Image recognition: Visual grouping, recognition, and learning

Vision extracts useful information from images. Reconstructing the three-dimensional structure of our environment and recognizing the objects that populate it are among the most important functions of our visual system. Computer vision researchers study the computational principles of vision and aim at designing algorithms that reproduce these functions. Vision is difficult: the same scene may give rise to very different images depending on illumination and viewpoint. Typically, an astronomical number of hypotheses exist that in principle have to be analyzed to infer a correct scene description. Moreover, image information might be extracted at different levels of spatial and logical resolution dependent on the image processing task. Knowledge of the world allows the visual system to limit the amount of ambiguity and to greatly simplify visual computations. We discuss how simple properties of the world are captured by the Gestalt rules of grouping, how the visual system may learn and organize models of objects for recognition, and how one may control the complexity of the description that the visual system computes.

Dual HLA class I and class II restricted recognition of alloreactive T lymphocytes mediated by a single T cell receptor complex

The alloreactive human T cell clone MBM15 was found to exhibit dual specificity recognizing both an antigen in the context of the HLA class I A2 molecule and an antigen in the context of the HLA class II DR1. We demonstrated that the dual reactivity that was mediated via a single clonal T cell population depended on specific peptide binding. For complete recognition of the HLA-A2-restricted specificity the interaction of CD8 with HLA class I is essential. Interestingly, interaction of the CD8 molecule with HLA class I contributed to the HLA-DR1-restricted specificity. T cell clone MBM15 expressed two in-frame T cell receptor (TCR) Vα transcripts (Vα1 and Vα2) and one TCR Vβ transcript (Vβ13). To elucidate whether two TCR complexes were responsible for the dual recognition or one complex, cytotoxic T cells were transduced with retroviral vectors encoding the different TCR chains. Only T cells transduced with the TCR Vα1Vβ13 combination specifically recognized both the HLA-A2+ and HLA-DR1+ target cells, whereas the Vα2Vβ13 combination did not result in a TCR on the cell surface. Thus a single TCRαβ complex can have dual specificity, recognizing both a peptide in the context of HLA class I as well as a peptide in the context of HLA class II. Transactivation of T cells by an unrelated antigen in the context of HLA class II may evoke an HLA class I-specific T cell response. We propose that this finding may have major implications for immunotherapeutic interventions and insight into the development of autoimmune diseases.

A growth-constrained environment drives tumor progression in vivo

We recently have shown that selective growth of transplanted normal hepatocytes can be achieved in a setting of cell cycle block of endogenous parenchymal cells. Thus, massive proliferation of donor-derived normal hepatocytes was observed in the liver of rats previously given retrorsine (RS), a naturally occurring alkaloid that blocks proliferation of resident liver cells. In the present study, the fate of nodular hepatocytes transplanted into RS-treated or normal syngeneic recipients was followed. The dipeptidyl peptidase type IV-deficient (DPPIV−) rat model for hepatocyte transplantation was used to distinguish donor-derived cells from recipient cells. Hepatocyte nodules were chemically induced in Fischer 344, DPPIV+ rats; livers were then perfused and larger (>5 mm) nodules were separated from surrounding tissue. Cells isolated from either tissue were then injected into normal or RS-treated DPPIV− recipients. One month after transplantation, grossly visible nodules (2–3 mm) were seen in RS-treated recipients transplanted with nodular cells. They grew rapidly, occupying 80–90% of the host liver at 2 months, and progressed to hepatocellular carcinoma within 4 months. By contrast, no liver nodules developed within 6 months when nodular hepatocytes were injected into the liver of recipients not exposed to RS, although small clusters of donor-derived cells were present in these animals. Taken together, these results directly point to a fundamental role played by the host environment in modulating the growth and the progression rate of altered cells during carcinogenesis. In particular, they indicate that conditions associated with growth constraint of the host tissue can drive tumor progression in vivo.

Related homing endonucleases I-BmoI and I-TevI use different strategies to cleave homologous recognition sites

A typical homing endonuclease initiates mobility of its group I intron by recognizing DNA both upstream and downstream of the intron insertion site of intronless alleles, preventing the endonuclease from binding and cleaving its own intron-containing allele. Here, we describe a GIY-YIG family homing endonuclease, I-BmoI, that possesses an unusual recognition sequence, encompassing 1 base pair upstream but 38 base pairs downstream of the intron insertion site. I-BmoI binds intron-containing and intronless substrates with equal affinity but can nevertheless discriminate between the two for cleavage. I-BmoI is encoded by a group I intron that interrupts the thymidylate synthase (TS) gene (thyA) of Bacillus mojavensis s87-18. This intron resembles one inserted 21 nucleotides further downstream in a homologous TS gene (td) of Escherichia coli phage T4. I-TevI, the T4 td intron-encoded GIY-YIG endonuclease, is very similar to I-BmoI, but each endonuclease gene is inserted within a different position of its respective intron. Remarkably, I-TevI and I-BmoI bind a homologous stretch of TS-encoding DNA and cleave their intronless substrates in very similar positions. Our results suggest that each endonuclease has independently evolved the ability to distinguish intron-containing from intronless alleles while maintaining the same conserved recognition sequence centered on DNA-encoding active site residues of TS.

Brain activity evidence for recognition without recollection after early hippocampal damage

Amnesic patients with early and seemingly isolated hippocampal injury show relatively normal recognition memory scores. The cognitive profile of these patients raises the possibility that this recognition performance is maintained mainly by stimulus familiarity in the absence of recollection of contextual information. Here we report electrophysiological data on the status of recognition memory in one of the patients, Jon. Jon's recognition of studied words lacks the event-related potential (ERP) index of recollection, viz., an increase in the late positive component (500–700 ms), under conditions that elicit it reliably in normal subjects. On the other hand, a decrease of the ERP amplitude between 300 and 500 ms, also reliably found in normal subjects, is well preserved. This so-called N400 effect has been linked to stimulus familiarity in previous ERP studies of recognition memory. In Jon, this link is supported by the finding that his recognized and unrecognized studied words evoked topographically distinct ERP effects in the N400 time window. These data suggest that recollection is more dependent on the hippocampal formation than is familiarity, consistent with the view that the hippocampal formation plays a special role in episodic memory, for which recollection is so critical.

Drosophila Heterochromatin Protein 1 (HP1)/Origin Recognition Complex (ORC) Protein Is Associated with HP1 and ORC and Functions in Heterochromatin-induced Silencing

Heterochromatin protein 1 (HP1) is a conserved component of the highly compact chromatin of higher eukaryotic centromeres and telomeres. Cytogenetic experiments in Drosophila have shown that HP1 localization into this chromatin is perturbed in mutants for the origin recognition complex (ORC) 2 subunit. ORC has a multisubunit DNA-binding activity that binds origins of DNA replication where it is required for origin firing. The DNA-binding activity of ORC is also used in the recruitment of the Sir1 protein to silence nucleation sites flanking silent copies of the mating-type genes in Saccharomyces cerevisiae. A fraction of HP1 in the maternally loaded cytoplasm of the early Drosophila embryo is associated with a multiprotein complex containing Drosophila melanogaster ORC subunits. This complex appears to be poised to function in heterochromatin assembly later in embryonic development. Here we report the identification of a novel component of this complex, the HP1/ORC-associated protein. This protein contains similarity to DNA sequence-specific HMG proteins and is shown to bind specific satellite sequences and the telomere-associated sequence in vitro. The protein is shown to have heterochromatic localization in both diploid interphase and mitotic chromosomes and polytene chromosomes. Moreover, the gene encoding HP1/ORC-associated protein was found to display reciprocal dose-dependent variegation modifier phenotypes, similar to those for mutants in HP1 and the ORC 2 subunit.

ATP bound to the origin recognition complex is important for preRC formation

The origin recognition complex (ORC) binds origins of replication and directs the assembly of a higher order protein complex at these sites. ORC binds and hydrolyzes ATP in vitro. ATP binding to the largest subunit of ORC, Orc1p, stimulates specific binding to origin DNA; however, the function of ATP hydrolysis by ORC is unknown. To address the role of ATP hydrolysis, we have generated mutants within Orc1p that are dominant lethal. At physiological ATP concentrations, these mutants are defective for ATP hydrolysis but not ATP binding in the absence of DNA. These mutants inhibit formation of the prereplicative complex when overexpressed. The dominant lethal phenotype of these mutant ORC complexes is suppressed by simultaneous overexpression of wild-type, but not mutant, Cdc6p. Our findings suggest that these hydrolysis-defective mutants inhibit growth by titrating Cdc6p away from the origin. Based on these observations, we propose that Cdc6p specifically recognizes the ATP-bound state of Orc1p and that ATP hydrolysis is coupled to preRC disassembly.

Intervention of carbohydrate recognition by proteins and nucleic acids.

Carbohydrates in biological systems are often associated with specific recognition and signaling processes leading to important biological functions and diseases. Considerable efforts have been directed toward understanding and mimicking the recognition processes and developing effective agents to control the processes. The pace of discovery research in glycobiology and development of carbohydrate-based therapeutics, however, has been relatively slow due to the lack of appropriate strategies and methods available for carbohydrate-related research. This review summarizes some of the most recent developments in the field, with particular emphasis on work from our laboratories regarding the use of chemoenzymatic strategies to tackle the carbohydrate recognition problem. Highlights include the study of selectin-carbohydrate and aminoglycoside-RNA interactions and development of agents for the intervention of these recognition processes.

Covalent HLA-B27/peptide complex induced by specific recognition of an aziridine mimic of arginine.

The class I major histocompatibility complex (MHC) glycoprotein HLA-B27 binds short peptides containing arginine at peptide position 2 (P2). The HLA-B27/peptide complex is recognized by T cells both as part of the development of the repertoire of T cells in the cellular immune system and during activation of cytotoxic T cells. Based on the three-dimensional structure of HLA-B27, we have synthesized a ligand with an aziridine-containing side chain designed to mimic arginine and to bind covalently in the arginine-specific P2 pocket of HLA-B27. Using tryptic digestion followed by mass spectrometry and amino acid sequencing, the aziridine-containing ligand is shown to alkylate specifically cysteine 67 of HLA-B27. Neither free cysteine in solution nor an exposed cysteine on a class II MHC molecule can be alkylated, showing that specific recognition between the anchor side-chain pocket of an MHC class I protein and the designed ligand (propinquity) is necessary to induce the selective covalent reaction with the MHC class I molecule.

Gene recognition via spliced sequence alignment.

Gene recognition is one of the most important problems in computational molecular biology. Previous attempts to solve this problem were based on statistics, and applications of combinatorial methods for gene recognition were almost unexplored. Recent advances in large-scale cDNA sequencing open a way toward a new approach to gene recognition that uses previously sequenced genes as a clue for recognition of newly sequenced genes. This paper describes a spliced alignment algorithm and software tool that explores all possible exon assemblies in polynomial time and finds the multiexon structure with the best fit to a related protein. Unlike other existing methods, the algorithm successfully recognizes genes even in the case of short exons or exons with unusual codon usage; we also report correct assemblies for genes with more than 10 exons. On a test sample of human genes with known mammalian relatives, the average correlation between the predicted and actual proteins was 99%. The algorithm correctly reconstructed 87% of genes and the rare discrepancies between the predicted and real exon-intron structures were caused either by short (less than 5 amino acids) initial/terminal exons or by alternative splicing. Moreover, the algorithm predicts human genes reasonably well when the homologous protein is nonvertebrate or even prokaryotic. The surprisingly good performance of the method was confirmed by extensive simulations: in particular, with target proteins at 160 accepted point mutations (PAM) (25% similarity), the correlation between the predicted and actual genes was still as high as 95%.

Involvement of Asn-293 in stereospecific agonist recognition and in activation of the beta 2-adrenergic receptor.

To investigate the molecular mechanism for stereospecific binding of agonists to beta 2-adrenergic receptors we used receptor models to identify potential binding sites for the beta-OH-group of the ligand, which defines the chiral center. Ser-165, located in transmembrane helix IV, and Asn-293, situated in the upper half of transmembrane helix VI, were identified as potential binding sites. Mutation of Ser-165 to Ala did not change the binding of either isoproterenol isomer as revealed after transient expression in human embryonic kidney (HEK)-293 cells. In contrast, a receptor mutant in which Asn-293 was replaced by Leu showed substantial loss of stereospecific isoproterenol binding. Adenylyl cyclase stimulation by this mutant after stable expression in CHO cells confirmed the substantial loss of stereospecificity for isoproterenol. In a series of agonists the loss of affinity in the Leu-293 mutant receptor was strongly correlated with the intrinsic activity of the compounds. Full agonists showed a 10-30-fold affinity loss, whereas partial agonists had almost the same affinity for both receptors. Stereospecific recognition of antagonists was unaltered in the Leu-293 mutant receptor. These data indicate a relationship between stereospecificity and intrinsic activity of agonists and suggest that Asn-293 is important for both properties of the agonist-receptor interaction.

Cleavage of lamin A by Mch2 alpha but not CPP32: multiple interleukin 1 beta-converting enzyme-related proteases with distinct substrate recognition properties are active in apoptosis.

Although proteases related to the interleukin 1 beta-converting enzyme (ICE) are known to be essential for apoptotic execution, the number of enzymes involved, their substrate specificities, and their specific roles in the characteristic biochemical and morphological changes of apoptosis are currently unknown. These questions were addressed using cloned recombinant ICE-related proteases (IRPs) and a cell-free model system for apoptosis (S/M extracts). First, we compared the substrate specificities of two recombinant human IRPs, CPP32 and Mch2 alpha. Both enzymes cleaved poly-(ADP-ribose) polymerase, albeit with different efficiencies. Mch2 alpha also cleaved recombinant and nuclear lamin A at a conserved VEID decreases NG sequence located in the middle of the coiled-coil rod domain, producing a fragment that was indistinguishable from the lamin A fragment observed in S/M extracts and in apoptotic cells. In contrast, CPP32 did not cleave lamin A. The cleavage of lamin A by Mch2 alpha and by S/M extracts was inhibited by millimolar concentrations of Zn2+, which had a minimal effect on cleavage of poly (ADP-ribose) polymerase by CPP32 and by S/M extracts. We also found that N-(acetyltyrosinylvalinyl-N epsilon-biotinyllysyl)aspartic acid [(2,6-dimethylbenzoyl)oxy]methyl ketone, which derivatizes the larger subunit of active ICE, can affinity label up to five active IRPs in S/M extracts. Together, these observations indicate that the processing of nuclear proteins in apoptosis involves multiple IRPs having distinct preferences for their apoptosis-associated substrates.

Sequence-specific recognition of cytosine C5 and adenine N6 DNA methyltransferases requires different deformations of DNA.

DNA methyltransferases modify specific cytosines and adenines within 2-6 bp recognition sequences. We used scanning force microscopy and gel shift analysis to show that M.HhaI, a cytosine C-5 DNA methyltransferase, causes only a 2 degree bend upon binding its recognition site. Our results are consistent with prior crystallographic analysis showing that the enzyme stabilizes an extrahelical base while leaving the DNA duplex otherwise unperturbed. In contrast, similar analysis of M.EcoRI, an adenine N6 DNA methyltransferase, shows an average bend angle of approximately 52 degrees. This distortion of DNA conformation by M.EcoRI is shown to be important for sequence-specific binding.

Mutational analysis of NM23-H2/NDP kinase identifies the structural domains critical to recognition of a c-myc regulatory element.

NM23-H2, a presumed regulator of tumor metastasis in humans, is a hexameric protein with both enzymatic (NDP kinase) and regulatory (transcriptional activation) activity. While the structure and catalytic mechanisms have been well characterized, the mode of DNA binding is not known. We examined this latter function in a site-directed mutational study and identified residues and domains essential for the recognition of a c-myc regulatory sequence. Three amino acids, Arg-34, Asn-69, and Lys-135, were found among 30 possibilities to be critical for DNA binding. Two of these, Asn-69 and Lys-135, are not conserved between NM23 variants differing in DNA-binding potential, suggesting that DNA recognition resides partly in nonconserved amino acids. All three DNA-binding defective mutant proteins are active enzymatically and appear to be stable hexamers, suggesting that they perform at the level of DNA recognition and that separate functional domains exist for enzyme catalysis and DNA binding. In the context of the known crystal structure of NM23-H2, the DNA-binding residues are located within distinct structural motifs in the monomer, which are exposed to the surface near the 2-fold axis of adjacent subunits in the hexamer. These findings are explained by a model in which NM23-H2 binds DNA with a combinatorial surface consisting of the "outer" face of the dimer. Chemical crosslinking data support a dimeric DNA-binding mode by NM23-H2.