Team:MIT/ResultsFoundational

From 2012.igem.org

(Difference between revisions)
m
Line 11: Line 11:
<div id="col_left">
<div id="col_left">
-
<div class= "section" id="b1bio">
+
 
-
<h1> Overview: </h1>
+
<h1>Overview: </h1>
<p> Will include cartoon on strand displacement with our best microscopy and link to find out more about how strand displacement works </p>
<p> Will include cartoon on strand displacement with our best microscopy and link to find out more about how strand displacement works </p>
<br>Will link in vitro work to nucleic acid delivery to in vivo work  
<br>Will link in vitro work to nucleic acid delivery to in vivo work  
<br>  
<br>  
 +
 +
 +
<!-- design -->
 +
<div class= "section" id="b4bio">
 +
<h1>Designing Strands for <em>In Vivo</em> Strand Displacement</h1>
 +
 +
<h2>Porting <em>In Vitro</em> DNA Sequenes to <em>In Vivo</em> RNA Strands</h2>
 +
 +
We used the DNA S6 domain and toehold sequences used by [http://www.sciencemag.org/content/332/6034/1196.abstract Qian ''et al.'' 2011], and converted them into RNA. Furthermore, we added 2'O Methyl modifications the ribose of each base. This is a widespread RNA modification (e.g. [http://www.ncbi.nlm.nih.gov/pubmed/19025401 Behlke ''et al.'' 2008]) to protect it from its own catalytic 2'OH, which can catalyze backbone cleavage. It is also beneficial, as a modified base can protect the strand from enzymatic cleavage by rendering it to a non-substrate.
 +
 +
<h2>Improvng Sequence Designs for <em>In Vivo</em> Strand Displacement</h2>
 +
 +
<h3>Motivation for Improved Sequence Designs</h3>
 +
 +
Initial results of <em>in vivo</em> strand displacement suggested that the strand sequences used by Qian ''et al.'' for in vitro experiments is not suitable for in <em>vivo work</em> due to the reporter complex coming apart inside cells (see [https://2012.igem.org/Team:MIT/ResultsFoundational#b3bio <em>In vivo</em> strand displacement]).
 +
 +
 +
There could be various reasons for the reporter complex coming apart. For example, the incubation temperature for cells (37&deg;C) might be too high for the annealed reporter, or enzymes might be degrading the strands. Thus we set out to design new sequences that:
 +
 +
 +
<ul>
 +
<li>have increased thermodynamic stability (greater T<sub>m</sub>, the melting temperature of a double-stranded complex)</li>
 +
<li>are orthogonal to RNAs expressed in HEK293 cells</li>
 +
<li>are more protected from nuclease activity</li>
 +
<li>are more protected from the RISC complex</li>
 +
</ul>
 +
 +
 +
We achieved this by:
 +
 +
<ul>
 +
<li>increasing the domain and toehold lengths</li>
 +
<li>increasing the G/C content</li>
 +
<li>analyzing the HEK293 transcriptome</li>
 +
<li>researching literature for protective modifications</li>
 +
</ul>
 +
 +
 +
<h3>Sequence Selection Algorithm</h3>
 +
 +
We analyzed HEK293 transcriptome data by first fetching the accession IDs from [http://webserver.mbi.ufl.edu/~shaw/293.html HEK293 Cell Database] and then the corresponding sequences from GenBank. Duplicate sequences were removed.
 +
 +
 +
We first chose an 8 nucleotide (nt) toehold sequence. The sequence space was restricted to a 3-base code (A/T/C) with the sequence ending in 5' CA 3' (clamp, see Qian ''et al.'' above). We calculated the [http://en.wikipedia.org/wiki/Hamming_distance Hamming distance] between each potential toehold and the transcriptome, along with the number of exact consecutive nucleotide matches from the beginning to rank each potential sequence. The resulting list can be sorted by deviation from the mean for each possible Hamming distance (0-8) and exact consecutive run (0-8). The best sequence would be one for which low Hamming distance instances are relatively (&gt;=1 standard deviation) less common and high Hamming distances are relatively more common. Similarly, shorter consecutive run instances are more favored over longer ones.
 +
 +
<code>
 +
5' UACUAC CA 3' // has repeats
 +
5' ACUAUC CA 3' // choose this one
 +
5' CCUAUA CA 3'
 +
5' CUAUAC CA 3'
 +
5' ACCUAU CA 3'
 +
</code>
 +
<em>Top 5 generated toeholds.</em>
 +
 +
 +
Then, the same approach was used to identify the 20 nt domain sequence. However, generating all sequences is very time intensive with roughly 3<sup>20</sup> combinations of 3-base code sequence to evaluate. Instead, a [http://en.wikipedia.org/wiki/Genetic_algorithm genetic algorithm] was used to generate sequences, evaluate them and to improve on the top scoring sequences by mutating them or performing crossovers with other sequences. After an overnight run of four threads of this algorithm, the best sequences were selected. Then the sequence with the highest G/C content and fewest repeats was chosen. In principle, the latter constraints can be added to the automatic scoring feature of the genetic algorithm.
 +
 +
 +
<h3>Protective Base Modifications</h3>
 +
 +
The original strands were modified with 2'O-Methyl groups on the ribose sugar. This is a widespread RNA modification (e.g. [http://www.ncbi.nlm.nih.gov/pubmed/19025401 Behlke ''et al.'' 2008]) to protect it from its own catalytic 2'OH, which can catalyze backbone cleavage. We kept these modifications on the newly designed strands.
 +
 +
 +
For strands without 3' fluorophore (ROX or Alexa 488) modifications, we added a phosphate group to the 3' end. According to [www.idtdna.com/catalog/Modifications/Modifications.aspx?catid=1 IDT], this protects from some 3' exonuclease degradation. Input strands received an IR800 modification on the 5' end for quantification and potential protection from 5' exonucleases.
 +
 +
<h3>Protective Sequence Modifications</h3>
 +
 +
To protect against the RISC complex, we made a variant of the reporter top strand that included a bulge 10 nucleotides from the 3' end ([http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3333889/ Iba ''et al.'', 2012]). There were 3 different bulges published by Iba ''et al.'', but only one of them, 5' CACU 3', used a 3-base code, so we chose this sequence as the bulge.
 +
 +
</div>
 +
<!-- /design -->
 +
 +
 +
<!-- in vitro strand displacement -->
 +
<div class= "section" id="b1bio">
<h1> RNA Strand Displacement In Vitro </h1>
<h1> RNA Strand Displacement In Vitro </h1>
<p>
<p>
Line 42: Line 117:
</div>
</div>
 +
<!-- /in vitro -->
 +
<!-- nucleic acid delivery -->
<div class= "section" id="b2bio">
<div class= "section" id="b2bio">
Line 114: Line 191:
</div>
</div>
 +
<!-- /nucleic acid delivery -->
 +
 +
<!-- in vivo strand displacement -->
<div class= "section" id="b3bio">
<div class= "section" id="b3bio">
Line 184: Line 264:
<br> What's next!
<br> What's next!
</div>
</div>
 +
<!-- /foundational -->
 +
<!-- end foundational -->
<!-- end foundational -->

Revision as of 22:43, 2 October 2012

DEPRECATED. DO NOT USE OR EDIT. If a page uses this template, relink with MIT-results2.

Overview:

Will include cartoon on strand displacement with our best microscopy and link to find out more about how strand displacement works


Will link in vitro work to nucleic acid delivery to in vivo work

Designing Strands for In Vivo Strand Displacement

Porting In Vitro DNA Sequenes to In Vivo RNA Strands

We used the DNA S6 domain and toehold sequences used by [http://www.sciencemag.org/content/332/6034/1196.abstract Qian ''et al.'' 2011], and converted them into RNA. Furthermore, we added 2'O Methyl modifications the ribose of each base. This is a widespread RNA modification (e.g. [http://www.ncbi.nlm.nih.gov/pubmed/19025401 Behlke ''et al.'' 2008]) to protect it from its own catalytic 2'OH, which can catalyze backbone cleavage. It is also beneficial, as a modified base can protect the strand from enzymatic cleavage by rendering it to a non-substrate.

Improvng Sequence Designs for In Vivo Strand Displacement

Motivation for Improved Sequence Designs

Initial results of in vivo strand displacement suggested that the strand sequences used by Qian ''et al.'' for in vitro experiments is not suitable for in vivo work due to the reporter complex coming apart inside cells (see [https://2012.igem.org/Team:MIT/ResultsFoundational#b3bio In vivo strand displacement]). There could be various reasons for the reporter complex coming apart. For example, the incubation temperature for cells (37°C) might be too high for the annealed reporter, or enzymes might be degrading the strands. Thus we set out to design new sequences that:
  • have increased thermodynamic stability (greater Tm, the melting temperature of a double-stranded complex)
  • are orthogonal to RNAs expressed in HEK293 cells
  • are more protected from nuclease activity
  • are more protected from the RISC complex
We achieved this by:
  • increasing the domain and toehold lengths
  • increasing the G/C content
  • analyzing the HEK293 transcriptome
  • researching literature for protective modifications

Sequence Selection Algorithm

We analyzed HEK293 transcriptome data by first fetching the accession IDs from [http://webserver.mbi.ufl.edu/~shaw/293.html HEK293 Cell Database] and then the corresponding sequences from GenBank. Duplicate sequences were removed. We first chose an 8 nucleotide (nt) toehold sequence. The sequence space was restricted to a 3-base code (A/T/C) with the sequence ending in 5' CA 3' (clamp, see Qian ''et al.'' above). We calculated the [http://en.wikipedia.org/wiki/Hamming_distance Hamming distance] between each potential toehold and the transcriptome, along with the number of exact consecutive nucleotide matches from the beginning to rank each potential sequence. The resulting list can be sorted by deviation from the mean for each possible Hamming distance (0-8) and exact consecutive run (0-8). The best sequence would be one for which low Hamming distance instances are relatively (>=1 standard deviation) less common and high Hamming distances are relatively more common. Similarly, shorter consecutive run instances are more favored over longer ones. 5' UACUAC CA 3' // has repeats 5' ACUAUC CA 3' // choose this one 5' CCUAUA CA 3' 5' CUAUAC CA 3' 5' ACCUAU CA 3' Top 5 generated toeholds. Then, the same approach was used to identify the 20 nt domain sequence. However, generating all sequences is very time intensive with roughly 320 combinations of 3-base code sequence to evaluate. Instead, a [http://en.wikipedia.org/wiki/Genetic_algorithm genetic algorithm] was used to generate sequences, evaluate them and to improve on the top scoring sequences by mutating them or performing crossovers with other sequences. After an overnight run of four threads of this algorithm, the best sequences were selected. Then the sequence with the highest G/C content and fewest repeats was chosen. In principle, the latter constraints can be added to the automatic scoring feature of the genetic algorithm.

Protective Base Modifications

The original strands were modified with 2'O-Methyl groups on the ribose sugar. This is a widespread RNA modification (e.g. [http://www.ncbi.nlm.nih.gov/pubmed/19025401 Behlke ''et al.'' 2008]) to protect it from its own catalytic 2'OH, which can catalyze backbone cleavage. We kept these modifications on the newly designed strands. For strands without 3' fluorophore (ROX or Alexa 488) modifications, we added a phosphate group to the 3' end. According to [www.idtdna.com/catalog/Modifications/Modifications.aspx?catid=1 IDT], this protects from some 3' exonuclease degradation. Input strands received an IR800 modification on the 5' end for quantification and potential protection from 5' exonucleases.

Protective Sequence Modifications

To protect against the RISC complex, we made a variant of the reporter top strand that included a bulge 10 nucleotides from the 3' end ([http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3333889/ Iba ''et al.'', 2012]). There were 3 different bulges published by Iba ''et al.'', but only one of them, 5' CACU 3', used a 3-base code, so we chose this sequence as the bulge.

RNA Strand Displacement In Vitro

Background:

In 2011, Lulu Qian and Erik Winfree, researchers at Caltech, published a paper entitled "Scaling Up Digital Circuit Computation with DNA Strand Displacement Cascades." This paper demonstrated how scalable logic circuits based on DNA strand displacement cascades in vitro are capable of processes as complicated as the square root function. See Motivation for more details.

MIT iGEM 2012:

Before our team attempted to bring the mechanism of strand displacement into an in vivo context, we first decided to assay strand displacement in vitro using RNA. We used 2'-O-methylated RNA strands, which had never before been shown to undergo strand displacement in vitro. Before creating our own constructs, we adapted sequences from the Qian/Winfree paper to RNA.

MIT iGEM Foundational Experiment:

Figure A shows a foundational in vitro RNA strand displacement experiment that was performed on a plate reader. The negative control, in black, is a well that received only an annealed reporter complex. The bottom strand of this complex is the gate strand, T*-S6*, with the 3' end tagged with the ROX fluorophore. The top strand of the complex is the output strand, S6. This is complementary to the S6* domain of the gate strand. The 5' end is tagged with the Iowa Black RQ quencher, which absorbs the ROX fluorescence; thus, when the two strands of the reporter are annealed, no fluorescence should be observed. The positive control, in red, is the input strand, T-S6, annealed to the gate strand, T*-S6* tagged with ROX. This is what we would expect the product of a strand displacement reaction to look like. We can see that in the experimental well, when the input is present, it can bind to the exposed T* domain of the reporter and displace the output strand, yielding a fluorescent complex and a waste strand.


Figure A

Nucleic Acid Delivery

In order to implement RNA strand displacement cascades in vivo, we first demonstrated our ability to deliver nucleic acids to mammalian cells. We have successfully delivered plasmid DNA, single-stranded modified RNA and double-stranded modified RNA to mammalian cells through both lipofection and nucleofection. In this section, we will review three different nucleic acid delivery techniques to mammalian cells, (1) Delivery of plasmid DNA, (2) Delivery of 2'-O-Methyl RNA and (3) Delivery of plasmids with inducible protein control.

(1) Delivery of Plasmid DNA to Mammalian Cells

Through the Gateway method, we have assembled many promoter-gene constructs as detailed on our Biobrick Parts Page. After construction of the plasmid, we delivered the plasmid DNA to HEK293 human embryonic kidney cells through transient transfection. In particular, we used lipofection with Lipofectamine 2000 reagent. See Materials and Methods for our Gateway and lipofection protocols.
Images of eYFP/mKate transfection from Confocal

This figure shows HEK293 cells which were transiently transfected with Hef1A:eYFP and Hef1A:mKate using Lipofectamine 2000. Equimole amounts of DNA were delivered, 500 ng total per 1.65 uL of reagent. Images were taken on a Zeiss microscope at 10X.

(2) Delivery of 2'-O-Me RNA to Mammalian Cells

For the purpose of our experiments, RNA oligos with chemical modifications that confer significant overall stability and increase in melting temperature is necessary to prevent spontaneous dissociation and rapid degradation by nucleases under in vivo conditions. One form of chemically modified RNA, 2’O-Methyl RNA, is a naturally occurring and nontoxic RNA variant found in mammalian ribosomal RNAs and transfer RNAs. These modified oligos are in most respects similar to RNA, but the 2' O-Methyl modification increases overall stability as the -OH functional group at the 2' position is replaced with a -OMe group, which can't perform cleavage of the RNA backbone. In addition to significant nuclease stability, the modification seem to confer increases in melting temperature, which minimizes the chance of the RNA strands dissociating upon introduction to a cellular environment. 1

Therefore, we need the ability to deliver our 2’-O-Methyl RNA constructs to mammalian cells and have developed protocols on our Methods Page to do so:



The movie above shows HEK293 cells expressing constitutive eYFP with a 2'-O-Methyl RNA strand labeled with ROX (5-carboxy-x-rhodamine) on the 3' end. As time passes, the complex/vesicles are uptaken by the cell, releasing their payload resulting in whole cell fluorescence. Each frame is 5 minutes, movie encompasses 200 minutes in 9 seconds.

Time point images taken at t = 0, 2, 3, and 4 hours post-transfection. Images taken at 10X on Zeiss microscope.



Once we demonstrated ability to deliver 2'-O-Me RNA to mammalian cells, we ran optimization experiments to optimize the ratio of 2'-O-Me RNA delivered to RNAiMAX (transfection reagent used) to achieve maximum transfection efficiency. The invitrogen protocol2 for Lipofectamine RNAiMAX reagent recommends 0.6 to 30 pmol of nucleic acid delivered per well per 24-well plate. Throughout our experiments we observed that we could see high transfection efficiency for 30 pmol of RNA but almost no transfection for 15 pmol, so we ran a titration experiment , testing out 20, 25 and 30 pmol of RNA with 1.0, 1.33 and 1.5 uL of transfection reagent and see saturation levels at 25 pmol. Therefore, the rest of our experiments were standardized around 25 pmol of RNA delivered with 1.33 uL of reagent:



This graph needs to be fixed! Then caption




(3) Inducible Control of Protein Expression


100,000 HEK293 cells were transfected with Hef1a:rTTA3, TreT:mKate, and Hef1a:TagBFP (as a transfection marker). After 48 hours, cells were imaged on a Leica Confocoal Microscope at 10x. We can see that we have transfection, since cells are fluorescing blue. Also, we can see that as we increase the concentration of DOX present, we see an increase in red fluorescence.


100,000 HEK293 cells were transfected with Hef1a:rTTA3, TreT:mKate, and Hef1a:TagBFP (as a transfection marker). After 24 hours, DOX was then added to 16 different concentrations ranging from .1 nM to 5000 nM. Lastly, cells were harvested for flow cytometry after 48 hrs and allowed to count 10,000 events. As we increase concentrations od DOX, the mean red fluorescence increases.

1 Behlke MA. (2008) Chemical modification of siRNAs for in vivo use. Oligonucleotides, 18(4):305–320.

2 Invitrogen. (2006) Lipofectamine RNAiMAX

In Vivo RNA Strand Displacement

** GRAPHIC SHOWING THE PROGRESS OF OUR STORY!**

We have successfully achieved RNA strand displacement in vivo and are extremely excited to bring such a powerful processing technology to mammalian cells. We attempted five iterations with many experiments each, of different approaches to achieve RNA strand displacement in vivo. We 1st used existing reporting sequences in an RNA context and transfected RNA strands to mammalian cells. We 2nd switched transfection reagents based on critical research. We 3rd added a transfection marker to our system so we could properly analyze the source of our error. We 4th designed DNA plasmids that could transcribe our input strands (link to U6-TetO:S6, U6-TetO:S1) so that we could have the ability to produce inputs in vivo.

We 5th redesigned new gate sequences based upon literature citing longer toehold lengths (link to Eerik's new page on our new reporter design) and through both nucleofection and transient transfection, observed the 1st RNA Strand Displacement Reaction In Vivo . Follow the story below from Strategy 1 through 5 or skip immediately to Strategy 5, Our Personal Favorite.

Strategy 1: Lipofectamine 2000 Transfection of RNA version of Reporter from Winfree/Qian 2011 Paper

Our first strategy to implement RNA strand displacement in vivo was to adapt the DNA sequences of inputs, gates and reporters from the Qian/Winfree, "Scaling Up Digital Circuit Computation with DNA Strand Displacement Cascades," 2011 Science Paper to 2’-O-Methyl RNA strands to transfect into mammalian cells. See our Motivation page for more details.

In the first foundational experiment, HEK293 (Human Embryonic Kidney cells) were used that constitutively expressed a yellow fluorescent protein (eYFP) in order to be easily visible in microscopy images. 200,000 HEK293 cells were seeded into four wells of a 24 well plate in supplemented DMEM without phenol red pH indicator. The negative control well did not receive any RNA. As a transfection reagent, each well received 1 uL of Lipofectamine 2000. The positive control well received 5 pmol of a gate strand tagged with a ROX fluorophore annealed to an input strand, to act as a product of a strand displacement reaction. The scrambled input well received 5 pmol annealed double stranded reporter with quenched ROX along with a 5 pmol of an input strand containing the correct toehold domain but the incorrect binding domain. Therefore, when both constructs are inside the cell, a strand displacement reaction should not occur, and the fluorophore remains quenched. In the final well, correct input, the cells received 5 pmol of double stranded reporter as well as 5 pmol of an input strand with the correct toehold domain and hybridization domain. Accordingly, we should expect that the toehold of the input strand binds to the complementary exposed toehold on the double stranded reporter, and will branch migrate and effectively kick off the output strand of the reporter that is tagged with a quencher. Therefore, the fluorophore will no longer be quenched, yielding red fluorescence.

Refer to this diagram to identify labeled strands



In the negative control well, 200,000 HEK293+eYFP cells are healthy and adherent. In the positive control well, we see localized red fluorescence in the form of vesicles as well as distributed, whole cell red fluorescence. In the scrambled input well, we see red vesicles as well as red whole cell fluorescence. In the correct input well, we see only whole cell red fluorescence.

Strategy 2: Switch Transfection reagent to RNAiMAX


From the first foundational experiment, we observed localized red fluorescence in vesicles as well as whole cell fluorescence. This indicates that our reporter complex is either melting, being degraded, being recognized by a specific enzyme etc as well as the reporter is coming apart inside of the lipofectamine vesicles. We researched better transfection reagents for double stranded RNA, and found that Lipofectamine RNAiMAX is designed specifically for the delivery of double stranded RNA, whereas Lipofectamine 200 is specifically designed for the delivery of plasmids.

Once we received the new transfection reagent, we set up experiments similar to the initial experiment but with an optimized protocol for RNAiMAX (See Materials and Methods).


These images were taken on a Zeiss microscope at 10 X. The row labeled Reporter received only our double stranded quenched ROX complex with an exposed toehold. We can see that as the time goes from T=0 minutes post transfection to T=60 (24 hours post transfection) the reporter is once again either being degraded or melting inside of the HEK293 cells. We see an interesting trend in the row labeled incorrect input. This well received the reporter complex along with our input with the correct toehold but the incorrect binding domain. It appears that the toehold of the incorrect input can bind to the exposed toehold of the reporter, somehow stabilizing the complex. Finally, in the well that received the correct input for a strand displacement reaction along with the reporter, we see identical results to the just reporter well, which makes our argument for strand displacement initially unclear. The success of this experiment, however, is the fact that we no longer observe red vesicles, which indicates that RNAiMAX is indeed a better transfection reagent.

Strategy 3: Tag RNA strand with an Alexa Fluorophore to act as a transfection marker

Since we continually observed a trend of the reporting strand coming apart inside of the cell (i.e. the quencher no longer quenching the fluorophore) we modified the reporter with an additional fluorophore, AlexaFluor488 to act as a transfection marker, so that we could observe the reporter entering the cell. AlexaFluor488 is a green dye, and we predict that if the two strands melt inside of HEK293 cells then we would see an initially green vesicle enter the cell and then localized green and red fluorescence in the cytoplasm, indicating that the strands came apart.


Nice Data showing reporter melting inside cell from 7/29 powerpoint


Strategy 4: Create DNA plasmids driving transcription of RNA inputs, while transfecting RNA Reporter


Strategy 5: Redesign RNA Reporter and Attempt Nucleofection and Transection


Our best new data from Transfection!

Our best new data from Nucleofection!

Ongoing Experiments:

What's next!