news and views a
b O
Cytoplasm +
+
R
+
+
siRNA
O OMe O P O B O O
S O
+
B
O
B = A, C, G, U
O
F
– +
Cleavage and siRNA release RISC assembly
+
© 2014 Nature America, Inc. All rights reserved.
–
–
+
+ +
–
–
+
–
–
–
– –
– –
+ –
+ +
–
TAT-conjugated siRNN
–
–
–
–
–
–
–
–
siRNA released following intracellular thioesterase cleavage
TAT-conjugated siRNN
mRNA
–
– –
–
–
mRNA cleavage
npg
+
–
–
+
–
mRNA loading
Thioesterase
–
+
+
RISC complex
+
+ TAT-mediated uptake
TAT peptide
–
TAT peptide
tBu-SATE (guide strand)
Guide strand (2′-F pyrimidine, 2′-OMe purine modified)
A-SATE (passenger strand)
Passenger strand (2′-F pyrimidine, 2′-OMe purine modified)
5′-DMB phosophotriester (passenger strand)
Phosphate group
Thioesterase
Figure 1 Internalization and cellular processing of TAT-conjugated siRNN. (a) Internalization of siRNNs is mediated by four TAT peptides conjugated to SATE groups. Upon internalization, the TAT-SATE modification is cleaved by intracellular thioesterases, releasing a functional negatively charged siRNA molecule. The deprotected siRNA is subsequently loaded into the RISC complex for target mRNA silencing. (b) TAT-conjugated siRNNs include four highly cationic TAT peptides (red) bound to a SATE-modified passenger strand RNA (white) containing four A-SATE linkers (dark gray) and a single irreversible 5′-dimethylbutyl (DMB) modification (light gray). The guide strand (teal) is decorated with six t-butyl SATE groups (black) and a 5′-phosphate (blue). Both RNA strands contain 2′-O-methyl purine and 2′-fluoro pyrimidine substitutions. The SATE modifications partially neutralize the highly negative charge of the 20-nucleotide RNA duplex, which is restored upon intracellular thioesterase conversion of the phosphotriester to a phosphatediester (mechanism shown in insert).
The most clinically advanced RNAi platform today is GalNAc conjugates. This technology— which combines fully stabilized siRNA8 and triple GalNac conjugation (first considered for oligonucleotide delivery in the early 90s9)— works extremely well in the liver because uptake depends on the cellular expression levels of asialoglycoprotein receptors, which are highly expressed on hepatocytes. In contrast, uptake of siRNNs is not cell-type specific and therefore might have much wider clinical applications. The ease of incorporating siRNN modifications will allow the community to perform a myriad of novel conjugations without the need for complicated chemistry. This capability will undoubtedly lead to the creation of novel classes of oligonucleotides. Naturally, further optimization will be needed to achieve ideal pharmacokinetic and pharmacodynamic behavior and acceptable toxicity, as indicated by the moderate in vivo efficacy shown by Meade et al.1 But clearly this work inaugurates a fundamentally new approach to overcome the challenge of cellular delivery of therapeutic oligonucleotides. Note: Any Supplementary Information and Source Data files are available in the online version of the paper .
1198
COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. 1. Meade, V. et al. Nat. Biotechnol. 32, 1256–1261 (2014). 2. Anonymous. RNAi roundtable: advances in delivery of RNAi therapeutics with enhanced stabilization chemistry (ESC)-GalNAc-siRNA conjugates http:// www.alnylam.com/capella/roundtables/esc-galnacsirna-conjugates/ (Alnylam, 2014). 3. Fawell, S. et al. Proc. Natl. Acad. Sci. USA 91, 664–668 (1994).
4. Périgaud, C. et al. Bioorg. Med. Chem. 3, 2521–2526 (1993). 5. Wiesler, W.T. & Caruthers, M.H. J. Org. Chem. 61, 4272–4281 (1996). 6. Gait, M.J. Cell. Mol. Life Sci. 60, 844–853 (2003). 7. Eguchi, A. et al. Nat. Biotechnol. 27, 567–571 (2009). 8. Morrissey, D.V. et al. Hepatology 41, 1349–1356 (2005). 9. Hangeland, J.J. et al. Bioconjug. Chem. 6, 695–701 (1995).
Lightening the load in synthetic biology Eric Klavins A new biological device known as a ‘load driver’ improves the performance of synthetic circuits by insulating genetic parts from each other. Synthetic biologists dream of designing genetic circuits as easily as electrical engineers design electronic circuits. Yet, unlike microchips, genetic parts such as transcription factors and Eric Klavins is at the Department of Electrical Engineering, University of Washington, Seattle, Washington, USA. e-mail: [email protected]
promoters rarely behave as intended when they are interconnected. Predictable performance is undermined by the nature of the cellular environment, including the genetic context of sequences encoding parts, the unanticipated effects of parts on each other and their nonspecific interactions with other molecules in the cell. These effects are compounded as the size of a circuit increases, preventing the
volume 32 number 12 DECEMBER 2014 nature biotechnology
news and views a
Source
Buffer
Load
Feedback
npg
© 2014 Nature America, Inc. All rights reserved.
b
Source
Buffer (load driver)
P
Load
P
Feedback
Downstream genes
Figure 1 Using buffers in electronics and synthetic biology. (a) When a high-impedance source is connected directly to multiple loads, its output is delayed and attenuated, concomitant with increases in the load. One common solution is to insert a unity gain buffer, shown here as a standard operational amplifier with feedback (dashed line), which effectively insulates the source from downstream circuitry. The output of the buffer rapidly equilibrates to the same voltage as the output of the source, but draws very little current. Furthermore, because the buffer is powered (not shown) it can supply considerably more current to the load than the source does. One crucial aspect of the buffer system is the feedback from the output to the inverted input of the amplifier, which minimizes the difference between its two inputs. (b) When a protein source (such as a transcriptional regulator) activates multiple downstream genes (the load), its effect on each downstream gene is attenuated and delayed owing to sequestration of the transcription factor by multiple gene promoters. The insertion of a genetic load driver insulates the source from the downstream load by quickly equilibrating the output transcriptional activator (yellow node) to the concentration of the source protein through the use of a phosphorylation cascade.The concentration of load driver proteins must be high, akin to the high power supplied to the unity gain buffer in electronics. Also, just as in the electronic version, the load driver incorporates feedback (dashed line) because the output protein is dephosphorylated at a rate that depends on how much of it is available.
construction of circuits with more than a handful of components despite the availability of large libraries of parts. In this issue, Mishra et al.1 tackle one aspect of the problem with a new part, called a load driver, that effectively insulates an upstream part from the downstream elements to which it is connected. This approach marks an important step toward the goal of being able to connect parts reliably in complex circuits. An engineered device is modular if its input and output characteristics do not depend on how it is connected to other modules. Open up a computer and you will find a great variety of modules that could equally well have been configured differently. Modularity is essential in engineered circuits to allow designers
to manage vast complexity through compartmentalization. In synthetic biology, the quest for modularity has long preoccupied researchers as they modified natural biological parts for reuse in engineered circuits. Whether any of the parts harvested from biological systems are inherently modular is, however, debatable2. Indeed, many organisms seem to exploit contextdependent gene expression as a means of fine-tuning cellular processes. These subtle features of regulatory control can undermine efforts to build circuits that perform as expected. For example, the sequence of a promoter affects how transcribed mRNA folds, which makes the promoter behave unpredictably when inserted into a multigene circuit.
nature biotechnology volume 32 number 12 DECEMBER 2014
To simplify circuit design, promoters can now be made modular, and easier to reuse, by incorporating an RNA processing step to clip off the 5′-untranslated region of mRNA so that the translation rate is independent of the promoter sequence3,4. ‘Loading’ is another way in which the modularity of genetic parts is subverted. When a downstream module has an unintended retroactive effect5 on an upstream module, the behavior of the upstream module—and thus of the entire circuit—is unexpected and unpredictable. Consider a transcription factor that controls a target gene by binding to and activating the target gene’s promoter. Ideally, the expression level of the target gene is a predictable function of the concentration of the transcription factor. However, when other downstream genes with the same promoter are present, a fraction of the transcription factor population will bind to them, thereby reducing expression of the target gene. Dynamic effects can also occur. Mishra et al.1 show that increases in target gene expression mediated by a transcription factor are both attenuated and delayed to a degree that correlates with the size of the load. Thus, the circuit designer cannot simply wire transcription factors into a circuit at will because the behavior of transcription factors changes depending on how they are wired. The challenge of loading is not unique to biology. It was once a common problem in electronics, and electrical engineers have developed a large set of tricks to address it. Almost exactly analogous to the unpredictable behavior of a transcription factor in a synthetic biological circuit, the signal from a high-output electrical impedance source can be highly sensitive to low-input impedance downstream devices. The standard solution to this problem is to place a ‘unity gain buffer’6 after the impedance source (Fig. 1a). Such a buffer can be fashioned simply from an operational amplifier with feedback. Because it has highinput impedance, it draws little current from the source device, while faithfully reproducing the source’s dynamic behavior for downstream devices. Modern electronic components, such as CMOS (complementary metal-oxide semiconductor)-based microchips, are now always packaged with similar isolating circuitry, and reliable interconnections—an art form in the early days of electronics—are now taken for granted. Precise analogies between biochemistry and electronics are enticing but are only beginning to be explored in synthetic biology. Although retroactive effects such as loading have been discussed in the literature and several buffer designs have been suggested5, the paper by 1199
npg
© 2014 Nature America, Inc. All rights reserved.
news and views Mishra et al.1 describes the first practical buffer device for the control of gene expression in eukaryotes7. The authors compare two gene circuits, unbuffered and buffered. The unbuffered circuit contains an inducible transcription factor that regulates a GFP reporter. When additional binding sites for the reporter are added, the response of GFP expression to time-varying changes in the level of transcription factor is significantly attenuated and delayed. In the buffered circuit, the authors introduce a ‘load driver’ between the input and output modules of the circuit. The load driver is implemented as a phosphotransfer cascade whose fast dynamics bridge the slower input and output stages. The input to the load driver is a protein kinase that initiates phosphorylation of the cascade. The output is a transcription factor that requires phosphorylation for activation. The device incorporates negative feedback because the output transcription factor is dephosphorylated at a rate that is proportional to how much of it is free rather than bound to downstream promoters (Fig. 1b). The negative feedback and the short timescale of protein phosphorylation, compared with the long timescale of gene expression, allow the load driver to quickly adjust its output to match its input. By increasing the abundance of protein components in the load driver, the system can be made almost insensitive to downstream load, much like increasing the power of the unity gain buffer in electronics. As proof that the scheme works, Mishra et al.1 show that
when the buffered circuit is loaded, it performs almost exactly as if it were unloaded. Many problems will have to be solved before synthetic biologists can routinely use load drivers to create large-scale gene circuits. For example, the load driver described by Mishra et al.1 can be used for only one connection in a given circuit, owing to the specificities of the proteins involved. In principle, many of the naturally
“By increasing the abundance of protein components in the load driver, the system can be made almost insensitive to downstream load, much like increasing the power of the unity gain buffer in electronics.” occurring phosphorylation cascades could be repurposed as load drivers, but ideally the proteins that serve as load drivers will be designed from the bottom up using custom DNA-binding, protein-protein interaction and phosphorylation domains so that engineers can build as many as needed. In the meantime, the authors list several phosphorylation cascades that differ in their protein interaction specificities and architectures and that might be repurposed to buffer multiple connections in a larger circuit. As the work of Mishra et al.1 shows, synthetic biology is moving from an initial focus on assembling libraries of parts to tackling major
roadblocks in the assembly of genetic circuits. Progress in the field to date has largely been based on a methodical engineering approach. The development of the genetic load driver was preceded by many theoretical papers detailing mathematical analyses of retroactivity8. Building on this foundation, Mishra et al.1 provide a thorough mathematical analysis of the performance and tunability of their load driver device, using timescale separation from nonlinear dynamical systems theory and disturbance rejection from control theory. It is increasingly clear that these fundamental concepts, and related principles from analog computation9, will underpin the development of a robust, modular synthetic biology, just as these same tools enabled robust electronic circuits. COMPETING FINANCIAL INTERESTS The author declares no competing financial interests. 1. Mishra, D., Rivera, P.M., Lin, A., Del Vecchio, D. & Weiss, R. Nat. Biotechnol. 32, 1268–1275 (2014). 2. Del Vecchio, D. Phys. Biol. 9, 045008 (2012). 3. Lu, T.K., Khalil, A.S. & Collins, J.J. Nat. Biotechnol. 27, 1139–1150 (2009). 4. Qi, L., Haruwitz, R.E., Shao, W., Doudna, J.A. & Arkin, A.P. Nat. Biotechnol. 30, 1002–1006 (2012). 5. Del Vecchio, D., Ninfa, A.J. & Sontag, E.D. Mol. Syst. Biol. 4, 161 (2008). 6. Horowitz, P. & Hill, W. The Art of Electronics. 2nd ed. (Cambridge Univ. Press, 1989). 7. Nilgiriwala, K.S., Jiménez, J., Rivera, P.M. & Del Vecchio, D. ACS Synth Biol. doi:10.1021/ sb5002533 (3 October 2014). 8. Del Vecchio, D. Annu. Rev. Control 37, 333–345 (2013). 9. Daniel, R., Rubens, J.R., Sarpeshkar, R. & Lu, T.K. Nature 497, 619–623 (2013).
Research Highlights Papers from the literature selected by the Nature Biotechnology editors. (Follow us on Twitter, @NatureBiotech #nbtHighlight) A three-dimensional human neural cell culture model of Alzheimer’s disease Choi, S.H. et al. Nature 515, 274–278 (2014) Precision microbiome reconstitution restores bile acid mediated resistance to Clostridium difficile Buffie, C.G. et al. Nature doi:10.1038/nature13828 (22 October 2014) Chimeric antigen receptor T cells for sustained remissions in leukemia Maude, S.L. et al. N. Engl. J. Med. 371, 1507–1517 (2014) CRISPR-Cas9 knockin mice for genome editing and cancer modeling Platt, R.J. et al. Cell 159, 440–455 (2014) Dominant drug targets suppress the emergence of antiviral resistance Tanner, E.J. et al. eLife doi:10.7554/eLife.03830 (3 November 2014)
1200
volume 32 number 12 DECEMBER 2014 nature biotechnology
b O
Cytoplasm +
+
R
+
+
siRNA
O OMe O P O B O O
S O
+
B
O
B = A, C, G, U
O
F
– +
Cleavage and siRNA release RISC assembly
+
© 2014 Nature America, Inc. All rights reserved.
–
–
+
+ +
–
–
+
–
–
–
– –
– –
+ –
+ +
–
TAT-conjugated siRNN
–
–
–
–
–
–
–
–
siRNA released following intracellular thioesterase cleavage
TAT-conjugated siRNN
mRNA
–
– –
–
–
mRNA cleavage
npg
+
–
–
+
–
mRNA loading
Thioesterase
–
+
+
RISC complex
+
+ TAT-mediated uptake
TAT peptide
–
TAT peptide
tBu-SATE (guide strand)
Guide strand (2′-F pyrimidine, 2′-OMe purine modified)
A-SATE (passenger strand)
Passenger strand (2′-F pyrimidine, 2′-OMe purine modified)
5′-DMB phosophotriester (passenger strand)
Phosphate group
Thioesterase
Figure 1 Internalization and cellular processing of TAT-conjugated siRNN. (a) Internalization of siRNNs is mediated by four TAT peptides conjugated to SATE groups. Upon internalization, the TAT-SATE modification is cleaved by intracellular thioesterases, releasing a functional negatively charged siRNA molecule. The deprotected siRNA is subsequently loaded into the RISC complex for target mRNA silencing. (b) TAT-conjugated siRNNs include four highly cationic TAT peptides (red) bound to a SATE-modified passenger strand RNA (white) containing four A-SATE linkers (dark gray) and a single irreversible 5′-dimethylbutyl (DMB) modification (light gray). The guide strand (teal) is decorated with six t-butyl SATE groups (black) and a 5′-phosphate (blue). Both RNA strands contain 2′-O-methyl purine and 2′-fluoro pyrimidine substitutions. The SATE modifications partially neutralize the highly negative charge of the 20-nucleotide RNA duplex, which is restored upon intracellular thioesterase conversion of the phosphotriester to a phosphatediester (mechanism shown in insert).
The most clinically advanced RNAi platform today is GalNAc conjugates. This technology— which combines fully stabilized siRNA8 and triple GalNac conjugation (first considered for oligonucleotide delivery in the early 90s9)— works extremely well in the liver because uptake depends on the cellular expression levels of asialoglycoprotein receptors, which are highly expressed on hepatocytes. In contrast, uptake of siRNNs is not cell-type specific and therefore might have much wider clinical applications. The ease of incorporating siRNN modifications will allow the community to perform a myriad of novel conjugations without the need for complicated chemistry. This capability will undoubtedly lead to the creation of novel classes of oligonucleotides. Naturally, further optimization will be needed to achieve ideal pharmacokinetic and pharmacodynamic behavior and acceptable toxicity, as indicated by the moderate in vivo efficacy shown by Meade et al.1 But clearly this work inaugurates a fundamentally new approach to overcome the challenge of cellular delivery of therapeutic oligonucleotides. Note: Any Supplementary Information and Source Data files are available in the online version of the paper .
1198
COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. 1. Meade, V. et al. Nat. Biotechnol. 32, 1256–1261 (2014). 2. Anonymous. RNAi roundtable: advances in delivery of RNAi therapeutics with enhanced stabilization chemistry (ESC)-GalNAc-siRNA conjugates http:// www.alnylam.com/capella/roundtables/esc-galnacsirna-conjugates/ (Alnylam, 2014). 3. Fawell, S. et al. Proc. Natl. Acad. Sci. USA 91, 664–668 (1994).
4. Périgaud, C. et al. Bioorg. Med. Chem. 3, 2521–2526 (1993). 5. Wiesler, W.T. & Caruthers, M.H. J. Org. Chem. 61, 4272–4281 (1996). 6. Gait, M.J. Cell. Mol. Life Sci. 60, 844–853 (2003). 7. Eguchi, A. et al. Nat. Biotechnol. 27, 567–571 (2009). 8. Morrissey, D.V. et al. Hepatology 41, 1349–1356 (2005). 9. Hangeland, J.J. et al. Bioconjug. Chem. 6, 695–701 (1995).
Lightening the load in synthetic biology Eric Klavins A new biological device known as a ‘load driver’ improves the performance of synthetic circuits by insulating genetic parts from each other. Synthetic biologists dream of designing genetic circuits as easily as electrical engineers design electronic circuits. Yet, unlike microchips, genetic parts such as transcription factors and Eric Klavins is at the Department of Electrical Engineering, University of Washington, Seattle, Washington, USA. e-mail: [email protected]
promoters rarely behave as intended when they are interconnected. Predictable performance is undermined by the nature of the cellular environment, including the genetic context of sequences encoding parts, the unanticipated effects of parts on each other and their nonspecific interactions with other molecules in the cell. These effects are compounded as the size of a circuit increases, preventing the
volume 32 number 12 DECEMBER 2014 nature biotechnology
news and views a
Source
Buffer
Load
Feedback
npg
© 2014 Nature America, Inc. All rights reserved.
b
Source
Buffer (load driver)
P
Load
P
Feedback
Downstream genes
Figure 1 Using buffers in electronics and synthetic biology. (a) When a high-impedance source is connected directly to multiple loads, its output is delayed and attenuated, concomitant with increases in the load. One common solution is to insert a unity gain buffer, shown here as a standard operational amplifier with feedback (dashed line), which effectively insulates the source from downstream circuitry. The output of the buffer rapidly equilibrates to the same voltage as the output of the source, but draws very little current. Furthermore, because the buffer is powered (not shown) it can supply considerably more current to the load than the source does. One crucial aspect of the buffer system is the feedback from the output to the inverted input of the amplifier, which minimizes the difference between its two inputs. (b) When a protein source (such as a transcriptional regulator) activates multiple downstream genes (the load), its effect on each downstream gene is attenuated and delayed owing to sequestration of the transcription factor by multiple gene promoters. The insertion of a genetic load driver insulates the source from the downstream load by quickly equilibrating the output transcriptional activator (yellow node) to the concentration of the source protein through the use of a phosphorylation cascade.The concentration of load driver proteins must be high, akin to the high power supplied to the unity gain buffer in electronics. Also, just as in the electronic version, the load driver incorporates feedback (dashed line) because the output protein is dephosphorylated at a rate that depends on how much of it is available.
construction of circuits with more than a handful of components despite the availability of large libraries of parts. In this issue, Mishra et al.1 tackle one aspect of the problem with a new part, called a load driver, that effectively insulates an upstream part from the downstream elements to which it is connected. This approach marks an important step toward the goal of being able to connect parts reliably in complex circuits. An engineered device is modular if its input and output characteristics do not depend on how it is connected to other modules. Open up a computer and you will find a great variety of modules that could equally well have been configured differently. Modularity is essential in engineered circuits to allow designers
to manage vast complexity through compartmentalization. In synthetic biology, the quest for modularity has long preoccupied researchers as they modified natural biological parts for reuse in engineered circuits. Whether any of the parts harvested from biological systems are inherently modular is, however, debatable2. Indeed, many organisms seem to exploit contextdependent gene expression as a means of fine-tuning cellular processes. These subtle features of regulatory control can undermine efforts to build circuits that perform as expected. For example, the sequence of a promoter affects how transcribed mRNA folds, which makes the promoter behave unpredictably when inserted into a multigene circuit.
nature biotechnology volume 32 number 12 DECEMBER 2014
To simplify circuit design, promoters can now be made modular, and easier to reuse, by incorporating an RNA processing step to clip off the 5′-untranslated region of mRNA so that the translation rate is independent of the promoter sequence3,4. ‘Loading’ is another way in which the modularity of genetic parts is subverted. When a downstream module has an unintended retroactive effect5 on an upstream module, the behavior of the upstream module—and thus of the entire circuit—is unexpected and unpredictable. Consider a transcription factor that controls a target gene by binding to and activating the target gene’s promoter. Ideally, the expression level of the target gene is a predictable function of the concentration of the transcription factor. However, when other downstream genes with the same promoter are present, a fraction of the transcription factor population will bind to them, thereby reducing expression of the target gene. Dynamic effects can also occur. Mishra et al.1 show that increases in target gene expression mediated by a transcription factor are both attenuated and delayed to a degree that correlates with the size of the load. Thus, the circuit designer cannot simply wire transcription factors into a circuit at will because the behavior of transcription factors changes depending on how they are wired. The challenge of loading is not unique to biology. It was once a common problem in electronics, and electrical engineers have developed a large set of tricks to address it. Almost exactly analogous to the unpredictable behavior of a transcription factor in a synthetic biological circuit, the signal from a high-output electrical impedance source can be highly sensitive to low-input impedance downstream devices. The standard solution to this problem is to place a ‘unity gain buffer’6 after the impedance source (Fig. 1a). Such a buffer can be fashioned simply from an operational amplifier with feedback. Because it has highinput impedance, it draws little current from the source device, while faithfully reproducing the source’s dynamic behavior for downstream devices. Modern electronic components, such as CMOS (complementary metal-oxide semiconductor)-based microchips, are now always packaged with similar isolating circuitry, and reliable interconnections—an art form in the early days of electronics—are now taken for granted. Precise analogies between biochemistry and electronics are enticing but are only beginning to be explored in synthetic biology. Although retroactive effects such as loading have been discussed in the literature and several buffer designs have been suggested5, the paper by 1199
npg
© 2014 Nature America, Inc. All rights reserved.
news and views Mishra et al.1 describes the first practical buffer device for the control of gene expression in eukaryotes7. The authors compare two gene circuits, unbuffered and buffered. The unbuffered circuit contains an inducible transcription factor that regulates a GFP reporter. When additional binding sites for the reporter are added, the response of GFP expression to time-varying changes in the level of transcription factor is significantly attenuated and delayed. In the buffered circuit, the authors introduce a ‘load driver’ between the input and output modules of the circuit. The load driver is implemented as a phosphotransfer cascade whose fast dynamics bridge the slower input and output stages. The input to the load driver is a protein kinase that initiates phosphorylation of the cascade. The output is a transcription factor that requires phosphorylation for activation. The device incorporates negative feedback because the output transcription factor is dephosphorylated at a rate that is proportional to how much of it is free rather than bound to downstream promoters (Fig. 1b). The negative feedback and the short timescale of protein phosphorylation, compared with the long timescale of gene expression, allow the load driver to quickly adjust its output to match its input. By increasing the abundance of protein components in the load driver, the system can be made almost insensitive to downstream load, much like increasing the power of the unity gain buffer in electronics. As proof that the scheme works, Mishra et al.1 show that
when the buffered circuit is loaded, it performs almost exactly as if it were unloaded. Many problems will have to be solved before synthetic biologists can routinely use load drivers to create large-scale gene circuits. For example, the load driver described by Mishra et al.1 can be used for only one connection in a given circuit, owing to the specificities of the proteins involved. In principle, many of the naturally
“By increasing the abundance of protein components in the load driver, the system can be made almost insensitive to downstream load, much like increasing the power of the unity gain buffer in electronics.” occurring phosphorylation cascades could be repurposed as load drivers, but ideally the proteins that serve as load drivers will be designed from the bottom up using custom DNA-binding, protein-protein interaction and phosphorylation domains so that engineers can build as many as needed. In the meantime, the authors list several phosphorylation cascades that differ in their protein interaction specificities and architectures and that might be repurposed to buffer multiple connections in a larger circuit. As the work of Mishra et al.1 shows, synthetic biology is moving from an initial focus on assembling libraries of parts to tackling major
roadblocks in the assembly of genetic circuits. Progress in the field to date has largely been based on a methodical engineering approach. The development of the genetic load driver was preceded by many theoretical papers detailing mathematical analyses of retroactivity8. Building on this foundation, Mishra et al.1 provide a thorough mathematical analysis of the performance and tunability of their load driver device, using timescale separation from nonlinear dynamical systems theory and disturbance rejection from control theory. It is increasingly clear that these fundamental concepts, and related principles from analog computation9, will underpin the development of a robust, modular synthetic biology, just as these same tools enabled robust electronic circuits. COMPETING FINANCIAL INTERESTS The author declares no competing financial interests. 1. Mishra, D., Rivera, P.M., Lin, A., Del Vecchio, D. & Weiss, R. Nat. Biotechnol. 32, 1268–1275 (2014). 2. Del Vecchio, D. Phys. Biol. 9, 045008 (2012). 3. Lu, T.K., Khalil, A.S. & Collins, J.J. Nat. Biotechnol. 27, 1139–1150 (2009). 4. Qi, L., Haruwitz, R.E., Shao, W., Doudna, J.A. & Arkin, A.P. Nat. Biotechnol. 30, 1002–1006 (2012). 5. Del Vecchio, D., Ninfa, A.J. & Sontag, E.D. Mol. Syst. Biol. 4, 161 (2008). 6. Horowitz, P. & Hill, W. The Art of Electronics. 2nd ed. (Cambridge Univ. Press, 1989). 7. Nilgiriwala, K.S., Jiménez, J., Rivera, P.M. & Del Vecchio, D. ACS Synth Biol. doi:10.1021/ sb5002533 (3 October 2014). 8. Del Vecchio, D. Annu. Rev. Control 37, 333–345 (2013). 9. Daniel, R., Rubens, J.R., Sarpeshkar, R. & Lu, T.K. Nature 497, 619–623 (2013).
Research Highlights Papers from the literature selected by the Nature Biotechnology editors. (Follow us on Twitter, @NatureBiotech #nbtHighlight) A three-dimensional human neural cell culture model of Alzheimer’s disease Choi, S.H. et al. Nature 515, 274–278 (2014) Precision microbiome reconstitution restores bile acid mediated resistance to Clostridium difficile Buffie, C.G. et al. Nature doi:10.1038/nature13828 (22 October 2014) Chimeric antigen receptor T cells for sustained remissions in leukemia Maude, S.L. et al. N. Engl. J. Med. 371, 1507–1517 (2014) CRISPR-Cas9 knockin mice for genome editing and cancer modeling Platt, R.J. et al. Cell 159, 440–455 (2014) Dominant drug targets suppress the emergence of antiviral resistance Tanner, E.J. et al. eLife doi:10.7554/eLife.03830 (3 November 2014)
1200
volume 32 number 12 DECEMBER 2014 nature biotechnology
