Rivka Adar

An autonomous molecular computer for logical control of gene expression

Early biomolecular computer research focused on laboratory-scale, human-operated computers for complex computational problems. Recently, simple molecular-scale autonomous programmable computers were demonstrated allowing both input and output information to be in molecular form. Such computers, using biological molecules as input data and biologically active molecules as outputs, could produce a system for ‘logical’ control of biological processes. Here we describe an autonomous biomolecular computer that, at least in vitro, logically analyses the levels of messenger RNA species, and in response produces a molecule capable of affecting levels of gene expression. The computer operates at a concentration of close to a trillion computers per microlitre and consists of three programmable modules: a computation module, that is, a stochastic molecular automaton; an input module, by which specific mRNA levels or point mutations regulate software molecule concentrations, and hence automaton transition probabilities; and an output module, capable of controlled release of a short single-stranded DNA molecule. This approach might be applied in vivo to biochemical sensing, genetic engineering and even medical diagnosis and treatment. As a proof of principle we programmed the computer to identify and analyse mRNA of disease-related genes associated with models of small-cell lung cancer and prostate cancer, and to produce a single-stranded DNA molecule modelled after an anticancer drug.

Programmable and autonomous computing machine made of biomolecules.

Devices that convert information from one form into another according to a definite procedure are known as automata. One such hypothetical device is the universal Turing machine, which stimulated work leading to the development of modern computers. The Turing machine and its special cases, including finite automata, operate by scanning a data tape, whose striking analogy to information-encoding biopolymers inspired several designs for molecular DNA computers. Laboratory-scale computing using DNA and human-assisted protocols has been demonstrated, but the realization of computing devices operating autonomously on the molecular scale remains rare. Here we describe a programmable finite automaton comprising DNA and DNA-manipulating enzymes that solves computational problems autonomously. The automatons hardware consists of a restriction nuclease and ligase, the software and input are encoded by double-stranded DNA, and programming amounts to choosing appropriate software molecules. Upon mixing solutions containing these components, the automaton processes the input molecule via a cascade of restriction, hybridization and ligation cycles, producing a detectable output molecule that encodes the automatons final state, and thus the computational result. In our implementation 1012 automata sharing the same software run independently and in parallel on inputs (which could, in principle, be distinct) in 120 μl solution at room temperature at a combined rate of 109 transitions per second with a transition fidelity greater than 99.8%, consuming less than 10-10 W.

Gly369Cys mutation in mouse FGFR3 causes achondroplasia by affecting both chondrogenesis and osteogenesis

Missense mutations in fibroblast growth factor receptor 3 (FGFR3) result in several human skeletal dysplasias, including the most common form of dwarfism, achondroplasia. Here we show that a glycine-to-cysteine substitution at position 375 (Gly375Cys) in human FGFR3 causes ligand-independent dimerization and phosphorylation of FGFR3 and that the equivalent substitution at position 369 (Gly369Cys) in mouse FGFR3 causes dwarfism with features mimicking human achondroplasia. Accordingly, homozygous mice were more severely affected than heterozygotes. The resulting mutant mice exhibited macrocephaly and shortened limbs due to retarded endochondral bone growth and premature closure of cranial base synchondroses. Compared with their wild-type littermates, mutant mice growth plates shared an expanded resting zone and narrowed proliferating and hypertrophic zones, which is correlated with the activation of Stat proteins and upregulation of cell-cycle inhibitors. Reduced bone density is accompanied by increased activity of osteoclasts and upregulation of genes that are related to osteoblast differentiation, including osteopontin, osteonectin, and osteocalcin. These data reveal an essential role for FGF/FGFR3 signals in both chondrogenesis and osteogenesis during endochondral ossification.

DNA molecule provides a computing machine with both data and fuel

The unique properties of DNA make it a fundamental building block in the fields of supramolecular chemistry, nanotechnology, nano-circuits, molecular switches, molecular devices, and molecular computing. In our recently introduced autonomous molecular automaton, DNA molecules serve as input, output, and software, and the hardware consists of DNA restriction and ligation enzymes using ATP as fuel. In addition to information, DNA stores energy, available on hybridization of complementary strands or hydrolysis of its phosphodiester backbone. Here we show that a single DNA molecule can provide both the input data and all of the necessary fuel for a molecular automaton. Each computational step of the automaton consists of a reversible software molecule/input molecule hybridization followed by an irreversible software-directed cleavage of the input molecule, which drives the computation forward by increasing entropy and releasing heat. The cleavage uses a hitherto unknown capability of the restriction enzyme FokI, which serves as the hardware, to operate on a noncovalent software/input hybrid. In the previous automaton, software/input ligation consumed one software molecule and two ATP molecules per step. As ligation is not performed in this automaton, a fixed amount of software and hardware molecules can, in principle, process any input molecule of any length without external energy supply. Our experiments demonstrate 3 × 1012 automata per μl performing 6.6 × 1010 transitions per second per μl with transition fidelity of 99.9%, dissipating about 5 × 10−9 W/μl as heat at ambient temperature.

Human Combinatorial Fab Library Yielding Specific and Functional Antibodies against the Human Fibroblast Growth Factor Receptor 3

The human combinatorial antibody library Fab 1 (HuCAL®-Fab 1) was generated by transferring the heavy and light chain variable regions from the previously constructed single-chain Fv library (Knappik, A., Ge, L., Honegger, A., Pack, P., Fischer, M., Wellnhofer, G., Hoess, A., Wölle, J., Plückthun, A., and Virnekäs, B. (2000) J. Mol. Biol. 296, 57–86), diversified in both complementarity-determining regions 3 into a novel Fab display vector, yielding 2.1 × 1010 different antibody fragments. The modularity has been retained in the Fab display and screening plasmids, ensuring rapid conversion into various antibody formats as well as antibody optimization using prebuilt maturation cassettes. HuCAL®-Fab 1 was challenged against the human fibroblast growth factor receptor 3, a potential therapeutic antibody target, against which, to the best of our knowledge, no functional antibodies could be generated so far. A unique screening mode was designed utilizing recombinant functional proteins and cell lines differentially expressing fibroblast growth factor receptor isoforms diversified in expression and receptor dependence. Specific Fab fragments with subnanomolar affinities were isolated by selection without any maturation steps as determined by fluorescence flow cytometry. Some of the selected Fab fragments completely inhibit target-mediated cell proliferation, rendering them the first monoclonal antibodies against fibroblast growth factor receptors having significant function blocking activity. This study validates HuCAL®-Fab 1 as a valuable source for the generation of target-specific antibodies for therapeutic applications.

Differential Activation of Cysteine-Substitution Mutants of Fibroblast Growth Factor Receptor 3 Is Determined by Cysteine Localization

Various human skeletal disorders are thought to be caused by mutations in fibroblast growth factor receptor 3 (FGFR3). These result in chronic FGFR3 hyperactivation and inhibition of bone growth. One such disorder, thanatophoric dysplasia, the most common form of sporadic, lethal dwarfism, is associated frequently with cysteine substitutions (G370C, S371C, and Y373C) in the extracellular juxtamembrane region of the receptor. These mutations have been suggested to induce disulfide‐mediated receptor dimerization and constitutive activation. An adjacent cysteine substitution (G375C) leads to a less severe form of human dwarfism, achondroplasia, suggesting that the intensity of FGFR3 activation by these cross‐links may be position dependent. To test this hypothesis, we have sequentially replaced each amino acid at positions 370‐375 of FGFR3 with cysteine. Expression of each of these mutant forms in 293T cells led to their spontaneous, ligand‐independent dimerization and increased basal phosphorylation. Wild‐type (WT) FGFR3 became dimerized and phosphorylated only on FGF stimulation. Among the mutants, only two (G370C and S371C) caused high basal phosphorylation with significantly increased constitutive levels of mitogen‐activated protein kinase (MAPK) phosphorylation and c‐fos transcription. This activity was probably caused by mutant homodimer pairs, because WT‐mutant heterodimers were observed only in the presence, but not in the absence, of FGF1. The high spontaneous activity of the mutants in positions 370‐371, unlike those in 372‐375, affirms their known involvement with thanatophoric dysplasia. We conclude that the G370C and S371C mutant receptors spontaneously dimerize in the correct spatial orientation required for effective signal transduction, whereas the 372‐5 mutants, like the WT receptor, may achieve this orientation only on ligand binding.

FGF receptors ubiquitylation: dependence on tyrosine kinase activity and role in downregulation

A crucial aspect of ligand‐mediated receptor activation and shut‐down is receptor internalization and degradation. Here we compared the ubiquitylation of either wild type or a K508A ‘kinase‐dead’ mutant of fibroblast growth factor receptor 3 (FGFR3) with that of its naturally occurring overactive mutants, G380R as in achondroplasia, or K650E involved in thanatophoric dysplasia. Fibroblast growth factor receptors ubiquitylation was found to be directly proportional to their intrinsic tyrosine kinase activity, both of which could be blocked using kinase inhibitors. Despite excessive ubiquitylation, both overactive mutants failed to be efficiently degraded, even when challenged with ligand or overexpression of c‐Cbl, a putative E3 ligase. We conclude that phosphorylation is essential for FGFR3 ubiquitylation, but is not sufficient to induce downregulation of its internalization resistant mutants.

Recursive construction of perfect DNA molecules from imperfect oligonucleotides.

Making faultless complex objects from potentially faulty building blocks is a fundamental challenge in computer engineering, nanotechnology and synthetic biology. Here, we show for the first time how recursion can be used to address this challenge and demonstrate a recursive procedure that constructs error‐free DNA molecules and their libraries from error‐prone oligonucleotides. Divide and Conquer (D&C), the quintessential recursive problem‐solving technique, is applied in silico to divide the target DNA sequence into overlapping oligonucleotides short enough to be synthesized directly, albeit with errors; error‐prone oligonucleotides are recursively combined in vitro, forming error‐prone DNA molecules; error‐free fragments of these molecules are then identified, extracted and used as new, typically longer and more accurate, inputs to another iteration of the recursive construction procedure; the entire process repeats until an error‐free target molecule is formed. Our recursive construction procedure surpasses existing methods for de novo DNA synthesis in speed, precision, amenability to automation, ease of combining synthetic and natural DNA fragments, and ability to construct designer DNA libraries. It thus provides a novel and robust foundation for the design and construction of synthetic biological molecules and organisms.

Cell lineage analysis of acute leukemia relapse uncovers the role of replication-rate heterogeneity and microsatellite instability

Human cancers display substantial intratumoral genetic heterogeneity, which facilitates tumor survival under changing microenvironmental conditions. Tumor substructure and its effect on disease progression and relapse are incompletely understood. In the present study, a high-throughput method that uses neutral somatic mutations accumulated in individual cells to reconstruct cell lineage trees was applied to hundreds of cells of human acute leukemia harvested from multiple patients at diagnosis and at relapse. The reconstructed cell lineage trees of patients with acute myeloid leukemia showed that leukemia cells at relapse were shallow (divide rarely) compared with cells at diagnosis and were closely related to their stem cell subpopulation, implying that in these instances relapse might have originated from rarely dividing stem cells. In contrast, among patients with acute lymphoid leukemia, no differences in cell depth were observed between diagnosis and relapse. In one case of chronic myeloid leukemia, at blast crisis, most of the cells at relapse were mismatch-repair deficient. In almost all leukemia cases, > 1 lineage was observed at relapse, indicating that diverse mechanisms can promote relapse in the same patient. In conclusion, diverse relapse mechanisms can be observed by systematic reconstruction of cell lineage trees of patients with leukemia.

Cell Lineage Analysis of the Mammalian Female Germline

Fundamental aspects of embryonic and post-natal development, including maintenance of the mammalian female germline, are largely unknown. Here we employ a retrospective, phylogenetic-based method for reconstructing cell lineage trees utilizing somatic mutations accumulated in microsatellites, to study female germline dynamics in mice. Reconstructed cell lineage trees can be used to estimate lineage relationships between different cell types, as well as cell depth (number of cell divisions since the zygote). We show that, in the reconstructed mouse cell lineage trees, oocytes form clusters that are separate from hematopoietic and mesenchymal stem cells, both in young and old mice, indicating that these populations belong to distinct lineages. Furthermore, while cumulus cells sampled from different ovarian follicles are distinctly clustered on the reconstructed trees, oocytes from the left and right ovaries are not, suggesting a mixing of their progenitor pools. We also observed an increase in oocyte depth with mouse age, which can be explained either by depth-guided selection of oocytes for ovulation or by post-natal renewal. Overall, our study sheds light on substantial novel aspects of female germline preservation and development.