A large body of scientific nanotechnology literature is dedicated to the biomedical aspect of nanoparticle delivery into cells and tissues. The functionalization of the nanoparticle surface is designed to insure their specificity at targeting only a certain type of cells, such as cancers cells. Other technological approaches aim at the cargo design, in order to ensure the targeted release of various biologically active agents: small pharmacological substances, peptides or entire enzymes, or nucleotides such as regulatory small RNAs or even genes. There is however a main limitation to this approach: though cells do readily take up nanoparticles through specific membrane-bound receptor interaction (endocytosis) or randomly (pinocytosis), these nanoparticles hardly ever truly reach the inside of the cell, namely its nucleocytoplasmic space. Solid nanoparticles are namely continuously surrounded by the very same membrane barrier they first interacted with when entering the cell. These outer-cell membrane compartments mature into endosomal and then lysosomal vesicles, where their cargo is subjected to low pH and enzymatic digestion. The nanoparticles, though seemingly inside the cell, remain actually outside. How so?

Dave Fernig, professor at the University of Liverpool, UK, and specialist for cell membrane signalling, explained this apparent conundrum in his email to me:

“Easiest way is to think of your body. The inside of your stomach (and indeed your entire gastrointestinal tract) is in fact outside your body: you don’t have to cross any barriers to get there. Same with a cell.  The lumen of the vesicles and tubes that comprise the endosomal apparatus, from the initial vesicle pinched off the plasma membrane through to the lysosome (the cell’s “stomach”, where enzymes break down the proteins, polysaccharides and nucleic acids brought in by endocytosis) is outside the cell. So to be inside, you have to physically cross a membrane”.

Yet while there are many studies showing the uptake of nanoparticles by cells, there seems to be little evidence that solid nanoparticles ever pass through the endosomal membrane and enter the nucleocytoplasmic space, where they could faithfully deliver their cargo to its correct destination. Occasionally however, novel research on exactly such technical advances is announced, and due to its potentially tremendous relevance to the field, such publications often feature in certain top-tier journals. Here, I will present the controversy surrounding two such approaches to nucleocytoplasmic nanoparticle delivery: the so-called striped nanoparticles from the laboratory of Francesco Stellacci, and the spherical nucleic acids, i.e., nanoparticles functionalised with nucleic acids, from the laboratory of Chad Mirkin.


The stripes which never were?

Raphaël Lévy, senior lecturer at the University of Liverpool, UK, was always interested in this possibility of nanoparticle delivery straight into the nucleocytoplasm. Thus, it was hardly surprising that his attention was caught by research from the Stellacci laboratory (then at the MIT in USA, now at EPFL in Switzerland). In 2004, a paper in Nature Materials (Jackson et al., Vol. 3:330-336, 2004) described “striped” nanoparticles, where alternating functional coatings were arranged on the gold nanoparticle core through a kind of self-assembly in a geometrical pattern (hence, striped). This was followed by another Nature Materials paper in 2008  (Verma et al., Vol. 7:588-595, 2008), where this striped peculiarity allegedly allowed the nanoparticles to enter the cytoplasm directly, while nanoparticles without this functionalisation ended up trapped inside endosomes.

Lévy has been challenging this concept for some time (Cesbron et al., Small 8:3714-3719, 2012), but in order to provide definitive evidence, he teamed up with Philip Moriarty, professor of physics at University of Nottingham and scanning probe microscopy specialist. The studies were led by Moriarty’s then-PhD student Julian Stirling, who published his results as corresponding author in PLOS One (Stirling et al., PLoS ONE 9:e108482, 2014). What Stirling et al did first, was to ask Stellacci to share the original microscopy data from his publications, which they then re-analysed. Their interpretation: the stripe morphology of alternating functionalizing ligands, as detected by Stellacci’s team, “can instead be explained by a combination of well-known instrumental artefacts, or by issues with data acquisition/analysis protocols”. Indeed, the stripes presented in Stellacci’s publication in the images of several adjacent nanoparticles were all equally oriented, a statistically unlikely situation. Yet as Moriarty explained to me, the stripes on all nanoparticles were perpendicular to the scan direction, which for him is evidence for certain technical negligence when using scanning tunnelling microscopy (STM). Together, the Moriarty team could reproduce very convincing stripe-like patterns on bare nanoparticles, where actually no ligands at all could have been involved into stripe formation.  All they had to do was to introduce a certain scanning error into their image acquisition. Moriarty added that, when other methods (namely transmission electron microscopy, nuclear magnetic resonance spectroscopy and small angle neutron scattering) were applied by Stellacci’s team to find further proof for the existence of stripes, other technical artefacts sneaked in there as well. To him and Lévy it is therefore obvious that the nanoparticles never had any stripes.  In fact, Stellacci’s own former graduate student, Predrag Djuranovic, has voiced his concerns about the instrumental-error origin of the stripes already in 2005, which he defined as “feedback oscillations in STM imaging”.

Stellacci in his email to me rejected outright the artefact claims by Moriarty and Lévy. He pointed out his reply to their criticisms in PLOS One (Ong & Stellacci, PLoS ONE 10:e0135594, 2015) as well as listed a number of studies where his data has been reproduced.  According to Stellacci, “the controversy is over”, since the striped phenotype was confirmed independently in four different labs (Biscarini et al., Langmuir 29:13723-13734, 2013; Ong et al., ACS Nano 7:8529-8539, 2013; Ong et al., Chemical Communications 50:10456-10459, 2014). However, Stellacci is last and corresponding author on all three of these publications, which is not what generally associated with truly independent replications. The biological aspect of the nucleoplasmatic penetrability of the striped nanoparticles has been also reported by other labs (Carney et al., Biointerphases 7:17, 2012; Sabella et al., Nanoscale 6:7052-7061, 2014), as referred by Stellacci, who is also corresponding author of these two papers as well. To him, “the concept [of nanoparticle cell penetration] is totally accepted by the literature nowadays”. Stellacci also stated in his email: “Whether or not the interpretation of my STM images is right, it is inconsequential on the biological behavior”. However, the key message of the relevant Stellacci paper reads that only the striped gold nanoparticles “penetrate the plasma membrane without bilayer disruption, whereas the [randomly coated ones] are mostly trapped in endosomes”. However, if the physical existence of the striped pattern is questioned, it may also directly concern the results of the biological cell penetrability of these very particles.

Smart Flares: unspecific signals?

The next family of allegedly nucleocytoplasmic nanoparticles which Lévy turned his attention to, was that of the so called “spherical nucleic acids”, developed in the lab of Chad Mirkin, multiple professor and director of the International Institute for Nanotechnology at the Northwestern University, USA. These so called “Nano-Flares” are gold nanoparticles, functionalized with fluorophore-coupled oligonucleotides matching the messenger RNA (mRNA) of interest (Prigodich et al., ACS Nano 3:2147-2152, 2009; Seferos et al., J Am. Chem.Soc. 129:15477-15479, 2007). The mRNA detection method is such that the fluorescence is initially quenched by the gold nanoparticle proximity. Yet when the oligonucleotide is displaced by the specific binding of the mRNA molecules present inside the cell, the fluorescence becomes detectable and serves thus as quantitative read-out for the intracellular mRNA abundance. Exactly this is where concerns arise. To find and bind mRNA, spherical nucleic acids must leave the endosomal compartments. Is there any evidence that Nano-Flares ever achieve this and reach intact the nucleocytoplasmatic space, where their target mRNA is?

Lévy’s lab has focused its research on the commercially available analogue of the Nano-Flares, based on the patent to Mirkin and Northwestern University and sold by Merck Millipore under the trade name of SmartFlares. These were described by Mirkin as “a powerful and prolific tool in biology and medical diagnostics, with ∼ 1,600 unique forms commercially available today”. The work, led by Lévy’s postdoctoral scientist David Mason, now available in post-publication process at ScienceOpen and on Figshare, found no experimental evidence for SmartFlares to be ever found outside the endosomal membrane vesicles. On the contrary, the analysis by several complementary approaches, i.e., electron, fluorescence and photothermal microscopy, revealed that the probes are retained exclusively within the endosomal compartments.

In fact, even Merck Millipore was apparently well aware of this problem when the product was developed for the market. As I learned, Merck performed a number of assays to address the specificity issue. Multiple hundred-fold induction of mRNA by biological cell stimulation (confirmed by quantitative RT-PCR) led to no significant changes in the corresponding SmartFlare signal. Similarly, biological gene downregulation or experimental siRNA knock-down had no effect on the corresponding SmartFlare fluorescence. Cell lines confirmed as negative for a certain biomarker proved highly positive in a SmartFlare assay.  Live cell imaging showed the SmartFlare signal to be almost entirely mitochondrial, inconsistent with reported patterns of the respective mRNA distributions.  Elsewhere however, cyanine dye-labelled oligonucleotides were found to unspecifically localise to mitochondria   (Orio et al., J. RNAi Gene Silencing 9:479-485, 2013), which might account to the often observed punctate Smart Flare signal.

Luke Armstrong, former Research & Development manager with the Merck Millipore division in California, described his experience in his email to me:

“My impression was that the signals shown in SmartFlare product documentation, while quite impressive in many cells, had nothing to do with the levels of their intended target mRNAs.  Instead, the product documentation indicated to me that the fluorescent oligos were being released from the quenching gold nanoparticles by another mechanism, which could be the acidic environment of the endosome, endosomal nucleases (Mirkin’s own studies (Seferos et al., Nano Letters 9:308-311, 2009) found that nucleases degrade gold-bound oligos, albeit less efficiently than free oligos), or cytosolic glutathione (if the SmartFlare particles are actually able to escape the endosome).  One publication from Mirkin’s own lab (Wu et al., J. Am. Chem. Soc.136:7726-7733, 2014) confirmed much of this, which one would have thought would put to rest the idea that SmartFlares actually detect mRNAs, but Mirkin himself and Merck Millipore have taken the position that those results are anomalous, and published reports of success with SmartFlares detecting mRNAs are legitimate”.

More recently, Mirkin lab has developed a novel version of spherical nucleic acids, named Sticky-Flares (Briley et al., PNAS 112:9591-9595, 2015), which has also been patented for commercial use. The claim is that “the Sticky-flare is capable of entering live cells without the need for transfection agents and recognizing target RNA transcripts in a sequence-specific manner”. To confirm this, Lévy used the same approach as for the striped nanoparticles: he approached Mirkin by email and in person, requesting the original microscopy data from this publication. As Mirkin appeared reluctant, Lévy invoked the rules for data sharing by the journal PNAS, the funder NSF as well as the Northwestern University. After finally receiving Mirkin’s thin-optical microscopy data by air mail, Lévy and Mason re-analyzed it and determined the absence of any evidence for endosomal escape, while all Sticky-Flare particles appeared to be localized exclusively inside vesicular membrane compartments, i.e., endosomes (Mason & Levy, bioRxiv 2015). The two authors submitted their evaluations to PNAS as a Letter to the Editor, which however was rejected by the Editor-in-Chief Inder Verma on the grounds that this “letter does not contribute significantly to the discussion of this paper”.

In the nutshell, the key issue with SmartFlares is the likely unspecific signal they produce, when the fluorescent nucleic acid probes become unquenched by lytic degradation and leak out of the endosomes, while the solid nanoparticle cores are apparently still trapped inside.

Mirkin declined to comment on this controversy publicly, and gave me no permission to cite from his confidential email in regard to Lévy’s criticisms.  Merck’s spokesperson, Jill DeCoste, initially promised to provide a statement, which never came, despite my reminders. Recently, Mirkin and Northwestern University have received from the US National Cancer Institute (NCI) 11.7 million USD funding specifically for the development of clinical treatments based on spherical nucleic acids, as part of its 5-year cancer nanotechnology program.

As Lévy pointed out, gold nanoparticles were already used in the early days of the cell membrane trafficking research in order to visualize endosomal pathways, since gold colloids, unlike viruses, remained permanently trapped inside endosomes (Epstein et al., JEM 119:291-302, 1964; Harford et al., J.Biophys.& Biochem. Cytology 3:749-756, 1957). Stephen Royle, professor at the University of Warwick and specialist for endocytosis and Membrane trafficking, described in his email to me the current scientific understanding as such: “endosomal escape has not been well documented for SmartFlares/StickyFlares or for other nanoparticles”.  He added that the difficulty to replicate the published results “highlights a real gap in our understanding of the cell biology behind endosomal processing by cells”. Royle concludes: “a mechanism for endosomal escape needs to be defined before we can use nanoparticles as reliable delivery systems”.

However, a brief look into scientific literature easily reveals that enough scientists seem to have already moved on: though never directly observed or mechanistically proven, endosomal escape of solid nanoparticles is dealt with as an established fact. In this regard it is not helpful that analytic methods like low resolution-light microscopy or even flow cytometry (FACS) are applied to provide evidence for the nanoparticles’ intracellular localization.

Especially the latter technology, though highly popular due to its quantitative advantage, offers no insights whatsoever whether the detected fluorescent signal stems from inside the nucleocytoplasmatic space, endosomal/lysosomal vesicles or even outer cell membrane. This fundamental FACS limitation did not discourage scientists around Steve McClellan from the Mitchell Cancer Institute at the University of South Alabama to offer to scientific community “a next generation cancer stem cell identification technique”, based solely on flow cytometrical detection of SmartFlares (McClellan et al., Methods 82:47-54, 2015).

Other researchers, like the research group of Harald Lahm from the German Heart Center in Munich, did apply both FACS and microscopy for their “live fluorescent RNA-based detection of pluripotency gene expression” (Lahm et al., Stem Cells 33:392-402, 2015), though, as Lahm admits, with insufficient resolution to properly analyse the SmartFlare localization inside cells. In such cases, certain negative control experiments are generally required, to verify the probe signal’s specificity. Lahm however commented on behalf of all authors in his email: “we did not explicitly perform any assays for the specificity of SmartFlares”. He advised me to contact Mirkin for questions in this regard.

Thus, scientists should always seek to obtain direct evidence via appropriate high-resolution methods such as electron microscopy, before any new claims of nucleocytoplasmatic penetrance of solid nanoparticles can be made.

11 thoughts on “Do nanoparticles deliver? Merck’s Smart Flares and other controversies

  1. Reblogged this on Rapha-z-lab and commented:
    Leonid Schneider’s article starts with a summary of the stripy controversy and then moves on to the SmartFlare. Of particular interest is the quote from Luke Armstrong, formerly at EMD Millipore, which demonstrates that the company ought to be well aware that the probes detect nucleases rather than mRNAs. This begs the question of why they are still selling and advertising this product. Unfortunately, they did not provide a statement to Leonid.


  2. The above mentioned publication from the German Heart Center (Lahm et al., Stem Cells 33:392-402, 2015) does offer an important negative control for SmartFlare specificity. It is shown in Fig 1B, where scramble SmartFlares are applied, showing only a very weak background signal.

    It seems an error in image assembly was made, as the overlay image is composed of two different fields: a monolayers of cells in “phase” is different from the cell colony shown in “Scramble SF”. Other overlay images seem correct, only the negative control seems affected by this assembly error. It is very important to provide the correct image for the negative control, since other labs (see above) reported a strong unspecific signal from Scramble SmartFlares.
    I have contacted Lahm and his co-authors Krane, Doppler and Dressen about this issue on Nov 20th, but received no reply so far, despite our earlier intense and successful communication.


    1. Hi Leonid,

      I think I have to disagree about the “figure assembly error” (which is a charmingly diplomatic term!). I agree that if you crank the levels you do see the big blob as you’ve shown above, but this will happen with any widefield microscope that has slightly uneven or off-centre illumination. When you push the levels so far, you start to discern that background illumination pattern (especially when combined with some compression artifacts) which you wouldn’t expect to see in the phase image (as it’s a different light source).

      If you don’t push the levels too far, you do see a low level of (cell-sized) punctate signal. See:


      1. Hi Dave, thanks for your comment. I however disagree with your interpretation, the purpose of boosting the background is to see the shape of the cell colonies, via their autofluorescence or background signal. to me it does look like a differently shaped cell colony, similar to the issue presented here: https://pubpeer.com/publications/1B7DE752C246EF1C74CF7970C0FA08#fb32037
        I think the issue can be easily clarified if the authors would consider to engage. Unfortunately, they do not.


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