MHC class II but not class
bilizing the T-cell receptor-problems in solution. Immunol. Today 70,
I molecules intersects the endocytic route. Cell 61, 171-183.
29-32.
Nuchtern, J. G., Bonifacino, J. S., Biddison, W. E., and Klausner,
Trowsdale, J., Hanson, I., Mockridge, I., Beck, S., Townsend, A., and
R. D. (1989). Brefeldin A implicates egress from endoplasmic reticulum
Kelly, A. (1991). Sequences encoded in the class II region of the MHC
in class I restricted antigen presentation. Nature 339, 223-228.
related to the “ABC’superfamily of transporters. Nature 348,741-743.
Nygard, N. R., Bono, C., Brown, L. R., Gorka, J., Giacoletto, K. S.,
Van Bleek, G. M., and Nathenson, S. G. (1990). Isolation of an endoge-
Schaiff, W. T., Graham, M. B., McCourt, D. W., Kabeer, M., Braciale,
nously processed immunodominant viral peptide from the class I H-2K”
V. L., Braciale, T. J., and Schwartz, 8. D. (1991). Antibody recognition
molecule. Nature 348, 213-218.
Empty HLA-DRl from Insect Ceils
477

Watts, T. H., and McConnell, H. M. (1966). High-affinity fluorescent

peptide binding to 1-A” in lipid membranes. Proc. Natl. Acad. Sci. USA
83, 9660-9664.
Wettstein, D. A., Soniface, J. J., Fleay, P. A., Schild, H., and Davis,
M. M. (1991). Expression of a class II major histocompatibility complex
(MHC) heterodimer in a lipid-linked form with enhanced peptidelsolu-
ble MHC complex formation at low pH. J. Exp. Med. 774, 219-226.
Yewdell, J. W., and Bennink, J. R. (1990). The binary logic of antigen
processing and presentation to T cells. Cell 62, 203-206.