Tài liệu: Ai cần tớ giúp....

Volume 185 Issue 1 Page 86 - July 2002

Perspective on antigen processing and presentation
Emil R. Unanue
Summary:The phenomenon of antigen processing and presentation and the concept that T cells recognize peptides resulting from the partial catabolism of proteins, are relatively new. These concepts were first recognized and developed at a time when lymphocyte immunity the adaptive system - and cellular immunity, with its major component of activated macrophages, were not perceived as part of one integrated system. To me, it was the fundamental findings on the role of major histocompatibility (MHC) molecules that set the framework for understanding how phagocytes and the antigen presenting cell (APC) system interact with the adaptive cellular system, in a truly symbiotic relationship (1). In this chapter we make a historical review of the developments that, in my biased opinion, led to the understanding of antigen presentation as a central event. I emphasize my own work, placing it in my perspective of how I saw the field moving.


Perspectives on the early studies on macrophage-lymphocyte interactions

Enter phagocytosis
Phagocytosis was the seminal contribution of Eli Metchnikoff who ascribed a major role for the phagocytic system in defense mechanisms (2). It was at about the same time that antitoxins were identified in blood, leading Paul Ehrlich to postulate the side chain theory, a major conceptual advance on how lymphocyte immunity could develop (3). From that point on, that is the early 1900s, these two fields of immunity were studied separately, as if they existed independently from each other. Those studying phagocytes followed one direction, in great part led by the findings in tuberculosis that resulted in the identification of the activated macrophages. Indeed, it was Koch who described first the infectious foci, as well as the tuberculin reaction, with the typical granulomas made of large macrophages (4). Eventually, the activated macrophage was defined by the pioneering investigations of Lurie in rabbits infected with Mycobacterium tuberculosis (5), and subsequently by the studies of Mackaness and associates in both tuberculosis and Listeria monocytogenes immunity (6). Resolution of infection required an activated macrophage that was highly bactericidal. (Other studies pointed to a second microbicidal mechanism, this time involving neutrophils taking up a different type of bacteria, the extracellular pyogenic bacteria opsonized by antibodies.) In the case of the cellular immunity to intracellular pathogens, it took many years to establish the relationship between lymphocyte responses and activated macrophages. One seminal paper was that of Chase establishing in the guinea pig the transfer by lymphocytes of tuberculin immunity (7). However, all these studies with phagocytes were considered part of the effector reactions of immunity. No mention was made that actually phagocytosis and intracellular processing of proteins could be linked to the initiation of the response.

Lymphocyte responses
Those studying lymphocyte immunity first gave a major emphasis to the examination of the antibody molecule, its secretion in the blood as a result of antigenic stimulation, its binding properties, their structure, and the analysis of its component chains and their interactions with antigens and effector functions. By the middle 1960s, when I entered the field, studies were starting to focus on lymphocyte recognition. These studies were sparked by Burnett's highly visible studies on clonal selection (8), and later by the seminal analysis of Miller and associates on the role of the thymus (9), and by the identification of T and B cells, which included contributions from several laboratories (10-12). Clonal selection and the concept that an antigen molecule would bind and select those cells with high affinity receptors dominated the whole thinking (reviewed in 13). The findings on affinity maturation of the antibody response supported the selection theories (14). Later, antigen molecules could be identified binding directly to lymphocytes (15). Hence a major influential concept arose that recognition preceded denaturation and catabolism of the protein antigen (16). In retrospect, all these findings were correct, but applied only to B cells. The initial thinking was that a similar recognition system was operating for T cells, but there were major difficulties in identifying the T-cell receptor (TCR) for antigen, which did not take place until the early 1980s (17-19).

Enter MHC molecules
The seminal findings that MHC molecules were involved in immune responses to proteins completely changed the field of immune recognition. These studies set the base that allowed eventually T cells to be linked to the recognition of a processed part of the protein antigen presented by an MHC molecule. But before the initial findings of Hugh McDevitt in the mouse (20), and Benacerraf's group (21) first in the guinea pig, there were already indications that the immune reactions of delayed sensitivity might follow a different pattern of recognition.
Delayed hypersensitivity reactions, which were subsequently linked to T cells, were found to be insensitive to denaturation of the protein antigen and were 'carrier dependent' (22). That is to say, that while the recognition by antibodies against haptens in hapten-protein conjugates was not dependent on the carrier, the recognition of the conjugate in skin sensitivity reactions, in contrast, was dependent on the carrier portion, even when the protein was denatured. The relevance of the early studies to immune recognition was not well appreciated: delayed reactions were difficult to measure and quantitate, their specificity was not clear, and the role of either antibodies as cell-bound antibodies or T cells was debated. The important studies from Mitchison, pointing to B- and T-cell collaboration in the hapten-carrier protein response, indicated that each component the hapten and the carrier protein could be recognized by distinct cells (23). Later studies testing T cells and antibodies made it clear that T cells recognized denatured proteins as well as conformationally intact proteins, while antibodies discriminated between both (for examples, see 24-26 and citations therein). However, the explanation for such differences was not forthcoming until it was proven see below that an intact protein eventually was always recognized as a denatured protein or fragment, as a result of intracellular processing by an APC.
McDevitt's studies were highly seminal in that his laboratory established, using the powerful tool of mouse genetics, a control of the antibody response by the MHC, and for the first time indicated a physiological role of this gene locus distinct from the classical role in allogeneic interactions (27). Their studies also led to the identification of a new genetic region, encoding the class II MHC genes. These were powerful results. Concerning the Benacerraf group, aside from the demonstration of the control of the polylysine gene response in guinea pigs (21), they pointed very early to the possible cell that was the locus of the genetic control, and that was to the T cell recognizing the carrier, perhaps associated with an APC (28). (Indeed, they had used hapten conjugated to synthetic peptides made of polylysine.) Since the carrier recognition, as mentioned above, was now tied up to T cells, it was then apparent that T cells recognized linear determinants under the control of MHC molecules.
The restriction of the cytolytic T-cell response by MHC molecules first shown by Zinkernagel & Doherty (29), and that of B- and T-cell collaborations (30, 31), and very importantly the T-cell proliferative response first noted by Rosenthal & Shevach (32) added a new dimension; the MHC molecules controlled all cellular interactions involving T cells. (As Jan Klein, who has thought much on these issues, said, 'the body adheres to the etiquette of a proper English home; a foreigner can be introduced only by a person known to the family' (33)). Rosenthal went on to postulate that the MHC was involved in the process that led to recognition of different segments of the antigen molecule, in their case the protein insulin, in what they termed 'determinant selection' (34).

==========
Fig. 1. A side view of the peptide 52-62 of HEL bound to I-Ak molecules. This study was done by Daved Fremont (73). The solvent accessible surface is shown as blue dots. Indicated are the P1, P4, P6, P7 and P9 pockets. The HEL peptide is indicated in a ball and stick model. Shown is the alpha 52 arginine (in green) at the base of P1, which forms a salt bridge with the Asp 52 of the peptide.
==========
MHC molecules bind peptides
The role of MHC molecules became central: did they serve as peptide recognition molecules or were they molecules modified in some form by the protein antigen so to become a neo-antigen (so called altered-self); or did they play a role in mediating cell to cell contact? Having failed to detect MHC interactions using radioactive proteins (52), we opted to identify from tryptic digests of HEL, the peptide that stimulated a highly reactive T-cell hybridoma. The peptide 46-61 contained part of the epitope (48). (We subsequently identified the natural epitope from the I-Ak molecule of APC fed with HEL: the processed peptides encompassed residues 48-63 (53) ( Table 2 ).
We conceived that 46-61 peptide of HEL, having some hydrophobic residues, could either interact with membranes or with I-Ak molecules. Following Occam's razor, since testing on cells was going to be messy, why not do it directly on purified MHC molecules? We used equilibrium dialysis with labeled peptides and MHC molecules purified from cell membranes. The experiments, made by Bruce Babbitt who had just joined us, gave beautiful results that were too good to be true (54, 55). Here was a saturable, homogeneous binding interaction where one peptide molecule bound per molecule of I-Ak and with an affinity in the àM range. A peptide with scrambled residues from 46 to 61 did not bind and, importantly, a self peptide from lysozyme also bound (55). Moreover, the complex could stimulate the T-cell hybridomas directly.
The finding that an autologous lysozyme peptide also bound to I-Ak with the same affinity as the foreign lysozyme peptide was telling, along two lines. First, the MHC was not making the self/nonself discrimination and, therefore, autoimmune reactions could target self peptides, as indeed was predictable. Second, autologous peptides could compete with the foreign peptide for processing and presentation.
Going back to the discussion on the catabolism of proteins to amino acids and how to integrate this finding with immunogenicity, it was apparent that through the interactions with MHC, the bound peptides were saved from degradation and that the complex constituted the substrate of the TCR. Later, Paul Allen and his group (56) found that the peptide-MHC complex was quite resistant to proteolytic enzymes. (More recently, we reevaluated the sensitivity of purified peptide-MHC complex to chemotryptic digestion. The sensitivity depended critically on the binding affinity of the peptide for the class II molecule. In fact, mutating a single anchor residue in the peptide and weakening the binding interaction resulted in a complex that became very sensitive to proteolysis and in which the class II molecule itself was actually the target of the enzyme (57). So, a strong interaction of a peptide with a class II MHC molecule forms a tight protein complex that is resistant to proteolysis, in contrast to the complex formed by weak peptides. The reader is also referred to the relevant studies from Germain's laboratory (58, 59)).
Grey's group confirmed our binding results and extended them noticeably, reporting binding kinetics and testing a variety of peptides (60, 61). Townsend's group then made the point that class I MHC recognition by CD8 T cells also involved peptide recognition (62). Thus, the raison d'etre of the elusive MHC molecules was to select peptides and create the antigenic epitope recognized by T cells.
Seminal, of course, was the X-ray crystallographic analysis that followed. Wiley's laboratory published the first results on a class I MHC molecule, with Jack Strominger providing the biochemical expertise, and with Pam Bjorkman as lead author (63, 64). Knowing that there was a binding groove, with its walls of alpha helices and its beta-pleated sheet platform, provided the framework for defining a T-cell immunogen and for studying the fine nature of different peptide interactions among various allelic forms of MHC class I or II molecules ( Fig. 1 ). The initial chemical and immunological analysis reported by us and also by Grey's group established that some amino acid side chains in the peptide were 'anchors' required for the binding to MHC, while side chains of other residues served as contact for the TCR (65, 66). These findings became very apparent as the structures of various peptide-MHC complexes were solved, all showing a stretched peptide in the binding groove. We refer to several studies in both class I and II MHC molecules (67-75) and to recent ones on the structure of the TCR contacting the peptide-MHC complex (76-79). Since the amino acids responsible for the allelic differences were located in the binding groove for peptide, it established that the evolutionary pressure that diversified the MHC was centered on the peptide binding properties of the molecule.
=============
Fig. 2. Processing pathways in the APC. The APC contains two distinctly functional processing vesicles. Class II molecules are transported into late, lysosomal-like vesicles with the invariant chain. At this site they meet with HEL, which is then unfolded and assembled under the catalytic influence of H-2 DM. Expression of HEL peptides thus involves nascent class II molecules and H-2 DM. The late compartment only generates type A complexes. In contrast, peptides can directly assemble in early endosomes, which can recycle MHC molecules. Some MHC molecules in these early endosomes contain weak peptides that can exchange with incoming HEL peptides ( 87). Presentation of peptides takes place in mature MHC molecules in an H-2 DM independent process. The endocytic vesicles can generate both type A and B complexes.
======================
Concay
ĐặỏằÊc Milou sỏằưa vào 06:27 ngày 18/06/2003

Volume 185 Issue 1 Page 86 - July 2002

Perspective on antigen processing and presentation
Emil R. Unanue
Summary:The phenomenon of antigen processing and presentation and the concept that T cells recognize peptides resulting from the partial catabolism of proteins, are relatively new. These concepts were first recognized and developed at a time when lymphocyte immunity the adaptive system - and cellular immunity, with its major component of activated macrophages, were not perceived as part of one integrated system. To me, it was the fundamental findings on the role of major histocompatibility (MHC) molecules that set the framework for understanding how phagocytes and the antigen presenting cell (APC) system interact with the adaptive cellular system, in a truly symbiotic relationship (1). In this chapter we make a historical review of the developments that, in my biased opinion, led to the understanding of antigen presentation as a central event. I emphasize my own work, placing it in my perspective of how I saw the field moving.


Perspectives on the early studies on macrophage-lymphocyte interactions

Enter phagocytosis
Phagocytosis was the seminal contribution of Eli Metchnikoff who ascribed a major role for the phagocytic system in defense mechanisms (2). It was at about the same time that antitoxins were identified in blood, leading Paul Ehrlich to postulate the side chain theory, a major conceptual advance on how lymphocyte immunity could develop (3). From that point on, that is the early 1900s, these two fields of immunity were studied separately, as if they existed independently from each other. Those studying phagocytes followed one direction, in great part led by the findings in tuberculosis that resulted in the identification of the activated macrophages. Indeed, it was Koch who described first the infectious foci, as well as the tuberculin reaction, with the typical granulomas made of large macrophages (4). Eventually, the activated macrophage was defined by the pioneering investigations of Lurie in rabbits infected with Mycobacterium tuberculosis (5), and subsequently by the studies of Mackaness and associates in both tuberculosis and Listeria monocytogenes immunity (6). Resolution of infection required an activated macrophage that was highly bactericidal. (Other studies pointed to a second microbicidal mechanism, this time involving neutrophils taking up a different type of bacteria, the extracellular pyogenic bacteria opsonized by antibodies.) In the case of the cellular immunity to intracellular pathogens, it took many years to establish the relationship between lymphocyte responses and activated macrophages. One seminal paper was that of Chase establishing in the guinea pig the transfer by lymphocytes of tuberculin immunity (7). However, all these studies with phagocytes were considered part of the effector reactions of immunity. No mention was made that actually phagocytosis and intracellular processing of proteins could be linked to the initiation of the response.

Lymphocyte responses
Those studying lymphocyte immunity first gave a major emphasis to the examination of the antibody molecule, its secretion in the blood as a result of antigenic stimulation, its binding properties, their structure, and the analysis of its component chains and their interactions with antigens and effector functions. By the middle 1960s, when I entered the field, studies were starting to focus on lymphocyte recognition. These studies were sparked by Burnett's highly visible studies on clonal selection (8), and later by the seminal analysis of Miller and associates on the role of the thymus (9), and by the identification of T and B cells, which included contributions from several laboratories (10-12). Clonal selection and the concept that an antigen molecule would bind and select those cells with high affinity receptors dominated the whole thinking (reviewed in 13). The findings on affinity maturation of the antibody response supported the selection theories (14). Later, antigen molecules could be identified binding directly to lymphocytes (15). Hence a major influential concept arose that recognition preceded denaturation and catabolism of the protein antigen (16). In retrospect, all these findings were correct, but applied only to B cells. The initial thinking was that a similar recognition system was operating for T cells, but there were major difficulties in identifying the T-cell receptor (TCR) for antigen, which did not take place until the early 1980s (17-19).

Enter MHC molecules
The seminal findings that MHC molecules were involved in immune responses to proteins completely changed the field of immune recognition. These studies set the base that allowed eventually T cells to be linked to the recognition of a processed part of the protein antigen presented by an MHC molecule. But before the initial findings of Hugh McDevitt in the mouse (20), and Benacerraf's group (21) first in the guinea pig, there were already indications that the immune reactions of delayed sensitivity might follow a different pattern of recognition.
Delayed hypersensitivity reactions, which were subsequently linked to T cells, were found to be insensitive to denaturation of the protein antigen and were 'carrier dependent' (22). That is to say, that while the recognition by antibodies against haptens in hapten-protein conjugates was not dependent on the carrier, the recognition of the conjugate in skin sensitivity reactions, in contrast, was dependent on the carrier portion, even when the protein was denatured. The relevance of the early studies to immune recognition was not well appreciated: delayed reactions were difficult to measure and quantitate, their specificity was not clear, and the role of either antibodies as cell-bound antibodies or T cells was debated. The important studies from Mitchison, pointing to B- and T-cell collaboration in the hapten-carrier protein response, indicated that each component the hapten and the carrier protein could be recognized by distinct cells (23). Later studies testing T cells and antibodies made it clear that T cells recognized denatured proteins as well as conformationally intact proteins, while antibodies discriminated between both (for examples, see 24-26 and citations therein). However, the explanation for such differences was not forthcoming until it was proven see below that an intact protein eventually was always recognized as a denatured protein or fragment, as a result of intracellular processing by an APC.
McDevitt's studies were highly seminal in that his laboratory established, using the powerful tool of mouse genetics, a control of the antibody response by the MHC, and for the first time indicated a physiological role of this gene locus distinct from the classical role in allogeneic interactions (27). Their studies also led to the identification of a new genetic region, encoding the class II MHC genes. These were powerful results. Concerning the Benacerraf group, aside from the demonstration of the control of the polylysine gene response in guinea pigs (21), they pointed very early to the possible cell that was the locus of the genetic control, and that was to the T cell recognizing the carrier, perhaps associated with an APC (28). (Indeed, they had used hapten conjugated to synthetic peptides made of polylysine.) Since the carrier recognition, as mentioned above, was now tied up to T cells, it was then apparent that T cells recognized linear determinants under the control of MHC molecules.
The restriction of the cytolytic T-cell response by MHC molecules first shown by Zinkernagel & Doherty (29), and that of B- and T-cell collaborations (30, 31), and very importantly the T-cell proliferative response first noted by Rosenthal & Shevach (32) added a new dimension; the MHC molecules controlled all cellular interactions involving T cells. (As Jan Klein, who has thought much on these issues, said, 'the body adheres to the etiquette of a proper English home; a foreigner can be introduced only by a person known to the family' (33)). Rosenthal went on to postulate that the MHC was involved in the process that led to recognition of different segments of the antigen molecule, in their case the protein insulin, in what they termed 'determinant selection' (34).

==========
Fig. 1. A side view of the peptide 52-62 of HEL bound to I-Ak molecules. This study was done by Daved Fremont (73). The solvent accessible surface is shown as blue dots. Indicated are the P1, P4, P6, P7 and P9 pockets. The HEL peptide is indicated in a ball and stick model. Shown is the alpha 52 arginine (in green) at the base of P1, which forms a salt bridge with the Asp 52 of the peptide.
==========
MHC molecules bind peptides
The role of MHC molecules became central: did they serve as peptide recognition molecules or were they molecules modified in some form by the protein antigen so to become a neo-antigen (so called altered-self); or did they play a role in mediating cell to cell contact? Having failed to detect MHC interactions using radioactive proteins (52), we opted to identify from tryptic digests of HEL, the peptide that stimulated a highly reactive T-cell hybridoma. The peptide 46-61 contained part of the epitope (48). (We subsequently identified the natural epitope from the I-Ak molecule of APC fed with HEL: the processed peptides encompassed residues 48-63 (53) ( Table 2 ).
We conceived that 46-61 peptide of HEL, having some hydrophobic residues, could either interact with membranes or with I-Ak molecules. Following Occam's razor, since testing on cells was going to be messy, why not do it directly on purified MHC molecules? We used equilibrium dialysis with labeled peptides and MHC molecules purified from cell membranes. The experiments, made by Bruce Babbitt who had just joined us, gave beautiful results that were too good to be true (54, 55). Here was a saturable, homogeneous binding interaction where one peptide molecule bound per molecule of I-Ak and with an affinity in the àM range. A peptide with scrambled residues from 46 to 61 did not bind and, importantly, a self peptide from lysozyme also bound (55). Moreover, the complex could stimulate the T-cell hybridomas directly.
The finding that an autologous lysozyme peptide also bound to I-Ak with the same affinity as the foreign lysozyme peptide was telling, along two lines. First, the MHC was not making the self/nonself discrimination and, therefore, autoimmune reactions could target self peptides, as indeed was predictable. Second, autologous peptides could compete with the foreign peptide for processing and presentation.
Going back to the discussion on the catabolism of proteins to amino acids and how to integrate this finding with immunogenicity, it was apparent that through the interactions with MHC, the bound peptides were saved from degradation and that the complex constituted the substrate of the TCR. Later, Paul Allen and his group (56) found that the peptide-MHC complex was quite resistant to proteolytic enzymes. (More recently, we reevaluated the sensitivity of purified peptide-MHC complex to chemotryptic digestion. The sensitivity depended critically on the binding affinity of the peptide for the class II molecule. In fact, mutating a single anchor residue in the peptide and weakening the binding interaction resulted in a complex that became very sensitive to proteolysis and in which the class II molecule itself was actually the target of the enzyme (57). So, a strong interaction of a peptide with a class II MHC molecule forms a tight protein complex that is resistant to proteolysis, in contrast to the complex formed by weak peptides. The reader is also referred to the relevant studies from Germain's laboratory (58, 59)).
Grey's group confirmed our binding results and extended them noticeably, reporting binding kinetics and testing a variety of peptides (60, 61). Townsend's group then made the point that class I MHC recognition by CD8 T cells also involved peptide recognition (62). Thus, the raison d'etre of the elusive MHC molecules was to select peptides and create the antigenic epitope recognized by T cells.
Seminal, of course, was the X-ray crystallographic analysis that followed. Wiley's laboratory published the first results on a class I MHC molecule, with Jack Strominger providing the biochemical expertise, and with Pam Bjorkman as lead author (63, 64). Knowing that there was a binding groove, with its walls of alpha helices and its beta-pleated sheet platform, provided the framework for defining a T-cell immunogen and for studying the fine nature of different peptide interactions among various allelic forms of MHC class I or II molecules ( Fig. 1 ). The initial chemical and immunological analysis reported by us and also by Grey's group established that some amino acid side chains in the peptide were 'anchors' required for the binding to MHC, while side chains of other residues served as contact for the TCR (65, 66). These findings became very apparent as the structures of various peptide-MHC complexes were solved, all showing a stretched peptide in the binding groove. We refer to several studies in both class I and II MHC molecules (67-75) and to recent ones on the structure of the TCR contacting the peptide-MHC complex (76-79). Since the amino acids responsible for the allelic differences were located in the binding groove for peptide, it established that the evolutionary pressure that diversified the MHC was centered on the peptide binding properties of the molecule.
=============
Fig. 2. Processing pathways in the APC. The APC contains two distinctly functional processing vesicles. Class II molecules are transported into late, lysosomal-like vesicles with the invariant chain. At this site they meet with HEL, which is then unfolded and assembled under the catalytic influence of H-2 DM. Expression of HEL peptides thus involves nascent class II molecules and H-2 DM. The late compartment only generates type A complexes. In contrast, peptides can directly assemble in early endosomes, which can recycle MHC molecules. Some MHC molecules in these early endosomes contain weak peptides that can exchange with incoming HEL peptides ( 87). Presentation of peptides takes place in mature MHC molecules in an H-2 DM independent process. The endocytic vesicles can generate both type A and B complexes.
======================
Concay
ĐặỏằÊc Milou sỏằưa vào 06:27 ngày 18/06/2003

Fig. 3. An example of the peptide isolation using antipeptide antibodies. A mixture of peptides extracted from I-Ak molecules from 109 APC lacking (panel A) or containing (panel B) HEL was passed through an antipeptide column containing a monoclonal antibody to peptide 31-45. After capture, the material was eluted and examined by electrospray tandem MS. About 30 different peptides were isolated. Shown are the ions detected in the mass to charge ratio of 779-1043. Taken from (95) experiments made by Raffi Gugasyan, Ilan Vidavasky and Michael Gross.
Concay

Fig. 3. An example of the peptide isolation using antipeptide antibodies. A mixture of peptides extracted from I-Ak molecules from 109 APC lacking (panel A) or containing (panel B) HEL was passed through an antipeptide column containing a monoclonal antibody to peptide 31-45. After capture, the material was eluted and examined by electrospray tandem MS. About 30 different peptides were isolated. Shown are the ions detected in the mass to charge ratio of 779-1043. Taken from (95) experiments made by Raffi Gugasyan, Ilan Vidavasky and Michael Gross.
Concay

Fig. 4. Quantitation of peptides isolated from APC by the anti peptide ELISA procedure (99). The top panel shows the results of 5 individual experiments (each is 1 dot) of 3 sets of HEL peptides extracted from a B-lymphoma line expressing HEL. Note the very large differences in content. On the lower panel the ratios of various peptides are plotted in peptides isolated from I-Ak molecules of B-lymphoma cell lines cultured with HEL (open squares), or the same line expressing a membrane form of HEL (dark squares); or spleen (open circle) and thymi (closed circle) of HEL transgenic mice. The ratios are the same in all APC. From Carlos Velazquez in [99].
Concay

Fig. 4. Quantitation of peptides isolated from APC by the anti peptide ELISA procedure (99). The top panel shows the results of 5 individual experiments (each is 1 dot) of 3 sets of HEL peptides extracted from a B-lymphoma line expressing HEL. Note the very large differences in content. On the lower panel the ratios of various peptides are plotted in peptides isolated from I-Ak molecules of B-lymphoma cell lines cultured with HEL (open squares), or the same line expressing a membrane form of HEL (dark squares); or spleen (open circle) and thymi (closed circle) of HEL transgenic mice. The ratios are the same in all APC. From Carlos Velazquez in [99].
Concay

Fig. 5. Total ion chromatogram of autologous peptides isolated from I-Ak molecules of a B-lymphoma line (top panel). The lower panel shows the results from the same line containing a membrane form of HEL. Indicated in large number is the area peak containing the 48-62 family of HEL peptides (Table 2). Experiment published in (96).
Concay

Fig. 5. Total ion chromatogram of autologous peptides isolated from I-Ak molecules of a B-lymphoma line (top panel). The lower panel shows the results from the same line containing a membrane form of HEL. Indicated in large number is the area peak containing the 48-62 family of HEL peptides (Table 2). Experiment published in (96).
Concay

Antigen processing and peptide selection-recent developments
The field of presentation then moved from the cell biology of the processing, to the biochemistry of the MHC molecules, to the chemistry of peptides, and more recently, back to the whole experimental animal.

Processing events
The APC by way of its MHC molecules informs the T-cell system of its internal milieu, of both its autologous peptides as well as of new peptides, resulting from the processing of an external moiety. We know now that the class I MHC system samples the cytosol, and the class II system samples the proteins of the vesicular system, but see below. There has been notable progress in delineating the various steps involved in the intracellular handling of protein for presentation by the class II molecules. The trajectory of class II MHC molecules from endoplasmic reticulum to Golgi to the class II bearing vesicles have been examined by many, and it is beyond us now to cover all the extensive analysis of the role of invariant chain and of the molecules like H-2 DM that help in the assembly of the peptide-MHC complex (80, 81).
Following on our early experiments on the processing event, we examined in which intracellular compartment of the APC HEL was processed. HEL, a tight globular protein with 4 disulfides, was known to be highly resistant to proteolysis unless it was reduced. In the APC, it required a step of reduction identified through cell fractionation studies to a late lysosomal vesicle (82, 83). We directly tested HEL encapsulated in liposomes of various compositions, each having different fates in the APC: HEL was highly immunogenic when included in liposomes that only opened in highly acidic lysosomal vesicles; in contrast, HEL packaged in liposomes that opened in early vesicles involved in recycling was less immunogenic, probably because less of it reached the deep vesicles where presumably reduction took place (83). Cresswell's laboratory has recently identified an interferon-gamma inducible thio-reductase involved in the denaturation of the protein that could be central in the processing (84, 85).
These experiments extended our early studies using lysosomotropic drugs, but they were still surprising. The initial thinking was that perhaps the processing for MHC interactions was going to take place in vesicles where the content of proteolytic enzymes could be limiting. However, targeting HEL directly to highly acidic vesicles called attention to the fact that presentation was taking place in an enzyme rich, acidic environment that received endocytosed molecules after a period of sojourn in the APC. Important studies have now defined vesicles rich in class II MHC molecules and also rich in lysosomal enzymes that receive endocytosed proteins (81, 86). Still, the fine anatomical stages of the assembly process of the protein antigen, in our case going from native HEL to denatured HEL to its MHC-bound fragment, needs to be firmly dissected.
We strongly favor a scenario where, in the lysosomal-like vesicle, the MHC samples the denatured protein and selects a segment with which to interact favorably. Once this segment is 'selected', it is protected from catabolism, and then trimmed by amino or carboxy peptidases (see below). This process of assembly from HEL takes place on nascent class II MHC molecules in which the invariant chain peptide is removed and where H-2 DM catalyzes the assembly (80, 81). Finally, it is important to emphasize that in the case of denatured or peptide fragments, the assembly with class II MHC molecules takes place in early recycling vesicles, in a process involving peptide exchange with mature MHC molecules and not involving H-2 DM (87). Thus, we favor two major sites of assembly, each having important functional implications, as discussed in the next section ( Fig. 2 ).

Peptide selection
'Peptide selection' is the process whereby peptides are captured in the APC and displayed by the MHC molecule, the final process that creates the T-cell epitope. In what chemical forms are peptides selected and in what amounts? Where are the peptides derived from, and what is the relationship between selection and display and the immune response? What is the nature of those peptides that trigger autoimmune responses? Only by obtaining fundamental information on these issues, and a mechanistic explanation, will we be able to correlate peptide selection to biological responses, including the responses to microbes, cancers, to autoimmune antigens, and to vaccines.
At present, there is no substitute for obtaining chemical data from the direct analysis of peptides extracted from MHC molecules. Another approach taken by many laboratories is to study peptides from large libraries for their binding property, and to extrapolate to the situation within an APC. Indeed, peptide binding motifs have been identified for some MHC genotypes. These are important approaches for understanding binding properties of MHC molecules, but they will not give the information on the biochemical nature of the MHC displayed peptides. At this point, the structure of naturally processed peptides is not predictable. Rammensee's group made the first attempts to identify peptides extracted from class I MHC molecules; they also sequenced bulk extracts of eluted peptides, obtaining information on favorable peptide sequences (88, 89). Other studies rapidly followed (53, 90-98).
An important breakthrough for analyzing MHC-bound peptides is electrospray tandem mass spectrometry (MS) combined with high performance liquid chromatography (HPLC) separation of peptides. This approach was championed particularly by Don Hunt and his associates, including Victor Engelhard, mainly studying class I MHC-bound peptides (92, 96). However, the identification of class II-bound peptides is considerably more difficult than that of class I-bound peptides. First, the peptides displayed by class II vary greatly in amounts expressed and in binding affinities; a peptide expressed at a small level and binding poorly will be difficult to identify particularly because the HPLC fractionation cannot resolve individual peptides to much satisfaction. Testing in antigen presentation assays, a poor binding peptide mixed with others is highly unsatisfactory. Second, and very importantly, class II peptides are not found as a single peptide bound to an MHC molecule, but rather as several peptides having in common a central core sequence with variable flanking extensions at either end.
To obviate the problems in identifying the class II-bound peptides, we devised a procedure in which monoclonal antibodies were made to the segments of the peptide that binds to the MHC groove (the core). Such monoclonals were used to identify peptides in free solution and also to capture them from complex mixtures (95, 97-100) ( Figs 3 and 4 ). This procedure allowed Carlos Velazquez to quantitate peptides (99) and to analyze various peptide families represented in very small numbers (about a few femtomoles per 109 APC) ( Fig. 4 ). (All our MS studies are done jointly with Michael Gross in our Department of Chemistry.)
What have we learned so far from the analysis of bound peptides from class II MHC molecules? The peptides bound by class II molecules are derived from internalized proteins, and also from self proteins that are normal components of the vesicles. Only recently have we appreciated the contribution of cytosolic proteins to the class II pathway of presentation (53, 89-92, 100-104). The cytosolic peptides have the same chemical properties as the peptides that derive from the processing of internalized proteins by the vesicular system, that is, they are more than 10 residues in length (see below). The mechanisms of transport from cytosol to the class II molecules on vesicles most likely involves some form of autophagy. Their biological relevance still needs to be determined.

Peptide cores and flanks
We describe first some of the binding features of peptides that have been isolated from MHC class II molecules, emphasizing our own studies with I-Ak. The next section contains a summary of these findings plus detailed studies on the diabetogenic I-Ag7 molecules.
As mentioned, peptides are selected as families with the core segment and the flanking residues. The members in a peptide family can be numerous, with as many as 30-40! (94) ( Fig. 3 ). The core segment has been identified by binding analysis and also, most convincingly, by X-ray crystallographic analysis on the bound peptides. The 9-residue central core extends from the residues that interact with the P1 binding pocket or site, up to the last pocket site or P9, and includes the P4, P6 and a shallow P7 site. These pockets will harbor the peptide amino acid side chains in interactions with the surrounding molecules the MHC anchor residues. The central core segment can also be identified by binding analysis, in which each residue has been changed ( Table 2 ). Core segments contribute to binding energy as a result of favorable interactions of one or more side chains and contribute two to three solvent-exposed residues that are available to contact the TCR.
The major peptide family selected from processing of HEL (see next section) has the core segment from 52 to 60 and binds strongly to I-Ak molecules, usually in the nM range. It contains a strong acidic residue positioned at the fourth or fifth residue from the amino end of the peptide (53, 73, 105). The acidic residue forms a salt bridge with an arginine alpha 52 that forms the base of the P1 allelic site (73) ( Fig. 1 ). No other residue in 52-60 provides binding strength to I-Ak molecules. An early study indicated that a polyalanine peptide would bind to HLA DR1 provided it contained a single residue, a tyrosine, to interact with the P1 site (106). We found this result also with the 52-60 peptide from HEL. A polyalanine peptide bearing an aspartic acid residue conferred binding, and the degree of binding was dependent on the length of the chain, a confirmation that the backbone peptide interaction contributed binding energy.
Many of the residues that face the MHC will contribute marginally to binding energy, but will be of considerable importance because they may favor or, in contrast, result in unfavorable interactions that inhibit the binding of the whole peptide (107-110). Moreover, there may be cooperativity among such hindering residues in that one or two singly displayed residues may allow for binding but will prevent it when together in the same peptide (110).
Regarding the flanking residues of class II peptides, these vary in their length on both the amino and the carboxy termini. This heterogeneity accounts for the number of members in the family. There are no obvious motifs for proteolytic enzymes on the ends of the peptides. In the case of the 52-60 core segment, it is presented in peptides that frequently start at residue 48 and end in 62 or 63 ( Table 2 ).
One finding is that there is a frequency of proline residues on the ends of the flanks (89). We mutated some of the residues flanking the 52-60 core sequence of HEL, placing prolines instead, and observed that indeed the peptides could be extended: proline inhibits the cleavage by amino and carboxypeptidases (94). The interpretation for these results given above is that HEL opens in the reducing vesicles, allowing for MHC to bind and protect a segment, which is subsequently trimmed by amino or carboxypeptidases. This interpretation does not rule out that the denatured protein could be first cut into large segments that are later trimmed after being bound to MHC molecules.
Finally, flanking residues contributed to binding energy, and influenced the time of persistence of the peptide-MHC complex on the APC surface (111). Importantly, the flanks contributed to the specificity of T cells in two ways. First, the flanks influenced the display of the TCR contact residues, resulting in T cells that may only react with the core segment if presented together with the flanks. Second, residues next to P1 and P9 contributed to specificity, perhaps by serving as contact residues, a point well studied in the Vignali laboratory (112).
In summary, there is a consensus that peptide binding to class II MHC molecules is determined by the following:
interactions of conserved residues of the class II-MHC molecule with the peptide backbone, which is influenced by the length of the peptide;

favorable interaction of MHC anchor residues with the allele specific sites;

the favorable constellation of the 'auxiliary' residues and the lack of hindering residues.

The specificity of T cells is determined by the TCR contact residues that are solvent exposed in the core segment and also by the peptide flanks, which add in a major way to the diversity of the peptide reactive T cells.

=============
Fig. 6. This figure indicates the distribution of acidic residues along the C termini of peptides isolated from I-Ag7 or I-Ag7 PD APC. The last 10 residues of all the peptides isolated from the MHC molecule were scored for the presence of glutamic acid or aspartic acid at each position. The bar represents the percentage containing the acidic residues from all peptides examined. The last residue was assigned as 10, the penultimate as 9, etc. Note the predominance of acidic residues from the two I-Ag7 lines compared to the I-Ag7 PD line. Experiment of Suri and Vidavsky in (100).
===================
Concay

Antigen processing and peptide selection-recent developments
The field of presentation then moved from the cell biology of the processing, to the biochemistry of the MHC molecules, to the chemistry of peptides, and more recently, back to the whole experimental animal.

Processing events
The APC by way of its MHC molecules informs the T-cell system of its internal milieu, of both its autologous peptides as well as of new peptides, resulting from the processing of an external moiety. We know now that the class I MHC system samples the cytosol, and the class II system samples the proteins of the vesicular system, but see below. There has been notable progress in delineating the various steps involved in the intracellular handling of protein for presentation by the class II molecules. The trajectory of class II MHC molecules from endoplasmic reticulum to Golgi to the class II bearing vesicles have been examined by many, and it is beyond us now to cover all the extensive analysis of the role of invariant chain and of the molecules like H-2 DM that help in the assembly of the peptide-MHC complex (80, 81).
Following on our early experiments on the processing event, we examined in which intracellular compartment of the APC HEL was processed. HEL, a tight globular protein with 4 disulfides, was known to be highly resistant to proteolysis unless it was reduced. In the APC, it required a step of reduction identified through cell fractionation studies to a late lysosomal vesicle (82, 83). We directly tested HEL encapsulated in liposomes of various compositions, each having different fates in the APC: HEL was highly immunogenic when included in liposomes that only opened in highly acidic lysosomal vesicles; in contrast, HEL packaged in liposomes that opened in early vesicles involved in recycling was less immunogenic, probably because less of it reached the deep vesicles where presumably reduction took place (83). Cresswell's laboratory has recently identified an interferon-gamma inducible thio-reductase involved in the denaturation of the protein that could be central in the processing (84, 85).
These experiments extended our early studies using lysosomotropic drugs, but they were still surprising. The initial thinking was that perhaps the processing for MHC interactions was going to take place in vesicles where the content of proteolytic enzymes could be limiting. However, targeting HEL directly to highly acidic vesicles called attention to the fact that presentation was taking place in an enzyme rich, acidic environment that received endocytosed molecules after a period of sojourn in the APC. Important studies have now defined vesicles rich in class II MHC molecules and also rich in lysosomal enzymes that receive endocytosed proteins (81, 86). Still, the fine anatomical stages of the assembly process of the protein antigen, in our case going from native HEL to denatured HEL to its MHC-bound fragment, needs to be firmly dissected.
We strongly favor a scenario where, in the lysosomal-like vesicle, the MHC samples the denatured protein and selects a segment with which to interact favorably. Once this segment is 'selected', it is protected from catabolism, and then trimmed by amino or carboxy peptidases (see below). This process of assembly from HEL takes place on nascent class II MHC molecules in which the invariant chain peptide is removed and where H-2 DM catalyzes the assembly (80, 81). Finally, it is important to emphasize that in the case of denatured or peptide fragments, the assembly with class II MHC molecules takes place in early recycling vesicles, in a process involving peptide exchange with mature MHC molecules and not involving H-2 DM (87). Thus, we favor two major sites of assembly, each having important functional implications, as discussed in the next section ( Fig. 2 ).

Peptide selection
'Peptide selection' is the process whereby peptides are captured in the APC and displayed by the MHC molecule, the final process that creates the T-cell epitope. In what chemical forms are peptides selected and in what amounts? Where are the peptides derived from, and what is the relationship between selection and display and the immune response? What is the nature of those peptides that trigger autoimmune responses? Only by obtaining fundamental information on these issues, and a mechanistic explanation, will we be able to correlate peptide selection to biological responses, including the responses to microbes, cancers, to autoimmune antigens, and to vaccines.
At present, there is no substitute for obtaining chemical data from the direct analysis of peptides extracted from MHC molecules. Another approach taken by many laboratories is to study peptides from large libraries for their binding property, and to extrapolate to the situation within an APC. Indeed, peptide binding motifs have been identified for some MHC genotypes. These are important approaches for understanding binding properties of MHC molecules, but they will not give the information on the biochemical nature of the MHC displayed peptides. At this point, the structure of naturally processed peptides is not predictable. Rammensee's group made the first attempts to identify peptides extracted from class I MHC molecules; they also sequenced bulk extracts of eluted peptides, obtaining information on favorable peptide sequences (88, 89). Other studies rapidly followed (53, 90-98).
An important breakthrough for analyzing MHC-bound peptides is electrospray tandem mass spectrometry (MS) combined with high performance liquid chromatography (HPLC) separation of peptides. This approach was championed particularly by Don Hunt and his associates, including Victor Engelhard, mainly studying class I MHC-bound peptides (92, 96). However, the identification of class II-bound peptides is considerably more difficult than that of class I-bound peptides. First, the peptides displayed by class II vary greatly in amounts expressed and in binding affinities; a peptide expressed at a small level and binding poorly will be difficult to identify particularly because the HPLC fractionation cannot resolve individual peptides to much satisfaction. Testing in antigen presentation assays, a poor binding peptide mixed with others is highly unsatisfactory. Second, and very importantly, class II peptides are not found as a single peptide bound to an MHC molecule, but rather as several peptides having in common a central core sequence with variable flanking extensions at either end.
To obviate the problems in identifying the class II-bound peptides, we devised a procedure in which monoclonal antibodies were made to the segments of the peptide that binds to the MHC groove (the core). Such monoclonals were used to identify peptides in free solution and also to capture them from complex mixtures (95, 97-100) ( Figs 3 and 4 ). This procedure allowed Carlos Velazquez to quantitate peptides (99) and to analyze various peptide families represented in very small numbers (about a few femtomoles per 109 APC) ( Fig. 4 ). (All our MS studies are done jointly with Michael Gross in our Department of Chemistry.)
What have we learned so far from the analysis of bound peptides from class II MHC molecules? The peptides bound by class II molecules are derived from internalized proteins, and also from self proteins that are normal components of the vesicles. Only recently have we appreciated the contribution of cytosolic proteins to the class II pathway of presentation (53, 89-92, 100-104). The cytosolic peptides have the same chemical properties as the peptides that derive from the processing of internalized proteins by the vesicular system, that is, they are more than 10 residues in length (see below). The mechanisms of transport from cytosol to the class II molecules on vesicles most likely involves some form of autophagy. Their biological relevance still needs to be determined.

Peptide cores and flanks
We describe first some of the binding features of peptides that have been isolated from MHC class II molecules, emphasizing our own studies with I-Ak. The next section contains a summary of these findings plus detailed studies on the diabetogenic I-Ag7 molecules.
As mentioned, peptides are selected as families with the core segment and the flanking residues. The members in a peptide family can be numerous, with as many as 30-40! (94) ( Fig. 3 ). The core segment has been identified by binding analysis and also, most convincingly, by X-ray crystallographic analysis on the bound peptides. The 9-residue central core extends from the residues that interact with the P1 binding pocket or site, up to the last pocket site or P9, and includes the P4, P6 and a shallow P7 site. These pockets will harbor the peptide amino acid side chains in interactions with the surrounding molecules the MHC anchor residues. The central core segment can also be identified by binding analysis, in which each residue has been changed ( Table 2 ). Core segments contribute to binding energy as a result of favorable interactions of one or more side chains and contribute two to three solvent-exposed residues that are available to contact the TCR.
The major peptide family selected from processing of HEL (see next section) has the core segment from 52 to 60 and binds strongly to I-Ak molecules, usually in the nM range. It contains a strong acidic residue positioned at the fourth or fifth residue from the amino end of the peptide (53, 73, 105). The acidic residue forms a salt bridge with an arginine alpha 52 that forms the base of the P1 allelic site (73) ( Fig. 1 ). No other residue in 52-60 provides binding strength to I-Ak molecules. An early study indicated that a polyalanine peptide would bind to HLA DR1 provided it contained a single residue, a tyrosine, to interact with the P1 site (106). We found this result also with the 52-60 peptide from HEL. A polyalanine peptide bearing an aspartic acid residue conferred binding, and the degree of binding was dependent on the length of the chain, a confirmation that the backbone peptide interaction contributed binding energy.
Many of the residues that face the MHC will contribute marginally to binding energy, but will be of considerable importance because they may favor or, in contrast, result in unfavorable interactions that inhibit the binding of the whole peptide (107-110). Moreover, there may be cooperativity among such hindering residues in that one or two singly displayed residues may allow for binding but will prevent it when together in the same peptide (110).
Regarding the flanking residues of class II peptides, these vary in their length on both the amino and the carboxy termini. This heterogeneity accounts for the number of members in the family. There are no obvious motifs for proteolytic enzymes on the ends of the peptides. In the case of the 52-60 core segment, it is presented in peptides that frequently start at residue 48 and end in 62 or 63 ( Table 2 ).
One finding is that there is a frequency of proline residues on the ends of the flanks (89). We mutated some of the residues flanking the 52-60 core sequence of HEL, placing prolines instead, and observed that indeed the peptides could be extended: proline inhibits the cleavage by amino and carboxypeptidases (94). The interpretation for these results given above is that HEL opens in the reducing vesicles, allowing for MHC to bind and protect a segment, which is subsequently trimmed by amino or carboxypeptidases. This interpretation does not rule out that the denatured protein could be first cut into large segments that are later trimmed after being bound to MHC molecules.
Finally, flanking residues contributed to binding energy, and influenced the time of persistence of the peptide-MHC complex on the APC surface (111). Importantly, the flanks contributed to the specificity of T cells in two ways. First, the flanks influenced the display of the TCR contact residues, resulting in T cells that may only react with the core segment if presented together with the flanks. Second, residues next to P1 and P9 contributed to specificity, perhaps by serving as contact residues, a point well studied in the Vignali laboratory (112).
In summary, there is a consensus that peptide binding to class II MHC molecules is determined by the following:
interactions of conserved residues of the class II-MHC molecule with the peptide backbone, which is influenced by the length of the peptide;

favorable interaction of MHC anchor residues with the allele specific sites;

the favorable constellation of the 'auxiliary' residues and the lack of hindering residues.

The specificity of T cells is determined by the TCR contact residues that are solvent exposed in the core segment and also by the peptide flanks, which add in a major way to the diversity of the peptide reactive T cells.

=============
Fig. 6. This figure indicates the distribution of acidic residues along the C termini of peptides isolated from I-Ag7 or I-Ag7 PD APC. The last 10 residues of all the peptides isolated from the MHC molecule were scored for the presence of glutamic acid or aspartic acid at each position. The bar represents the percentage containing the acidic residues from all peptides examined. The last residue was assigned as 10, the penultimate as 9, etc. Note the predominance of acidic residues from the two I-Ag7 lines compared to the I-Ag7 PD line. Experiment of Suri and Vidavsky in (100).
===================
Concay