| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Centro Nacional de Biología Fundamental, Instituto de Salud Carlos III, Madrid, Spain
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The capacity of MHC class I molecules to present epitopes is linked,
among other factors (7), to the efficient generation and
function of CTL. Little is known about what determines the quantity of
pathogen-derived peptides presented by MHC class I molecules and the
possible existence of any rate-limiting step in this endogenous pathway
of processing and presentation. Theoretically, there is a number of
factors that may quantitatively affect the outcome of this pathway.
Some reports have indirectly addressed this question, but their
conclusions are only partial either because few experimental points
were tested, or because the possible effect of regions flanking the
epitope (8, 9) was not considered. One of the first
studies with murine CMV showed an increase in mice protection when the
yield of antigenic peptides from vaccine constructs also increased
(8). This and another report with three strains of
Listeria (10) suggest that the more Ag is
expressed, the better the recognition by CTL will be. On the other
hand, other reports indirectly suggest that other factors may limit the
cell capacity to present epitopes. Further studies with
Listeria suggest that increased Ag processing does not
result in better CTL induction in vivo (11). Also, reports
comparing the efficiency of peptide generation from either cytosolic or
ER-targeted minigene products with that from longer polypeptides show a
remarkably higher efficiency of peptide production with the former
constructs (12, 13), suggesting a rate-limiting step
either in the processing or in the transport of antigenic peptides.
Other studies report that substrate ubiquitination or the proteasome
activity are limiting for Ag presentation or MHC class I assembly,
respectively, but only after IFN-
stimulation (14, 15),
a treatment that alters the basal balance of the cellular components of
the pathway by up-regulating several of them (16, 17).
Results with murine CMV show that IFN- governs the yield of
antigenic peptides also in vivo, demonstrating that factors other than
Ag synthesis limit the efficacy of Ag processing (18). The
last six reports (11, 12, 13, 14, 15, 18) thus suggest that the
endogenous pathway of processing and presentation has a rate-limiting
step, in contrast with the first two reports mentioned above.
Our report contributes to the present knowledge by quantitatively studying the endogenous pathway of processing and presentation over a wide range of different amounts of starting Ag expressed in infected cells, thus aiming to reach saturation. Recombinant vaccinia viruses (rVV) with point mutations in the promoter were used to intracellularly express increasing amounts of the model Ag ß-galactosidase (ß-gal) (19) that eventually can enter the endogenous processing and presentation pathway. Because ß-gal had the same sequence in all constructs tested (20), the possible effect of flanking regions on the efficiency of processing of the antigenic epitope (8, 9) was avoided. ß-gal is proteolytically cleaved by proteasomes, its TPHPARIGL (9ß-gal) epitope binds to the Ld allele (14, 21), and it is able to induce specific CTL when BALB/c mice are immunized with an appropriate vector (22). The MHC class I molecule Ld is characterized by its weak association with ß2m, its slow intracellular transport, and its low cell surface expression (23). ß2m plays a significant role in control of Ld expression (24, 25), but it is the processed peptide that induces the proper folding and increases dramatically Ld surface expression (26). A weak interaction with peptides has been suggested from the crystal structure of Ld (27). Indeed, the paucity of antigenic peptides able to bind to Ld is suggested as the main reason for retention of Ld molecules in the ER (26), in that there is a reservoir of non-peptide-bound Ld molecules that can be specifically detected by mAbs and that are retained in substantial amounts in the ER (28).
In this report, we have quantitatively compared the number of molecules of synthesized protein expressed by a series of 13 rVV with promoters of different strengths, with the number of molecules of processed antigenic peptide bound to MHC class I molecules at any place in the cell, and with surface recognition by CTL of cells infected with each rVV. Our results show that with increasing amounts of starting protein, there was a rise in presentation to CTL, reaching a point above which saturation was revealed. The same saturation point was found in the number of fully processed antigenic peptides bound to MHC class I molecules. The limiting step was located in the binding of the antigenic peptide to MHC or before this event, either in the processing or transport of peptides or in the availability or avidity of MHC class I molecules. The estimated maximum efficiency of the endogenous processing and presentation pathway was 1/3900; i.e., one antigenic peptide was produced of 3900 protein molecules, giving rise to a maximum of some 40 processed peptide molecules per infected cell.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
P815 cells (H-2d) and their derivative P13.1, which are transfected with the lacZ gene encoding ß-gal (29), as well as the T2/Ld cells, which are transfected with the Ld gene (30), were maintained in Iscove’s modified Dulbecco’s medium with 10% FCS.
rVV from the vMJ series were provided by Dr. B. Moss (National Institutes of Allergy and Infectious Diseases, Bethesda, MD). They express varying amounts of ß-gal due to the different point mutations introduced in the early region of the VV early-late promoter 7.5 kDa (20). vSC8 is a rVV with the ß-gal gene under the control of the late promoter P11 (31), which was used as a positive control in some experiments. As a negative control, wild-type WR virus was used. rVV stocks were grown in CV1 cells in DMEM supplemented with 10% FCS and consisted of clarified sonicated extracts of pellets of infected cells. When used, purification involved centrifugation through a 36% sucrose cushion.
In all experiments described below, infection of P815 cells was performed as described (9), using a multiplicity of infection of 5–10 PFU/cell. Infection of T2/Ld required a multiplicity of 20, as described (3, 32). After 1 h of adsorption, the virus inoculum was thoroughly washed; this was taken as time 0 h postinfection. In all types of experiments, infection was allowed to proceed for 7 h.
Protein analysis
For metabolic labeling, infected P815 cells were starved of methionine and cysteine during the viral adsorption period and then labeled with 100 µCi/ml [35S]methionine + cysteine (Amersham, Arlington Heights, IL) in medium free of both amino acids for the 7-h infection period. Proteins were then resolved by 7.5% SDS-PAGE. Similar gels with unlabeled samples were blotted to Immobilon (Millipore, Bedford, MA) and incubated with rabbit Ab anti-ß-gal (Cappel, Westchester, PA) at a 1/5,000 dilution and goat anti-rabbit Ab labeled with horseradish peroxidase (Southern Biotechnology Associates, Birmingham, AL), at 1/25,000. Bands were detected with ECL (Amersham).
Quantitation of ß-gal from rVV-infected P815 cells
For each rVV, lysates of pooled infected P815 cells and culture
medium were prepared. Quantitation of ß-gal in serial dilutions of
these lysates was done by luminometry using the Luminescent
ß-Galactosidase Genetic Reporter System II (Clontech Laboratories,
Palo Alto, CA) and measuring for 15 s in an Optocomp I luminometer
(GEM Biomedical). Only values in linear regions of the dose-response
curve were taken for extrapolation on the titration curve, which was
prepared with serial dilutions of ß-gal (Sigma, St. Louis, MO). Mean
values and SEs of ß-gal expressed in units were transformed to
molecules/cell by considering the m.w. and specific activity of
ß-gal. Background resulting from enzyme present in the input virus
was measured on samples harvested immediately after the 1-h adsorption
period. It had an average value of 14% of actual expression by each
individual rVV after 7 h of infection. In sharp contrast, in some
viruses (see Results) this background was 
65%, so that
we could not conclude that there was significant evidence of expression
by these rVV. This background, as well as the background detected in
WR-infected cells (see legends to figures), were subtracted from each
sample.
CTL cultures and cytotoxicity assays
Female 7-wk-old BALB/c (H-2d) mice were
bred in our animal care facility and were immunized by i.v. injection
with 5 x 107 PFU of vMJ360 in 0.1–0.2 ml
PBS. Polyclonal ß-gal-specific CTL lines were generated as follows.
At least 3 wk after immunization, 107
splenocytes/ml from immunized mice were restimulated in vitro with
105/ml mitomycin C-treated P13.1 cells or with
10-8 M 9ß-gal peptide and cultured in 
-MEM
supplemented with 1% 2-ME and 10% FCS. IL-2, generously provided by
Hoffmann-LaRoche, was added after 5 days at a final concentration of 25
U/ml. Specificity of CTL cultures was tested starting 2 days later.
Long term CTL cultures were maintained in the presence of 100 U/ml IL-2
by weekly restimulation with mitomycin C-treated P13.1 or with
10-8 M 9ß-gal peptide and mitomycin C-treated
splenocytes prepulsed for 20 min with 10-5 M
9ß-gal peptide. Following this protocol, no vaccinia-specific CTL
were ever selected. Occasionally, the ß-gal-specific CTL line 0805B
(22) was used.
For cytolytic assays, after 4 h of infection, cells were pulsed with Na51CrO4 for 90 min at 37°C in the presence or in the absence of exogenous peptide, washed, and combined with CTL. Total infection time until the addition of CTL was 7 h. 51Cr release cytolytic assays were performed for 3–4 h as described (2, 33). When used, cycloheximide (Sigma) at 50 µg/ml, brefeldin A (Sigma) at 1 µg/ml, or lactacystin (E. J. Corey, Harvard University, Cambridge, MA) at 200 µM was added after the adsorption period, kept until targets were combined with CTL, and replaced by brefeldin A 1 µg/ml during the CTL assay itself.
Quantitation of naturally processed antigenic peptides from rVV-infected P815 cells
P815 cells (5 x 108) were infected in parallel with different rVV, and 7 h later cell extracts were prepared in parallel as described (3, 8). Reversed phase HPLC was performed using a Smart equipment (Pharmacia, Piscataway, NJ) and a Sephasil C18-5 µm SC2.1/10 column and eluting with a rather flat gradient of 12.95 to 28.7% CH3CN in 0.1% trifluoroacetic acid. As internal standard in all HPLC runs, a ß-gal-unrelated peptide was included. Fractions of 30 µl were collected and fully used to prepare triplicate serial dilutions that were added to 96-well plates at 4°C in medium without FCS. Uninfected P815 cells that had been incubated overnight at 26°C to maximize expression of peptide-receptive Ld molecules (34) were labeled with 51Cr at 26°C, washed, added to the fraction-containing wells, and incubated for 15 min at room temperature in the presence of 2.5% FCS before adding effector CTL. At this point, the resulting final dilution of the fractions was at least 20-fold to avoid interference of solvents with the CTL assay. The concentration of antigenic peptide recovered in the titrated HPLC fractions was calculated by extrapolation from parallel titration curves of purified synthetic peptide 9ß-gal, synthesized, and sequenced in Applied Biosystems (Foster City, CA) equipment.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
rVV have been extensively used for both target formation and induction of CTL and have been shown to present a large variety of Ags both in vitro and in vivo (19). The vMJ series of rVV expresses ß-gal under the control of 7.5K-related promoters of different strength, all of which drive the expression of the same native ß-gal gene (20). A set of 13 of these viruses that differ only in point mutations in the early phase region of this promoter was used in our study. Kinetics of expression from these promoters is the same, and they only differ in the efficiency of transcription, which results in varying amounts of expressed ß-gal. Other viruses that differ in the late phase part of this promoter were not used because of possible interference by vaccinia late phase gene products with Ag presentation (35, 36). The widely used P815 cells were chosen to express our model Ag ß-gal, frequently used in infectious and tumor systems (36), and to quantitatively study the endogenous Ag processing and presentation pathway.
Vaccinia virus replication in P815 cells proceeded at a low level,
because expression of neither ß-gal nor vaccinia virus proteins was
detected after metabolic labeling of infected P815, which essentially
synthesized the same proteins as uninfected cells (Fig. 1
A). The more sensitive
Western blot analysis allowed detection of ß-gal expression (Fig. 1B). This was all the result of new synthesis, because no
ß-gal was detected in cells harvested immediately after the 1-h virus
adsorption period (lane 0 h pi). Thus, rVV
were able to direct the synthesis of new ß-gal molecules in infected
P815 cells, without having a measurable effect on host macromolecular
synthesis and without representing a massive load to the
cell-biosynthetic machinery.
|
Quantitation of the antigenic protein ß-gal in infected cells
was performed by enzymatic luminometry. Enzymatic colorimetry,
previously used to detect expression in infected CV1 cells
(20), was not sensitive enough for P815. The results are
shown in Fig. 2. Evidence of significant
expression above virus input background was detected for P815 infected
with vMJ337, with the value of 93,000 ± 19,000 ß-gal
molecules/cell, and for all viruses expressing higher amounts. Four
other viruses, including vMJ102 shown in Fig. 1
A, are not
included in Fig. 2 for clarity, because they expressed no detectable
ß-gal above the background defined by WR. The level of ß-gal
expression ranged between the value of vMJ337 and the highest in the
vMJ series, vMJ356, with 1,400,000 molecules/cell; i.e., a spectrum
encompassing at least a 15-fold range of Ag expression. This wide and
continuous range of expression of the starting protein allowed us later
to determine the quantitative influence of the Ag synthesized in the
infected cell on the outcome of the processing and presentation
pathway.
|
The endogenous nature of the ß-gal processing and presentation
pathway to CTL was next characterized. Although little evidence of
protein expression or enzymatic activity was detected in cells in which
viral replication was not allowed to proceed (see above), this point
was further checked at the level of target cell formation. Indeed, the
former assays would not have detected minor amounts in the input virus
inoculum of already processed ß-gal or of ß-gal that could be
processed exogenously. These minor amounts might nevertheless result in
exogenous loading of surface MHC class I molecules. To this end, CTL
recognition of cells harvested immediately after the 1-h virus
adsorption period was checked and found to be negative (data not
shown). Also, no differences were found when purified virus inoculum
was used, when the virus inoculum was extensively washed after
adsorption, or when this procedure was performed at 4°C to prevent
virus internalization (data not shown). Altogether, these data
indicated that the virus inoculum was not a source of peptide
recognized by CTL. In addition, treatment with the protein synthesis
inhibitor cycloheximide after adsorption significantly decreased target
formation (Fig. 3A). Residual
recognition might be caused by synthesis initiated from the first virus
particles entering the cell during the 1-h adsorption period.
Collectively, these results indicated that de novo-synthesized ß-gal
was the source for processing and presentation to CTL.
|
B), it was concluded that presentation of ß-gal
followed the endogenous route. Further, the involvement of two
components of the classical endogenous processing pathway was
confirmed. First, proteasomes were involved in ß-gal proteolytic
cleavage, because treatment of infected cells with the
proteasome-specific inhibitor lactacystin (38) completely
blocked recognition by CTL (Fig. 3C). Second, no
presentation was detected in T2/Ld mutant cells
lacking TAP but transfected with the gene expressing the presenting
molecule Ld (Fig. 3D), as compared
with infected P815 cells. As a positive control, these
TAP- T2/Ld cells were able
to present to specific MHC class I-restricted CTL the rVV-expressed
CMV-derived chimeric protein sC-A9A, which is endogenously synthesized
but fully processed in the secretory pathway by an alternative route
(3) (data not shown). In conclusion, recognition of
ß-gal by CTL in rVV-infected P815 cells required endogenous synthesis
of the Ag and traffic through the Golgi apparatus to the plasma
membrane of the presenting complex and involved the two main components
of the endogenous pathway, proteasomes and TAP transporters. CTL specifically recognize and discriminate between different Ag quantities
Recognition by CTL of P815 target cells infected with each of the
13 rVV from the vMJ series was assayed next. Average results from three
experiments are shown in Fig. 4. The
additional four rVV that gave ß-gal expression levels
indistinguishable from that of WR were also negative in the CTL assay
and are not included in the figure for clarity. The 100% value
corresponded to lysis of wild-type-infected target cells incubated with
excess exogenous peptide at the end of the infection time, just before
combination with CTL. The fact that no vMJ-infected sample reached a
normalized value of 100% demonstrated that the capacity of CTL to
recognize antigenic peptide at the infected cell surface was not
saturated, because they were able to recognize higher quantities of
antigenic peptide when it was added in excess. The first samples that
differed statistically from background were P815 infected with vMJ195,
vMJ179, or vMJ337, with values of 7% normalized lysis. In the
t test, p was <0.05 in each of the three
separate experiments, when compared with wild-type WR. CTL recognition
was thus significantly more sensitive than enzymatic detection of the
protein. Half-maximal lysis among virus-infected cells was marked by
the vMJ172 and vMJ177 viruses. Fig. 4 also shows that the CTL exhibited
different degrees of recognition of the cells infected with the
different rVV, which implies that different quantities of antigenic
peptides were exposed to CTL recognition and that the CTL were able to
discriminate between them within a certain range.
|
), it was found that the
increasing Ag recognition obtained with growing ß-gal protein
expression reached a point where saturation in recognition by CTL
occurred (see also Fig. 6). In other words, within a range of
differences in Ag expression, up to vMJ243, CTL were able to respond
with increasing lysis. From here on, as few as 140,000 molecules of
synthesized protein in vMJ243-infected cells sufficed to give the
maximal CTL response, and up to 10-fold higher synthesis rates in
vMJ356 did not contribute to a better recognition of infected cells at
the cell surface. These data fitted very well with the standard
saturation curve (Fig. 6, curved line). Because the CTL were not
saturated, as discussed above, it was concluded that there was a
saturation point in the endogenous pathway of ß-gal processing and
presentation and that this was located in the infected cell.
|
|
A likely hypothesis is that the saturation detected occurred intracellularly. Besides testing it, we also wanted to study whether or not the rate-determining step was placed before or after the binding of peptide to MHC class I in the ER. Isolation of the previously described (21) ß-gal antigenic peptide 9ß-gal from infected cells could help solve these questions. Such isolation is dependent on 9ß-gal binding to the MHC class I molecule Ld (39). Particularly, we wanted to know whether the saturation point defined by vMJ243 at the target cell surface also applied when the naturally processed peptides were quantified or whether a virus with higher expression would now mark the saturation point. In the latter case, intracellular saturation would be located after complex formation between MHC and the fully processed peptide. Therefore, the biochemical analysis of endogenous 9ß-gal was performed on vMJ243 and viruses of higher expression.
First of all, specificity and yield of the peptide extraction and HPLC
procedures were tested. As shown in Fig. 5A, the synthetic 9ß-gal
peptide was resolved in our HPLC gradient system and specifically
recognized by CTL, whereas P815 cells did not generate any antigenic
peptide in detectable amounts (Fig. 5B). In contrast, an
antigenic peak was detected from P13.1 cells, which constitutively
express, accumulate, and present ß-gal. This peak coeluted with the
synthetic 9ß-gal, suggesting that the endogenously generated peptide
was indeed 9ß-gal (Fig. 5C). A sample of P815 infected
with WR (Fig. 5D), to which a known amount of synthetic
peptide was added just before starting the biochemical isolation of
cellular peptides (Fig. 5E), was used to calculate the
efficiency of the experimental procedure. Recovery was 1.8% of the
starting antigenic activity. Care was taken to use an initial ratio of
synthetic peptide molecules to WR-infected cell of 120, close to that
actually found in vMJ-infected cells (see below), so as to mimic
closely the conditions met by the natural peptides. It was thus also
confirmed that the procedures did not alter the HPLC behavior of
9ß-gal.
|
, F–H.
The antigenic peak from all infected samples coeluted with synthetic
9ß-gal, within the range of experimental error. Pure synthetic
peptides repeatedly injected in the HPLC gave a similar range of
variation. Quantitation of 9ß-gal in the samples was next performed
by serial dilution followed by extrapolation on a parallel titration
curve of synthetic peptide. Peptide recovery as measured earlier was
taken into account. Peptide yield was similar in all analyzed samples
with a maximum value of 40 (average, 36 ± 4) molecules of
antigenic peptide per infected cell (Table I). The vMJ360 sample
depicted in Fig. 5G showed the lowest repeatability. Cells
infected with viruses such as vMJ172 and vMJ177 would be expected to
contain less total peptide than vMJ243, because if they were to have
the same amount, one would reasonably expect them to be recognized by
CTL to the same extent as those viruses in the saturation region.
Attempts were made to quantify peptides from samples not located in the
saturation region of CTL recognition, but they were unsuccessful, in
spite of using all the material from the HPLC runs for the assay.
Finally, data on whole cell natural peptide quantitation are plotted vs
the number of synthesized ß-gal molecules in Fig. 6. These results indicated that the
saturation point found at the vMJ243 virus by probing the cell surface
with CTL was also found at the level of intracellularly processed
peptides. Also, the results confirmed that saturation occurred
intracellularly, because several viruses that expressed increasing
amounts of protein nevertheless produced the same amount of processed
peptide. Further, these data suggested that the rate-limiting step was
located in the binding of peptide to the available MHC class I
molecules or before this event, as will be discussed below.
Efficiency of endogenous Ag processing
Once the values of extracted antigenic peptides were obtained, the
rate of Ag processing was determined by comparing the value of steady
state ß-gal molecules with the number of antigenic peptides obtained
in each infected sample. Because peptide content from viruses in the
nonsaturated region of the curve was below our detection limit, we
could not accurately determine the maximum efficiency of processing.
Our closest estimation is to assume that vMJ243 is just the turning
point from the nonsaturated to the saturated segments of the curve.
Based on this assumption, the total efficiency of the endogenous
processing pathway had an estimated maximum value of 0.03% (Table I).
In other words, the cell needed 3900 molecules of steady state starting
Ag to generate 1 antigenic peptide bound to MHC class I molecules. The
efficiency in rVV expressing more ß-gal decreased accordingly,
because the processing pathway was saturated.
In the nonsaturated segment of the saturation curve shown in Fig. 6, one might expect a linear relationship between the amount of 9ß-gal
peptide in infected cells, and the percentage specific lysis by CTL of
the same cells. Based on this assumption, the value of recognition by
CTL of P815 cells infected with vMJ172 or vMJ177, which were the
samples giving half-maximal lysis, was extrapolated in the antigenic
peptide scale. This extrapolation showed that about 18 molecules of
total antigenic peptide would be needed for recognition by CTL of
one-half of the cells. The same approach, applied to vMJ195-, vMJ179-,
or vMJ337-infected P815, which were the first positive samples,
suggested that 4 molecules of total antigenic peptide in the
infected cell would be needed to trigger recognition by CTL. This is
the first time that the minimum quantity of total antigenic peptide in
a VV-infected cell needed to detect a CTL response is roughly
estimated.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
40 processed 9ß-gal peptide molecules were bound to MHC
per infected cell, giving an estimated maximal rate of processing of
1/3900 in this system. Few studies have fully addressed quantitative aspects of the endogenous pathway of Ag presentation as a whole in living cells. Although in an earlier study by us with two murine CMV antigenic constructs (8) and in another report with three different expression levels of a Listeria Ag (10), saturation of their presentation to CTL was not reached, several other reports are compatible with a saturation of the pathway (11, 12, 13, 14, 15, 18). At present it is not clear whether the different infectious agents used in all of these reports, including ours, contribute themselves to potentiate or diminish the saturation of the pathway. Saturation in our system was detected by CTL at the infected cell surface and independently revealed to occur intracellularly after biochemical purification of processed peptides from whole infected cells. The fact that we now did find saturation probably relates to the wider range of Ag expression levels studied by us, but might also be explained by the different systems used.
Location of the intracellular rate-limiting step
When varying amounts of a protein are synthesized in the cytosol, a given fraction will be processed to yield peptides bound to MHC class I molecules, following a dose-response curve that eventually may reach saturation levels, as would any biological process. CTL that recognize this processed peptide at the cell surface will follow this response; i.e., the more peptide is processed, the more recognition and lysis will be detected. Theoretically, if transit of peptide/MHC complexes from the ER to the plasma membrane were a rate-limiting step, then CTL would reach their maximal recognition at protein expression levels lower than those that lead to saturation of the whole cell content of processed peptides, including both intracellular and cell surface peptides. In other words, in this scenario, even if more processed peptides were available intracellularly, they would not be presented to the CTL. On the contrary, if traffic through the secretory pathway is not a rate-limiting step, CTL recognition parallels the amount of processed peptide, and if saturation is ever achieved, both reach saturation levels at the same level of original antigenic protein, as we have found in our model. Indeed, the levels of peptide molecules per infected cell for all viruses in the CTL saturation area were the same. Therefore, we conclude that the rate-limiting step is located either at binding of MHC class I and peptide, or before this event, whereas complex migration to the plasma membrane is not saturated.
Our data indicated that there was no limitation in peptide/MHC class I complex traffic through the secretory pathway. A recent report indicated that such a control step exists for Kb and Db molecules (40). It may well exist also for Ld as suggested from earlier experiments (41) but, at least for ß-gal in VV-infected cells, this second control point would not be operative because saturation occurs at a lower Ag level at an earlier step. Because the level of processed peptides was saturated, the rate-limiting step was unequivocally located within the infected cell. Any cellular component involved in the pathway until or in the ER may contribute to the limitation. These include ubiquitination (14), proteasomes (15) or other enzymes involved in proteolytic cleavage (2, 3), the TAP complex, including calnexin, calreticulin, and tapasin (4), chaperonins gp96 and protein disulfide isomerase (42), ß2m, and MHC class I molecules themselves. Little can be said about other processing enzymes and tapasin, because they have been described recently and little is known about their availability in cells. Protein disulfide isomerase, gp96, and calreticulin have been shown to bind peptides, although none of them yet with high selectivity (42, 43). On the other hand, gp96 and calnexin have been already shown not to determine the rate of Ag presentation (44, 45), so that the implication of all these molecules in limiting Ag presentation can be taken as unlikely. On the contrary, it cannot be excluded that ß2m may be the limiting factor (46).
Proteasomes and TAP have been reported as being sequence selective in
that certain sequences in certain contexts make processing by
proteasomes in vitro and in vivo (47, 48) and transport by
TAP (49) less efficient. That makes them good candidates
as cellular components that may prevent unlimited presentation of the
ß-gal epitope. Because proteasome activity is very tightly regulated
(1), it is not difficult to believe that it might
represent a bottleneck in the pathway, particularly when its primary
function is probably not Ag processing. Some articles support this
idea, as they show increased yields of MHC-bound peptides when
proteolytic cleavage is by-passed with minigene constructs in
rVV-infected cells (12, 13) or that ubiquitination
and the proteasome activity limit presentation (50) and
the assembly of MHC class I molecules, respectively, albeit only under
IFN- stimulation (14, 15).
On the other hand, TAP appears at first sight not to be quantitatively limiting. Thus, the amount of peptides that TAP can transport to the ER is well above the numbers that can be bound by MHC class I molecules (51), and the recovery of peptides bound to MHC class I from constructs with and without signal sequence was the same (12). However, TAP shows a particular difficulty in transporting peptides with proline in the second position (49). Such peptides, which include the 9ß-gal epitope, TPHPARIGL, happen to be those with highest affinity for Ld, the anchor residues of which are proline at position 2 and a hydrophobic amino acid at position 9 (52). Thus, TAP might indeed be functionally rate limiting for transporting peptides for Ld. This is in line with speculations as to why unfolded Ld molecules accumulate in the ER (28), an accumulation that can be overcome with an adequate excess supply of peptides (26).
Whether MHC class I molecules can represent a rate-limiting step in Ag processing and presentation is under some controversy. Several findings favor this possibility. In some tumor systems, increased elimination in vivo is correlated with increased MHC class I levels (53). Also, TAP can transport to the ER more peptides than MHC can bind (51). Additionally, MHC class I displays a broad difference in affinities to peptides (54). Finally, the Ld molecule exhibits some peculiarities. Its recently published crystallographic structure reveals an unusually weak interaction with ß2m and peptide (27), which suggests that its affinity to peptides could be relatively low, and even in a situation of high antigenic peptide concentration in the ER, the interaction would be poorly efficient, and therefore Ld could be functionally rate limiting. Because this is the first quantitative report on endogenous levels of an Ld-presented peptide, further studies with other Ld-restricted epitopes are needed to shed light on the issue of whether the limiting step and low endogenous numbers found for the ß-gal epitope are a property of Ld or of the 9-ßgal/Ld complex, such as a putative low binding affinity.
On the contrary, some pieces of evidence discard the MHC as the limiting factor. First, there is a reservoir of non-peptide-bound Ld molecules in the ER (28), and addition of antigenic peptide can induce proper folding and increase Ld surface expression (26). Also, a lack of limitation in TAP and MHC availability is suggested from the experiments where very high numbers of peptides are presented from minigene products (12).
Finally, in vivo experiments with murine CMV show that IFN- governs
the yield of CMV peptides (18), implying that
IFN--inducible cellular components are rate limiting in Ag
processing. These include the three that we have just discussed, namely
some proteasome subunits, TAP, and MHC class I molecules (16, 17). Our results are consistent with this idea because the
limiting step was located at the binding of antigenic peptide to MHC
class I or before this point, either in the proteolytic cleavage,
transport, or availability or avidity of MHC class I.
Efficiency of the endogenous Ag processing and presentation pathway
We estimated that the maximum efficiency of Ag processing
in our system was 1/3900, i.e., that one correctly processed peptide
molecule was produced of 3900 steady state ß-gal protein molecules.
This low number may represent a low energetic cost for the cell and
thus allow that not only nonproperly folded proteins, as has been
suggested (55), but also functional proteins enter the
endogenous processing pathway. In the fully different
Listeria system, efficiency was defined as the number of
correctly processed epitopes produced from the subset of protein
molecules that are degraded, and expectedly found to be higher, of 1/35
(10). In our system, with the widely used and potent P815
targets, it was surprising that as few as 40 processed 9ß-gal
peptides were produced. This number of peptide molecules bound to MHC
per infected cell lies at the lower end but is within the range that
has been found for other Ags in other systems, i.e., between 10 and
85,000 molecules/cell (10, 12, 13, 56)., and even with
this limited Ag expression and peptide production, it was remarkable
that saturation was reached. This low number found in cell culture
allows nevertheless efficient induction of CTL in vivo. In contrast, in
the Listeria system, constructs that produce at least 1000
molecules in vitro are needed in vivo (11). Finally, our
results suggest that triggering of our CTL may require at least 4
antigenic peptides in the whole cell, both inside and at the cell
surface, a result in agreement with the recently estimated minimum
triggering number of 1 antigenic molecule presented by
Ld at the cell surface (57).
We think that for different combinations of presenting cells (58), Ags, expression vectors, MHC class I alleles, etc., slightly different rate-limiting or saturation conditions may apply. However, our results probably truly reflect the situation in vaccinia virus-infected cells in vivo and may thus probably apply for vaccine development. Indeed, the results are in agreement with data obtained with two of these recombinant viruses in vivo, because vMJ360, a virus in the saturation region, was found to confer better protection than vMJ177, a virus giving half-maximal detection by CTL, when assayed in a tumor system using ß-gal as surrogate Ag (36).
Also, given the moderate to high sensitivity to Ag sequence differences
of many cellular components of the processing pathway, it is likely
that functionally saturating conditions may occur with certain
frequency. Thus, one implication of this findings is that competition
between Ags.5 is
expected whenever there is a shortage of any given component of the
pathway. Another important point is that our studies can propose ways
to overcome this deficit in presentation. It is very possible that
codelivering or coexpressing cytokines such as IFN- with different
types of vaccines may improve their efficacy, because in vivo IFN-
can enhance both presentation by infected cells and the function of
professional APC (18) by up-regulating the steps that we
have identified to be rate limiting. Our results also open the way to
characterize in more detail the precise step that is rate determining
for the whole pathway, and thus to the development of more specifically
targeted strategies for vaccine improvement.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Current address: Edward Jenner Institute for Vaccine Research, Compton, Berkshire, U.K.
3 Address correspondence and reprint requests to Dr. Margarita Del Val, Centro Nacional de Biología Fundamental, Instituto de Salud Carlos III, Ctra. Pozuelo, km 2, E-28220 Majadahonda (Madrid), Spain. E-mail address:
4 Abbreviations used in this paper: ß2m, ß2-microglobulin; ß-gal, Escherichia coli ß-galactosidase; 9ß-gal, 876–884 epitope (TPHPARIGL) from ß-gal; ER, endoplasmic reticulum; rVV, recombinant vaccinia virus.
5 D. López, Y. Samino, U. H. Koszinowski, and M. Del Val. Submitted for publication.
Received for publication December 7, 1998. Accepted for publication June 9, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
stimulation. J. Immunol. 159:5896.[Abstract]
is a prerequisite for optimal antigen processing of viral peptides in vivo. J. Immunol. 158:3303.[Abstract]
This article has been cited by other articles:
|
|
 |
|
  S. Bulik, B. Peters, and H.-G. Holzhutter Quantifying the Contribution of Defective Ribosomal Products to Antigen Production: A Model-Based Computational Analysis J. Immunol., December 15, 2005; 175(12): 7957 - 7964. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Palliser, E. Guillen, M. Ju, and H. N. Eisen Multiple Intracellular Routes in the Cross-Presentation of a Soluble Protein by Murine Dendritic Cells J. Immunol., February 15, 2005; 174(4): 1879 - 1887. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Samino, D. Lopez, S. Guil, P. de Leon, and M. Del Val An Endogenous HIV Envelope-derived Peptide without the Terminal NH3+ Group Anchor Is Physiologically Presented by Major Histocompatibility Complex Class I Molecules J. Biol. Chem., January 9, 2004; 279(2): 1151 - 1160. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Fruci, G. Lauvau, L. Saveanu, M. Amicosante, R. H. Butler, A. Polack, F. Ginhoux, F. Lemonnier, H. Firat, and P. M. van Endert Quantifying Recruitment of Cytosolic Peptides for HLA Class I Presentation: Impact of TAP Transport J. Immunol., March 15, 2003; 170(6): 2977 - 2984. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. N. Mueller, C. M. Jones, W. Chen, Y. Kawaoka, M. R. Castrucci, W. R. Heath, and F. R. Carbone The Early Expression of Glycoprotein B from Herpes Simplex Virus Can Be Detected by Antigen-Specific CD8+ T Cells J. Virol., February 15, 2003; 77(4): 2445 - 2451. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Lopez, B. C. Gil-Torregrosa, C. Bergmann, and M. Del Val Sequential Cleavage by Metallopeptidases and Proteasomes Is Involved in Processing HIV-1 ENV Epitope for Endogenous MHC Class I Antigen Presentation J. Immunol., May 15, 2000; 164(10): 5070 - 5077. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Ramírez, M. M. Gherardi, and M. Esteban Biology of Attenuated Modified Vaccinia Virus Ankara Recombinant Vector in Mice: Virus Fate and Activation of B- and T-Cell Immune Responses in Comparison with the Western Reserve Strain and Advantages as a Vaccine J. Virol., January 1, 2000; 74(2): 923 - 933. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
),
and the number of whole cell naturally processed peptide molecules from
each rVV-infected sample (Table I
), are
plotted against the measured number of ß-gal molecules per infected
cell (Fig. 2
), as well as two 2-fold serial dilutions
thereof); B, uninfected P815 cells (1.4 x
107 initial cell equivalents per assay well);
C, ß-gal-transfected P13.1 cells (1.2 x
107); D, WR-infected P815 cells (4.2 x
107); E, WR-infected P815 cells that had
received synthetic 9ß-gal just before acid extraction (4.2 x
107 initial cell equivalents and 5 x 109
initial peptide molecules per assay well, corresponding to a ratio of
120 peptide molecules per cell (