Discovery of novel p-arylthio cinnamides as antagonists of leukocyte function-associated antigen-1/intercellular adhesion molecule-1 interaction. 4. Structure-activity relationship of substituents on the benzene ring of the cinnamide.

We have shown that p-arylthio cinnamides can inhibit the interaction of LFA-1 and ICAM-1, which is involved in cell adhesion and the inflammatory process. We now show that 2,3-disubstitution on the aryl portion of the cinnamide results in enhanced activity over mono substitution on the ring. The best 2,3-substituents were chlorine and trifluoromethyl groups. Compounds 39 and 40 which contain two CF3 groups have IC(50) values of 0.5 and 0.1 nM, respectively, in inhibiting JY8 cells expressing LFA-1 on their surface, from adhering to ICAM-1. The structure-activity relationship (SAR) was examined using an NMR based model of the LFA-1 I domain/compound 31 complex. One of our compounds (38) was able to reduce cell migration in two different in vivo experiments.

Discovery of potent antagonists of leukocyte function-associated antigen-1/intercellular adhesion molecule-1 interaction. 3. Amide (C-ring) structure-activity relationship and improvement of overall properties of arylthio cinnamides.

The interaction of LFA-1 and ICAM-1 plays an important role in the cell adhesion process. On the basis of previously reported SAR and structural information on the binding of our p-arylthiocinnamide series to LFA-1, we have identified the cyclic amide (C-ring) as a site for modification. Improvement in potency and, more importantly, in the physical properties and pharmacokinetic profiles of the leading compounds resulted from this modification. One of the best compounds (11f) is also shown to reduce myocardial infarct size in rat.

Discovery and SAR of diarylsulfide cyclopropylamide LFA-1/ICAM-1 interaction antagonists.

Diarylsulfide cyclopropylamides were synthesized and evaluated as LFA-1/ICAM-1 interaction antagonists. A substituent pattern was identified which maximized potency and minimized protein binding as exemplified by antagonist 30 (IC50 = 5 nM).

Novel p-arylthio cinnamides as antagonists of leukocyte function-associated antigen-1/intracellular adhesion molecule-1 interaction. 2. Mechanism of inhibition and structure-based improvement of pharmaceutical properties.

The interaction between leukocyte function-associated antigen-1 (LFA-1) and intracellular adhesion molecule-1 (ICAM-1) has been implicated in inflammatory and immune diseases. Recently, a novel series of p-arylthio cinnamides has been described as potent antagonists of the LFA-1/ICAM-1 interaction. These compounds were found to bind to the I domain of LFA-1 using two-dimensional NMR spectroscopy of 15N-labeled LFA-1 I domain. On the basis of NOE studies between compound 1 and the I domain of LFA-1, a model of the complex was constructed. This model revealed that compound 1 does not directly inhibit ICAM-1 binding by interacting with the metal ion dependent adhesion site (MIDAS). Instead, it binds to the previously proposed I domain allosteric site (IDAS) of LFA-1 and likely modulates the activation of LFA-1 through its interaction with this regulatory site. A fragment-based NMR screening strategy was applied to identify small, more water-soluble ligands that bind to a specific region of the IDAS. When incorporated into the parent cinnamide template, the resulting analogues exhibited increased aqueous solubility and improved pharmacokinetic profiles in rats, demonstrating the power of this NMR-based screening approach for rapidly modifying high-affinity ligands.

Discovery of novel p-arylthio cinnamides as antagonists of leukocyte function-associated antigen-1/intracellular adhesion molecule-1 interaction. 1. Identification of an additional binding pocket based on an anilino diaryl sulfide lead.

The interaction between leukocyte function-associated antigen-1 (LFA-1), a member of the beta(2)-integrin family of adhesion molecules, and intracellular adhesion molecule ICAM-1 (cd54) is thought to play a critical role in the inflammatory process. On the basis of an anilino diaryl sulfide screening lead 1, in combination with pharmacophore analysis of other screening hits, we have identified an adjacent binding pocket. Subsequently, a p-ethenylcarbonyl linker was discovered to be optimal for accessing this binding site. Solution-phase parallel synthesis enabled rapid optimization of the cinnamides for this pocket. In conjunction with fine-tuning of the diaryl substituents, we discovered a novel series of potent, nonpeptide inhibitors of LFA-1/ICAM-1 interaction, exemplified by A-286982 (28h), which has IC(50) values of 44 and 35 nM in an LFA-1/ICAM-1 binding assay and LFA-1-mediated cellular adhesion assay, respectively.

Anti-melanoma cytotoxic T lymphocytes (CTL) recognize numerous antigenic peptides having 'self' sequences: autoimmune nature of the anti-melanoma CTL response.

A line of tumor-infiltrating lymphocytes (660TIL) specifically lysed the autologous HLA-A2+ melanoma (660MEL) and also most A2+ melanoma cell lines. We immunoprecipitated A2 from a large number (>10(12)) of 660MEL cells, extracted naturally processed peptides, fractionated them by HPLC, screened the fractions for recognition by 660TIL, and found a single predominant and a minor peak of activity. Although too little was recovered of the major 660MEL peptide to establish its sequence, HPLC fingerprinting showed that it did not correspond to any of the known A2-associated melanoma peptides recognized by T cells, including peptides from tyrosinase, MART-1/Melan-A, gp100 and MAGE-3. The major 660MEL antigenic peptide appears to be derived from MART-1/Melan-A but is neither AAGIGILTV nor ILTVILGVL nor any other MART-1/Melan-A peptide containing the A2 consensus motif. The multiplicity of melanoma peptides recognized by CD8+ T cells, most of which are non-mutated (including most likely the present 660MEL peptide), suggests the existence of unknown mechanisms, perhaps similar to those operating in autoimmune disorders, whereby T cells that recognize normal 'self' sequences become activated.

Dissolution rate of cholesterol and palmitic acid mixtures in cholelitholytic cosolvent systems.

The dissolution rates and solubilities of cholesterol monohydrate, palmitic acid, and their mixtures in the cholelitholytic solvents monooctanoin (MO) and methyl tert-butyl ether (MTBE) and mixtures of these two solvents were determined. The dissolution rates obtained were consistent with the diffusion-controlled two-component noninteracting model. The addition of MTBE as cosolvent to MO resulted in an increase in the solubility of both cholesterol monohydrate and palmitic acid; in the case of the former, the solubility peaked at 80% MTBE. Neither solute exhibited a log-linear solubility relationship on addition of MTBE as cosolvent. Furthermore the increases in the dissolution rates of both components were much larger than could be explained by the solubility increases alone. Mass transfer coefficients increased dramatically with increasing MTBE content of the solvent, were consistently higher for palmitic acid, and reflected the decline in solvent viscosity. Incorporation of relationships among solubility, viscosity, and cosolvent composition into the two-component noninteracting model gave good correlation between predicted and observed rates over nearly 3 orders of magnitude.

Antibody-targeted interleukin 2 stimulates T-cell killing of autologous tumor cells.

A genetically engineered fusion protein consisting of a chimeric anti-ganglioside GD2 antibody (ch14.18) and interleukin 2 (IL2) was tested for its ability to enhance the killing of autologous GD2-expressing melanoma target cells by a tumor-infiltrating lymphocyte line (660 TIL). The fusion of IL2 to the carboxyl terminus of the immunoglobulin heavy chain did not reduce IL2 activity as measured in a standard proliferation assay using either mouse or human T-cell lines. Antigen-binding activity was greater than that of the native chimeric antibody. The ability of resting 660 TIL cells to kill their autologous GD2-positive target cells was enhanced if the target cells were first coated with the fusion protein. This stimulation of killing was greater than that of uncoated cells in the presence of equivalent or higher concentrations of free IL2. Such antibody-cytokine fusion proteins may prove useful in targeting the biological effect of IL2 and other cytokines to tumor cells and in this way stimulate their immune destruction.

GD3-reactive antibodies can inhibit the lysis of autologous tumor cells by tumor-infiltrating lymphocytes.

GD3 is expressed in high concentrations on melanoma cells and may serve as a useful target antigen for mAb-mediated immunotherapy. Monoclonal antibodies (mAbs) against GD3 stimulate cell-mediated immune responses against tumor cells in vitro and this activity may contribute to antitumor effects in patients with melanoma treated with GD3-reactive mAbs. In the present study the effects of GD3-reactive mAbs on autologous tumor cell lysis by a human melanoma-derived tumor-infiltrating lymphocyte (TIL) population were examined. Unlike results reported for other GD3+ T cells isolated from melanoma patients, the tumor-specific lytic activity of the TIL line was inhibited by incubation with mAbs against GD3. Other melanoma-reactive mAbs, including those against GD2 and the high-molecular-weight melanoma-associated Ag, had no effect on the TIL lytic activity. Overall, these results indicate that mAbs against GD3 may have different effects on T cell/tumor cell interactions.

Increased tumor-specific CTL activity in human tumor-infiltrating lymphocytes stimulated with autologous tumor lines.

Two long-term tumor-infiltrating lymphocyte (TIL) lines and their autologous tumor lines have been established from solid tumors derived from different patients with metastatic melanoma. In 4-hr 51Cr release assays, each TIL culture lysed only the autologous cryopreserved fresh or established melanoma line, but failed to lyse other melanoma tumors or K562 cells. Repeated stimulation of TIL with the autologous melanoma lines resulted in significant increases in anti-tumor CTL activity with no apparent loss in specificity. Stimulated cells have retained cytotoxic activity for up to 5 months in culture. Tumor cell CTL activity for both long-term TIL lines is inhibited by several mAbs, including those against CD3, CD8, and class I MHC molecules, indicating that the effector cells are class I-restricted CD8+, CTL. Furthermore, recognition of Ag on one of the established melanoma lines by TIL is restricted by HLA A-2. The availability of autologous tumor lines may prove clinically useful for the selective stimulation and expansion of cells with anti-tumor activity within a heterogeneous TIL population.

The use of microcapsules for high density growth of human tumor infiltrating lymphocytes and other immune reactive T cells.

The use of microcapsules to achieve high density growth of tumor infiltrating lymphocytes (TIL) and other antigen-specific human T cells is described. Whereas human T cells in suspension cultures usually do not exceed 1-2 x 10(6) cells/ml, densities approaching that found in living tissues (greater than 10(8) cells/ml) have been observed for microcapsule cultures. TIL and human peripheral blood-derived T cells can be routinely recovered from microcapsules with viabilities greater than or equal to 90%. The recovered cells retain their antigen reactivities and bear cell surface phenotypes identical to their counterparts grown in suspension culture. These findings suggest that microcapsule technology could prove valuable in generating the vast numbers of cells required for TIL therapy and other forms of adoptive immunotherapy with T cells.

Serine esterase and hemolytic activity in human cloned cytotoxic T lymphocytes.

Target cell lysis by most murine cytotoxic T lymphocytes appears to be mediated by a complement (C9)-like protein called perforin, contained in high-density cytoplasmic granules. These granules also contain high levels of serine esterase activity, which may also play a role in cytolysis. Analysis of 17 cloned human cytotoxic T lymphocytes revealed the presence of serine esterase that is very similar to its murine counterpart in substrate and inhibitor specificities, pH optimum, and molecular mass; dot blot hybridization with synthetic oligonucleotides corresponding to the active sites of two known murine CTL esterases suggests homology to the murine enzyme HF. However, serine esterase was present at only approximately 10% of the level found in murine CTLs, and was not secreted during CTL-target cell interaction; moreover, hemolytic activity could not be detected in any of the seven cell lines tested. The results suggest that the human CTLs examined here kill their target cells by a mechanism different from that used by most cloned murine CTLs.

Tissue-specific expression of functionally rearranged lambda 1 Ig gene through a retrovirus vector.

To explore the potential use of retrovirus vectors for the transfer of genomic DNA sequences into mammalian cells, recombinant retroviral genomes were constructed that encode a functionally rearranged murine lambda 1 immunoglobulin gene. Several of these genomes could be transmitted intact to recipient cells by viral infection, although successful transmission depended both on the orientation of the lambda 1 sequences and on their specific placement within vector sequences. The lambda 1 gene transduced by viral infection was expressed in a cell lineage-specific manner, albeit at lower levels than endogenous lambda 1 gene expression in cells from the B-lymphocyte lineage. Vectors yielding integrated proviruses that lacked viral transcriptional enhancer sequences were used to show that neither viral transcription nor the viral transcriptional sequences themselves had any effect on the tissue specificity of lambda 1 gene expression or the absolute amount of lambda 1 transcription. Vector transcription did, however, dramatically decrease the amount of lambda 1 protein that could be detected in tranduced cells. These results suggest that retrovirus vectors may be useful reagents not only for the expression of complementary DNA sequences but also for studies of tissue-specific transcription in mammalian cells.

A functional gamma gene formed from known gamma-gene segments is not necessary for antigen-specific responses of murine cytotoxic T lymphocytes.

Structural similarities between surface immunoglobulins (s Ig) on B cells and antigen-specific receptors on T cells suggest that a T cell, like a B cell, should express only two immunoglobulin-like genes, one for each subunit of the disulphide-linked, heterodimeric, antigen-specific (alpha beta) T-cell receptor. However, cytotoxic T lymphocytes (Tc cells) and immature thymocytes also contain RNA transcripts of a third immunoglobulin-like gene, called gamma (refs 1-4). A polypeptide corresponding to the gamma gene has not yet been identified and the function of this gene remains an enigma. Judging from its nucleotide sequence, the rearranged gamma gene is expected to encode an integral membrane polypeptide chain, and gamma complementary DNAs from two cloned Tc cell lines have previously been found to have different sequences around the V-J (variable region-joining region) junction, suggesting that, in these cells, the gamma-gene product is a clonally diverse surface structure that may form part of an as yet unidentified, antigen-specific receptor. To analyse further the extent of diversity of the gamma-gene product, we have determined the partial sequences of 11 gamma cDNA clones from three other cloned Tc cell lines, and report here that the sequences are indeed clonally diverse, but in all instances they are out-of-phase in the region of the V-J junction. This finding and the pattern of gamma-gene rearrangements in these cell lines indicate that a polypeptide product of the previously reported gamma gene, V2J2-C2, is not expressed in them and is, therefore, not necessary for the antigen-specific cytotoxic and proliferative responses of these mature T cells.

Diversity at the variable-joining region boundary of lambda light chains has a pronounced effect on immunoglobulin ligand-binding activity.

By recombining lambda light (L) chains having known variable (V) region amino acid or nucleotide sequences with a heavy (H) chain from a myeloma protein or a monoclonal antibody, we obtained reconstituted Igs that differed from each other in sequence by only one or a few amino acid substitutions at known L chain positions. Differences in affinity of the reconstituted Igs for 2,4-dinitrophenyl (DNP) ligands revealed a pronounced effect on Ig binding activity of amino acids at the V-J boundary of the lambda chains. In one instance, two reconstituted Igs that differed about 1000-fold in affinity for epsilon-DNP-aminocaproate differed in primary structure by only a single tyrosine-phenylalanine substitution at the V-J junction (position 98) of their lambda 2 chains--i.e., by only one out of approximately 660 amino acid residues (L + H chains). By focusing on affinity changes, chains with unusual V lambda-J lambda junctional residues were identified. It is possible that because of a critical effect on tertiary structure junctional amino acid variations arising from gene segment assembly (V/J and perhaps V/D/J) constitute an important source of ligand-binding diversity of antibodies.

Amino acid and nucleotide sequences of variable regions of mouse immunoglobulin light chains of the lambda 3-subtype.

To learn about the V lambda gene segments that are expressed in lambda 3 light chains, the most recently identified lambda-chain subtype in inbred mice, we determined partial amino acid sequences of the V regions of two of these chains, L5-8 and Lc49 . The partial sequences were extended by establishing the complete V region sequence of cDNA for the lambda-chain mRNA from the hybridoma (RZ 5-8) and the myeloma ( CBPC -49) that produce these chains. The primer extension method used to sequence the cDNA is described in detail, because the same primer (a synthetic heptadecadeoxynucleotide ) can be used for sequencing cDNA for lambda-chains of all three subtypes of inbred mice and probably for lambda-chains from some other vertebrate species as well. The results confirm earlier preliminary findings that for both chains the V region is encoded by the V lambda 1 and J lambda 3 gene segments. The unmutated germ-line sequences of these gene segments are present in both chains, but the two chains, nevertheless, differ at codon 97, the V lambda-J lambda boundary. A T/G difference in the third position of this codon resulted in a codon for histidine (CAT) in one chain (L5-8) and for glutamine (CAG) in the other chain ( Lc49 ). This difference can be accounted for by variation in the site of V lambda 1-J lambda 3 recombination. Though the V region amino acid sequences of L5-8 and Lc49 differ only by the His/Gln substitution at position 97, the two chains have been shown (manuscript in preparation) to differ in their ability to form an effective combining site for the 2,4-dinitrophenyl group.

Frequency of lambda light chain subtypes in mouse antibodies to the 2,4-dinitrophenyl (DNP) group.

Of the three lambda chain subtypes made by inbred mice, chains of the lambda 1 subtype are much more frequent than those of the other subtypes (lambda 2,lambda 3) in antibodies (Ab) to those few antigenic structures that are known to elicit responses, in which lambda chains are the predominant type of light chain [(4-hydroxy-3-nitrophenyl)acetyl (NP) and dextran]. The reason for the frequency differences are not understood, and the large difference between the lambda 1 and lambda 3 frequencies is particularly puzzling, because in nearly all (about 95%) chains of these subtypes the N-terminal 97 or 98 amino acids are endoded by the same V lambda-gene segment. In an effort to identify an Ab response that has different lambda subtype frequencies, we analyzed the light chains of the Ab made by BALB/c and B6 mice in response to 2,4-dinitrophenylated chicken gamma globulin (DNP-CGG). We found that approximately 40% of the elicited anti-DNP molecules had lambda chains and of these approximately 40% were of the lambda 2 or lambda 3 subtype. Polyacrylamide gel electrophoresis indicated that the lambda 2 and lambda 3 chains were about equally abundant. Similar lambda subtype frequencies were found in the anti-DNP Ab produced by the hybridoma made with spleen cells from the same immunized mice. In the anti-DNP Ab elicited by DNP-CGG and in the anti-NP Ab elicited by NP-CGG the different lambda subtype frequencies (lambda 1/lambda 2 + lambda 3 = ca. 1.0-1.5 in anti-DNP and ca. 30 in anti-NP) were unaffected by immunizing mice with each of these antigens alone or with a mixture of the two. This finding, though preliminary, suggests that isotype-specific regulatory T cells are not responsible for the markedly different lambda subtype frequencies in anti-DNP and anti-NP Ab.

Restricted association of V and J-C gene segments for mouse lambda chains.

The frequencies of diverse rearrangements of variable (V)lambda to joining (J)lambda gene segments were examined by Southern blot hybridization in 30 murine B-cell lines, each producing an immunoglobulin lambda light chain of known subtype (lambda 1, lambda 2, or lambda 3). For 11 out of 12 lambda 1 chains, the rearrangement was V lambda 1----J lambda 1; for 9 out of 9 lambda 2 chains, it was V lambda 2----J lambda 2; and for 8 out of 9 lambda 3 chains, it was V lambda 1----J lambda 3. Similar results were obtained by considering the partial or complete sequences at the amino acid or cDNA level of 44 other lambda chains (24 previously described): for 43 of these chains the rearranged V-J gene segments were evidently V lambda 1-J lambda 1 for 28 lambda 1 chains, V lambda 2-J lambda 2 for 10 lambda 2 chains, and V lambda 1-J lambda 3 for 5 lambda 3 chains. Of the combined total of 74 chains there were 3 with unusual V lambda rearrangements, all involving the V lambda 2 gene segment: for 2 of these unusual chains, the encoding segments were V lambda 2-J lambda 1-C lambda 1 and for one they were V lambda 2-J lambda 3-C lambda 3. Thus, the results for all 74 lambda chains show that, in contrast to the apparently unrestricted V kappa----J kappa rearrangements for kappa chains, for each of the 3 murine lambda-chain subtypes V-J recombination is severely restricted: the V lambda gene segment expressed in lambda 1 and lambda 3 chains was nearly always V lambda 1 (95% and 93%, respectively), whereas in lambda 2 chains it was without exception V lambda 2 (19 out of 19 chains). Therefore V lambda-J lambda combinatorial variation is not a significant source of amino acid sequence diversity of lambda chains of inbred mice. If the order of the lambda gene segments is 5' V lambda 2-J lambda 2C lambda 2J lambda 4C lambda 4-V lambda 1-J lambda 3C lambda 3J lambda 1C lambda 1 3', as suggested previously and by the present findings, it appears that (i) when a V lambda gene segment rearranges in a developing B cell it ordinarily recombines with a J lambda gene segment in the nearest downstream (3') cluster of J lambda C lambda segments, and (ii) V lambda rearrangement to the upstream (5') cluster is very rare and possibly may not take place at all.