Pattern classification by adaptive machines.

Man's intelligent behavior is due in part to his ability to select, classify, and abstract significant information reaching him from his environment by way of his senses. This function, pattern recognition, has become a major focus of research by scientists working in the field of artificial intelligence. At the lowest level, pattern recognition reduces to pattern classification, which consists of the separation, into desired classes, of groups of objects, sounds, odors, events, properties, and the like; the separations are based on sets of measurements made on the entities being classified. The pattern classifier is composed of a data filter and a categorizer. The data filter selects the distinguishing features and represents them as sets of real numbers; each set is termed a pattern. The categorizer assigns each pattern to one of several desired classes. Patterns can be represented geometrically as points in an n-dimensional space; the n coordinates of each point are the numerical values of the features selected to represent the pattern. A pattern classification system separates an n-dimensional space into regions, each of which ideally contains points of only one class. One method to effect this separation is by means of ldquo;trainablerdquo; categorizers-major components of adaptive machines. They consist of networks whose internal parameters are varied according to a set of fixed rules during a training cycle. A statistically large sample of known patterns are presented, one at a time, to the networks; internal corrections are made each time a pattern is erroneously classified. Classifica-tion performance tends to improve as the set of known patterns is cycled repetitively through the machine. Finally, the adequacy of adaptation is tested by a separate set of similar patterns which have not been used in the training process. A number of different machine organizations and training rules have been developed and are being applied successfully to numerous classification problems. More difficult recognition problems requiring the aid of logioal tests and analysis, search and association, use the digital computer programmed to supplement the functions of the adaptive classifier.

A cDNA for a protein that interacts with the human immunodeficiency virus Tat transactivator.

The human immunodeficiency virus (HIV) tat protein (Tat) is a positive regulator of virus gene expression and replication. Biotinylated Tat was used as a probe to screen a lambda gt11 fusion protein library, and a complementary DNA encoding a protein that interacts with Tat was cloned. Expression of this protein, designated TBP-1 (for Tat binding protein-1), was observed in a variety of cell lines, with expression being highest in human cells. TBP-1 was localized predominantly in the nucleus, which is consistent with the nuclear localization of Tat. In cotransfection experiments, expression of TBP-1 was able to specifically suppress Tat-mediated transactivation. The strategy described may be useful for direct identification and cloning of genes encoding proteins that associate with other proteins to modulate their activity in a positive or negative fashion.

Trans-acting transcriptional activation of the long terminal repeat of human T lymphotropic viruses in infected cells.

The transcription initiation signals for retroviruses lie within the long terminal repeat (LTR) sequences that flank the integrated provirus. Two subtypes of human T lymphotropic virus (HTLV) are associated with different disease phenotypes. In this article it is shown that marked differences exist in the ability of LTR sequences of these subtypes to function as transcriptional elements in differentiated cell types. It is also shown that trans-acting regulatory factors present in HTLV-infected cells stimulate gene expression directed by these LTR sequences in a type-specific manner. These results have implications for understanding the diverse biological effects of HTLV infection.

Secondary structure is the major determinant for interaction of HIV rev protein with RNA.

A region in the human immunodeficiency virus (HIV) env message, with the potential to form a complex secondary structure (designated RRE), interacts with the rev protein (Rev). This interaction is believed to mediate export of HIV structural messenger RNAs from the nucleus to the cytoplasm. In this report the regions essential for Rev interaction with the RRE are further characterized and the functional significance of Rev-RRE interaction in vivo is examined. A single hairpin loop structure within the RRE was found to be a primary determinant for Rev binding in vitro and Rev response in vivo. Maintenance of secondary structure, rather than primary nucleotide sequence alone, appeared to be necessary for Rev-RNA interaction, which distinguishes it from the mechanism for cis-acting elements in DNA. Limited changes within the 200 nucleotides, which preserved the proper RRE conformational structure, were well tolerated for Rev binding and function. Thus, variation among the RRE elements present in the diverse HIV isolates would have little, if any, effect on Rev responsiveness.

Trans activation of the bovine leukemia virus long terminal repeat in BLV-infected cells.

The transcription initiation signals for retroviruses lie within the long terminal repeat (LTR) sequences that flank the integrated provirus. This study shows that factors present in cells infected with bovine leukemia virus (BLV) mediate transcriptional trans activation of the BLV LTR. This phenomenon is similar to that reported for the human T-cell leukemia virus LTR and establishes both structural and functional criteria for inclusion of BLV and human T-cell leukemia viruses in the same family of transforming retroviruses.

Identification of a naturally occurring transforming variant of the p65 subunit of NF-kappa B.

Transcription factor NF-kappa B comprises two proteins, p50 and p65, that have sequence similarity to the v-rel oncogene. In primary hematopoietic cell populations an alternatively spliced form of NF-kappa B p65 mRNA was observed that encoded a protein designated p65 delta. Expression of the p65 delta cDNA in Rat-1 fibroblasts resulted in focus formation, anchorage-independent growth in soft agar, and tumor formation in athymic nude mice, effects not obtained with expression of p65 or a p65 delta mutant that contains a disruption within the transcriptional activation domain. Thus, p65 delta, which associated weakly and interfered with DNA binding by p65, may sequester an essential limiting regulatory factor or factors required for NF-kappa B function.

Cellular transcription factors and regulation of IL-2 receptor gene expression by HTLV-I tax gene product.

Expression of the interleukin-2 receptor (IL-2R alpha) gene is activated by the transcriptional activator protein, Tax (previously referred to as the tat gene product), encoded by the human T-cell leukemia virus (HTLV-I). Multiple protein binding sites for specific DNA-protein interactions were identified over the upstream IL-2R alpha transcriptional regulatory sequences. However, only one region, which includes the sequence motif GGGGAATCTCCC, was required for activation by both the tax gene product and mitogenic stimulation. Remarkably, this sequence also bound the nuclear factor NF kappa B, which is important for induction of kappa-immunoglobulin gene expression. A model is presented whereby regulation of cellular gene expression by the HTLV-I tax gene product occurs via an indirect mechanism that may involve a post-translational modification of preexistent cellular transcription factors.

Isolation of a rel-related human cDNA that potentially encodes the 65-kD subunit of NF-kappa B.

A DNA probe that spanned a domain conserved among the proto-oncogene c-rel, the Drosophila morphogen dorsal, and the p50 DNA binding subunit of NF-kappa B was generated from Jurkat T cell complementary DNA with the polymerase chain reaction (PCR) and degenerate oligonucleotides. This probe was used to identify a rel-related complementary DNA that hybridized to a 2.6-kilobase messenger RNA present in human T and B lymphocytes. In vitro transcription and translation of the complementary DNA resulted in the synthesis of a protein with an apparent molecular size of 65 kilodaltons (kD). The translated protein showed weak DNA binding with a specificity for the kappa B binding motif. This protein-DNA complex comigrated with the complex obtained with the purified human p65 NF-kappa B subunit and binding was inhibited by I kappa B-alpha and -beta proteins. In addition, the 65-kD protein associated with the p50 subunit of NF-kappa B and the kappa B probe to form a complex with the same electrophoretic mobility as the NF-kappa B-DNA complex. Therefore the rel-related 65-kD protein may represent the p65 subunit of the active NF-kappa B transcription factor complex.

Isolation of a rel-related human cDNA that potentially encodes the 65-kD subunit of NF-kappa B.

The sequence reported in our 22 March 1991 report "Isolation of a rel-related human cDNA that potentially encodes the 65-kD subunit of NFkappaB" [Science 251, 1490 (1991)], contained some errors. Resequencing under strong denaturing conditions revealed three insertions at nucleotide positions 1194, 1212, and 1220, which changed the AA sequence from RSAR-PRLGP to QISQASALAP (residues 372 to 380), thus accounting for some of the divergence in this region. A corrected sequence has been sent to GenBank.

Mutation of a mutL homolog in hereditary colon cancer.

Some cases of hereditary nonpolyposis colorectal cancer (HNPCC) are due to alterations in a mutS-related mismatch repair gene. A search of a large database of expressed sequence tags derived from random complementary DNA clones revealed three additional human mismatch repair genes, all related to the bacterial mutL gene. One of these genes (hMLH1) resides on chromosome 3p21, within 1 centimorgan of markers previously linked to cancer susceptibility in HNPCC kindreds. Mutations of hMLH1 that would disrupt the gene product were identified in such kindreds, demonstrating that this gene is responsible for the disease. These results suggest that defects in any of several mismatch repair genes can cause HNPCC.

Trans-activator gene of HTLV-II induces IL-2 receptor and IL-2 cellular gene expression.

The human T-lymphotropic viruses types I and II (HTLV-I and -II) have been etiologically linked with certain T-cell leukemias and lymphomas that characteristically display membrane receptors for interleukin-2. The relation of these viruses to this growth factor receptor has remained unexplained. It is demonstrated here that introduction of the trans-activator (tat) gene of HTLV-II into the Jurkat T-lymphoid cell line results in the induction of both interleukin-2 receptor and interleukin-2 gene expression. The coexpression of these cellular genes may play a role in the altering T-cell growth following retroviral infection.

Major glycoprotein antigens that induce antibodies in AIDS patients are encoded by HTLV-III.

Antibodies from the serum of patients with the acquired immune deficiency syndrome (AIDS) or with the AIDS-related complex and from the serum of seropositive healthy homosexuals, recognize two major glycoproteins in cells infected with human T-cell lymphotropic virus type III (HTLV III). These glycoproteins, gp160 and gp120, are encoded by the 2.5-kilobase open reading frame located in the 3' end of the HTLV-III genome, as determined by amino terminus sequence analysis of the radiolabeled forms of these proteins. It is hypothesized that gp160 and gp120 represent the major species of virus-encoded envelope gene products for HTLV-III.

Item reduction of the voice handicap index based on the original version and on European translations.

OBJECTIVE: Constructing an internationally applicable short-scale of the Voice Handicap Index (VHI). METHODS: Subjects were 1,052 patients with 5 different types of voice disorder groups from Belgium, France, Sweden, Germany, Italy, The Netherlands, Portugal, and the USA. Different 9- and 12-item subsets were selected from the 30 VHI items using (1) the first factor of an unrotated factor analysis (narrow range subsets) and (2) the first three factors after promax rotation (broad range subsets). Country-specific subsets were selected to test deviations from the international subsets. For all subsets, reliability was investigated using Cronbach's alphas and correlations with the total VHI. Validity was investigated using regression on voice disorder groups. All analyses were performed for the total and for all country-specific subject samples. RESULTS: Reliability was high for all item subsets. It was lower for the international compared to the country-specific subsets and for the broad range compared to the narrow range subsets. Validity was best for the broad range subsets. Validity was better for the international than for the country-specific subsets. For all statistics the 12-item subsets were not essentially better than the 9-item subsets. CONCLUSION: The international broad range 9-item subset forms a scale which approximates well the total VHI.

Validation of the voice handicap index by assessing equivalence of European translations.

OBJECTIVE: We aimed to assess the equivalence of translations of the Voice Handicap Index (VHI). PATIENTS AND METHODS: Confirmatory factor analysis was used to assess equivalence of the US version and several translations including (1) Dutch, (2) Flemish Dutch (Belgium), (3) UK English, (4) French, (5) German, (6) Italian, (7) Portuguese and (8) Swedish. VHI questionnaires were gathered from 1,281 subjects. Patients were classified into 11 voice lesion categories. Patients with incomplete response (4%) and patients within voice lesion categories with small numbers were excluded from further analyses, leaving a cohort of 1,052 patients from 8 countries. RESULTS: The internal consistency of the VHI proved to be good. Confirmatory factor analysis across countries revealed that a 3-factor fixed measurement model best fitted the data; the 3 subscales appeared to highly intercorrelated, especially in the US data. The underlying structure of the VHI was also equivalent regarding various voice lesions, but distinct groups were recognized with respect to the height of the VHI scores, indicating that various voice lesions lead to a diversity of voice problems in daily life. CONCLUSION: The US VHI and the translations appeared to be equivalent, which means that the results from studies from the various included countries can be compared.

Regulation of HIV gene expression by RNA-protein interactions.

Human immunodeficiency virus (HIV) gene expression is tightly controlled through the interaction of trans-acting regulatory proteins with the many cis-acting elements present in viral DNA and RNA. Two proteins encoded by HIV, referred to as Tat and Rev, are essential positive regulators of gene expression. Recent work shows that these proteins control HIV gene expression through interaction with RNA target elements present within the 5' untranslated leader sequence and envelope gene, respectively. There is evidence that these interactions in themselves are not sufficient to confer regulation without the presence of additional host cell factors.

Transgenic dissection of HIV genes involved in lymphoid depletion.

Transgenic mice carrying an HIV provirus, with selective deletion of all three structural genes, developed extensive lymphoid depletion which was detected not only in the spleen and lymph nodes but also in the thymus. Mice with a high level of HIV gene expression developed acute disease which resulted in premature death, and mice with a low level of viral transcripts developed chronic disease with long-term survival. Neither HIV replication nor the envelope glycoprotein (gp120) was required for T cell depletion. Despite abundant viral gene expression early in life, cell death did not become evident until about the time of full lymphoid maturation, suggesting that thymopoiesis was not affected. The more mature T cells in the peripheral lymphoid organs and in the thymic medulla were less sensitive to the apoptotic process than the immature T cells in the thymic cortex. Gradual depletion of the T cell compartment in the peripheral lymphoid organs was intimately accompanied by the reciprocal expansion of the B cell compartment, resulting in the almost complete replacement of T lymphocytes with B immunoblasts in lymph nodes. Unlike T cells, which showed abundant HIV gene expression, B cells did not. The transgenic approach may help identify the HIV nonstructural gene(s) responsible for immune deficiency and help facilitate dissection of its role in inducing apoptosis.

Vascular cell adhesion molecule-1 (VCAM-1) gene transcription and expression are regulated through an antioxidant-sensitive mechanism in human vascular endothelial cells.

Oxidative stress and expression of the vascular cell adhesion molecule-1 (VCAM-1) on vascular endothelial cells are early features in the pathogenesis of atherosclerosis and other inflammatory diseases. Regulation of VCAM-1 gene expression may be coupled to oxidative stress through specific reduction-oxidation (redox) sensitive transcriptional or posttranscriptional regulatory factors. In cultured human umbilical vein endothelial (HUVE) cells, the cytokine interleukin 1 beta (IL-1 beta) activated VCAM-1 gene expression through a mechanism that was repressed approximately 90% by the antioxidants pyrrolidine dithiocarbamate (PDTC) and N-acetylcysteine (NAC). Furthermore, PDTC selectively inhibited the induction of VCAM-1, but not intercellular adhesion molecule-1 (ICAM-1), mRNA and protein accumulation by the cytokine tumor necrosis factor-alpha (TNF alpha) as well as the noncytokines bacterial endotoxin lipopolysaccharide (LPS) and double-stranded RNA, poly(I:C) (PIC). PDTC also markedly attenuated TNF alpha induction of VCAM-1-mediated cellular adhesion. In a distinct pattern, PDTC partially inhibited E-selectin gene expression in response to TNF alpha but not to LPS, IL-1 beta, or PIC. TNF alpha and LPS-mediated transcriptional activation of the human VCAM-1 promoter through NF-kappa B-like DNA enhancer elements and associated NF-kappa B-like DNA binding proteins was inhibited by PDTC. These studies suggest a molecular linkage between an antioxidant sensitive transcriptional regulatory mechanism and VCAM-1 gene expression that expands on the notion of oxidative stress as an important regulatory signal in the pathogenesis of atherosclerosis.

NF-kappa B subunit-specific regulation of the interleukin-8 promoter.

Interleukin-8 (IL-8), a chemotactic cytokine for T lymphocytes and neutrophils, is induced in several cell types by a variety of stimuli including the inflammatory cytokines IL-1 and tumor necrosis factor alpha TNF-alpha. Several cis elements, including a binding site for the inducible transcription factor NF-kappa B, have been identified in the regulatory region of the IL-8 gene. We have examined the ability of various NF-kappa B subunits to bind to, and activate transcription from, the IL-8 promoter. A nuclear complex was induced in phorbol myristate acetate-treated Jurkat T cells which bound specifically to the kappa B site of the IL-8 promoter and was inhibited by addition of purified I kappa B alpha to the reaction mixture. Only antibody to RelA (p65), but not to NFKB1 (p50), NFKB2 (p50B), c-Rel, or RelB was able to abolish binding, suggesting that RelA is a major component in these kappa B binding complexes. Gel mobility shift analysis with in vitro-translated and purified proteins indicated that whereas the kappa B element in the human immunodeficiency virus type 1 long terminal repeat bound to all members of the kappa B/Rel family examined, the IL-8 kappa B site bound only to RelA and to c-Rel and NFKB2 homodimers, but not to NFKB1 homodimers or heterodimers of NFKB1-RelA. Transient transfection analysis demonstrated a kappa B-dependent expression of the IL-8 promoter in a human fibrosarcoma cell line (8387) and in Jurkat T lymphocytes. Cotransfection with various NF-kappa B subunits indicated that RelA and c-Rel, but neither NFKB1 nor heterodimeric NFKB1-RelA, was able to activate transcription from the IL-8 promoter. Furthermore, cotransfection of NFKB1 and RelA, although able to support activation from the human immunodeficiency virus type 1 long terminal repeat, failed to activate expression from the IL-8 promoter. Antisense oligonucleotides to RelA, but not NFKB1, inhibited phorbol myristate acetate-induced IL-8 production in Jurkat T lymphocytes. These data demonstrate the differential ability of members of the kappa B/Rel family to bind to, and activate transcription from, the IL-8 promoter. Furthermore, while providing a novel example of a kappa B-regulated promoter in which the classical NF-kappa B complex is unable to activate transcription from the kappa B element, these data provide direct evidence for the role of RelA in regulation of IL-8 gene expression.

Conservation of transcriptional activation functions of the NF-kappa B p50 and p65 subunits in mammalian cells and Saccharomyces cerevisiae.

The NF-kappa B transcription factor complex is composed of a 50-kDa (p50) and a 65-kDa (p65) subunit. Both subunits bind to similar DNA motifs and elicit transcriptional activation as either homo- or heterodimers. By using chimeric proteins that contain the DNA binding domain of the yeast transcriptional activator GAL4 and subdomains of p65, three distinct transcriptional activation domains were identified. One domain was localized to a region of 42 amino acids containing a potential leucin zipper structure, consistent with earlier reports. Two other domains, both acidic and rich in prolines, were also identified. Of perhaps more significance, the same minimal activation domains that were functional in mammalian cells were also functional in the yeast Saccharomyces cerevisiae. Coexpression of the NF-kappa B inhibitory molecule, I kappa B, reduced the transcriptional activity of p65 significantly, suggesting the ability of I kappa B to function in a similar manner in S. cerevisiae. Surprisingly, while the conserved rel homology domain of p65 demonstrated no transcriptional activity in either mammalian cells or S. cerevisiae, the corresponding domain in p50 was a strong transcriptional activator in S. cerevisiae. The observation that similar domains elicit transcriptional activation in mammalian cells and S. cerevisiae demonstrates strong conservation of the transcriptional machinery required for NF-kappa B function and provides a powerful genetic system to study the transcriptional mechanisms of these proteins.

Selection of optimal kappa B/Rel DNA-binding motifs: interaction of both subunits of NF-kappa B with DNA is required for transcriptional activation.

Analysis of the p50 and p65 subunits of the NF-kappa B transcription factor complex has revealed that both proteins can interact with related DNA sequences through either homo- or heterodimer formation. In addition, the product of the proto-oncogene c-rel can bind to similar DNA motifs by itself or as a heterodimer with p50 or p65. However, these studies have used a limited number of known kappa B DNA motifs, and the question of the optimal DNA sequences preferred by each homodimer has not been addressed. Using purified recombinant p50, p65, and c-Rel proteins, optimal DNA-binding motifs were selected from a pool of random oligonucleotides. Alignment of the selected sequences allowed us to predict a consensus sequence for binding of the individual homodimeric Rel-related proteins, and DNA-protein binding analysis of the selected DNA sequences revealed sequence specificity of the proteins. Contrary to previous assumptions, we observed that p65 homodimers can interact with a subset of DNA sequences not recognized by p50 homodimers. Differential binding affinities were also obtained with p50- and c-Rel-selected sequences. Using either a p50- or p65-selected kappa B motif, which displayed differential binding with respect to the other protein, little to no binding was observed with the heterodimeric NF-kappa B complex. Similarly, in transfection experiments in which the selective kappa B binding sites were used to drive the expression of a chloramphenicol acetyltransferase reporter construct, the p65- and p50-selected motifs were activated only in the presence of p65 and p50/65 (a chimeric protein with the p50 DNA binding domain and p65 activation domain) expression vectors, respectively, and neither demonstrated a significant response to stimuli that induce NF-kappa B activity. These findings demonstrate that interaction of both subunits of the heterodimeric NF-kappa B complex with DNA is required for DNA binding and transcriptional activation and suggest that transcriptional activation mediated by the individual rel-related proteins will differ dramatically, depending on the specific kappa B motifs present.