rus
Issue #7-8/2018
D.Ryazantsev, E.Kuznetsov
Biomolecular computer: current state and trends of development
Biomolecular computer: current state and trends of development
Existing silicone computer technologies are coming to theoretical limit of scaling. The new breakthrough technologies with the help of which it will be possible to overcome these limits of scaling in the devices of information processing are required. One of the alternative approaches is to apply biomolecular chemical reactions for computation. Implementations of such computation based on DNA, proteins and molecular systems are reviewed.
Теги: biochemical information processing dna nanotechnology molecular computation molecular machine биоэлектроника днк-нанотехнологии молекулярные вычисления молекулярные машины
Nowadays the large-scale miniaturization of components of integrated circuits, so called Moore’s Law, continues to be the main driving force for development of different applications of handling or processing of information – inputting, transforming, storing, processing and transmitting of data. As far as the existing computer technologies based on silicone in the nearest future will come across the fundamental physical limits of scaling, it is necessary to search for conceptually new technologies for processing information, capable of substituting or expanding existing ones.
Rapid progress in such disciplines as cytology, sub-cellular systems and system biology has lately been observed. This progress is largely connected with deeper understanding and comprehension of informational aspects of biological processes which as a result has given an impulse in the researches of new methods for developing systems of information processing on the basis of biochemical reactions – a "biological" or molecular computer.
Nowadays "a biological computer" along with the computations on the basis of quantum logical gates systems ("quantum computer") is considered to be the main alternative approach to information processing (so called non-traditional methods of computing) in future computation systems. As a rule in concurrence with this the quantum computer is associated with powerful stationary computation systems and the biological computer is associated, at least in the nearest future, with small autonomous intellectual devices.
And if implementation of the quantum computer is still open to question, the living organisms created by nature and already surrounding us can be regarded as biological computers. Their "hardware" and "software" securing survival and adaptation of organisms to their environment largely surpass in effectiveness the existing technologies based on silicone. Thus in the work [19] bacterium coli (E.coli) is reviewed as a biological information processor. It is compared with the marginal hypothetical silicon computer of the same capacity in 1µm³. The lower threshold of information volume necessary for the construction of a new cell during its division is evaluated at 10¹¹ bits, the speed of information processing in such a processor is 107 bps with the consumption of energy per one operation of order 10 kT (k – being Boltsman constant, T- temperature), while in silicone transistor structure this value is limited to 104–105 kT. Hence it implies that "the marginal CMOS processor" may nor be compared with the information processor of a living cell neither in density of memory and logic elements, nor in speed or energy consumed.
Non-traditional methods of computation on the basis of biochemical reactions in the solution volume include both new computing paradigms, algorithms and new elements of computation systems circuits, out of them it is possible to distinguish elements on DNA (dsoxirybonucleic acid) structures, elements based on enzymes, and also elements on the basis of nanoparticles or molecular nanomaterials. Separately as the class of elements, and separate devices one can consider molecular nanosystems which distinguishing feature, besides information processing, is the presence of sensory, transformational and executive functions.
In Fig.1 there is a general scheme of "biocomputer" and the fields of science related to it. Also there are hierarchical levels of such a system as compared with hierarchical levels of a "silicon" computer.
By now a great number of publications dedicated to computation on DNA structures have been released.
DNA is the carrier of the genetic information in organisms. It is composed of the chains of nucleotides which are divided into 4 types A, G, T, C (adenine, guanine, thymine, cytosine). Where A/T and G/C are complementary to each other. The double helix is the basis of the DNA structure and is formed due to hybridization of two hydrogen bonds between A and T and tree hydrogen bonds between G and C.
The human genome contains approximately 3 billions of bases pairs, which define about 20,488 genes, arranged into 23 pairs of homologous chromosomes.
All the pairs of DNA bases of one human cell have the total length of 2.6 m when uncoiled and spread out but get compressed in the nucleus to the size of 200 µm.
For the first time Leonard Aldeman from the University of Southern California used the model of DNA for solving the seven-node problem of a travelling salesman in 1994. Cities and distances were encrypted using series of DNA and operations for the solution were produced with the use of polymer chain reactions.
Majority of the demonstrated computations on DNA are based on intermolecular reactions and the most famous approaches are the reactions on the basis of the displacement of DNA strands [1, 2] and reactions with biocatalysts based on DNA [3, 4] (so called deoxyribozymes or DNAzymes – artificially synthesized DNA with greater biological stability and specific emzymatic properties).
Displacement of DNA strands is the process where two strands with partial or complete complementation hybridize with each other displacing one or several previously hybridized strands. Nowadays realizations of several logic[5] and analogue circuits [7] and neural networks [6] are proposed on the basis of this reaction of displacement of strands of DNA.
The speed of the method of displacement of strands exponentially depends on the length of the strand. In Fig.2 the mechanism of work of DNA strands displacement method is shown on the example of the construction of two digital logic gates (AND and OR) [1]. The function is based on three functional elements. Gates "seesaw" (is included in gates 2,5 and 6) displaces the free strand of DNA (x1 and x2) and releases it is strand (works slowly). The threshold gate (is included into gate 5, works quickly due to having a reduced strand) contains DNA interacting with the input strands and transforming them into reaction waste (parameter th is set by the concentration of the threshold gate). The output gate (gate 6) is based on the mechanism of the "seesaw" gate, contains rhodamine (fluorescent mark) and rhodamine suppressant. Whenever there is a necessary free strand, the suppressant is displaced by another strand and the mark is activated. At the output the optical signal is formed.
To the benefits of computation using DNA it is possible to ascribe the ability to perform large parallel computation with high selectivity and great potential for schemes of little capacity. To drawbacks it is possible to ascribe intermolecular interactions on DNA, which due to their nature have the background and side reactions. For their minimization low nanomolar concentrations of reagents are used. As the speeds of the reaction and consequently the time of computation are proportionate to the absolute concentration of reagents, theses processes are performed slowly and can last from several hours to several days, which complicates their practical implementation [8].
For the solution of the "speed" problem live organisms use molecular patterns such as cytoskeleton, cellular membranes and proteins-hubs for selection of network components, thus increasing their effective concentration without the increase of the total quantity of molecules. Other methods of increasing the kinetics of the DNA strand displacement are based on the use of the 38th kilodalton protein for restoration and keeping of DNA as the catalyst [8] and cationic copolymers[9] for improvement of kinetics through stabilization of complex formation. Recently it has been proposed the use of the structure of three-dimensional triangular prism which according to the opinion of the authors possesses greater stability in comparison with regular double strand of DNA for the method of DNA displacement [10].
The field of prospective implementations of DNA computation is really wide – from molecular machines, robots and networks, performing binary operations and computations on the molecular level till absolutely autonomous systems emulating lifelike functions.
Catalytic reactions on the DNA systems are functionally limited in variations and effectiveness, and are inferior to the proteins from the point of view of affinity and diversity of ligands which DNA can identify [11]. Also enzymes (proteins possessing specific catalytic properties) are selective and sensitive receptors. They’re known as the best catalysts which allow to boost the speed 1017 times in comparison with non-catalyzed reactions [12]. Computations based on enzymes are limited to the operations of several logic gates, and the complexity of the network is limited by cross-reactivity of enzymes and noise [13].
The example of the enzymic system (Fig.4) is BioLogicChip – the concept of the device based on integration in the chip of biomolecular gates with biosensor/actuator system. The system contains a digital transducer of glucose, a hydrogel valve with the heating element for the liberation of the medicinal preparation out of the closed reservoir in the micro liquid channel under the application of pressure to the hydrogel, the reservoir for the medicinal preparation,an impedance transducer (for detection of hydrogel compression) and an insulin transducer (for the insulin release control) [14].Enzymic logic gate activates the heater and induces changes in hydrogel. Hydrogel works as the valve for the actuator mechanism (opens or closes the reservoir) and can release insulin for lowering the level of glucose. The insulin transducer controls the liberation of insulin (dosage and time) and the level of glucose.
Combining of enzymic and DNA computing systems in the circuit enzyme-DNA will allow the realization of (1) highly-selective identification of the diverse range of biological molecules and disease markers; (2) amplification of the catalytic signal; (3) massive quick and parallel processing of data and (4) complex processing of computing information for biologically induced signals.
Mixed computing systems based on DNA and enzymes have functional limitations on the level of enzyme properties.
In DNA-enzymic combined computing systems there are enzymes which affect directly only DNA, e.g. DNA -polymerases, DNA-ligases, endonucleases and so on. But enzymes, processing DNA cannot identify biomarkers of a disease in the form of small biological molecules of sugars, proteins and etc. Biocomputer systems based on common enzymes (not connected with DNA) are successfully used for logic processing and binary recognition of different combinations of physiological biomarkers. That is why there should be a more universal interface for connecting enzymatic output signal with the circuits for DNA processing [15].
Majority of methods used for reading the results of computation include the use of probes or complexes for reading that carry fluorescent marks which seriously restricts opportunities for multiplexing and complicates the process of measuring. These limitations can be taken off in the hybrid bioelectronic systems, where biomolecular computation are integrated into regular electronics through immobilization of DNA-machines on the surface of electronic circuits [16, 17].
Molecules of DNA and proteins are the basic structural elements of molecular nanosystems. In the foundation of any significant biological process there are nanomotors and machines of the molecular level: transporting cargo around the cell (kinesin), stimulating the motion of organisms (bacterial flagellum propellers), synthesizing proteins (rybosoms) and separating strands of DNA (helicases).
Molecular systems can be built both on the basis of biological systems and as well on the basis of microscopic objects of non-biological origin, for example, "molecular elevators" of J. F. Stoddart, "molecular pistons", "nanocars" and so on. It is impossible to say unambiguously which of the approaches will be used for building molecular systems in future.
In 1980–1990 the first methods of creating mechanically connected synthetic molecular systems were invented (catenanes – Jean Pierre Sauvage, rotaxanes – James Frazer Stoddart, molecular motor – Bernard Lucas Feringa). In 2016 this team received the Nobel Prize in Chemistry for designing and synthesizing molecular machines. They developed molecules the movements of which can be controlled.
For realization of the molecular systems potential, several main tasks should be solved:
It is necessary to create chemically controlled autonomous synthetic small-molecular systems;
It is necessary to develop systems with multiple integrated components, where each small-molecular machine performs a certain role;
It is necessary to take into consideration the stochastic nature of molecular dynamics. Biological machines are capable of making "mistakes" in contrast to the machines of the macroscopic world.
It is necessary to create the small-molecular "Turing machine", i.e. the molecular machine, which can read information from the symbol coded molecular chain.
In future, on the basis of biomolecular systems it will be possible to create compound systems of "smart" materials. And binding of electronic circuits and biomolecular devices may enable the creation of low-power biocompatible intellectual systems for highly-parallel and potentially fault-tolerant processing of information [20].
Within the last 20 years a great amount of biomolecular systems on the basis of DNA/RNA and enzymes/proteins have been developed, although the methods for creating a universal biomolecular computing machine similar to a classic computer haven’t been developed. These new biomolecular systems find its implementation in real biomedical applications (theranostics) [21]. Step by step humanity is approaching the creation of compound heterogeneous systems, where microelectronic, biochemical systems and smart materials will be combined. ■
The article was prepared with the financial support of the Ministry of Education and Science (Minobrnauka) within the frames of performing state assignment 16.12976.2018/8.9.
Rapid progress in such disciplines as cytology, sub-cellular systems and system biology has lately been observed. This progress is largely connected with deeper understanding and comprehension of informational aspects of biological processes which as a result has given an impulse in the researches of new methods for developing systems of information processing on the basis of biochemical reactions – a "biological" or molecular computer.
Nowadays "a biological computer" along with the computations on the basis of quantum logical gates systems ("quantum computer") is considered to be the main alternative approach to information processing (so called non-traditional methods of computing) in future computation systems. As a rule in concurrence with this the quantum computer is associated with powerful stationary computation systems and the biological computer is associated, at least in the nearest future, with small autonomous intellectual devices.
And if implementation of the quantum computer is still open to question, the living organisms created by nature and already surrounding us can be regarded as biological computers. Their "hardware" and "software" securing survival and adaptation of organisms to their environment largely surpass in effectiveness the existing technologies based on silicone. Thus in the work [19] bacterium coli (E.coli) is reviewed as a biological information processor. It is compared with the marginal hypothetical silicon computer of the same capacity in 1µm³. The lower threshold of information volume necessary for the construction of a new cell during its division is evaluated at 10¹¹ bits, the speed of information processing in such a processor is 107 bps with the consumption of energy per one operation of order 10 kT (k – being Boltsman constant, T- temperature), while in silicone transistor structure this value is limited to 104–105 kT. Hence it implies that "the marginal CMOS processor" may nor be compared with the information processor of a living cell neither in density of memory and logic elements, nor in speed or energy consumed.
Non-traditional methods of computation on the basis of biochemical reactions in the solution volume include both new computing paradigms, algorithms and new elements of computation systems circuits, out of them it is possible to distinguish elements on DNA (dsoxirybonucleic acid) structures, elements based on enzymes, and also elements on the basis of nanoparticles or molecular nanomaterials. Separately as the class of elements, and separate devices one can consider molecular nanosystems which distinguishing feature, besides information processing, is the presence of sensory, transformational and executive functions.
In Fig.1 there is a general scheme of "biocomputer" and the fields of science related to it. Also there are hierarchical levels of such a system as compared with hierarchical levels of a "silicon" computer.
By now a great number of publications dedicated to computation on DNA structures have been released.
DNA is the carrier of the genetic information in organisms. It is composed of the chains of nucleotides which are divided into 4 types A, G, T, C (adenine, guanine, thymine, cytosine). Where A/T and G/C are complementary to each other. The double helix is the basis of the DNA structure and is formed due to hybridization of two hydrogen bonds between A and T and tree hydrogen bonds between G and C.
The human genome contains approximately 3 billions of bases pairs, which define about 20,488 genes, arranged into 23 pairs of homologous chromosomes.
All the pairs of DNA bases of one human cell have the total length of 2.6 m when uncoiled and spread out but get compressed in the nucleus to the size of 200 µm.
For the first time Leonard Aldeman from the University of Southern California used the model of DNA for solving the seven-node problem of a travelling salesman in 1994. Cities and distances were encrypted using series of DNA and operations for the solution were produced with the use of polymer chain reactions.
Majority of the demonstrated computations on DNA are based on intermolecular reactions and the most famous approaches are the reactions on the basis of the displacement of DNA strands [1, 2] and reactions with biocatalysts based on DNA [3, 4] (so called deoxyribozymes or DNAzymes – artificially synthesized DNA with greater biological stability and specific emzymatic properties).
Displacement of DNA strands is the process where two strands with partial or complete complementation hybridize with each other displacing one or several previously hybridized strands. Nowadays realizations of several logic[5] and analogue circuits [7] and neural networks [6] are proposed on the basis of this reaction of displacement of strands of DNA.
The speed of the method of displacement of strands exponentially depends on the length of the strand. In Fig.2 the mechanism of work of DNA strands displacement method is shown on the example of the construction of two digital logic gates (AND and OR) [1]. The function is based on three functional elements. Gates "seesaw" (is included in gates 2,5 and 6) displaces the free strand of DNA (x1 and x2) and releases it is strand (works slowly). The threshold gate (is included into gate 5, works quickly due to having a reduced strand) contains DNA interacting with the input strands and transforming them into reaction waste (parameter th is set by the concentration of the threshold gate). The output gate (gate 6) is based on the mechanism of the "seesaw" gate, contains rhodamine (fluorescent mark) and rhodamine suppressant. Whenever there is a necessary free strand, the suppressant is displaced by another strand and the mark is activated. At the output the optical signal is formed.
To the benefits of computation using DNA it is possible to ascribe the ability to perform large parallel computation with high selectivity and great potential for schemes of little capacity. To drawbacks it is possible to ascribe intermolecular interactions on DNA, which due to their nature have the background and side reactions. For their minimization low nanomolar concentrations of reagents are used. As the speeds of the reaction and consequently the time of computation are proportionate to the absolute concentration of reagents, theses processes are performed slowly and can last from several hours to several days, which complicates their practical implementation [8].
For the solution of the "speed" problem live organisms use molecular patterns such as cytoskeleton, cellular membranes and proteins-hubs for selection of network components, thus increasing their effective concentration without the increase of the total quantity of molecules. Other methods of increasing the kinetics of the DNA strand displacement are based on the use of the 38th kilodalton protein for restoration and keeping of DNA as the catalyst [8] and cationic copolymers[9] for improvement of kinetics through stabilization of complex formation. Recently it has been proposed the use of the structure of three-dimensional triangular prism which according to the opinion of the authors possesses greater stability in comparison with regular double strand of DNA for the method of DNA displacement [10].
The field of prospective implementations of DNA computation is really wide – from molecular machines, robots and networks, performing binary operations and computations on the molecular level till absolutely autonomous systems emulating lifelike functions.
Catalytic reactions on the DNA systems are functionally limited in variations and effectiveness, and are inferior to the proteins from the point of view of affinity and diversity of ligands which DNA can identify [11]. Also enzymes (proteins possessing specific catalytic properties) are selective and sensitive receptors. They’re known as the best catalysts which allow to boost the speed 1017 times in comparison with non-catalyzed reactions [12]. Computations based on enzymes are limited to the operations of several logic gates, and the complexity of the network is limited by cross-reactivity of enzymes and noise [13].
The example of the enzymic system (Fig.4) is BioLogicChip – the concept of the device based on integration in the chip of biomolecular gates with biosensor/actuator system. The system contains a digital transducer of glucose, a hydrogel valve with the heating element for the liberation of the medicinal preparation out of the closed reservoir in the micro liquid channel under the application of pressure to the hydrogel, the reservoir for the medicinal preparation,an impedance transducer (for detection of hydrogel compression) and an insulin transducer (for the insulin release control) [14].Enzymic logic gate activates the heater and induces changes in hydrogel. Hydrogel works as the valve for the actuator mechanism (opens or closes the reservoir) and can release insulin for lowering the level of glucose. The insulin transducer controls the liberation of insulin (dosage and time) and the level of glucose.
Combining of enzymic and DNA computing systems in the circuit enzyme-DNA will allow the realization of (1) highly-selective identification of the diverse range of biological molecules and disease markers; (2) amplification of the catalytic signal; (3) massive quick and parallel processing of data and (4) complex processing of computing information for biologically induced signals.
Mixed computing systems based on DNA and enzymes have functional limitations on the level of enzyme properties.
In DNA-enzymic combined computing systems there are enzymes which affect directly only DNA, e.g. DNA -polymerases, DNA-ligases, endonucleases and so on. But enzymes, processing DNA cannot identify biomarkers of a disease in the form of small biological molecules of sugars, proteins and etc. Biocomputer systems based on common enzymes (not connected with DNA) are successfully used for logic processing and binary recognition of different combinations of physiological biomarkers. That is why there should be a more universal interface for connecting enzymatic output signal with the circuits for DNA processing [15].
Majority of methods used for reading the results of computation include the use of probes or complexes for reading that carry fluorescent marks which seriously restricts opportunities for multiplexing and complicates the process of measuring. These limitations can be taken off in the hybrid bioelectronic systems, where biomolecular computation are integrated into regular electronics through immobilization of DNA-machines on the surface of electronic circuits [16, 17].
Molecules of DNA and proteins are the basic structural elements of molecular nanosystems. In the foundation of any significant biological process there are nanomotors and machines of the molecular level: transporting cargo around the cell (kinesin), stimulating the motion of organisms (bacterial flagellum propellers), synthesizing proteins (rybosoms) and separating strands of DNA (helicases).
Molecular systems can be built both on the basis of biological systems and as well on the basis of microscopic objects of non-biological origin, for example, "molecular elevators" of J. F. Stoddart, "molecular pistons", "nanocars" and so on. It is impossible to say unambiguously which of the approaches will be used for building molecular systems in future.
In 1980–1990 the first methods of creating mechanically connected synthetic molecular systems were invented (catenanes – Jean Pierre Sauvage, rotaxanes – James Frazer Stoddart, molecular motor – Bernard Lucas Feringa). In 2016 this team received the Nobel Prize in Chemistry for designing and synthesizing molecular machines. They developed molecules the movements of which can be controlled.
For realization of the molecular systems potential, several main tasks should be solved:
It is necessary to create chemically controlled autonomous synthetic small-molecular systems;
It is necessary to develop systems with multiple integrated components, where each small-molecular machine performs a certain role;
It is necessary to take into consideration the stochastic nature of molecular dynamics. Biological machines are capable of making "mistakes" in contrast to the machines of the macroscopic world.
It is necessary to create the small-molecular "Turing machine", i.e. the molecular machine, which can read information from the symbol coded molecular chain.
In future, on the basis of biomolecular systems it will be possible to create compound systems of "smart" materials. And binding of electronic circuits and biomolecular devices may enable the creation of low-power biocompatible intellectual systems for highly-parallel and potentially fault-tolerant processing of information [20].
Within the last 20 years a great amount of biomolecular systems on the basis of DNA/RNA and enzymes/proteins have been developed, although the methods for creating a universal biomolecular computing machine similar to a classic computer haven’t been developed. These new biomolecular systems find its implementation in real biomedical applications (theranostics) [21]. Step by step humanity is approaching the creation of compound heterogeneous systems, where microelectronic, biochemical systems and smart materials will be combined. ■
The article was prepared with the financial support of the Ministry of Education and Science (Minobrnauka) within the frames of performing state assignment 16.12976.2018/8.9.
Readers feedback
