How smart are SmartFlares?

This post is co-authored by Raphaël Lévy and Dave Mason.

Note: We contacted Chad Mirkin and EMD Millipore for comments. Chad Mirkin replied but did not allow me to share his comments as he prefers to discuss his work in peer reviewed manuscripts rather than blogs. EMD Millipore has provided a response (reproduced below) and is keen to further engage in the discussion.  They wrote that they “look forward to responding to [the questions you pose at the end of the post] after your blog is posted so other researchers who may have the same questions can follow our discussion online.” [Update: they have gone into silent mode after that comment].

Update: The results of our study have now been published.


To image proteins in cells, biologists have powerful tools based on the Green Fluorescent Protein (GFP) for which Osamu Shimomura, Martin Chalfie and Roger Y. Tsien  obtained the 2008 Nobel Prize in Chemistry. RNA molecules play crucial roles in cells such as coding, decoding, regulation, and expression of genes, yet they are much more difficult to study. SmartFlares are nanoparticle-based probes for the detection and imaging of RNA in live cells. Could they become the GFP of the RNA world?

Many certainly believe this to be the case. SmartFlare ranked second in TheScientist top ten 2013 innovations, with one of the judges, Kevin Lustig, commenting “These new RNA detection probes can be used to visualize RNA expression in live cells at the single-cell level.”  The following year, SmartFlare won an R&D100 award. The technology comes from Chad Mirkin’s lab at Northwestern University. Chad Mirkin is the winner of numerous prestigious prizes and a science adviser to the President of the United States. The scientific articles introducing the SmartFlare concept (under the name of Nano-Flare) were published in the Journal of the American Chemical Society in 2007, ACS Nano in 2009, etc. In 2013, the SmartFlare technology was licensed to EMD Millipore. Here is one of their promotional video:

For a molecular sensor to work, it needs a detection mechanism. The principle of the SmartFlare is explained from 0:45. A capture oligonucleotide (i.e. DNA) is bound to the gold nanoparticles. A reporter strand is bound to the capture strand. The reporter strand carries a fluorophore but that fluorophore does not emit light because it is too close to the gold (the fluorescence is “quenched”). In the presence of the target RNA, the reporter strand is replaced by the target RNA and therefore released, quenching stops, and fluorescence is detected. The release is shown at 2:05. Simple and convincing. Gold nanoparticles are indeed excellent fluorescence quenchers (we have used this property in a couple of papers).

But, for a molecular sensor to work, it also needs to reach the molecule it is supposed to detect. The SmartFlares are shown at 1:40 entering the cells via endocytosis, a normal mechanism by which the cell engulfs extracellular material by entrapping them into a bag made of cell membrane. Molecules and particles which enter the cell by endocytosis normally remain trapped in this bag. This entrapping is essential to protect us from viruses and bacteria by preventing them from accessing the cell machinery. Here, however, at 1:45 – 1:46, something truly remarkable happens: the endosome (the bag) suddenly fades away leaving the particles free to diffuse in the cell and meet their RNA targets. This is a promotional video so you might say that the demonstration of, and explanation for, this remarkable endosomal escape is to be found in the primary literature but that is not the case.

SmartFlares_scheme

There is an extensive body of literature (not related to SmartFlare) dealing with endosomal escape. Some bacteria (like Listeria which can cause food poisoning) and viruses (like Influenza or HIV) use proteins to destabilise the endosome, escape and cause disease. Other mechanisms involve altering the ion balance in the endosome to pop it like an over-inflated balloon (you can read more about the ‘Proton Sponge Effect’ in this review). The problem is that none of these scenarios are applicable to gold nanoparticles conjugated to oligonucleotides. The problem is compounded by the choice of techniques used to analyse SmartFlare uptake into cells. Most of the published papers (for examples see here, here and here) characterise “uptake” and do so largely via Flow Cytometry or Mass Spectrometry (to measure the gold content of the cells). These papers certainly support NanoFlares being taken up into endosomes, but don’t offer any evidence for endosomal escape. A systematic unbiased electron microscopy study would enable to gather an estimate of how many nanoparticles have escaped the endosomes. Alternatively, fluorescence microscopy can be used to visualise a diffuse (released) instead of punctate (still in endosomes) distribution of intensity. While there are some images of cells having taken up NanoFlares, the sort of resolution required to discern distribution is not afforded by publication-size figures.

Wouldn’t it be nice if we had access to the original data? Researchers are often left squinting at published figures and all too often have to rely on the author’s interpretation of the data. One solution to this problem is to make supporting data available after publication. This is the idea behind the JCB Dataviewer; allowing authors to upload the original data to support papers published in the Journal of Cell Biology. The other option is to make the data available before publication, in what is called Open Research. This has the huge advantage of opening up a discussion about data, its interpretation and meaning before going through the formal peer-review process.

It is this latter technique that we are currently using to share our study of the use of NanoFlares as VEGF RNA reporters in cells. Our Open Science Notebook gives an overview of the experimental design, results and discussion, while our OMERO server is being used to host all of the original data for anyone to access. The project is still in progress, however our main findings so far are that:

In all conditions where fluorescence is seen, the distribution is consistently punctate (see all of the data here ).

So far these findings have left us with several questions, the most interesting of which are:

  1. Why do we see punctate fluorescence with the VEGF SmartFlares? If the SmartFlares are still in endosomes, they shouldn’t be able to interact with mRNA and thus fluorescence should be quenched.
  2. Why do we see signal at all in the scrambled control?
  3. Why do different cells take up varying amounts of SmartFlares? Fluid phase dextran shows approximately equal uptake in all cells.

We’re presently investigating these and other questions. As we find out more, we will continue to post the data and update the blog.


RESPONSE from EMD Millipore:

In their response, EMD Millipore pointed to a number of relevant publications suggesting that we should revise the post after having considered this evidence. We had already seen those references and we have not altered the post, but we reproduce EMD Millipore’s response below:

 Oligo-modified nanoparticle internalization and endosomal release:

·         Oligonucleotide modified nanostructures are taken in through an endocytotic mechanism.  http://www.pnas.org/content/110/19/7625.long

·         These highly anionic structures attract a counterbalancing salt cloud.  http://pubs.acs.org/doi/pdf/10.1021/jp205583j

·         This is thought to be the mechanism of release from endosomes (via osmotic pressure) 

 Observation of punctate fluorescence:

·         At short time points, when these structures are indeed in the endosomes,  or at low detection gains on a microscope (where you are adjusting for the brightest points) the staining appears punctate.  (For example- the light in a room comes from the bulb, which is the brightest, but the room is still lit.  Keeping only the brightest point in a picture would only show you the bulb.)

·         Therefore, with regards to the experiment you’ve already performed, our first suggestion would be to turn up the gain to see cytoplasmic fluorescence.

·         Here for example are some pictures showing nice cytoplasmic stain  http://www.pnas.org/content/109/30/11975/F1.expansion.html

 

Also may be of interest: 

http://www.nature.com/mt/journal/v22/n6/full/mt201430a.html

It may be worth noting some of our more recent examples of SmartFlare in the literature, spanning across cancer and stem cell research on a variety of detection platforms (flow & microscopy).  Here the punctate fluorescence is also observed, but you can also see nice cytoplasmic staining.

·         Seftor et al. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4026856/

 

·         Mehta et al.  http://www.ncbi.nlm.nih.gov/pubmed/24623279
 
Advertisements

63 comments

  1. I’d advice more controls to test for SmartFlare sensitivity. Using knock-out cells is an opinion, but one has to be careful that they also produce no mRNA in question or to design probes that match wt-mRNA only.
    When comparing the reliability of relative signal intensity (like in your VEGF experiment) or when any other known stimulus to obtain an effect on expression of a certain gene: consider running qRT-PCR in parallel to assess the change in relative mRNA levels and compare this to SmartFlare read-out.

    Liked by 1 person

      1. KOs could be a different approach, so if anyone was questioning your methods when using scramble or DMOG would be finally convinced then. HCT116 are available as different KOs.

        Like

  2. You asked me to leave a comment so here it is:
    I really like what you’re doing here and sharing the data as you get it is revolutionary. I’m also puzzled as to how the probes get out of the endosome. I’m not aware of a way to get nanoparticles out of endosomes. We’d really love to do this for a number of experiments. As you replied to me on Twitter probably the only way to deliver NPs into the cytoplasm is by microinjection.

    Like

  3. It is my understanding that the same vesicle can be used for endocytosis and exocytosis. It is also my understanding that vesicles are not complete emptied during exocytosis. I could be wrong about both of these points and would gladly be corrected. If my understanding is correct then could it be possible that RNA could be within exocytosing vesicles and that this RNA could then combine with the SmartFlares as they are endocytosed? That way everything would happen within the vesicles only.

    Like

    1. It is true that large dense core vesicles can exocytose and then be recaptured (because of their dense core), but smaller vesicles likely fuse fully with the plasma membrane. However, the RNA to be detected is *inside* the cell, so the mechanism of exo and endo is not really relevant here.
      Once an endocytosed vesicle is inside the cell it can take several paths e.g. fuse with other vesicles, but critically what is on the inside stays on the inside.
      I hope I understood your comment correctly.

      Like

  4. Thanks for the comment Bob. That does not correspond to the picture given in Mirkin’s papers nor EMD Millipore advertising and would not explain our results (we see no difference between the scramble control and the VEGF target). You are correct that exchanges of vesicles from cells to cells occur in vivo and in cell culture too and it is a way of cell-to-cell communication. Exosomes have even been reported to carry mRNAs and microRNAs. Still seems a very far fetched explanation, but it could be tested with high quality/high magnification EM + microscopy. In any case, even if that was true, it would still mean that SmartFlare cannot be used to image in real time mRNA levels in the cytosol…

    Liked by 1 person

    1. Thanks Raphael

      Yep, my interest was not so much explaining the results of your tests but simply wondering if RNA and NPs could possibly come in contact without the vesicle dissolving once endocytosed. There appears to be a lot of claims of how NPs could interact with internal structures within cells and many are not plausible (as you have pointed out before) because of the vesicle wall problem. So I was just wondering if in this case there might be another way to bring the interacting components together. Of course, this is certainly not how EMD Millipore claims they interact.

      Has there been any response from EMD Millipore or Chad Mirkin about the results of your scramble control? Are you looking at the “bulb” rather than the “lit” room in your tests?

      Like

      1. I have copied EMD’s response in the post. They were supposed to comment further but have not done so yet. Chad Mirkin replied to my emails but, as noted above, did not allow me to share his comments as he prefers to discuss his work in peer reviewed manuscripts rather than blogs. https://twitter.com/brembs/status/595588276611895296/photo/1

        I am not sure I understand your last question about bulb and lit room; can you explain?

        Like

  5. yes, of course, sorry… That analogy is both problematic and interesting.

    First, contrarily to all of the data available in the literature, you can judge for yourself because we are sharing the data in a format where you can change the contrast of each channel, saturating the bulbs to reveal the lit, etc; really, have a play: http://cci02.liv.ac.uk/gallery/show_project/1217/

    In short (and this will need more analysis), there does not seem to be more of a diffuse “background” in any of the conditions. EMD suggestion “to turn up the gain to see cytoplasmic fluorescence” certainly requires a lot of caution and integrity. Sure by turning the gain at some point you will have some (auto)fluorescence in the background… Dave Mason, who acquired these images will have more to say I am sure about imaging parameters when he is back.

    Finally, in experiments such as FACS, which appears to be the main application of the SmartFlares at the moment, it is the overall fluorescence of a cell which is measured, independently of intracellular localisation. It this was dominated by non-mRNA specific signal from the cells, it would be a bit of a problem too.

    Thanks again for engaging in this discussion. Given the number of papers and citations, it would be really nice to see more contributions, even, why not, some non-anonymous 😉 from bionano people and from SmartFlare users.

    Like

  6. Thanks Raphael

    I have a bit of a bonehead question. Do you know how the oligonucleotide is bound to the gold NPs? Are there “binding sites” at which other reagents could compete for attachment to the NP and end up displacing a oligonucleotide with a bound reporter. I assume that the reporter would still fluoresce if this were to occur since into would be distant from the quenching NP.

    Like

  7. Hi Raphael,

    I have been following the project (admittedly sporadically, so I may have missed some data), but I have enjoyed reading the developments of the project in almost “real time”. Also I was aware of these probes, but not so familiar with the literature.

    I am wondering, although you may have done this already, but have you guys tried to assess the stability/release of oligos from the SmartFlares, for example, in buffers of different pH? Maybe in the range of that in the endosomes, both early and late? If I remember correctly I think oligonucleotides/DNA are not so happy at acidic pH’s, and could explain the release in the endosome.

    Liked by 1 person

  8. Hi Dan, Bob

    @Dan: Great to see you here! (If you want to write a short (or long) blog post about your move to Australia, that would be very nice too).

    @Dan I completely agree with your suggestions. The problem is that this stuff is pretty expensive and these sort of experiments in a cuvette do use quite a lot of materials… (if EMD Millipore is reading, maybe they want to send us some more so that we can do that?). We will do a couple of other things first but it is definitely on the agenda.

    @Bob: the oligonucleotides are attached to the gold nanoparticles via a S-Au bond. With a single S-Au, there is the possibility of ligand exchange. According to EMD, they use a disulphide, i.e. two S-Au per ligand so ligand exchange is less likely. However, it is still possible and we should do the kind of experiments suggested by Dan. Another possibility is nucleases; I don’t know how much of those are present in endosomes? We have shown peptide degradation on peptide-capped nanoparticles upon uptake:
    https://www.researchgate.net/profile/Paola_Nativo/publication/26787950_Cathepsin_L_digestion_of_nanobioconjugates_upon_endocytosis/links/0912f507823daef4d9000000.pdf

    Raphael

    Liked by 1 person

    1. Yes. Good point about the possibility of endosome nucleases.Do you suspect that there would be enough room for the nuclease to cleave the oligonucleotide between the NP and the fluorophore?

      Like

      1. Cool. Thanks for that link. So it appears that some magic is, at least in a former version, occurring in the endosome.

        Like

      2. My interpretation of the data is that SmartFlare detects endosomal nucleases, hence the majority of studies finding the fluorescence in organelle-like punctae with barely detectable signal in cytosol, and your finding that the VEGF SmartFlare was unable to detect a 35-fold increase in VEGF RNA upon DMOG treatment. In the link you mention below, the differences in conditions between the Mirkin article and the “more representative” article that EMD Millipore cites are inconsequential. The “positive” data that they cite, purporting to show increased levels of mRNAs related to cancer cell phenotype, are probably just reflecting increased degradative capacity of aggressive cancer cells.

        Liked by 2 people

  9. @Luke @Dan, I think the possibility of endosomal nucleases is a really interesting one. As such, I’ve been looking into the pH requirements of endosomal nucleases, with the hope to block endosomal acidification (with chloroquine, ammonia or a plecomacrolide like bafilomycin), to see if we still see SmartFlares as puncta. More on that coming up on the blog.

    Like

  10. Very nice work, Raphaël and co.! Curious to see if it really works.
    Quick note from my side: it can be hard to keep gold NP’s from agglomerating, especially at elevated temperatures and higher concentrations. Do these NPs lose their quenching properties when they become larger?

    I could imagine the endosome to create an environment which leaves the smartflares in very close proximity indeed (high concentration), which could lead to agglomeration of the NPs and a loss of quenching ability.

    Good luck with the upcoming experiments!

    Like

    1. Alright, thanks for the clarification! I assumed the NPs had to be of a particular size to quench a particular wavelength.

      Like

  11. Hi everybody,
    as the correspending author of a Stem Cell paper in which we have used the SmartFlares on different pluripotent cells of human, murine and porcine origin I want to reply to two of the above mentioned questions.

    Why do we see a signal at all in the scramble control?
    I think one cannot expect a negative control which does not produce a fluorescent signal at all. The fluorophore may not be quenched by 100% and may be subject to degradation, especially when applied for a longer time (two days or more). Nevertheless, within 16 to 24 hours after the application of the nanoparticles we see a clear-cut difference of fluorescence intensity when comparing scramble control and gene-specifc Smart Flares.
    http://www.ncbi.nlm.nih.gov/pubmed/25335772

    We believe that this difference is reliable and specific. We have selected freshly reprogrammed murine iPS cells based on their Nanog-specific fluorescence intensity in situ. In downstream experiments we could confirm that only colonies with a high fluorescence intensity expressed higher amounts of endogenous pluripotency factors and showed a superior capacity to differentiate. Therefore, we belive that these functional data strongly support the idea that the fluorescence intensity was indeed correlated to a specific interaction with the Nanog mRNA in these clones.

    Why do different cells take up varying amounts of SmartFlares?
    I think this difference is not surprising as the nanoparticles are engulfed by endocytosis. This process is influenced by the cell type, the differentiation status and the cellular ability to perform phago- and macropinocytosis. Therefore, we think that a uniform uptake rate cannot be expected.

    Liked by 2 people

    1. Hi Harald, thanks for the comments! It’s great to have other people who have used SmartFlares contributing to the discussion.

      On your last point; this was also my original thinking regarding heterogeneous uptake, but wouldn’t this also affect the rate of 10kDa dextran uptake? (which we’ve always seen to be consistent and homogeneous between cells)

      I’m really trying to get my teeth into the nuances of caveolae-mediated endocytosis which may also help to explain the differences.

      Like

      1. Hi Dave,
        I am not a specialist in endocytosis, therefore, I can’t judge whether dextran uptake will be influenced. However, what I can tell you is that we have performed comparative experiments on undifferentiated and differentiated murine ES cells. In our hands, we see a decrease in the uptake in the differentiated cells using equal amounts of SmartFlare nanoparticles.

        Like

    2. Hi Harald

      Thanks again for commenting here and sorry for the delay in replying. It is interesting that you see some differences but the big question that remains is how could the technology possibly work?

      It can only work if the particles escape endosomes, but: 1) there is no reason why they should, 2) this problem is not discussed in the original publication introducing the technology, 3) there is no direct evidence in the literature that it happens, and, 4) all the data we are accumulating indicates that the particles are indeed in vesicular compartments (more on this soon on the open notebook as we have just had our cell electron microscopy results this week).

      The images shown in your articles are low mag overviews of many cells and therefore the resolution does not allow to discuss any cellular localization. Do you have any higher resolution images that you could share? Do you have any (direct) evidence and/or proposed mechanism for endosomal escape?

      The unequal distribution of uptake (cell to cell variability) is also a big concern. I don’t believe that it relates to differences between rate of uptake of different cells. Such differences would average over an 18 hour period and they should also be seen in the dextran uptake. A possible interpretation would be some degree of nanoparticle association/aggregation before interaction with the cells (this is to be tested experimentally).

      Raphael

      Like

      1. Hi Raphael,
        let me answer your two questions:
        1) Higher resolution images: The ones we have presented is the most we can do with the fluorescence microscope in our lab.
        2) Endosomal escape: We have not done any experiments to evidence the endosomal escape or to unravel any possible mechanisms involved in that.
        Best
        Harald

        Like

  12. Very nice blog. Your images clearly show that the signaling oligo’s (or clusters thereof) are still trapped in the vesicles. A hallmark of escape of labeled oligos into cytoplasm is their accumulation in the nucleus. Soon after their introduction into cell, oligonucleotides get sequestered in the nucleus (see for example Tyagi S, Alsmadi O (2004) Imaging native beta-actin mRNA in motile fibroblasts. Biophys J 87: 4153-4162). Detection of mRNAs in live cells with probes is very difficult because most mRNAs are expressed at 10-100 copies per cell, are constantly moving around, and are coated with proteins. The limited conditions under which native mRNAs have reliably been detected in live cells are where mRNAs were heavily expressed, localized in subcellular zones, or contained tandemly repeated sequences which could bind to many probes simultaneously (PNAS 100: 13308-13313, 2003 and 102: 17008-17013, 2005). The probes also have to be chemically modified so they and their hybrids with mRNA can withstand cellular nucleases.

    Like

    1. Hi Sanjay, Thanks for the interesting comments and reference. I had always assumed that when released, the fluorescent reporter strands would be cytosolic and diffuse. Needless to say, we don’t see them in the Nucleus either!

      Like

  13. Hi! Great blog, thanks for the bunch of information. Have you already figured out why the scramble give so high signal? (this is the case in my experiments as well, especially when Smartflare probes are incubated with adherent cells). Best!

    Liked by 1 person

    1. Hi Paulina, we’ve not really delved mechanistically into this, although we suspect that the signal we see is from degradation of the oligos, thus releasing the reporter strand (or the fluorophore alone).

      Interestingly the MIrkin group themselves have published a JACS paper (doi: 10.1021/ja503010a) attributing such degradation to DNaseII and other late endosomal enzymes.

      Do you also see a punctate distribution? Thanks for your comments.

      Liked by 1 person

      1. Thank You Dave and Thank You, Raphael. The DNAse problem sounds reasonable. I also get more punctuate signal than diffused one (altough I still observe the latter one as well, but regardless the type of the probe: uptake, scramble or specific). Generally, I use the protocol Milipore is suggesting yet I have also tested higher concentration of smart flares but the effect is the same. I will post if anything new comes out of my experiments. Regards.

        Liked by 1 person

    2. Hi Paulina
      Great to hear from a SmartFlares users. One of the big problem in science is lack of sharing of negative data: we do not know how many people have bought these particles, got negative data, and then put these in a drawer. If you want to share more details of your attempts, including images, you are most welcome to write a guest post for this blog. We think that the degradation (and high signal) comes from DNAse. This hypothesis is actually backed by a paper… from Mirkin’s group (http://dx.doi.org/10.1021/ja503010a).
      Raphael

      Like

    1. We have not, but I have no doubt that it works: the first step of RNA Scope, like other FISH methods, is fixation & permeabilisation. It does not rely on the probe getting magically out of the endosomes. Of course, that also means that it cannot look at live cells.

      Like

  14. Hi, I´m not an expert in the topic at all, but I think it is an interesting discussion.
    I wonder if it would be of any help to introduce peptides like TAT-peptides or nuclear localization signals into the smartflares?

    Like

  15. Did anyone use Cas9 FISH experiments beforee? Do they work? Also what is the best and most consistent method to do RNA or DNA fFISH presently?

    Like

  16. Study that uses Smartflare that I came across. Any thoughts on Figure 2. Shows pretty specific fluorescence – not something you would see with general endosomal uptake and nuclease driven fluorescence. – Studies like this make one wonder if Smartflare works… sometimes.

    Like

    1. Hi James, thanks for the comment and the link to the paper. I’m not sure what you mean by “specific fluorescence”. Regardless, I don’t think I can agree with the follow up comment. There are so many things that change the rate of endocytic uptake, assuming you’re seeing nuclease-driven fluorescence what you could be measuring is any one of these (serum, temperature, pH etc). This is why the Uptake Control is so critical in these experiments (and notably absent in this paper).

      Regadring the Figure itself, the magnification is too low, the signal too saturated and the image quality too poor (when will publishers learn!) to really make out distribution. I find the cell-to-cell heterogeneity interesting (we recently reported the same observation even with the uptake-control SmartFlares), although in this case this could represent genuine variation in the levels of miR-182/183 between cells.

      Finally, I find the variation in background intensity between the Scrambled and Experimental panels slightly concerning (for the same acquisition settings you would expect very similar background levels) but as it’s a composite image, this could represent a higher background in the nuclear channel.

      Out of interest I’ve made an original data request to Prof Takeshima for the Figure 2 data. I’ll post back if I hear anything.

      Like

    1. Hi James,
      I contacted the authors of the study and they were very forthcoming in providing the data. Interestingly but unsurprisingly (to us), the signal is largely punctate (something you would never know from the published figure). For fun, you can see the two compared here:

      http://cci02.liv.ac.uk/webgateway/render_image/33451/

      Definitely another support case for Open Data!

      I really wish we could get a comment from Millipore on why everyone is publishing SmartFlare fluorescence in puncta.

      Liked by 1 person

  17. Thanks James. I agree with Dave’s remarks. It is impossible to really determine the localization of the signal from these images. I guess the only intriguing thing is the difference between the scramble control and the other smartflares, but, as Dave notes, there could be a number of explanations for this difference.
    You write: “Studies like this make one wonder if Smartflare works… sometimes.”
    Broken clocks also give the exact time twice a day. There is in this story the unseen side of the iceberg: how many groups in the world pay Merck-Millipore, do these experiments and then do not publish the results as they can see that it does not work, or, they don’t get the results they want. Potentially an interesting case of publication bias.

    Like

  18. Hi Raphaël.
    This is an interetsing blog that I was not aware of previously until I was notified about your pubpeer comment on our recent article in scientific reports. Thank you for pointing out the typo regartding peptide reporter, it is obviously a nucleotide reporter.
    I would be genuinely interested to hear your comments on the data produced. In our paper it is mostly based on RNA message, sorting cells with different levels of fluorescence and then confirming with PCR. We show the validation experiments and we describe the reponse in multiple human donors, with individual donor data shown in the supplementary figures. We also then sorted cells based on their relative fluorescence and demonstrated they behave as functionally expected. How would you interpret the results?

    Like

    1. Hi Martin,

      Thanks for your comment.

      Note that Mirkin’s own paper shows degradation (nuclease) in the endosomes [1] and that Millipore’s own (internal and unpublished) research shows that SmartFlare do not detect mRNA [2]. The few published SmartFlare articles which show good quality microscopy images of cells (i.e. sensible magnification, intensity not saturated, etc) also reveal a punctate distribution characteristic of endocytosis. Finally, our own study [3], show that they do not detect mRNA and the electron microscopy pictures reveal that all observed particles are within membranous compartments. In short, the onus is on Millipore and Mirkin to answer some very simple questions such as: 1) how and how much of the SmartFlare do get out of endosomes? 2) how much of the particles that stay trapped in endosomes do get degraded giving rise to false positive signal?

      Regarding your own study, I cannot give you an explanation. One hypothesis would be that you are simply sorting on endocytosis and that this correlate with the other functions you are measuring. That hypothesis is somewhat supported by the fact that your two signals correlate so well as seen in Fig 2A and 2B.

      It would be very helpful if you could share the results of the scramble controls (mentioned in the methods section but not shown in Figure 1) and also if you could share good quality microscopy images of those cells that would enable to ascertain intracellular localization.

      All the best,

      Raphael

      [1] http://dx.doi.org/10.1021/ja503010a
      [2] https://raphazlab.wordpress.com/2015/11/20/do-nanoparticles-deliver-mercks-smartflares-and-other-controversies/
      [3] https://www.scienceopen.com/document?vid=a6754b9a-273e-4ccb-b965-2c98d96ac087

      Like

      1. Hi Raphael,
        The scramble controls also have a level of signal, I would have expected this as the reporter strand is held in place using simple hydrogen bonds, so there will always be a degree of spontaneous dissociation. From the microscopy we have done we also see a level of punctate staining.
        Have you also looked into molecular beacons? They have also been used to investigate mRNA signals in live cells, but as they are not nanoparticles they need other mechanisms to gain cellular access. Most images from that mRNA technology also show punctate staining. As two systems with different delivery mechanisms both show similar punctate staining, I have always assumed it was due to some form of complexes forming within the cells. It’s not really my primary area of research so I have never investigated it further, but it might be something for you to look into.
        The controversy regarding Smartflare is new to me, but is odd as the data we obtain makes sense when investigating other genes that we didn’t sort for. One way for us to look at your suggestion would be to sort by scramble controls and see if the data obtained look similar.
        Thanks, Martin

        Liked by 1 person

  19. Hi Martin,

    Our entire DNA is held together via hydrogen bonds but the degree of spontaneous dissociation is null if the complementary sequence is long enough (and therefore the number of hydrogen bonds high enough). It is the same here. The complementary sequence has been designed such that there should not be any spontaneous dissociation, but only competition replacement in the presence of the target. In vitro, it works well. The fact that the scramble shows some signal indicate that another process takes place, most likely chemical degradation by a nuclease (see reference 1 in my previous comment). Thanks for confirming that the distribution you observe has a level of punctate staining. Was the intensity of the scramble similar to the targets or very different?

    The experiment you suggest is definitely a good one. I wonder if Millipore will offer you the reagents to do those complementary controls…

    See also Dave Fernig relevant post on off target effects of siRNA.
    https://ferniglab.wordpress.com/2015/06/12/of-nanoparticles-cells-and-polyanions/

    Raphael

    Like

    1. Martin, I spent about 8 months working with SmartFlare at Millipore, and concluded that SmartFlares were not detecting specific RNAs, but instead were reporting an off-target event that caused release of the probe from the nanoparticle in the endosomal pathway (e.g. cleavage of the labeled oligo by endosomal nuclease, acid- or glutathione-mediated dissociation of the probe from the gold particle, etc.). My experience with the “scramble” SmartFlare probe was that it always exhibited lower signal than target-specific probes, even for target-specific probes directed towards RNAs expressed at minimal levels in the cells being analyzed. The consistently lower “scramble” signal could be the result of the premature release of the labeled oligo from the particle having some degree of sequence preference. A better control probe would be a target known to be absent from the cells, or one that is known not to change with differentiation. Another informative experiment for you would be to transfect the cells with recombinant Runx2 or Sox9 and see what happens to the respective SmartFlare signals. I did a transfection experiment like that and saw no change in SmartFlare signal for the transfected target (and immunostaining showed that the transfection was successful). That was “game over” for me and I left that position soon after.

      Like

  20. Hello All, I recently bought the SmartFlare kit and tried the starter probes on HeLa cells. After a number of optimization runs (with Flow and Microscopy and trying various probe concentrations, incubation times and cell densities), one dissatisfying observation has been very high scramble control signal and almost covering the entire range of the fluorescence distribution as the 18S housekeeping probe. I wonder if the scramble is being recognized somehow in HeLa cells. I am hesitating to try some target probes that are transcribed after cytokine stimulation. Any thoughts?

    Liked by 1 person

    1. Hi Matthew,

      Firstly, thanks for commenting. Publication bias makes it really hard to tell how many people are in the situation of buying, trying then binning products like these. I implore you to get these data out into the public domain for further discussion! Blogs / Figshare / Open Data repositories are all great choices.

      There is an incalculably small chance that your scrambled SF is recognised within the cell. In our hands, as you will see from the comments and links above, we found that the SmartFlares were retained in endosomes, degraded and thus produced signal regardless of the existence of a specific target.

      Because the degradation is largely by an endocytic nuclease (DNAse II: from Mirkin’s own work), a hallmark of this is the signal being largely punctate. I would love to see the data for myself but in lieu of that, is this what you see from your microscopy?

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s