Rationally engineered proteins or antibodies with absent or reduced immunogenicity.

One challenge associated with the clinical use of protein therapeutics destined for chronic administration is the potential for the development of unwanted anti-drug immune reactions. The molecular basis for this reactivity is the binding of peptide fragments (epitopes) derived from the breakdown of the protein drug to the HLA receptors expressed by the patient's immune cells. If these epitopes are recognized as "foreign" by the immune system, specific helper T lymphocytes (HTL), are activated, which initiate and direct the formation of antibodies against the protein drug. These antibodies can bind and neutralize the protein drug, resulting in either decreased efficacy or total ineffectiveness of the drug. Moreover, various safety concerns, such as allergic reactions and other adverse events, are also frequently associated with the formation of anti-drug antibodies. Herein, we describe the development of "ImmunoStealth", an integrated bioinformatics, biochemical and cellular immunology approach that specifically addresses the issue of unwanted immune responses against protein therapeutics. Unwanted HTL epitopes are identified using in silico sequence analysis methods and high throughput in vitro biochemical evaluations and thereafter confirmed using cellular immunogenicity assays. The "offending" epitopes within the drug are then rationally modified to alter their HLA binding capacity, and thus render them non-recognizable by the immune system. This technology will ultimately facilitate the design of safer, more potent and more economical drugs.

Optimizing vaccine design for cellular processing, MHC binding and TCR recognition.

In this review we describe the methods and processes that our group have developed while aiming to test and design multiepitope vaccines for infectious diseases and cancer. Testing the performance of vaccines composed of epitopes restricted by human leukocyte antigen (HLA) molecules is accomplished by in vitro antigenicity assays, as well as in vivo immunogenicity assays in HLA transgenics. The efficiency by which multiepitope vaccines are processed is optimized by spacer residues, which are designed to facilitate generation by natural processing of the various class I- and class II-restricted epitopes. Methods and strategies to test and optimize HLA binding affinity, patient coverage from the vaccine construct, and TCR recognition of HLA/epitope complexes are also discussed.

Structural features of peptide analogs of human histocompatibility leukocyte antigen class I epitopes that are more potent and immunogenic than wild-type peptide.

Certain peptide analogs that carry substitutions at residues other than the main major histocompatibility complex anchors and are surprisingly much more antigenic than wild-type peptide (heteroclitic analogs). To date, it was unknown how frequently wild-type epitopes could be modified to obtain heteroclitic activity. In this study, we analyzed a large panel of analogs of two different human histocompatibility leukocyte antigen (HLA)-A2.1-restricted epitopes and found that heteroclitic analogs were associated with higher magnitude responses and increased (up to 10(7)-fold) sensitivity to antigen, and corresponded to conservative or semiconservative substitutions at odd-numbered positions in the middle of the peptide (positions 3, 5, or 7). These findings were validated by performing additional immunogenicity studies in murine and human systems with four additional epitopes. The biological relevance of heteroclitic analogs was underlined when predicted analogs of the p53.261 epitope was shown to induce cytotoxic T lymphocytes (CTLs) that recognize low concentrations of peptide (high avidity) in vivo and demonstrate in vitro antitumor recognition. Finally, in vitro immunization of human peripheral blood mononuclear cells with two heteroclitic analogs resulted in recruitment of more numerous CTLs which were associated with increased antigen sensitivity. In conclusion, heteroclitic analogs were identified in each of the six cases studied and structural features were defined which allow identification of such analogs. The strong CTL immunity elicited by heteroclitic epitopes suggest that they could be of significant value in vaccination against tolerant or weakly immunogenic tumor-associated and viral antigens.

Presentation of peptide antigens by mouse CD1 requires endosomal localization and protein antigen processing.

Mouse CD1(mCD1) molecules have been reported to present two types of antigens: peptides or proteins and the glycolipid alpha-galactosylceramide. Here, we demonstrate that a protein antigen, chicken ovalbumin (Ova), must be processed to generate peptides presented by mCD1 to CD8(+) T cells. The processing and mCD1-mediated presentation of chicken Ova depend on endosomal localization because inhibitors of endosomal acidification and endosomal recycling pathways block T cell reactivity. Furthermore, a cytoplasmic tail mutant of mCD1, which disrupts endosomal localization, has a greatly reduced capacity to present Ova to mCD1 restricted cells. Newly synthesized mCD1 molecules, however, are not required for Ova presentation, suggesting that molecules recycling from the cell surface are needed. Because of these data showing that mCD1 trafficks to endosomes, where it can bind peptides derived from exogenous proteins, we conclude that peptide antigen presentation by mCD1 is likely to be a naturally occurring phenomenon. In competition assays, alpha-galactosylceramide did not inhibit Ova presentation, and presentation of the glycolipid was not inhibited by excess Ova or the peptide epitope derived from it. This suggests that, although both lipid and peptide presentation may occur naturally, mCD1 may interact differently with these two types of antigens.

Antigen-presenting function of mouse CD1: one molecule with two different kinds of antigenic ligands.

Mouse CD1 (mCD1) is an antigen-presenting molecule that is constitutively expressed by most bone marrow-derived cells. Peptides with a hydrophobic binding motif can bind to mCD1, and the peptide-CD1 complex is recognized by CD8+ cytolytic T cells. In contrast, NK1.1+ T cells, which are CD8-, are autoreactive for mCD1 molecules. This autoreactivity, along with the ability of NK T cells to rapidly produce large amounts of cytokine, has led to the suggestion that these cells may be immunoregulatory. We have shown that the mCD1-autoreactive T cells can distinguish between different cell types that express similar levels of mCD1, suggesting that mCD1-bound autologous ligands may be critical for T-cell stimulation. Consistent with this, some of these mCD1-restricted T cells can recognize the glycolipid alpha-galactosylceramide presented by mCD1, while others do not respond. The mCD1 crystal structure reveals a deep and narrow hydrophobic antigen-binding site which can more easily bind lipid antigens than the long hydrophobic peptides that we have defined as mCD1 antigens. The ability of mCD1 to bind and present two different types of ligands raises the question as to how mCD1 can accommodate both types of antigens.

Mouse CD1-autoreactive T cells have diverse patterns of reactivity to CD1+ targets.

Humans and mice contain significant populations of T cells that are reactive for autologous CD1 molecules. Using a panel of five mouse CD1 (mCD1)-autoreactive T cell hybridomas, we show here that this autoreactivity does not correlate with the level of CD1 expression. In some cases, these autoreactive T cells can distinguish between different cell types that express the same CD1 molecule, suggesting that some factor in addition to CD1 expression is critical for autoreactive T cell stimulation. To determine whether a CD1-bound ligand may be required, we expressed mutant mCD1 molecules that are defective for the putative endosomal localization sequence in the cytoplasmic domain. We demonstrate that mCD1, like its human CD1 homologues, is found in endosomes, and that it colocalizes extensively with the DM molecule. We further demonstrate, by site-directed mutagenesis, that the tyrosine in the cytoplasmic sequence is required for this endosomal localization. A T cell hybrid expressing Vbeta8 and Valpha14, the major TCR expressed by NK1+ T cells, exhibited greatly diminished reactivity to mutant CD1 molecules that do not traffic through endosomes, although the reactivity of other T cell hybrids to this mutant was not greatly affected. Therefore, we propose that at least some of the autoreactive T cells require endosomally derived CD1-bound ligands, and that they are capable of distinguishing between a diverse set of such self-ligands, which might be either autologous lipoglycans or peptides.

Antigen-presenting function of the mouse CD1 molecule.

CD1 molecules are distantly related to major histocompatibility complex (MHC)-encoded class I molecules, and they are coexpressed with beta2 microglobulin (beta2m). In the mouse, CD1 is expressed by intestinal epithelial cells and also by some cells in spleen and lymph node. We have shown that surface expression of mouse CD1 (mCD1) is not dependent upon a functional transporter associated with antigen processing (TAP). This, and other data, suggest that mCD1 may acquire peptides in an intracellular compartment other than the endoplasmic reticulum, where classical class I molecules bind peptide. mCD1 molecules also are distinct from classical class I molecules with regard to the types of peptides that they bind. We have demonstrated that mCD1 molecules preferentially bind peptides much longer than the 8-9 amino acids typical of the peptides that bind to classical class I molecules. The sequence motif for mCD1 peptide binding is characterized by the presence of bulky and hydrophobic amino acid side chains. We have generated mCD1-restricted and peptide-specific T-cell lines, thereby demonstrating the immunologic relevance of peptide binding to mCD1. The reactive T cells are TCR alphabeta+ and CD8+, a phenotype typical of many lymphocytes in both lymph node and intestinal mucosae. We speculate that mCD1 molecules may be capable of sampling peptides from the gut lumen and presenting them to mucosal T lymphocytes. In this way, they may function in the maintenance of normal mucosal homeostasis, and perhaps also in the induction of systemic tolerance to antigens delivered by the oral route. In summary, CD1 molecules are a novel category of antigen-presenting molecules that have features in common with class I molecules, features in common with class II, and properties distinct from either subset of antigen-presenting molecules. Further studies of the antigen-presenting function of these molecules are certain to yield new insight into immune regulation and perhaps also into the mechanism of oral tolerance.

CD1-restricted CD4+ T cells in major histocompatibility complex class II-deficient mice.

Rather unexpectedly, major histocompatibility complex class II-deficient mice have a significant population of peripheral CD4+ T lymphocytes. We have investigated these cells at the population and clonal levels. CD4+ T lymphocytes from class II-deficient animals are thymically derived, appear early in ontogeny, exhibit the phenotype of resting memory cells, are potentially functional by several criteria, and have a diverse T cell receptor repertoire. They do not include substantially elevated numbers of NK1.1+ cells. Hybridomas derived after polyclonal stimulation of the CD4+ lymphocytes from class II-deficient animals include a subset with an unusual reactivity pattern, responding to splenocytes from many mouse strains including the strain of origin. Most members of this subset recognize the major histocompatibility complex class Ib molecule CD1; their heterogeneous reactivities and T cell receptor usage further suggest the involvement of peptides and/or highly variable posttranslational modifications.

Antigen-presenting function of the TL antigen and mouse CD1 molecules.

The hallmark of all the nonclassical antigen-presenting molecules, including nonclassical class I and nonclassical class II (Karlsson et al. 1992) molecules, is their lack of polymorphism. It is presumed, therefore, that these nonclassical molecules must have a distinct antigen-presenting function in which polymorphism is not advantageous. In some cases this may involve presentation of a nonpeptide antigen, as has been demonstrated for human CD1b. It is possible that a molecule adapted to present bacterial lipids would remain relatively nonpolymorphic, because a lipid, which is the end product of a complex biosynthetic pathway, is likely to evolve less rapidly than a short stretch of amino acid sequence containing a T-cell epitope. Alternatively, the lack of polymorphism could reflect the presentation by these molecules of relatively invariant peptides, such as those derived from heat shock proteins. It also is possible that a nonpolymorphic molecule could be selected for the presentation of modified peptides. An example of this is the M3 molecule, which can bind even short peptides as long as they have a formylated N-terminus (Fischer Lindahl et al. 1991). Based upon their structural differences, we believe it is likely that the TL antigen and mCD1 are likely to present different types of ligands. The presence in the TL antigen of the conserved amino acids, which in class I normally from hydrogen bonds with peptides, suggests that the TL antigen also can present nanomeric peptides. A peptide antigen-presenting function also is suggested by the expression of the TL antigen by at least one antigen-presenting cell type, the epithelial cell of the intestine, and by the ability of alloreactive T cells to recognize the TL molecule. While we favor the hypothesis that the TL antigen presents peptides, the data cited above do not constitute formal proof of any kind of antigen-presenting function, and it remains possible that the TL antigen does something else. As noted above, no attempts to elucidate the structure of the ligands bound to the TL antigen have so far succeeded, including the screening of bacteriophage display libraries (Castaño, A.R., Miller, J.E., Holcombe, H.R., unpublished data). In contrast, our recent work has demonstrated that mCD1 presents relatively long peptides with a structured motif distinct from classical class I molecules. This mCD1-binding motif, which is present in a wide range of proteins, does not by itself provide a simple explanation for the lack of mCD1 polymorphism and, as noted above, it remains possible that the natural ligand for mCD1 is a nonpeptide structure. Besides their lack of polymorphism, the TL antigen and mCD1 molecules share two additional features in common which might give insight into their their biological role. First, their surface expression does not depend upon the presence of a functional TAP transporter, and they probably can reach the cell surface as empty molecules. Second, both molecules are expressed by epithelial cells in the intestine. This leads to the speculation that these two nonclassical class I molecules could be involved in sampling or uptake of lumenal peptides for their ultimate presentation to cells of the systematic immune system. For example, longer lumenal peptides could be taken up by mCD1, and perhaps by the TL antigen, and then further processed to nonamers for presentation by classical class I molecules. They also could be transported across the epithelial cell by the TL antigen or mCD1 and subsequently presented by either class I or class II molecules expressed by cells in the lamina propria. This sampling or uptake mediated by either the TL antigen or mCD1 could play a role in the induction of immune responses, or more likely perhaps, in the induction of systemic oral tolerance to peptide antigens.(ABSTRACT TRUNCATED)

Peptide binding and presentation by mouse CD1.

CD1 molecules are distantly related to the major histocompatibility complex (MHC) class I proteins. They are of unknown function. Screening random peptide phage display libraries with soluble empty mouse CD1 (mCD1) identified a peptide binding motif. It consists of three anchor positions occupied by aromatic or bulky hydrophobic amino acids. Equilibrium binding studies demonstrated that mCD1 binds peptides containing the appropriate motif with relatively high affinity. However, in contrast to classical MHC class I molecules, strong binding to mCD1 required relatively long peptides. Peptide-specific, mCD1-restricted T cell responses can be raised, which suggests that the findings are of immunological significance.

Maternal anti-placental reactivity in natural, immunologically-mediated fetal resorptions.

Observations on maternal recognition of the fetus and the demonstration of the effects of cytokines on reproductive events led to the "immunotrophism" model, which suggests that maternal immune recognition of fetally-derived Ags results in the release of cytokines that promote the growth of the placenta; any disturbance in this balance of cytokines could result in deleterious consequences for the placenta and, in turn, the fetus. We have focused our attention on the murine CBA/J x DBA/2 model of spontaneous abortions and compared them with normal CBA x BALB/c pregnancies. Our results indicate that the extent of stimulation of maternal strain lymphocytes in response to stimulator placental cells in mixed lymphocyte-placenta reactions (MLPR) was much higher in the normal mating combination compared with the abortion-prone mating combination. Cytokine analysis of the supernatants from MLPR indicates that there is significantly higher production of TNF-alpha, IFN-gamma, and IL-2 in supernatants from the abortion-prone combination than in supernatants from the normal combination. Furthermore, MLPR-stimulated cells induce resorptions in normal pregnant mice; maternal strain lymphocytes stimulated by placentas from the abortion-prone combination induce high rates of fetal resorptions, but lymphocytes stimulated with placentas from the normal combination do not. Together, these results suggest that immunologically mediated fetal resorptions probably result from improper or inappropriate maternal responses to placental Ags. Our observations also suggest that such effects are probably mediated by cytokines.

Expression of cytokines in placentas of mice undergoing immunologically mediated spontaneous fetal resorptions.

It is clear that the immune system and the reproductive system interact with and influence each other and that the immune system can have positive and negative regulatory effects on the outcome of pregnancy. The discovery of murine models of immunologically mediated spontaneous fetal resorptions has proved to be very useful for the study of immunological influences on pregnancy. In an attempt to elucidate the mechanisms underlying pregnancy impairment in one such "natural" model of pregnancy loss, we compared the expression of the cytokines tumor necrosis factor alpha, interferon tau, and interleukin-2 in placental tissue from a resorption-prone strain combination with the expression from a normal combination. We found significantly enhanced expression of these three cytokines in placentas from the resorption-prone combination using dot-blot hybridization and Northern hybridizations. Since these cytokines are abortifacients in vivo and have detrimental effects on the placenta, and hence on fetal development and survival, our demonstration of enhanced expression of these deleterious cytokines may give insight into the mechanisms involved in immunologically mediated spontaneous abortions.

Recombinant fusion protein identified by lepromatous sera mimics native Mycobacterium leprae in T-cell responses across the leprosy spectrum.

Pooled polyvalent sera from lepromatous leprosy patients were used to screen a lambda gt11 recombinant DNA expression library of Mycobacterium leprae in order to identify the relevant antigens recognized by the human immune response. Of the 300,000 phages screened, 4 clones were identified that coded for fusion proteins of the same molecular mass. The fusion protein from clone LSR2 was tested for immunoreactivity in assays using peripheral blood cells and sera from 11 laboratory personnel and 105 patients across the leprosy spectrum. LSR2 protein appears to be predominantly a T-cell antigen. It evokes similar lymphoproliferative responses as the native bacillus both at the individual level and in the leprosy spectrum as a whole. Though only 50% of patient sera with anti-M. leprae antibodies reacted with the fusion protein, the pattern of reactivity in the antibody responses was also similar for the various clinical types. The coding regions of clones LSR1 and LSR2 are identical. They show no homology with sequences stored in data banks and encode a protein of 89 amino acids with a calculated molecular mass of approximately 10 kDa.