Note: I am trying to write on this blog as a New Year’s resolution, and today’s google doodle made me write up some thoughts and memories of Khorana.

In the spring of 2002, I was in my first year of graduate school and joined Drew Endy’s lab, who had just started as a fellow on the sixth floor of the biology building (building 68). Our floor then included several senior faculty included Khorana and Boris Magasanik. When I was started, I had no idea who they were and they just seemed to be old people that I tried to avoid. I still remember my first meeting with Khorana, which in retrospect was mortifying.

I was working at the computer, in a basically empty lab space, and a slight, old man with a gigantic cup of hot tea held in both hands walked by slowly taking a long look. Then he walked by again going the other way, and then back again, each time peering in looking curious. I kind of thought it was one of those Indian things where the older Indian generation still seemed shocked every time they see another Indian person in America. I’m thinking mostly that I should probably say hi and help him move along:

Me: Hi, my name is Sri, can I help you?
Khorana: Hi, I’m Khorana
me: …
Khorana: I like seeing new labs start. What are you working on?

I had no idea who he was. I just saw a nice old man that at the time I thought was wasting my time. I was working on a simulator for gene expression, to build a model for a simple virus that infects E. coli called bacteriophage T7. I didn’t really want to discuss with him what I was working on, because I didn’t know where to begin, and I was 22 and pretty confident I had nothing to learn from him. I was luckily rescued by Drew, who brought him into his office.

In the ensuing years, I began diving a lot deeper in the biology of T7 and models of gene expression, which led me to many papers that were published in the Cold Spring Harbor Symposia series on Quantitative Biology that were organized by Demerec (who also first isolated T7). It was really amazing reading the history of your field through these symposia, and how people thought about the emergence of molecular biology. There was an especially eventful symposium, 1961 on regulatory mechanisms, where many of the luminaries like Jacob and Monod, Brenner, Crick and others were formulating early ideas around the central dogma. I could not find them online, nor in the library, and for some reason that day I knocked on Boris Magasanik’s office. I asked if he had a copy of a particular paper from that issue. He said, sure, he was there, and pulled out the book. I was pretty stunned. We talked for something like 3 hours in his office that day about the symposium and what people were talking about. I ended up having many talks with him, and took his course on bacterial physiology (which I loved). I dove deeper into the symposia, as Boris had most of them, and came upon the famous 1966 meeting, where I ran first ran into Khorana’s work. Around the same time, I began reading the Eighth Day of Creation, where I read more about the excitement around that time and Khorana’s formidable accomplishments. Over the next several years, I talked with Boris, and Gobind, but also Maury Fox, Vernon Ingram and others. I feel both blessed to have experienced that, but also regretful that I didn’t spend more time talking with all of them, especially Khorana.

Our lab’s work today is deeply inspired by Khorana’s work and approaches to use nucleic acid synthesis and gene assembly to explore important aspects of biology. It didn’t matter that he was solving the genetic code using unnatural sequences, in the end it was still the genetic code. Moreover, it turned out to be easier to break this code by using synthesis, rather than by genetics or inference from natural sequences. These are concepts that have taken me a while to understand, but Khorana clearly understood it decades before I was born, and doggedly pursued those goals producing spectacular work that changed the world.

My explorations of Khorana’s works were really fun. For example, there is the synthesis of the first artificial gene, a yeast tRNA, with the interesting title: “Studies on polynucleotides. 103. Total synthesis of the structural gene for an alanine transfer ribonucleic acid from yeast.” That 103 is amazing. It’s actually the 103rd paper with the same title, starting from the first in 1958, to some of the last ones I could find that end in 1976 at 143. He did this a lot apparently. Another great example is the synthesis of the second gene, published in a numbered series of 17 back-to-back papers entitled: “Total synthesis of the structural gene for the precursor of a tyrosine suppressor transfer RNA from Escherichia coli”. Back to the first series, Ian Molineux was on my thesis committee, and worked for Khorana as a postdoc. He pointed me to one that he was an author on, #96, entitled “Studies on polynucleotides: XCVI. Repair replication of short synthetic DNA's as catalyzed by DNA polymerases.” It basically describes PCR, and according to Ian, the method was routinely run a full decade prior to it’s invention by Mullis, but they didn’t think it was worth publishing. Read the final paragraph for instance:

“The principles for extensive synthesis of the duplexed tRNA genes which emerge from the present work are the following. The DNA duplex would be denatured to form single strands. This denaturation step would be carried out in the presence of a sufficiently large excess of the two appropriate primers. Upon cooling, one would hope to obtain two structures, each containing the full length of the template strand appropriately complexed with the primer. DNA polymerase will be added to complete the process of repair replication. Two molecules of the original duplex should result. The whole cycle could be repeated, there being added every time a fresh dose of the enzyme.”

Anyways, Khorana continued to be the only person building synthetic genes for a while, and it spawned an entire generation of people and an approach to biology that is still alive today. Part of the reason I cringe when people refer to me as a synthetic biologist, is because it seems to erase a history that is as old as molecular biology itself. It is very weird to me that I worked down the hall from him, as much of what he did seems to be from antiquity. I wish I had a chance to talk with him again, but at least I still have many chapters of Studies of Polynucleotides to get through.

—Sri

We are excited today to release DropSynth to the world in a new paper published online in Science Magazine. Briefly, DropSynth is a simple, low-cost method to build thousands of genes from microarray-derived oligos in a single reaction. In the published work, we build thousands of homologs of two essential bacterial genes from oligo arrays (~450-675bp in length), and then characterize homologs of the essential enzyme PPAT for their ability to complement in E. coli.

We expect this to be the first report of a method which we will continue to improve in scale, cost, length, and ease over the coming years. DropSynth is a fairly straightforward protocol that should be easy for individual labs to perform with no specialized equipment. We will gather useful documentation for the method at http://dropsynth.org. In the meantime, we’ve been asked questions by many others while presenting the work to colleagues, and thought we’d detail some of those answers below.

What is DropSynth?
DropSynth is a new gene synthesis method whereby large gene libraries are assembled from microarray-derived oligo libraries within water-in-oil droplets. Previous gene synthesis methods require isolating individual gene assemblies in different reactions, which becomes cost-prohibitive for assembling thousands of genes and requires expensive automation equipment. DropSynth overcomes this barrier by isolating individual assembly reactions in a vortexed emulsion, allowing for the entire gene library to be assembled in one pot.

How does it work?
DropSynth uses a set of barcoded beads, such that each bead pulls down the required oligos for a particular gene’s assembly. The beads are then emulsified, thereby isolating and concentrating the oligos into a picoliter-sized droplet. Then the barcodes are removed, and gene assembly takes place inside of those droplets by polymerase chain assembly (basically PCR). We then break the emulsion and recover our library of assembled genes.

How does it work in more detail?
An animation of DropSynth can be found at the bottom of this post. Briefly, the genes to be synthesized are first bioinformatically split into several fragments, such that each fragment can fit on an oligo. Restriction sites and a microbead barcode are added to each oligo. All of the oligos needed to assemble one particular gene are given the same microbead barcode. Typically, the library is split into batches of 384 genes, each with a unique pair of sub-pool amplification primers. After the oligo design is completed, the library is ordered from a commercial oligo pool vendor. Upon arrival, each sub-pool is PCR amplified from the pool with a biotinylated primer. The amplified oligos are then digested at high-temperature to expose the microbead barcode as a single-stranded DNA overhang. Streptavidin-coated beads bind and remove the small biotinylated fragment from the digestion. Processed oligos are mixed with a pool of 384 barcoded microbeads, with each microbead containing only one complementary barcode sequence. Complementary oligos hybridize and are ligated to the microbeads. Any excess or unbound oligos are washed away. The loaded beads are then mixed with PCR reagents, a restriction enzyme and some fluorinated oil. This mixture is then vortexed for several minutes to form a water-in-oil emulsion, which is placed into a thermocycler where the restriction enzyme displaces the oligos from the bead and the gene assembly reaction takes place inside the droplets. Upon completion, the aqueous solution containing the assembled genes is recovered from the emulsion and PCR-amplified again for downstream applications.

What can you build with DropSynth and what are the limitations?
We show in the paper that we can build thousands of genes of length ~450-675bp. It’s likely you can build longer genes, but the limitation is the error rate. Because oligos coming off an array have large error rates (only 50% or so are perfect), assembly of 4 or 5 oligos leads to a maximal rate of perfect synthesis of ~1-5%. The more oligos you use to build a gene, the less likely they it is to be perfect. Gene synthesis companies that use microarray-derived oligos often use enzymatic error correction to get passed these errors, but is difficult to do within the droplets (though we are working on it). Thus, DropSynth is suited for applications where the error rate isn’t as important, for example when you have a multiplexed functional assay or are doing a selection.

Why did you develop DropSynth?
It is difficult to reduce costs of gene synthesis below their current costs of a few cents a base. If you want access to much larger libraries of DNA to characterize that would be currently cost-prohibitive for an individual lab to do; e.g., all promoters of many yeast species, or tens of thousand synthetic gene designs, or genetic variants of thousands of human exons, or in our current example thousands of homologs of a protein from across the tree of life. In our lab, we do this a lot as we have developed many methods to characterize large libraries of genes in multiplex for sequences controlling transcription, translation, splicing, and gene function (this work). Right now, what limits these approaches is the length of oligo libraries to input into these functional assays.

How much does it cost?
The initial outlay is primarily in making the barcoded microbeads. In our hands, this costs around $3400 to create a pool of 384 barcoded microbeads, sufficient for around 200 DropSynth reactions (¢4 per gene). The costs of the oligos and other reagents required are about $1-$2 per gene, and are shown in Table S4 of our publication.

What is it useful for?
Since assembled gene libraries are pooled together, DropSynth-generated libraries are particularly useful as an input to multiplex assays, where many DNA encoded hypotheses are barcoded and tested together. In addition, since error rates are high, the method is useful when your screening methodology is very cheap, so you can screen 10-100x more variants than your library size. High-throughput sequencing is then used to evaluate each hypothesis by tracking the relative proportion of each barcode after exposure to some functional screening.

What is it not useful for?
Given their pooled nature, DropSynth libraries are less useful if each gene must be recovered and individually tested. Although we show we can use techniques like Dial-Out PCR to recover individual genes using the barcoded plasmids, extracting each gene out of a DropSynth pool would require significant investment in robotic automation to get large libraries of perfect constructs.

What are the error rates?
Error rates are primarily dependent on the oligo source, number of oligos in the assembly, and the polymerase used. We typically see a median of 2% to 4% perfect assemblies for 4-5 oligo assemblies using the Kapa Robust polymerase.

Where can I get DropSynth reagents?
All DropSynth reagents can be obtained through major suppliers. Microarray-derived oligo pools can be purchased from Agilent Technologies, Twist Bioscience or CustomArray, Inc. Primers used in sub-pool amplification and assembly can be obtained from IDT, DNA polymerases can be obtained from KAPA Biosystems, and restriction enzymes can be obtained from NEB. We are currently exploring avenues to make our pool of 384 DropSynth microbeads available to the scientific community. As of now, these microbeads can be generated using Protocol 1 on http://dropsynth.org, and can be used for up to 200 pooled assembly reactions before being depleted.

Who can I contact with more questions about DropSynth?
We intend for DropSynth to exist as a living protocol, with continuous input and optimization from the scientific community at large. Questions and comments can be made in the Discussion section on http://dropsynth.org, and additional queries can be made to Calin (plesa@ucla.edu), Angus (sidore@ucla.edu) or Sri (sri@ucla.edu).