Saturday, October 14, 2006

philosophy

There is an exception to everything.

Sunday, May 21, 2006

David Baltimore's perspectives on Synthetic Biology

Here's a rough transcript of Dr. Baltimore's talk:

I helped organize Asilomar. That was more than 30 years ago; none of you were born then, but perhaps you've heard of it. Back then we had no experience with recombinant technology. We didn't know what to expect of it and what kind of guidelines should be in place, and after three days of Asilomar, we still didn't know. Then we essentailly handed the problem off to the NIH, and in retropspect we were very very lucky legislation was a non-player. Guidelines, rules, were developed, but because they turned out to be concerns that were invalid, it was relatively easy to roll them back. That wouldn't have been the case if they had been fixed in legislation.

We focused on safety. We did not focus on ethical issues or bioweapon issues. We didn't have the situation we have today. Terrorists are divorced from central states and certainly will not be concerned with any non-weapon treaties (which we thought, and were, at least in the US, preventing the development of biowepons).

We had some lawyers there. Those were the one's who gave the frightening talks about the liabilities we would be taking home; none of us had really conceived of those until then. But we didn't have ethicists or philosophers. We didn't feel like we were speaking for the public, but from an expert positon.

I've mentioned I have three perspectives on synthetic biology. One of them is that I was involved with Asilomar, with which and tomorrow's policy discussions resonate. Secondly, I'm on a panel at the national academy of science called cscans [spelling?]. We are trying to balance open access to information and research and its potential conflict with national security. And so these two connections are sort of tied together.

The third connection I have is that I think its very likely that viruses are the major concern in infectious systems, and that's because they are esentially small and easily engineered [and lots of nasty templates are naturally available], until I saw this [stuff on bacterial genome refactoring], which potentially makes bacteria just as easy to work with as viruses.

To be honest I'm most concerned with smallpox. It's not a backyard experiment to constitute it, and in fact you would need a very safe laboratory to work with it, so you wouldn't accidentally infect yourself, so I'm not really worried about its construction in clandestine labs, but I think [got behind here... I'll listen to my recording and fill it in soon.]

Now, what is the real danger? What is the danger today? It's the stuff we know about, that exists, the stuff like smallpox that we know can infect people and is extremely deadly and infectuous. I don't put much stock in concerns over designing new, even more dangerous organisms. Nature is a very tough critic, and I think if we tried I would guess we would fail. In retrospect I think those sentiments were present at and help explain some of the complacency that characterized Asilomar.

Take ebola virus. It comes out of its resevouir - bats or whatever - [not very often. You have to wait for that to happen, steal it from a protected lab, or constitute it directly from the sequence online. It's just not that easy to get.]
Bird flu is really not likely to spread until it mutates to a somewhat lower pathogenicity until you go to bed and wake up with the symptoms of flu instead of dying in the night. Viruses are finely tuned to our particular lifestyle, and it's something to remember that [missed the point about non-resevoir human infections]. All I'm trying to say is that we should focus on what we have right now - enough threats already exsits, we don't need to fabricate any more to worry about.

Questions:
Tom Knight: One of the things that has changed is that it's no longer really a question of what's in your freezer, but what sequence you can get online.
David Baltimore: No no, I agree, but there really are not that many things that are not availible in the natural world but are availible in GenBank, and in a clandestine fashion you could recreate that virus.

Unknown Questioner: what can be done now to mitigate [that] problem?
DB: lots of people now have become rich by being involved with biology. So the first thing to do is to admit the reality, but then point out that there's a deeper goal here - the service and betterment of humanity. I can remember when I was getting involved with microbiology, we would write exactly the same thing in NIH grant proposals as you write today, we just never said when. We had faith! But it will happen, and is happening.

Drew Endy: I wanted to get your thoughts on the scope of the community. I remember seeing papers from the 1970s on how to brew botulinum[?] toxin in your kitchen, and it seems there would be a strong negative selection for that behavior. How is it different now, given new [?].
DB:Well I certainly hope we'll see a push towards distributed access, because that's the science, and that's the way we'll be able to release the imagination of people to do great and interesting things. I don't think we should say wholesale "but there is concern." We need to really think about [?]. We find at CalTech today that there is biology in every division, high school students in the laboratory, and what they've been doing, what they know, it's spectacular. No, I think that's all great; I think if there's concern to be had, smallpox. That's my concern.

Rune: There's a lot of people from different communities coming into this, but as someone who's been doing this fo forty years, newness is fine, especially for attracting more interest in the field. So I guess what I'm asking is this: what is new, really? What are its downsides? Its upsides?
DB: Well, you can convince venture capitalists, funding agencies, etc. to get on board, there's definitely that. But I think there really is someting new here, and it's a turning around from sequecing to synthesis, and I must say I'm impressed with the extrordinary facility researchers have in knowing what all the specifics do, what lambda does, this or that promoter does; ten years ago, we knew that would happen. I used to say in those days that there is no question in biology that we cannot answer - it may take a long time - but it can be answered. So it may not be surprising, at least to me, but I think it is new.

Summaries of selected 10-minute short talks

Caroline Ajo-Franklin (Harvard Medical School) presented work on a cell-cycle counter.

Steffen Mueller (SUNY Stony Brook) spoke about synthetic virus design.
Live attenuated viruses are usually generated through big serial dilutions in a costly and time consuming process, and the resulting mutations that attenuate it are usually few in number. Instead, perhaps one could attenuate a virus by de-optimizing the codon distribution in its genome (i.e. so that it encoded codons that were rare in the host). Redesigned two capsid replacement cassettes (2643 nucleotides) to maximize de-optimal codons (934 silent mutations). Strangely, the titer of the deoptimized viruses went down, but the number of virions built per infected cell remained roughly the same, suggesting that the changes had actually made them less infective. This is an interesting and exciting approach to attenuating a virus because it is thought that many many mutations introduced would be nearly impossible for the virus to evolve around.

Jack Newman (Amyris Biotechnologies) explained some of the market incentives that surround synthetic biology. He used the Keasling lab's artemisinin work as an example. Malaria kills 3 million+ people per year. Treating it would require something like 500+ million treatments, i.e. 400+ tons of artemisinin, i.e. 600+ tons of Artemisia annua plant material each year. Biological systems can be engineered to produe massive quantities of otherwise costly chemicals, such as artemisinin. He also mentioned Modular Expression Design (MED), which is a paradigm essentially for fast debugging of biocircuits. I wish I had more of the details of it. Lastly, he encouraged all us young, talented, bushy-tailed synthetic biologists to come work at Amyris, which is hiring.

Samantah Sutton
(MIT Endy lab) introduced her phosphoregulators (?) for post-translational logic. I unfortunately missed lots of this while finishing the previous entry... more later.

Trevor Swartz (UC Santa Cruz) talked about light-activated LOV-domain histidine kinases.
Interestingly, the LOV domain can be found in many bacteria that have no apparent light sensitivity. Erythrobacter has a 368 bp LOV-HPK that apparently is a light activated histidine kinase, and that a particular sulfide bond seems to be the photosensitive element that alters the kinase activity. Additionally, it is likely that all LOV domains are photoactive.

Brian Yeh (UCSF) thinks that protein based signaling circuits are in many ways superior to other signaling pathways because they operate on fast timescales and ___(?). One problem is ___. Another problem is that the output o one node must act as an input of the subsequent node. "One solution is to use common currencies of biological information." Intersectin is a Cdc42-specific GEF. Fuse PDZ + DH-PH GEF domain + PDZ-binding peptide(?) to make a signaling node with modular input. Yeh built a working proof-of-concept circuit composed of 2 synthetic GEFs in mammalian cells.

Peter Carr (MIT Media lab) & Farren Isaacs (Harvard Medical School) presented an overview of Whole Genome Engineering they called "rE. coli." (Recoding E.coli: rE.coli). So why synthetic genomes? All codons can be (re)optimized and the genome can be made much more stable. A technology platform is under development that employs both bottom-up and top-down techniques to make genome-scale refactoring expedient and practical. Multiplex DNA synthesis and Polymerase Assembly Multiplexing (PAM) can be combined to provide a synthesis technique with throughput on the order of megabases. Polony-based sequencing, DNA error correction, computation tools (genome annotation & analysis, genome sequence parsing, and PAM assembly simulation) and a variety of other techniques are being incorporated into the rE. coli toolbox.

"Synthetic retrotransposons" - Jef Boeke

  • Retrotransposition is one of the two basic types of transposition we find in biology. It's sort of fun to survey where transposons are in the tree of life: in the case of more eukaryotes, most of the genome is transposons; little of microbial genomes are retrotransposons.

LINEs (L1) transposon sequences comprise ~ 21% of the human genome, and generally dominate mammalian genomes.
  • The RNA is oth mRNA and "genome" (template) RNA
  • Most of the ~500,000 L1 copies are severely 5' truncated and live 1) in introns or 2) between genes
  • normally inserts only in the germ line - not somatically.
  • The promoter is entirely embedded in the mRNA and hence is taken along with the transposon.
Synthetic biology of retrotransposons:
make synthetic ORF2 and ORF1. Properties:
  • 25% of the nucleotides altered - new species of retrotransposons
  • adenosine content reduced 40%.
Built "ORFeus" retrotransposon which is driven by an external promoter, so it should jump just once, because it doesn't "take its promoter with it." Average number of insertions/progeny ~0.42 (pretty good). These are germ line insertions; there are even more somatic insertions. These insertions are very random across genes and chromosomes.

Besides mutagenesis, another application might be cancer gene discovery: they've developed a method of activating retrotransposition in any tissue of interest.

Anyhow, they've built 7+ new transposons (thanks DNA 2.0). And now for something completely different: the systems biology of yeast is the best of any, os if we know so much, can we use that knowledge to refactor it and make a "synthetic" strain?

Boeke's lab is designing and synthesizing yeast chromosome 3. They're doing so in an iterative fashion, refactoring the chromosome in chunks of 30Kb at a time, so they can check their work as they go. It's a straightforward procedure; get thousands of colonies from ~ug of refactored DNA.

Changes to make; want to make a super-stable genome:
  • remove all transposons
  • remove all introns
  • downsize telomeres
  • relocate all tRNA genes
Unfortunately, fitness has a tendancy to evaporate with each recombination. Can this be overcome with "genome swarms" and conditional evolution? Hmm.

P.S. - Joel Bader is writing a Genome Revision control System!

"Synthetic immunology: engineering immunity" - David Baltimore

"The world's major killers are killers because they elude immune attack: HIV, malaria, tuberculosis," etc. What are needed are new approaches to stopping these pathogens. "The don't work because they haven't worked" - and what I mean by that is that the immune system is gonna fail when attacked by these pathogens, so trying to fix it with traditional methods is inherently a poor solution.

Let's reprogram the immune system using gene therapy to direct synthesis though the immune system of monoclonal antibodies or TCRs; let's not rely on its natural detection and synthesis.
Modify hematopoietic stem cells (from which derive all blood cells) so that our reprogramming is widely effective.

Program T cells for anti-tumor immunity:[presented too quickly to summarize]

Safety Concerns (gene therapy):
three of nine childre who received sucessful gene therapy for the X-SCID disease develiped leukemia... but maybe that was an effect of X-SCID, and not of the lentiviruses.

Consider selective inhibition of CCR5 in primary CD4 T-cells by siRNA introduced by the lentivirus vector - in principle, if no T-cells express CCR5, then none should be infected by HIV.

This will be a grand challenge to implement in humans (but none are really conceptual, just technical). We won't go into them.

Targeting of lentivirus vectors (lentivectors):
Infection has two components:
1)Binding/entrance into the cell, and 2)endosomal membrane fusion. One thing that will bind nicely to some cell surface protein (like CD20) is an antibody. Lentiviruses, and all retroviruses, envelope themselves with the surface of the cell they infect, so if you modify host cells to express a membrane-bound antibody, lentiviruses that infect and bud off from them can be targeted to whatever host receptor that antibody binds.

"Secreting Spider Silk in Salmonella" - Chris Voigt

Let's engineer a "super" salmonella that can construct and extrude silk filaments from its membrane. Spider silks have a wide variety of desirable materials properties. Dragline silk is equal strength of kevlar, but 10 times more elastic. Silks are amino acid biopolymers with elastic and crystalline elements, and thread can be spun simply from a solution of silk monomers.

Will need a complicated secretion tran-membrane protein, and the silk monomers will need an N-terminal secretion signal to direct them to the secretion "needle." Oh, and we'll also need a complicated regulatory network to turn things on in the correct order (i.e. only start producing the monomers after successful construction of the complicated secretion needle.

Salmonella encodes just such a needle that it uses to inject proteins into a host cell that induces it to endocytose the entire salmonella bacterium. The genes and control network for this needle are encoded in a so-called "pathogenicity island." The dymanics of this control network can be teased out by constructing plasmids with one of the promoters and a GFP gene, and then inducing the transformed cells to activate the pathogenicity island.

As of friday, an engineered strain of salmonella was producing a couple mg/l of silk monomers.

"Synthetic mammalian gene networks" - Martin Fussenegger

For any therapeutic intervention, dosing is key.

Consider a system for gas-inducible transcription control composed of a fungal trans-activator (expected to be functional in mammalian cells), AlcR, and an operator, activated by small concentrations of a chemical found in tabacco smoke.

Some of their work (biotechnol bioeng. 83:810 2003) demonstrated that different amounts of the final product is expressed depending on which level of the control cascade is repressed, such that repressing the level directly responsible for expressing the desired product results in very little expression, while repression of increasingly distant levels of the cascade allows more and more expression of the product. This makes sense, as just a little bit of constituative expression or wobble in the initial signalling genes could be magnified though the control cascade.

Digital Electronics - Digital Genetics
Electronic circuit:
input:switch -> Buffer:capacitor -> delay:resistor -> switch:transistor -> output:lamp
Biological implementation:
?

HEALTH SESSION: "Directed evolution of new viruses for gene delivery" - David Schaffer

Overview of Health & Synthetic Biology by Wedell Lim:
The ability to systematically and flexibly manipulate and control cellular systems is the threshold to a new era in medicine, hence the importance of synthetic biology

synthetic organic chemistry -> intellectual capital: understanding of biochemical reactions; tangible benifits: construction of novel molecules

Synthetic biology -> intellectual capital: understanding cellular networks;cells with modified & novel functions;

Schaffer's talk:

Eytemology of virus: essentially "poison" (roots in latin, sanskrit).
serotypes are . Numerous serotypes of Adeno-Associated viruses have been isolated. The require the presence of an adenovirus to replicate, and they're non-pathogenic (90% of human population have already been exposed, and it's highly efficient).

Crosses cell membrane though interaction with heparan sulfate and delivers its genetic payload into the nucleus.

However, "viruses did not evolve in nature to be used as human therapies," and hence we will need to evolve them to our own ends. To re-evolve the virus to have the properties we desire, libraries of the genomes of different serotypes of the virus can be constructed, modified (i.e. introduce point mutations), and then screened.

A large fraction of the human population harbors antibodies and neutralizing antibodies against numerous AAV serotypes, which can signifiantly reduce gene delivery efficiency. How do you get around this? Evolve the virus by selecting for those that can still infect in the presence of a neutralizing antibody. [What were the unexpected effects of injecting AAVs with serum?] Interestingly, the two most important mutations that seem to allow mutants to evade rabbit antiserum are in a region that binds with the target cell receptors. More interestingly, mutations that allow AAVs to evade human serum are not in the same places; one is not exposed, but actually inside the virius at the interface between a receptor spike and other internal machinery.

"Towards the design oand synthesis of an artificial cell" - Jack Szostak

First point: "I really don't think this has any practical applications, at least in the foreseeable future."

Goal is to build a simple cell based on a replication vesicle for compartmentalization and a replicating genome. If the cell is to be simple, the environment must be correspondingly complex. Here are some technical considerations relevant to the construction of such a protocell:
  • No biochemical machinery - all processes must be spontaneous
  • nucleic acid template-directed copying, strand separation
  • membrane dynamics: growth, division, permeability
Building the membrane:
Fatty acid structures can actually self-organize into liposomes given particular environmental conditions. At a pH greater than 10, myristoleate will from into single layer micelles. There is a threshold at a lower pH at which this micelles reorganize into double-layered vesicles.
Additionally, osmotic pressures can be manipulated to drive membrane growth. This presents a spontaneous, downhill pathway for building vesicles.
Thirdly, larger vesicles can be divided into smaller vesicles by forcing it through solid channel smaller in width than the vesicle.

nucleic acid replication:
Need more energy to make the rate of the spontaneous polymerization reaction faster. How? Make ATP portion of the nucleotides more reactive: mess with the nucleobases, change the leaving groups, practically anything. Changing the nucleophile seems most promising, i.e. changing the organization sugar backbone.

[get more details about this]
Over two days, without an enzyme, one can extend a primer naturally...

Phosphoramidate DNA, ending not with a 3'OH, but with a 3'amidazolide[?] group.

Interestingly, MA:GMM vesicles are stable up to 90+ degrees C, and additionally nucleotide permeability goes up, facilitating nucleic acid separation and replication.

So, the experiment everyone has wanted to do for years: can we add nucleotides to the outside of such a vesicle containing a prebuilt primer, watch them diffuse in, and then polymerize? Yes! In an elementary manner, at least.

"Chemical tools for probing the glycome" - Carolyn Bertozzi

Chemical tools for probing the glycome - Carolyn Bertozzi

Glycosyltation is the most complex form of posttranslational modification. polysaccharaides attached to membrane-bound proteins can facilitate cell-cell (or membrane-membrane, at least) communication by interacting with corresponding membrane-bound receptors.

Fundamental "glycomics" questions:
  • Which proteins are modified with what kinds of glycans (50% of eykaryotic proteomes)?
  • what are the functional consequences?
  • How does the glycosylation vary with changes in physiology?
Chemical tools are needed for perturbing and monitoring glycosylation in living systems! Tools to modify and incrementally change existing systems, and then to report out the effects of those changes.

Glycosyltransferases use 9 (in higher mammals) monosaccharide building blocks (derived from metabolized sugars from diet) to glycosylize proteins in the endoplasmic reticulum and golgi compartment. There are essentially two forms of glycosylation: N-linked glycans and O-linked glycans. Both forms contain a "conserved core" with a GalNAc structure. This conserved core is linked to the N-group of asparagine for the N-linked form of glycans and to the oxygen of a serine or threonine in O-linked glycans.

Want to design inhibitors for O-linked glycans (the perturbing agent). (Diphosphate sugars are bad because they're charged and polar and prevent the compound from getting into the cell.) Built a 1338 uridine-based library to screen for ppGalNAcT inhibitors. Found two. Both had broad spectrum inhibitory activity. Importantly, none of the compounds inhibited any of the other glycosyltransferases.

It's hopefully self evident why you would want to visualize the presence and location of a molecule in a living system. You can usually fuse GFP onto a protein to track it, but what about the non-proteins components of the cell? Bioorthogonal reporters for particular glycan chains are needed and have been designed (the term "bioorthogonal" indicates the reporter should not interfere with normal biological activity).

Note, cancer cells have aberrent glycosylation... many possess lots of sialic acid glycan chains. How do you detect them? Modify the metabolic precursor (ManNAz) in an orthogonal way that can be detected with a reporter (synthesized in or ex vivo). For instance, radiolabeled and self-quencing "smart" probes have been developed.

"New bacterial communication lines by laboratory evolution of LuxR" - Frances Arnold

Let's make a new communication system by utilizing an existing communication system.

"I'm just the parts lady; I'm the one who has to sit there figure out how to define parts to... and so LuxR is my chosen substrate."

Novel idea for using signal pathways: predator-prey system, consensus consortium - something activates only when two cells are in proximity (i.e. in a biofilm).

Overview of Lux system:
V. fisheri synthase LuxI diffuses into medium, signal accumulates, binds to LuxR, activates gene transcription, turn on luciferase.

Here's some more information about LuxR and it's many cousins, which respond to a variety of acyl-homserine lactones.

The challenge is to evolve LuxR to respond to other signal molecules, multiplying the number of signalling pathways available to biological circuits.

Dual selection systems to evolve mutants (easy to make mutants, hard to find the ones that have the desired mutation; main principle of solution is to provide a selection pressure that only allows mutants with the desired solution to survive or to report themselves. Generalized dual selection system for maintaining specificity).

LuxR is modular, it is possible to recombine LuxR to respond different promoters, so in principle it should be possible to evolve new domain-swapped activators (why hasn't this worked before?)

Conclusion: "LuxR can give rise to a set of "standardized parts" for programmed cell-cell communications."

7-word summary of synth biology:
"Genome. Great story! Hard to write..."
Need a good editor. Evolution is one of the best! Massively parallel selection screens could be the "answer."

Collins et al nature biotech may or june

Day 2 - Chemistry Session: Protein Biosynthesis: Encoding Unnatural Amino Acids and Evolving in Cells - Lei Wang

"Proteins are involved in virtually every life process." 20 amino acids are encoded by 61 of the 64 codons. To generate interesting protein properties, there are two avenues: to increase the building blocks (i.e. genetically encode unnatural amino acids), and 2. to permute and recombine the existing AAs (?)

1. Is it possible to expand the genetic code to include new amino acids? The Amino acids, aminoacyl-tRNA synthetase, tRNA, and codon all must have a high specificity to insure fidelity and minimize cross talk. They tried many different tRNA/aaRS pairs, finally settling on one already existing in M. jannaschii.

Here is a general approach for the generation and improvement of orthogonal tRNAs:
generated libraries of the synthetase to evolve tRNAs that both 1) work in

final product had translational fidelity near 99%, comprable to natural tRNAs.

Labeling in vitro?
Also could make homogenous glycoproteins (help understand glycan dymanics?)

(check nature article about redesigning firefly luciferase to be more red)

Replace group with heavy metal to facilitate x-ray structure determination?

Could incorporate photo-active groups for reporting certain activities, or metal chelators to manipulate electron transport chains, etc.

sequence variants of a 100-aa "protein" 20^100 = 10^130 > number of atoms in the universe (~10^80).

B-Cells encounter and become activated by antigen causes introduction of point mutations in V region of lg at a rate 100 times greater than baseline mutation rate.

protein evolution using SHM?

Ratio Sorting for Red-Shifted cells (using FACS?) Longer wavelengths have better tissue penetration.

This was fairly technical and I plan on tracking down Dr. Wang later for clarification.

Saturday, May 20, 2006

DNA Synthesis Panel

J. D. Kittle - Coda genomics
  • CODA (Computationally Optimized DNA Assembly), also incorporates its patented Translation Engineering TM into the design of the gene.
John Danner - Codon Devices
  • Mission: to eliminate construction as a barrier to synthetic biology (Drew's point #4).
  • Key enabling technologies: CAD design environment -> multiplexed oligo synthesis & purification -> assembly...
  • The bottom line is that if you want to go bigger with synthesis, you've got to multiplex. For instance, you can get a large number of variants (for cents/bp) if you design a library well. Instead of doing blind mutagenesis, lets use the knowledge we have! Codon's BioFAB(tm) will make length concerns a non-issue in synthesis.
Hans Buegl - GeneArt
  • The market for customers of synthesis tech. has doubled in the last year, and is split nearly into thirds between Big Pharma, academic/research institutions, and ?
  • Largest synthesis ~21kb. They have software which attempts to select the optimal codon distribution by examaning and scoring each codon in turn in a given sequence.
  • They've optimized the synthesis workflow with lots of automation.
Jeremy Minshull - DNA 2.0 (president)
  • Gene Designer is a free program for optimizing codon usage for different organisims (but requires registration).
Questions:
What are the current bottlenecks in synthesis? Cost of synthesis, cost of sequencing (consider personal genome project), and perhaps surprisingly time required for synthesis. As Drew pointed out, a "compile time" of at least four to eight weeks is completely impractical, especially for students in a competition. But what market pressures exist to drive down the time required? What will spur innovation here?

"On a system for engineering genetic machines" - Drew Endy

  1. Recombinant DNA
  2. PCR
  3. Sequencing
  4. Synthesis
  5. Standardization
  6. Abstraction

We've got the first three. Four is coming, and that means we need to talk about five and six. When synthesis becomes trivial, the barriers to innovation will be largely informational, dependant on our ability as a community to organize and communicate and work together. The establishment of design standards is essential to innovation because it enables potential innovators to rely on simplifications of existing knowledge and devote the majority of their time to understanding the complexities surrounding their innovation, of the true unknown.

This is the purpose of synthetic biology.

"Engineering nucleic acid-based molecular sensors for probing and programming cellular systems" - Christina Smolke

If we are interested in programming cellular behavior, then we need to be concerned with the network of biomolecules that interact to cause certain behaviors. DNA -> pre-mRNA->mRNA -> [export] -> [translation complex] -> proteins all fall under the auspices of "biomolecules."

"We tend to think of the molecules we design as input/output devices." Output domains are areas of nucleic acids (themselves or others) that have endo- and exo-nucleic catalytic activity. Nucleic acid domains that can bind via base-pairing to transcription regulatory regions are also designated output domains. Additionally, one can design in interference

Input domains are areas of a nucleic acid chain that can "sense" a chemical with high specificity. Specifically, they are composed of an aptamer domain, which is the region of nucleotides to which the input chemical directly binds, within a longer strand of nucleotides that change conformationally

RNAs can be built that act as a switch triggered by a particular ligand. The design of the RNA can be described abstractly as a sensing domain coupled with an output domain in such a way that binding of a ligand to the aptamer within the sensing domain causes a conformational change that exposes the output domain. Output domains have been designed to bind to an mRNA and prevent its translation. These devices are dubbed "antiswitches."

Lots of potential exists for genetic programming with antiswitches and riboswitches. Computer programs designed to predict RNA secondary structure can be utilized to design the particulars of an antiswitch. Usually the output domain enters or exits from a hairpin structure upon activation of the input domain. Work is in progress in making the two domains completely independant of each other, but currently the output domain must be fine-tuned to work with a particular input domain.

Here's Smolke and Bayer's Programmable ligand-controlled riboregulators of eukaryotic gene expression.

"Cell-free gene expression in synthetic vesicles" - Vincent Noireaux

To build a cell you could start from the top down and reduce it's genome to the minimum size: ~400 genes.

Or you could take a bottom-up approach and attempt to build a cell molecule-by-molecule with the "molecules of life."

The idea is to develop an artificial cell as a programmable phospholipid vesicle, with the DNA as software and translation/transcription machinery as hardware. To follow with this computer science characterization, if you consider the entire cell a computer, then the idea is to simply format it's "natural" hard-drive and install your own rudimentary operating system, presumably because the computers behavior would be much easier to control and predict. Note that this approach to engineering useful systems is in direct conflict with Dr. Gardner's philosophy of utilizing the genetic circuits that have already successfully evolved.

Actually, that analogy is somewhat misleading, because Noireaux's work was really about building the simplest possible "synthetic cell." They developed or extended an emulsion technique so that they could incorporate into the the little "synthetic" phospholipid bubble different proteins and chemicals, such that they experimentally produced a synthetic vesicle containing translational machinery to produce green fluorescent protein. They observed its expression last for 4 days in lab.

How do you make a synthetic vesicle?
  1. Reaction in microfuge tube (CFE-DNA-RNAP)
  2. Add oil
  3. vortex to make many little droplets of phospholipid monolayers
  4. some trick to get all that translational machinery into the emulsion.

"A novel genome vector for giant DNA assembly" - Mitsuhiro Itaya

Megacloning.

The purpose of Itaya's work was push DNA cloning to the limit by cloning the entire 3.2 Mb genome of Synechocystis into B. subtilis. The work took several years, and is summarized at the Proceedings of the National Academy of Sciences, which also published the paper.

"Engineering artificial cytoskeletons" - Dyche Mullins

"The fundamental problem of cellular medicine: More is different"

Consider a chemotactic cell. The cell's motility is derived from a motile actin network - constant polymerization and protrusion and debranching and depolymerization. Not enough is known about the "strokes" of the motility engine - that is to say, how filaments are organized to drive in a particular direction.

Well, take an ActA-coated polystyrene bead and expose it to a solution of the Arp2/3 complex and the capping protein. For reasons that were explained but that I can't paraphrase, if one adds the right tag molecules, one can observe a filamentous shell nucleate around the bead, grow uniformly, and then suddenly rupture and concentrate around one area of the bead. The resulting disequilibrium forces the bead to move off in a particular area.

The plasmid R1 par operon is a 3 component system. ParM is a prokaryotic protein, so it could potentially be used in a eukaryote without dire side effects, and the operon inherently doesn't have a bunch of regulatory elements.

I had difficulty following all the elements of this talk. There were some fanatastic movies involved, and therefore you would gain so much more by watching the presentation. I'll try and beef this up later.

"DNA Origami" - Paul Rothemund

Richard Feynman once challenged the world to produce print the information in the encyclopedia britannica in a space no larger than that of a head of a pin. Science has since developed techniques that could potentially be used to accomplish this goal, but may be "impractical" for a variety of reasons - too expensive, too difficult of a fabrication process, etc. This talk isn't about printing the encyclopedia, but simple shapes.

This is a method of constructing nanometer scale shapes with DNA based on a long single-stranded "scaffold" strand and many little "staple" strands that crosslink portions of the scaffold, essentially weaving it into a particular secondary structure. The success of this method and the final shape of the depends sequence of the scaffold.

Practically speaking, take M12mp18 viral genome (~7kb), use a computer program to analyze its sequence and generate short staple sequences that will bend the scaffold sequence into the desired shape. Almost any arbitrary shape can be built; the staples are synthesized and added to a solution with the scaffold sequence and about 2 hours later many of the desired shape should be floating around, waiting to be set down on a sheet of mica and examined with an electron microscope.

This approach can generate nanoscale structures of equal or better compexity to those possible with more "conventional" methods, but whereas each of the latter must be produces one at a time, the origami method generates billions of the structure at once.

Future directions: "Nanobreadboard" - a way of organizing any nanoscale size componenents (quantum dots, fluorophores, gold balls, molecular switches) in a generalized, arbitrary way.

Perhaps using this technique could be applied towards engineering a synthetic cytoskeleton in vivo.

"Biologically inspired nanofabrication" - Dan Morse

Consider the silica structures that comprise the shells of diatoms. The precision of nanoscale architectual control of the silica shell fabrication process exceeds human technology. The fabrication process is very mild, by human standards: low temperatures, high yields, etc. The scale of production is enourmous - "gigatons!"

Much of the current work has been focused on sponge organisms that naturally built silica structures. One of these sponges that can be dissolved into essentially a handful of glass needles. Research has indicated proteins form a backbone in the center of these needles. Further research indicated that silicateins (the axial proteins within the needle) are homologous to hydrolases, suggesting that they may be enzymatically active - as catalysts! "These proteins, at neutral pH and low temperature, catalyze and template synthesis of silica from SI-alkoxides." (The source of silicon in biological fabrication systems is silicic acid.)

So, on might ask, could this protein do the same from metal oxide precursors? Perhaps.

How does the silacatein work? It perhaps self-assembles into a giant irregular array in which the precursors bind and are catalyzed into the crystal. Now then, don't blink: the catalyst is the template! "That's a design principle; I don't know of any other enzymes, not even DNA polymerase, that acts as its own template." (what about prions?)

"Could we extract the principles that nature has evolved, and do it without the biology? And does that later intersect with Synthetic Biology? That is my question to you."

Dr. Morse then went on to describe how he used those principles to construct a novel material that was comprised of a conductive back plate from which sprouted a forrest of nearly perpendicular cobalt-hydroxide plates. It's a p-type semiconductor. It's got a high surface area, and an extremely high purity, because the process never needed them (it was just a vapor deposition of a modified catalyst onto the oxide precursor... I think) - perhaps it could be very useful in energy applications. Because they contain no organics, the fabrication methods are "fully integrable with existing nanoscale (CMOS) methods."

Dr. Morese concluded with octopuses... or more specifically that octopuses posses adaptive, flexible, multifunction arrays in three layers in their skin for adaptive camouflage. This could be a great foundation for a new kind of flexible display system.

This was a great talk, a little technical at times, but very fascinating. Dr. Morese, like Arash Komeili before him, is an extremely good speaker, and he was able to present the material in a lucid and engaging way. Check out the webcasts of their presentations.

"Magnetite biominerlization in bacteria" - Arash Komeili

Thus begins the MATERIALS SESSION and ends the ENERGY SESSION.

Magnetotactic bacteria have organelles called magnetosomes with which they can orient in (geo)magnetic fields. They prefer oxic-anoxic interfaces in their environment. By orienting against the geomagnetic field, they can control for lateral motion and search simply in a vertical direction for the interface.

Magnetite has also been found in protists, fish, and potentially honeybees. Magnetite (Fe3O4) is a single domain magnet, and usually of uniform size and shape within an organism. Applications? "Magnetofossils" can be used as biomarkers; contaminating metals could accumulated within magnetosomes to facilitate bioremidiation; perhaps it could be used for nucleic acid and protein purification; and perhaps most exciting of all, magnetite could be used as a contrast agent for MRI.

How are the magnetosomes built? Magnetosomes are not just a single-domain magnetic crystal - they are also surrounded by a lipid membrane, a membrane with certain unique surface proteins. What are the mechanisms of biominerlization? How are organelles developed in eykaryotic cells (very little is known about the endomembrane system in eukaryotes)? Understanding the mechanisms behind magnetosomes could help us understand.

Two ways to explore the question: High resolution imaging and marking. Use Magnetospirillum sp. AMB-1 in the lab (it's a microaerophile, 4-6 doubling time; magnetite production dependant iron concentration). TEM imaging indicates the magnetosome membrane is present before the crystal forms. Collaborated to conduct Cryo-electron tomography (so awesome!!!). Imaging of magnetosomes seem to indicate they are not actually vesicles, but continuous with the cell membrane - every magnetosome had a membrane neck continuous with the cell membrane visible at some angle. This goes for both empty magnetosomes and those filled with the magnetite crystal.

A 100kb "magnetosome island" has been identified that seems to code for the majority of the magnetosome system. They focused on MamK (MreB homologe, Actin-like protein; homologues involved in cell shape determination, plasmid segregation). The structures of MamK and ParM have both been solved. Fillaments can be identified running along the magnetosome chain, and it is hypothesized that MamK is involved in keeping the chain organized; when it is deleted, the chains become broken and disorganized.

Komeili finished by showing us a simply awesome 3D visualization of the magnetosome chain from the interior of the cell based on the cryo-electron tomography data. This was an awesome talk. Go watch the stream of it right now!

"Shotgun mapping of transcription regulation" - Timothy Gardner

Largely focused on network biology. Antibiotic development (IDed potentiators of cipro), bioenergy, and synthetic biology. In particular focusing on techniques to speed up determining what these networks look like. Techniques for characterizing metabolic networks are better developed than those for transcription.

The RS latch is the fundamental element in computer circuits. Why not implement it biologically? Construct a plasmid with two genes that cross-inhibit eachother... it worked, and others have built with it (Elowitz, Weiss, Simpson, the biophotographic film); it's been great.

But there's a problem - almost all these devices have been constructed with only two or three parts - the same two or three parts (Tet, CI, and LacI). Even if we have the capability to synthesize and assemble genetic systems of unlimited size, we don't have many parts to build with and there's way too much cross-talk with endogenous circuits. Maybe what we should focus on is how to interface with the existing "endogenous circuitry."

Shewanella oneidensis can serve as catalysts without mediators; they could potentially produce current from any carbon source: sugar, sewage, cellulose. But they don't produce alot of power... it takes a suitcase of bacteria to power a single portable electronic device.

One can find similar metabolic pathways, like those that could produce current in S oneidensis, through homology searches. How can we build genome scale transcription and translation models?

Fortunately, a database of expression and transcriptional controls exists for E. coli, which facilitated the creation of a mathematical model of genome-wide gene expression and transcriptional control based on Affymetrix microarray data. This model is driven by the CLR algorithm and based on analysis with mutual information instead of correlation.

The best way to identify the regulatory networks is to stress and test a given microbe in a great diversity of conditions, hence the term "Shotgun mapping."

"When I think about synthetic biology, I think it's more about the biology of synthesis, the synthesis of fuels, drugs, etc., and so I think it's important right now to focus on ways of understanding and controling existing biological circuits and machines." (paraphrased).

"Ethanol Production" - Nancy Ho

Nancy Ho is the leader of the Laboratory of Renewable Resources Engineering at Purdue Univerity and has developed recombinant strains of yeast with 20-30% increases in yield of ethanol fermentation (I think that's how the moderator introduced her).

Plants are abundant across the plaent, and cellulose comprises about 45% of the plant; over 75% of the cellulose can be converted to sugars, which in turn can be converted by yeast into ethanol. "There is no other natural organism better at fermenting sugars into ethanol than yeast." However, Saccharomyces is missing some metabolic machinery needed to utilize xylose, which is a significant componant of those sugars that can be made from cellulose.

I have a hard time understanding the details, but apparently Dr. Ho caused the gene for D-Xylulose (and other genes related to it in the metabolic pathway? KK, AR, KD?) to be greatly overexpressed by transforming industrial strains of yeast with a plasmid. Experiments showed ethanol concentration increased as glucose and xylose decreased as the transformed strains fermented the sugars; it worked. Glucose is almost completely used within the first 10 hours of fermentation, whereas xylose isn't depleted for more than 30 hours. Dr. Ho explained that xylose is transported into the cell by glucose transporters, which bind glucose 100 times more strongly than xylose - this explains the slower usage of xylose.

Dr. Ho mentioned some kind of legal pressures drove them to work with a new strain of yeast to avoid letting a single company control the product of their research.

Dr. Ho's powerpoint consisted of approximitely 400 graphs, capped by intro and thank you slides (and there may have been a couple non-graphs slides sprinkled in between).

Synthetic Biology 2.0

I'm in Berkeley at Synthetic Biology 2.0. I was taking notes by hand, but I think it would be more fruitful to just make a bunch of posts in near-real time. If you're reading this between

Craig Venter is telling us about the incredible diversity of microbial life; a diversity that is ripe for inspiring new innovation and techniques. Knocking out individual elements of biological pathways with transposons "isn't gonna cut it" anymore, so Venter is working on rebuilding a genome in a manner that is more conducive to manipulation.

He says "Designer Viruses" are over a decade away.
Eukaryotic cells are a bit closer... perhaps within 2 years.

How do you make a self-assembling chromosome from little oligonucleotide fragments? D. radiodurans knows who... it can reassemble its genome within 20 hours after getting disintegrated by tons of radiation.

So, what do you do with a synthetic chromosome once you get it? Perhaps introduce it into a cell with liposomes. (Keep in mind how difficult / poorly understood bacterial transformation is.)

These are hard tasks, but there is great incentive to solving them: The societies of the world need energy, and our current sources won't last. We need to design organisms that take CO2 and make methane, biopolymers, sugars, & proteins... nowadays we *make* CO2 from burning oils.

Venter played us a little video clip from the Discovery Channel: "...new nano life has the potential to make new energy sources available... Craig's vision is to sail and sample...[ha ha]." He said that in the video clip he got to drive Shell's hydrogen minivan around the block after filling it up from the first hydrogen station in DC, and then they loaded it back onto the flatbed truck they had brought it in on and returned it to the lab; the tank was empty when they got there. The challenges are great.

Tuesday, May 02, 2006

Quest for Laziness 1: Completion!

I think I've gotten an autologin script figured out. I'm not sure how useful it will be to non-davidson students, but who knows. It is triggered by automatic system events that occur in response to changes in network connectivity. Big props to wombert for his brilliant use of Kicker.xml and configd! No infinite loops here, no sir.

It's at pastebin to preserve formatting and encourage spontaneous improvement.

edit: It got accepted at macosxhints.com. Yay.

Tuesday, March 14, 2006

Quest for laziness 1: an end without means

The network authentication servers serve up a series of webpages which:
  1. recieve a user's username and password
  2. authenticate the user
  3. check for acceptance of usage agreement
  4. allow access
There must be a way to automate the series of HTTP requests which convey a user through this process, but first, we need to trace exactly what those requests are. To do so, we'll use the unix tool curl, with which we can practically craft HTTP requests by hand to send to the authentication servers, and then see everything they send back. We'll essentially be reverse-engineering the flow of a user's credentials through the authentication websites. Later, we can apply the fruits of these labors (hopefully) to the design of an automated greasemonkey script.

The first page a user typically encounters when reconnecting to the Davidson College campus network after an absence of some time is one that looks like the following, with a base url similar to https://cha0-na-a.davidson.edu/auth/perfigo_weblogin.jsp. The cha0-na-a subnet is different around the campus - bakn-na-a seems to work in the student union, and cha0-na-a in chambers, and there are probably others.

Authentication involves several .jsp pages, all of which share a root similar to https://bakn-na-a.davidson.edu/auth/, depending on the particular subdomain. This root will henceforth be omitted for the sake of brevity. Anyway, supplying the initial page with a successful username and password will bring the user to /perfigo_validate.jsp, which displays an usage agreement and accept / decline buttons (the terms and buttons are actually in two separate frames, each generated by a different .jsp page). The accept button loads a url similar to
/perfigo_cm_agree.jsp?userkey=10.10.10.10_XXXXXXXXXXXXXXXX&OtherFormData
where the 10.10.10.10 is the user's current IP address and the 16 Xs are an all-uppercase alphanumeric string. /perfigo_cm_agree.jsp creates a page of two inline frames. The top frame presents a usage agreement, and the bottom accept and decline buttons. Pressing the accept button leads to perfigo_cm_policy.jsp which, if all went well, announces the user has successfully logged on and loads their original destination in an annoying pop-up - if I didn't mention it before, all navigation is simply redirected to the authentication pages until successful login.

Let's begin exploring by checking out the source code of /perfigo_weblogin.jsp. From it we can see 10 forms elements, 6 of which are hidden, that are used to pass information like the username and password to /perfigo_validate.jsp via a POST method. This is where curl starts to come in really handy - we can use it with the -d flag to essentially POST our own form data to /perfigo_validate.jsp. The -d expects a string representation of the form data like this: uri=&cm=&userip=10.10.10.10&os=MAC_OSX& etc. As soon as I figure out how to display longer code snippets in a way that doesn't overlay them across the sidebar and doesn't require manually inserting line break (which would probably break a command if it was copy and pasted into the terminal, anyway), I'll post all the curl commands I used.

A quick wrap up for now: I've been able to successfully authenticate myself solely using curl - the last command involves passing the names and values of the hidden form data of the frame with the network usage agreement accept button to perfigo_cm_policy.jsp using curl's -d option and a custom string. I think only the dynamic elements in the form are named userkey and userip - the other 5 are static (maybe they can be omitted). We can get the userkey value by posting the hidden form data (username, password, ip, intercepted IP, operating system, etc.) to perfigo_validate.jsp and looking in the src="..." url for the bottom frame (name="agree").

It's fairly simple to trace the flow and structure of information from the initial login site to the final using curl. To find out more, check out Using cURL to automate HTTP jobs. Look here for an exhaustive reference on HTTP/1.1 (rfc 2616), for a great way of getting at the guts of javascripts, definitely try the Javascript Shell, explained a bit at DiveIntoGreaseMonkey.org (we'll visit more of that site next time).

Quest for laziness 1: Davidson network auto-login

Users attempting to access the Davidson College network from a computer running Microsoft XP (and perhaps other windows variants) are compelled to install a program called Perfigo CleanConnect, which subsequently installs more programs among them one called the ePolicy Orchestrator Agent, before being granted access. These programs essentially give the network administrators the power to make a user's system do certain things, such as always running a virus scanner in "on-access read/write" mode and installing the latest update from microsoft. Systems lacking the compulsory software or reported as infected, out of date, etc. can then easily be identified and appropriately quarantined or denied access.

Essentially speaking, the citizens of the network trade a bit of their freedom for increased security and reliability.

Furthermore, from a practical-slash-laziness perspective, Perfigo CleanConnect can be set to store a user's credentials and automatically log on to the Davidson Network whenever it is connected (ironically decreasing security). As a user, this is great: authentication becomes transparent and internet connectivity, via ethernet or wifi, seems to become a persistent and continuous quality of the computer.

I use an apple MacBook, however, and while I appreciate not being forced to install draconian software just to access the internet, clicking through the alternative web authentication interface every single time my computer wakes up from sleep is getting a little boring. In fact, if it takes about 5 seconds of my attention four times a day, then I'm losing roughly 6.53 minutes of my life each 14-week semester (thanks calcoolate.com).

There must be a better way. You can read about it here when I find it.

Wednesday, March 08, 2006

A list of future topics

  • Working with common ground in New Orleans over Spring Break
  • Printing the Davidson College logo with Fluorescent DNA
  • Bioinformatics project 2: a program to enhance the Standard Registry of Biological Parts
  • A little perl/applescript to automatically log into DavidsonWLAN
  • Finding a laser scanner for the Salamander Waterloss project
  • Other interesting events in my life...