[Demelerlab] Proposed method -- next iteration

[Pahara, Justin (AAFC/AAC)]

Good Afternoon Borries,

Please see some comments in line:

Hi Justin,

thanks for the additional explanation, the process is now much clearer to me. The way I see it we will do the following:

  1.  attempt to separate bound aptamer from unbound aptamer in a negative selection experiment performed just with regular e-coli, where we just need to know where the free aptamers sediment in a sucrose gradient so we can recover them.  This process is repeated multiple times. The thought is that the unbound aptamers will sediment the slowest, since they have the highest friction and lowest molar masses compared to bound complexes. Here we will be able to identify the sedimentation speed of the unbound aptamer in whatever sucrose concentration you plan to use - so let us know what sucrose concentration you plan to use, that part is definitely doable.

Exactly – for the sucrose gradient, will we determine this after AUC for the sedimentation coefficient?


  1.  After amplification of the free aptamers found in step 1 that don't bind to generic e-coli proteins, we continue by reversing the process and selecting for the complexed aptamers that were exposed to the positive sample (presumably everything sedimenting faster than unbound aptamer).

Exactly – after a couple negative selection rounds we move to positive selection


  1.  Now what? You potentially have millions of different aptamers at this stage that bind to something in the positive sample, but mostly not to plain e-coli proteins, including molecules of actual interest. How do you further distinguish your antigen of interest among those millions of "hits"?

Typically after ~20 rounds of positive and negative selection the pool of unique aptamers is quite low < 100, often on the order of ~10. We will sequence the pool, and re-order each individual aptamer so we have pure aptamer pools. From there we propose to immobilize each aptamer pool, likely on magnetic beads. We then mix the immobilize aptamers with the engineered E. coli lysate and elute. We can either use SDS-PAGE to look at bound molecules, or at that point we would theoretically have relatively pure fractions that could be analyse in AUC. The absolute best outcome would be to see fluorescence signal in the eluted sample, as this would indicate the aptamer binds to the fluorescent protein.

I don't get the part of the fluorescent protein. If you are looking for aptamers that bind to the fluorescent protein, why not do your selex against the purified fluorescent protein? It would be so much easier.

The reason is that during our hunt for aptamers with anaplasma infected blood samples, there are not really any good known candidates. If this proposed experiment works, then we can apply it there and both identify aptamers and unique antigens that are selective for anaplasma in bovine blood. This experiment is a practice experiment to see if it is possible. Using a pure protein is indeed a good way to get selective aptamers, it just doesn’t reflect our scientific problem with anaplasma.

And what does the fluorescent protein have to do with the selected aptamers in the positive selex experiment? There are presumably a bunch of other selected aptamers present that sediment very differently from the one aptamer that bound to the fluorescently labeled molecule.

The fluorescent protein is unique in the engineered E.coli lysate compared to the non-engineered lysate that we use for negative selection. Same with CAT. If the negative selection works, then aptamers that bind E.coli will be removed. This means that aptamers that bind fluorescent protein/CAT will dominate during positive selection. The fluorescent protein may be a good tool for final analysis as described above.

Maybe one day I will understand how this is supposed to work...

Hahah, it is a complex process and I’ve had lots of time to digest it in advance.  It would be great to chat in person about this next week. Can I drop by?

Have a great weekend!

Best,
Justin


-Borries


On Thu, Jul 27, 2023 at 9:46 AM Pahara, Justin (AAFC/AAC) <justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>> wrote:
Good Morning Borries,

Please see some further comments below:



Since aptamers will bind to many different molecules, this will include small positively charged molecules as well as many bigger things. We will get a broad spectrum of different molar masses for the complex, some complexes that are almost the same size as the free aptamers, ranging to others that are much bigger. How do you know which size is the right one? If you separate based on free and some complexed MW that is easy to separate, how do you know that you don't completely miss the relevant complex because it is smaller than what you can separate from free aptamer?

It sounds like you expect to be able to tell the difference by doing your screening somehow by comparing lysates/serum from infected vs. negative control samples. This part is what I can't follow right now. How are you telling which aptamer is bound to the "right" antigen? Sorry, but if I cannot understand the downstream approach, I cannot design the right design for this experiment. I need to gain a better understanding first.

Specifically:

  1.  We do negative selection: Mix the aptamers with E. coli lysate to get rid of aptamers-E. coli derived interactions (the “heavy aptamer fraction”). You’re right, these will include aptamers binding to lots of E. coli stuff and so we want to get rid of those and keep the unbound fraction of “light aptamer fraction”. The assumption is that the “light fraction” of unbound aptamers do not interact well with E. coli stuff, however, some could interact with the fluorescent protein.
So you hypothesize that some aptamers will bind only to Ecoli and therefore you can remove them from the pool by telling the fractions that sediment faster than unbound aptamer, correct? Once you amplify the remaining pool, you will not introduce further variability? If you did, you would be back to square one. If you don't, how do you have enough variability left?

Precisely. As one continues through rounds of SELEX, they yield a diminished number of unique aptamer sequences, however, you enrich/amplify during each step so the total number of aptamers is high. In the end, you ideally have a dozen or so unique aptamer sequences from the quadrillion or so that you started with.  Because you start with ~quadrillion aptamer sequences,  we hypothesizes there will be enough variability left over after negative selection to also get some hits during the rounds of positive selection. This is the basis of SELEX and has worked for others, and so we hope it also works here.


  1.  We enrich/amplify the pool of remaining unbound aptamers so we can do further selection rounds
Whatever is left, may just bind weakly to your target, and you are back to square one, because it may bind just as weakly to some other proteins.

True, this is a possibility. Because our end-application requires that we have aptamers that only bind to anaplasma targets and not generic bovine targets, we need to do the negative selection steps and remove lots of aptamers. Otherwise our biosensors could send false positives from an aptamer that binds both a target and a bovine antigen.


  1.  We then do positive selection: Mix the enriched aptamer library with engineered E. coli. The assumption is that the only biochemical difference from (1) is the presence of the engineered proteins. Therefore, we assume that any heavy bound aptamer fraction is interacting with the engineered proteins.
So your hope is that whatever aptamer is left from your negative screen contains the one aptamer that actually can bind to your target, and strongly?

Yes. From the ~quadrillion we start with, there will likely still be thousands/millions/billions of unique aptamer sequences that don’t bind to E. coli stuff, but have the potential to bind to the fluorescent protein or chloramphenicol acetyltransferase protein, or the plasmid, or the mRNA encoding the proteins. Any of these targets can be considered unique from native E. coli stuff and could be used in a biosensor.


  1.  As you saw in the suggested protocol, we need to due several SELEX cycles of negative and positive selection and at the end, we hypothetically have an enriched pool of aptamers that specifically bind to the engineered proteins. We sequence these, order purified oligos, and test to confirm specific interactions.
So how do you plan to get around the heterogeneity that starts with very small molecules binding to the DNA aptamers that sediment at almost the same rate as unbound aptamers? What about the question about heating a fluorescently labeled DNA molecule to 95 degree and hoping the fluorescence survives, has that been tested? So I guess your plan is for us to first measure unbound aptamer in a sucrose gradient to find out how fast it is sedimenting so you can separate unbound aptamer from everything else by running the same gradient, correct? I just don't understand how you could come up with specificity if you don't even know what you plan to bind with your aptamer by this approach. I don't understand the basis by which you are separating the "correct" complex from all the others that will form if you don't even know what the correct complex looks like. Can you explain that further?

This is where we are a new to the ultracentrifugation and hope to understand feasibility from you and your team.

We suggest first getting the sedimentation coefficient of the pure aptamers using absorbance UC (A260), then create sucrose gradients and confirm how fast/where the unbound aptamers run using absorbance UC (e.g. 3 mm in 2 hours – I have no idea about this, just an example). Using this knowledge we know where the unbound aptamer “light fraction” will be and therefore where the unbound aptamer fraction will not be. The bound aptamer “heavy fraction” is assumed to be anywhere past where the unbound aptamers run (see image below).

Can we separate in crude fractions? We don’t need to accurately obtain pure band fractions, rather we need crude fractions (see the diagram below). For example, in the negative selection steps at the beginning we could aspirate the top few mm of the centrifuged samples (green). It is possible that we also get a bit of the “heavy fraction”, however, this isn’t a big deal because we get rid of most of the “heavy fraction” and we repeat this selection process through several rounds of SELEX.

During the positive selection we need the heavy fraction.  We would aspirate the top few millimeters of light fraction (green), and discard it, and then collect the rest of the sample that we consider heavy fraction (bound aptamers, red).

We’ll be doing phenol/chloroform purification to concentrate the DNA and then during the enrichment/amplification our primers are specific to the aptamers and so only the aptamers are enriched compared to any other E. coli DNA present.


[cid:image001.png at 01D9C14E.A658D630]


As we near the end of the many positive/negative selection SELEX rounds we should have enriched only few unique aptamer sequences. A large proportion of which should interact with the fluorescent protein (because it is one of the targets we are selecting for). For the latter stages of this mock experiment we then sequence the small pool of aptamers to get their sequences and order them in their pure form to test for their interaction with purified fluorescent protein. We could immobilize the aptamers on magnetic beads to do pull down assays to see if they bring the fluorescent protein with them.


Do you think ultracentrifugation will work?? Pls let me know what you think.

Best,
Justin.

-Borries

On Wed, Jul 26, 2023 at 1:48 PM Pahara, Justin (AAFC/AAC) <justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>> wrote:
Good Afternoon Borries,

Thank you for the feedback. I’ve included some comments/clarifications below:
I read through your document and am not at all convinced by your SELEX plan. The main problem is that you will not be able to "find" your antigen-bound aptamers of interest if you "train" them on a gemisch instead of a pure antigen. ssDNA will bind to anything, including RNA. In your step 1 experiment we have a whole bunch of labeled molecules (if the fluorophore actually survives the 95C denaturation cycle) which bind to anything in the lysate, potentially giving a very heterogeneous, non-specific mixture. Even if we have a fluorescent target to follow, we are clueless what all these different fluorescently labeled molecules are binding to. Potentially, each aptamer will bind to something else. How do you find the right one?

Agreed – the aptamer library which is comprised of a quadrillion possibilities will bind to many things, both naturally derived from E. coli, as well as *hopefully* to the engineered fluorescent protein. In the case of this mock experiment, the “right ones” would ultimately be enriched aptamers in a fluorescent protein fraction at the end of the SELEX rounds:


  1.  We do negative selection: Mix the aptamers with E. coli lysate to get rid of aptamers-E. coli derived interactions (the “heavy aptamer fraction”). You’re right, these will include aptamers binding to lots of E. coli stuff and so we want to get rid of those and keep the unbound fraction of “light aptamer fraction”. The assumption is that the “light fraction” of unbound aptamers do not interact well with E. coli stuff, however, some could interact with the fluorescent protein.
  2.  We enrich/amplify the pool of remaining unbound aptamers so we can do further selection rounds.

  1.  We then do positive selection: Mix the enriched aptamer library with engineered E. coli. The assumption is that the only biochemical difference from (1) is the presence of the engineered proteins. Therefore, we assume that any heavy bound aptamer fraction is interacting with the engineered proteins.
  2.  As you saw in the suggested protocol, we need to due several SELEX cycles of negative and positive selection and at the end, we hypothetically have an enriched pool of aptamers that specifically bind to the engineered proteins. We sequence these, order purified oligos, and test to confirm specific interactions.

For gathering the fractions, we don’t need to be precise. We simply need to have enough separation between bound/unbound fractions so we can aspirate, and then amplify with PCR (which is very sensitive and works in crude mixtures).


My proposal was different: pick a very specific and highly purified antigen target and perform your SELEX screen on that only. Once you got your DNA sequence narrowed down, fluorescently label the specific DNA molecule and mix it with the cell extract to see if it binds. Antigen is present = binding, antigen is not present = hopefully no binding. For that, the fluorescent approach will work fine, the nonspecific approach doesn't make sense to me. Maybe I am missing something?

I hope the above clarifies. We are hoping to achieve the same ending. The reason for suggesting that approach is that in the anaplasma infected blood samples there are few/no good targets for sensing it and we’ll need to find aptamers that bind to unknown anaplasma targets. The suggested experiment, is a first approximation of how we could do that in the anaplasma samples. If it doesn’t work in this mock experiment, then it is unlikely to work in our blood samples.

The experiment I propose is to first make a proof of concept: take a fluorescent protein (eGFP would be best) and see if we can train an aptamer sequence for eGFP using SELEX. Next, have one cell line that expresses eGFP, and another that doesn't, and then show that the DNA molecule actually binds to eGFP when it is endogenously expressed in a cell lysate.  If that works, we optimize the method to train an aptamer on your antigen of interest.

The limitation is that with anaplasma we ultimately do not have a good antigen of interest, so this proposed experiment has the potential to discover new ones whilst also identifying prospective aptamers.

Pls let me know what you think. I can drop the Uni to chat this or next week.

Best,
Justin.





On Wed, Jul 26, 2023 at 9:56 AM Pahara, Justin (AAFC/AAC) <justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>> wrote:
Good Morning Borries,

Just checking in to see if you’ve had a chance to check through the proposed SELEX, AUC method.

I hope your lab retreat was fun and energizing.

Best,
Justin.

From: Borries Demeler <demeler at gmail.com<mailto:demeler at gmail.com>>
Sent: Thursday, July 20, 2023 10:20 AM
To: Pahara, Justin (AAFC/AAC) <justin.pahara at AGR.GC.CA<mailto:justin.pahara at AGR.GC.CA>>
Subject: Re: Proposed method -- next iteration

CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
ATTENTION: Ce courriel provient de l’extérieur de l’organisation. Ne cliquez pas sur les liens et n’ouvrez pas les pièces jointes à moins que vous ne reconnaissiez l’expéditeur et que vous sachiez que le contenu est sûr.
Hi Justin,
We are currently on our lab retreat in Montana and at a workshop, I'll get back to you next week in more detail.
Thanks for your patience, -Borries

On Thu, Jul 20, 2023, 08:32 Pahara, Justin (AAFC/AAC) <justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>> wrote:
Good Morning Borries, Amy,

Please see the attached document outlining the proposed experiment. Thank you for the back and forth discussion as it has helped us to better understand AUC, however, we still have much to learn 😊. Please let us know if there is anything that will not work.

I do have one remaining question, however, to see if we can do this entirely without a fluorophore:

If we are able to characterize the aptamers S value using absorbance UC, create an appropriate sucrose gradient, and then confirm the sedimentation rate of the unbound aptamers using absorbance UC again would we know with fairly high accuracy the position of the unbound aptamer fraction even in a whole cell lysate mixture? Therefore we would know the distance the fraction travels as a function of time and be able to crudely remove it? All without fluorescence?

@Borries, I know you’d prefer using a well defined and pure target, but I’m hoping we could run an experiment that reflects what we aim to do with the anaplasma in blood. If you feel the proposed method will not work, we are happy to pivot.


Thank you,
Justin.


Dr. Justin Pahara

Research Scientist and Project Lead

Nanotechnology (Biotic Stresses and Adaptation)
Agriculture and Agri-Food Canada / Government of Canada
justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>

Nanotechnologie (Adaptation et Contraintes Biotiques)
Agriculture et Agroalimentaire Canada / Gouvernement du Canada
justin.pahara at agr.gc.ca<mailto:justin.pahara at agr.gc.ca>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.biophysics.uleth.ca/pipermail/demelerlab/attachments/20230728/9efd454c/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 554381 bytes
Desc: image001.png
URL: <http://www.biophysics.uleth.ca/pipermail/demelerlab/attachments/20230728/9efd454c/attachment-0001.png>