Part of our project was collaborating with iGEM Marburg 2021 on a "Remote Lab" project. We aimed to build
a platform to connect future iGEM teams to share resources and equipment for remote collaborations. This way, a team with access to
top rate equipment could collaborate with teams with less resources for better results and characterization.
As a proof of concept of the possibilities given through remote collaboration, we tried to standardize a high-throughput method for
Golden Gate assembly using a Beckman Coulter Echo™ liquid handler. This machine is capable of transferring nanodroplets from a microplate
to another using sound waves. Fast and highly accurate, it also limits cross-contamination. Such a machine can be used to set up faster and
more accurate Golden Gate assembly reactions, accurately while saving resources since the assembly can be performed by using volumes as small
as 100nL per part.
Both teams performed all experiments to ensure reproducibility and in-depth troubleshooting.
Remote Echo Golden Gate Assembly
Each 2021 iGEM team has access to a total of 30kb in DNA synthesis, giving the opportunity for characterization
and optimization of their DNA constructs at low costs. Teams are not limited to test a single promoter or RBS upstream of
their coding sequence with this amount of DNA synthesis; they can optimize their constructs for maximum yield or limited
leakiness. However, when assembly reactions are set up by hand for many constructs, the process becomes tricky, time and
resource consuming. Having a Beckman Coulter Echo™ liquid handler provides a high-throughput alternative for setting up Golden
1. Process for Remote Echo Golden Gate assembly
However, only a few iGEM teams have access to such a machine. Therefore, this is the situation: on one side,
there is a team interested in performing a high-throughput Golden Gate assembly experiment but without access
to an Echo machine and on the other there is a team with access to a Beckman Coulter Echo™ liquid handler that
is willing to perform the Golden Gate assembly protocol. For the sake of simplification, they will be referred to as
team A and B respectively. Our goal was then to connect both teams and implement the following process:
The first step is to connect both teams, a DNA synthesis company and a shipping company. Team A should be
able to contact team B and prepare the constructs in a standardized manner. A platform will connect all actors
to facilitate exchanges and define clear rules. The second step is exchanging parts. Once team B receives all of
the DNA parts, they follow standard protocols for sample preparation, Golden Gate assembly using a Beckman Coulter
Echo™ liquid handler, high-throughput transformation and plating. During the third and last step, the colonies
containing the constructs of interest are sent for sequencing and sent back to the receiving team.
This process needs to be troubleshot and standardized for it to be scaled up and accessible to all
raises the following problems:
How does team A provide the Echo team with the parts for the Golden Gate assembly? What about the primers
for verification of the constructs?
How much can the steps be optimized, troubleshot and simplified to reduce the amount of work of
How can the constructs be verified?
Do the constructs need to be sent back to the interested team or can they be characterized by team B
2. Providing the DNA parts
Once team A designs the constructs to be assembled, they need to provide team B with the DNA sequences.
The sequences can be distinguished in two different categories:
Standard parts such as basic promoters, ribosome binding sites (RBS), reporter proteins and
Atypical parts such as specific coding sequences.
Standard parts can be found in the Golden Gate collection kit we created while atypical parts are to
be ordered by team A from a DNA synthesis company and shipped directly to team B. Thanks to the partnerships between iGEM and DNA synthesis companies such as IDT or Twist, ordering such sequences would be free of charge.
In the future, we would create a code to create the Cherry Pick protocol automatically.
3. Standardizing and simplifying the Golden Gate assembly
Our goal was to simplify and reduce the steps to be performed by team B. All steps of the process have been
troubleshot and standardized.
We created a very detailed protocol
on how to set up the Golden Gate assembly with the Echo machine.
High-throughput transformation and plating:
Once the plasmids are assembled, they need to be transformed into competent cells. For our proof of concept,
we decided to use NEB® turbo E. coli competent cells but any competent cells can be used.
First, 20uL of competent cells are added directly to the PCR wells containing assembled products for two different reasons:
The products are at the nanoliter scale and cannot be pipetted efficiently;
The transformation is set in the same wells as the Golden Gate reaction.
The transformation steps are then automated in a thermal cycler as follows:
200µL of LB broth is then added to the wells, the plate is sealed and placed to incubate at 37°C with shaking for 1 to 2 hours. Simultaneously, a previously prepared OmniTray® (with LB agar and required antibiotics) is
incubated at 37°C. This allows for fast liquid absorption by the medium and also ensures the samples will not
mix when plated multiple at a time using a multi-channel pipette. 5uL of cells are then plated onto the OmniTray®.
Once the medium has completely absorbed the cultures, it is incubated overnight at 37°C. Single colonies can be observed the next day (Figure1).
Figure 1) Picture of an Omnitray with single colonies after high-throughput transformation of minipreped plasmids.
Unfortunately, we were not able to obtain similar results with constructs assembled with the Echo machine.
As the volume of each construct is very low, the transformation efficiency was not high enough to obtain at least one assembled and successfully transformed plasmid per 5uL. Traditional plating of 200uL on Petri dishes
was done in parallel and allowed for all constructs to be recovered.
The protocol for high-throughput transformation needs to be improved to face this problem.
Here is the protocol for
the HTP transformation.
Verification step and sending the constructs back:
The last steps consist in verifying the constructs and sending them back to team A. We explored various techniques
for product verification. The first was colony PCR which presents various drawbacks. As many steps must be performed
(cell lysis, reaction set up, amplification cycles, electrophoresis), it is time and resource consuming especially if
many constructs are handled by a single team. It was then decided that sending the products for sequencing would be
the fastest option. Samples can be sent for sequencing:
As PCR products. The drawbacks are similar to those of the colony PCR;
As purified plasmids. Requires for the products to be transformed, the cells grown (on agar and in a liquid
culture) to be finally purified as miniprep products;
As cell cultures;
As single colonies.
In order to reduce to a minimum the steps to be performed by team B, it was decided the last two options were the best.
The companies Genewiz and MicroSynth were contacted with the goal of forming a partnership for future years.
After a meeting with a MicroSynth representative, we were given the opportunity to send samples for sequencing for a
reduced price in a format called Ecoli NightSeq®. A colony is inoculated in a tube containing a special buffer provided by the company and the samples are dropped into a MicroSynth dropbox. The samples are then Sanger sequenced and the
results come in the next day. Standard primers provided by the company can be used free of charge. Therefore, all constructs should contain these primers. This is a great option because not only do the results come in fast, but also
they reduce the number of steps to be performed by team A to one, inoculating a colony.
Once the results from the sequencing arrive, the samples can be sent back to the receiving team as stab
cultures in which a colony is inoculated. All the shipments will be grouped and centralised to reduce environmental
and financial costs.
Here is the protocol
to send the samples back to the user.
4. Troubleshooting steps and lessons learned
When experimenting on this project, various steps required troubleshooting, especially the Golden Gate assembly
using the Beckman Coulter Echo™ liquid handler. Many mistakes were made. The list below shows the lessons learned.
Upon reception of plasmids directly in an Echo source plate, transform 1uL to create a glycerol stock of
them. This will ensure their availability in case of an incident with the source plate.
The Echo machine can only be used with low-profile skirted PCR plates.
A PCR plate compatible with both the Beckman Coulter Echo™ liquid handler and the thermal cycler
available should be used. Plates for the Beckman Coulter Echo™ liquid handler must be skirted and
have a standard height whereas the thermal cycler we used could only fit non-skirted plates. To solve
this problem, a 3D printed piece was designed to adapt the PCR plate to the Beckman Coulter Echo™
Figure 2) Picture of the 3D printed adaptor to fit a simple PCR plate both in a thermalcycler and an Echo machine.
The PCR plate must be sealed tightly directly after the reactions have been set up. The reaction volumes
are around 1uL, low enough for the water to evaporate very quickly. Aluminium seals can be used to
prevent evaporation from the plate.
The Cherry Pick software varies depending on the version of the machine. The initial protocol was set up
by iGEM Marburg for an Echo 525 but was not compatible with the Echo 550 accessible to our team. The plate
type 384_AQ_CP should be favored as it allows calibration for a wide range of fluid types. The protocol
should be set up with the following column names: Source plate name, Sample group, Source well, Destination
plate name, Destination well, Transfer volume and Part name (optional). For each liquid transfer, the sample
group should be defined. It should be 384PP_Plus_AQ_GP2 for the enzymes (T4 ligase and BsaI-HFv2) and
384PP_Plus_AQ_SP2 for the DNA parts and buffer. The calibration of the machine varies depending on the
fluid viscosity. Samples in water require less pulse than samples stored in glycerol which are more viscous.
Enzymes are typically stored in 50% glycerol.
A trial using various fluids should be run beforehand to ensure accurate volume transfer by the
machine depending on liquid viscosity. A simple protocol can be done using water and 50% glycerol.
Various volumes can be transferred to a plate sealed with an aluminium seal. The seal enables a good
visualization of the drop transferred and helps determine accurate pipetting. For each fluid type,
various volumes can be transferred to different wells and compared.
We decided to test our protocols with a useful experiment, the characterization of a RBS library. The
Community Collection of RBS (1) makes available RBS which are suitable for general protein expression in E. coli
or other prokaryotes. The collection is known to cover a range of translation initiation rates. Our goal was to further
characterize them using the following construct:
Figure 3) Plasmid map of the general construct. Only the RBS is changed. The CDS is sfGFP of sfGFPcat.
BBa_J23101 is a medium strong promoter of the Anderson library of promoters. Like the Community Library, they allow for a
wide range of transcription rates for fine tuning of protein expression.
Two different coding sequences (CDS) were characterized: superfolder green fluorescent protein (sfGFP) and sfGFP with a
chloramphenicol tag upstream (CamTag-sfGFP). The sfGFP sequence was codon optimized. It was reported that the identity of
the amino acids encoded by codons 3 to 5 impact protein yield (2). The first 5 amino acids of the chloramphenicol resistance
gene and a linker sequence were added at the 5' end of the sfGFP sequence. They should enable a higher protein expression as
well as enabling similar translation rates for different CDS. It has been shown that the gene sequence downstream of the RBS
has a clear impact on the resulting translational profiles (3). By adding this CamTag, the translational rates under a given
RBS are similar whatever the downstream gene is. This is to be proven by comparing the expression levels of sfGFP and
CamTag-sfGFP in the above construct.
Transcription is stopped by the double terminator BBa_B0015 (4). It is an efficient and standard terminator used in many constructs.
The transcriptional unit is inserted in a backbone with a pMB1 origin of replication and kanamycin resistance.
Before conducting the actual measurement we input our constructs online in the RBS calculator tool (5) to see if changes in
translation initiation rates could be detected. We plotted the gained data to be able to compare them with our final data we
gained from the experiment.
Figure 4) Bar plot of the predicted strength of the analyzed RBS Collection from the RBS calculator.
Once built using the Echo Golden Gate protocol previously described, a timecourse experiment is performed.
Constructs were verified by sequencing using the method previously described. Growth (through OD600 measurement) and fluorescence is measured over 9 hours. The
plate reader was calibrated with fluorescein using the. standard iGEM protocol.
2. Results of the experiment
Figure 5) Bar graph of the normalized fluorescence of the different RBS upstream of sfGFP or sfGFPcat in NEB Turbo. Data taken after 8 hours of incubation at 37°C as the mean of 14 replicates.
As expected, the dummy RBS does present very low fluorescence (due to the autofluorescence of the cells). However,
the other results are different from what was expected. The fluorescence should be increased with the cat tag.
This is the case only for B0029 and B0036. The opposite is observed for the other RBS. We expected B0030 and B0035 to be
the strongest but the data presents them as respectively, a dummy and a weak RBS.
However, the standard deviations are acceptable.
When taking a closer look at the OD data, we observed a steep decrease in OD at around 3 hours for all of the cells
until approximately 5 hours. Growth then starts again as normal. Because this drop affected all samples, we hypothesize a malfunction of the machine to have affected the samples.
However, this should not have affected the relative strengths of the RBS.
We plan on performing the experiment again to either confirm our results or to obtain results closer to the expectations.
3. Troubleshooting and lessons learned
During this step we also ran into several problems, which we did not account for in the beginning. Some of these issues
were particularly difficult to troubleshoot, as there is only a limited amount of literature on it. That is why we think
it is even more important to mention these problems here.
First of all, we unfortunately had to realize that there is no standardized M9 media recipe available for the growth of E.coli.
The 2 teams had different protocols. Furthermore, some of the recipes available do not account for the problem of precipitation of the added calcium chloride. If this compound is added too quickly and to low amounts of liquid it will precipitate and therefore be inaccessible to the cell. Therefore, we added calcium chloride as the last ingredient and while agitating using a magnetic stir bar.
After extensive comparison of different M9 recipes we decided on one media recipe in order to keep the data more comparable between
the two labs.
With this M9 medium we then compared the viability of different E. coli strains (NEB10 beta, NEB stable and NEB Turbo) in this medium.
To our surprise, we saw that the NEB10 beta strain did not grow at all on this medium, which we could not explain from its genotype.
We then settled to use the NEB Turbo strain as it was the most accessible one to both teams.
When conducting the actual measurement using a 384 well plate, we ran into the problem of calibrating the gain for the fluorescence signal of the plate reader. We tried to let the plate reader determine the best gain for the measurement by directing it to a well supplemented with 10µM solution of fluorescein. Unfortunately, the plate reader software encountered a bug, which resulted in a well being chosen as a reference with a very low fluorescence signal. Consequently, the growth curve resulted in the saturation of fluorescence signal for most wells, making the data unusable.
One of the major challenges we were aware of from the get-go was the extensive planning necessary to get the genetic parts
from team A to team B. While the generous DNA synthesis offer from companies like IDT and Twist are incredibly helpful in
setting up such an endeavor, it still requires a variety of logistical considerations that are both time-intensive and require
long-distance shipping. Since many of the constructs built by iGEM teams utilize a set of most common parts, the use of a
standardized part collection could address these issues while significantly reducing both overall cost and time.
One of the obvious candidates for such a collection is the iGEM Distribution, which is sent to all participating teams each year.
Since its initial creation in 2006, the distribution has been an essential part of the iGEM experience and builds the foundation of projects from teams all over the world. As the community has grown, so has the distribution, which now contains over 2000 of
the best parts in the competition. However, in a scientific discipline as rapidly evolving as synthetic biology, it is evident
that after 15 successful years, a new generation of distribution is needed.
An opinion that is also shared by the iGEM Foundation, which has announced its goal to create the second generation of the
distribution, the world-best biotechnology toolkit supporting both education and innovation. With such ambitious goals in mind,
the iGEM Foundation turned to the community and asked this year's teams for their honest feedback on the current distribution and their wishes for Distribution 2.0.
We recognized this unique opportunity and thoroughly investigated the current distribution. While doing so, we were not only
interested in examining the distribution in the context of our project, but also in potential obstacles that could hinder the
widespread adoption of such a collection.
Our goal was to create a new core distribution kit for future iGEM teams using Type IIS assembly. This kit would give the teams access to a collection of promoters, RBS, CDS, terminators and backbones ready for Golden Gate assembly. Receiving teams would
be able to use the standard parts of the kit in their constructs without having to send them to the Echo team, which would already
have them in stock.
The kit is composed of the iGEM 100 most used parts trimmed to remove similar parts. Other useful parts are added. A separate
collection for origins of replication and antibiotic resistance was created to produce tailored backbones. A connector library
is added for multiple level assembly. All parts are cloned in a pSC1B3 derivative backbone, modified to fit type IIS assembly.
Many parts of the kit were inspired from the Marburg Collection (6). The kit is documented on our GitHub repository. All information about the guide can be found on the Type IIS collection spreadsheet and all Genbank files
are accessibles for download.
For this analysis, we reflected on our own decisions during the planning stage of our projects and asked ourselves the
questions of why we did not incorporate the 2021 distribution to a greater extent in our project. We quickly realized
that one of the main reasons is the relatively difficult and non intuitive way in which teams currently interact with
the distribution online experience. An opinion that was also echoed by several other teams we’ve met during this year's meetups. Encouraged by these
discoveries, we have taken an in-depth look at this problem and identified 2 key issues:
Missing or conflicting relevant information;
Lack of user-centered design.
To get information about the parts contained on the 6 distribution plates, teams currently have to click through a series
of sites to arrive at a plate-specific page that contains the relevant data.
Albeit useful, the information contained here is rather limited and is mostly of use to find the location of already selected
parts. Somewhat concealed, users can find the link “Get a detailed Excel file for this plate” which links to the corresponding
Curious about possible differences between the two versions, we compared them against each other and found that one contains in
fact more parts than the other. This discrepancy originates from parts that are no longer included in the distribution due to
copyright concerns. The finding was immediately reported to the registry. Although the CSV file contains more information than
the html table, a closer look reveals that a large part of the information is still missing.
Out of the total 15 columns, 4 columns (Resistance, Gel Overall, Quantity, Seq Comment) contain no information, 2 columns
(Sequencing, Well Status) contain just 1 unique value and 1 column (Plasmid) is duplicate.
One interesting point to raise is the sequencing results available in the datasheets. The registry generally differentiates
sequencing results into one of 7 categories.
Partially Confirmed - software is only able to partially confirm sequence, most likely due to one read being poor
Long Part - length of sequence reads are insufficient to cover the middle of the part
Inconsistent - part does not match its target sequence, may have a single bp mutation or not match at all
Bad Sequence - usually caused by low DNA concentration or incorrect primers
No Sequencing Information
Interested in how often the respective categories occur, we have created a visualization tool that colors the respective positions on the 384well plate for each of the parts. Considering that even a single point mutation can substantially change the behavior of parts, we have decided to simplify the graph and split it into one of 2 categories - confirmed and not confirmed.
Figure 6) Sequencing data of the current iGEM Distribuion Kit. Wells in green are comfirmed while wells in red are not.
The resulting graph illustrates, in our opinion, a major problem of the current distribution. Given that teams may build their project on constructs that use one or more of these parts, it is of the utmost importance to ensure that they can rely upon the
sequence given on the part page. Failure to do so can, in the worst case, lead to non-functioning constructs and a substantial
time loss due to troubleshooting the associated problems.
Putting this issue aside, we once again concentrated on the part overview. As previously mentioned, we believe that the
other major issue with the current distribution is a lack of user-centered design. In particular, we want to highlight
the discrepancy between the information design as it currently exists and the workflows that teams want to use.
One of the main reasons for this is the lack of important information such as part type, source organism, and cloning compatibility in both the HTML as well as CSV overview. To use the distribution to its full extent currently requires downloading and analyzing all 6 CSV files, searching and opening up each of the >2000 part pages, and cross-referencing all these relevant resources. Evidently, something that is virtually impossible without the use of additional tools.
Adding to this, many users want an easy way to download the parts included in the distribution in standard formats such as
GenBank or SBOL, something that is currently only possible with the help of third-party providers. The absence of plasmid
maps further poses an unnecessary hurdle for many teams, something which should be addressed by future distributions.
To address some of these issues, we created an in silico version of the 2021 distribution that contains all parts, including
sequence annotation, as a GenBank file that can be found on our GitHub.
Furthermore, we searched for all plasmids that are used in the distribution, downloaded the corresponding sequences, and
reannotated them using bioinformatics tools. Something that was necessary since many plasmids had either missing or no
annotations at all. The resulting plasmids can also be downloaded from our GitHub.
To allow the tool to be both updated as well as adapted in a different context, we decided to make the data represented here
also freely available as a CSV file.
With this analysis in hand, we set out to contact the iGEM HQ and arranged a zoom call with the director of the registry -
Vinoo Selvarajah. In this meeting, we summarized our results in a short 30-minute presentation ( link to the slides).
The ensuing discussion gave us valuable insights into the complexity of distribution, its inner workings, and steps that are currently taken to improve it.
We were excited to see that our concerns and analyses were taken and how open iGEM is to feedback. Moreover, we were very
impressed by how seriously the wishes and opinions of the community are treated with regard to the new distribution.
Ultimately, we hope that our work this year will have a positive impact on the new distribution and enable future iGEM teams
and scientists around the world to utilize this toolkit to its fullest extent.
(1) Rubin, Weiss Kelly Bryant (n.d.).Ribosome Binding Sites/Prokaryotic/Constitutive/Community Collection.http : / / parts . igem . org / Ribosome _ Binding _ Sites / Prokaryotic / Constitutive / Community _Collection. Accessed: 2021-09-01
(2) Verma Choi, Cottrell et al. (2019). “A short translational ramp determines the efficiency of protein synthesis”.In:Nat Commun10.5774.doi:https://doi.org/10.1038/s41467-019-13810-1.
(3) Thiel Mulaku, Dandapani et al. (2018). “Translation efficiency of heterologous proteins is significantly affected by the genetic context of RBS sequences in engineered cyanobacterium Synechocystis”. In:Microb CellFact17.34.doi:https://doi.org/10.1186/s12934-018-0882-2.
(4) Shetty, Reshma (2003).Part:BBaB0015.http://parts.igem.org/Part:BBa_B0015. Accessed: 2021-09-01.
(5) Salis, H., Mirsky, E. & Voigt, C. Automated design of synthetic ribosome binding sites to control protein expression. Nat Biotechnol 27, 946–950 (2009). https://doi.org/10.1038/nbt.1568
(6) Stukenberg, D., Hensel, T., Hoff, J., Daniel, B., Inckemann, R., Tedeschi, J. N., Nousch, F., & Fritz, G. (2021). The Marburg Collection: A Golden Gate DNA Assembly Framework for Synthetic Biology Applications in Vibrio natriegens. ACS Synthetic Biology, 10(8), 1904–1919. https://doi.org/10.1021/acssynbio.1c00126