Page 1 of 2

Nanoparticles in CyTOF

PostPosted: Wed Nov 25, 2015 12:17 pm
by KalleB

Our group are interested in measuring the uptake of nanoparticles consisting of different combinations of lanthanides in cells. The nanoparticles are around 5-10nm in diameter and concentrations in 10-100ug/ml range. Since we are rather new to this I have some questions:
1, Will the nanoparticles be completly disintegrated in the ICP to single ions?
2, Is there any risk of too high concentration of ions? in other words what is roughly the "highest concentration" recomended of a sample?
3, Any general tips/risk that we should consider is greatly appreciated =)

Best regards

Re: Nanoparticles in CyTOF

PostPosted: Mon Nov 30, 2015 4:44 pm
by mleipold
Hi Kalle,

There are a few issues to discuss.

1. The size of the nanoparticle. Ultimately, it's the size of the inorganic element of the nanoparticle: an organic layer to passivate it and make it more stable in aqueous solution doesn't really matter to the argon plasma. So, for example, if the nanoparticle (including coating) is 10nm but the inorganic particle is only 5nm in total diameter, then you would only consider the particle to be 5nm for purposes of calculating the number of atoms.

Ultimately, it's the inorganic size/number of atoms of metal that are important to consider when determining whether a given particle is likely to be completely atomized and ionized by the plasma. Based on the density of the particle, the radius (and therefore volume), and Avogadro's number, you can calculate the theoretical number of metal atoms per particle.

I'll let Fluidigm make an official answer on this, but in past discussions with them, inorganic particles of a few nanometers (leading to several hundreds of atoms/particle) should be fine. But there definitely is an upper limit to the size that can be burned properly....probably <100nm in diameter.

2. Concentration of nanoparticle used isn't necessarily the relevant measure. You need to use a concentration of the necessary titer: one that gives you specificity without having nonspecific background and/or streaking.

More relevant is the expected number of copies of the epitope and therefore the total number of nanoparticles/cell. There *is* an upper limit to the amount of signal the CyTOF can measure accurately. Therefore, you would generally want to use nanoparticles for relatively low-abundance markers: for example, I see little reason why you would ever use a nanoparticle to measure CD57 or CD45, while ICOS, PD-1, TCRgd, etc would be decent targets.

3. Could you describe more about your intended assay, particularly when you say "measuring the uptake of nanoparticles consisting of different combinations of lanthanides in cells"?

Using nanoparticles to stain intracellular epitopes can be tricky: intracellular often has a higher background than surface epitopes, and nanoparticles often have a higher background than non-particle reagents, so added together, your background can be significant.

Are you coating your nanoparticles with something that would cause it to be taken up by the cell in a specific manner? Minimally, a lot of control experiments would need to be done to check for specificity and resolution: for example, an uptake assay performed in the presence and absence of an uptake inhibitor to confirm the mechanism of uptake (eg, specific endocytosis).


Re: Nanoparticles in CyTOF

PostPosted: Mon Nov 30, 2015 10:01 pm
by ErinSimonds
I'd just like to add that Qdots can be measured quite nicely on the CyTOF. You can even purchase Qdot-conjugated antibodies off-the-shelf from Invitrogen and use them on CyTOF. But beware that naturally-occurring cadmium contains 106, 108, 110, and 113 Da isotopes, which could interfere with palladium or indium channels that you may be using.

Qdots contain mostly cadmium and are spiked with various amounts of tellurium or selenium, depending on the desired fluorescence. I'd recommend measuring them on Cd114, which is the most abundant cadmium isotope. You can also get fancy and sum all of the signal from the cadmium isotopes.

Qdots vary in size quite a bit, from 4 nm spheres (Qdot 525) to 8 x 15 nm rods (Qdot 655). The general rule is that larger Qdots emit at a longer wavelength. I did some tests a long time ago and decided that Qdot655 had the best balance of high signal and nonspecific binding. I've used Qdot605-conjugated and Qdot655-conjugated anti-human CD3 from Invitrogen in the past with good success. To your question, Kalle, about whether these are "completely disintegrated to single ions", I'm not sure, but they are definitely ionized to some extent, and the amount of signal did correlate as expected with particle size.

Sean Bendall and I spent a lot of time in 2010-2012 trying to conjugate lanthanide nanoparticles to antibodies. Mark Nitz's group has published a bit on this, but they hit the same major challenge that we did --- lanthanide nanoparticles are incredibly sticky. We tried various coatings, but never managed to suppress the background enough to make them useful reagents. Perhaps if you're measuring cellular uptake (i.e. by active processes, such as phagocytosis), maybe the stickiness won't be as much of an issue.

Kalle, in response your question #2 -- yes, it is possible for the ion concentration to be "too high". The detector on the CyTOF/Helios has a finite lifetime of a certain number of atoms that can hit it before it needs to be replaced. A typical antibody-stained cell prepared for CyTOF analysis will contain between 1e7 and 4e8 heavy metal atoms (i.e. >90 Da). To be safe, start small and design your titrations so you're exposing the instrument to the lowest amount of metal, preferably below the range I just specified. Then, run your titration samples in increasing order, until you begin to see signal. If you saturate a channel, it's usually not the end of the world, but there's no way to know exactly how much metal you just hit the detector with (because it was saturated, and therefore not accurately quantified).

Good luck!

Re: Nanoparticles in CyTOF

PostPosted: Tue Dec 01, 2015 1:58 pm
by KalleB
Thanks for the answers!

We do not use antibodies in our setup (atleast not at the moment) only the nanoparticles. We use the nanoparticles for enhancement of signal in MRI etc and/or for targeting certain cells. So we "don't really care" exactly where in the cell they end up, just if they do or not and how many. We use different coatings for different purposes but they are not large/thick in comparison to the particle itself.

From what I've read our <10nm should not face the problem of not being totally ionized. But if we do, is it possible to measure in several "windows" at the same time? i.e measure 100-190da and at the same time 400-490da but give less spectras for each window?

What is the smartest way to correlate the signal to numbers of nanoparticles? Would it work to simply run a solution with know concentration of particles and use that as a reference?

Erin: When you did the experiments with the Qdots how did you correlate the signal to amount of Qdots if they did not completly disintegrate? I interpret this as you had lower signal in the 114 channel that you expected?


Re: Nanoparticles in CyTOF

PostPosted: Tue Dec 01, 2015 4:26 pm
by mleipold
Hi Kalle,

I'm not sure what you mean by "is it possible to measure in several "windows" at the same time? i.e measure 100-190da and at the same time 400-490da but give less spectras for each window?"

The CyTOF acquisition window as normally set up for a v2 or a Helios runs from about masses 80Da to 209Da (so, slightly below Y up through and including Bi), so a width of about 140 Da. I think there's some confusion about terminology here: by the use of "mass" or "Da", we are talking about the mass of an *individual* metal ion (protons+neutrons), not the molecular weight of an organic molecule or the weight of a nondissociated nanoparticle/metal cluster. As such, there's no measuring up to 400-490Da, as there's no isotope of any element that has a mass of 400 Da.

As I understand it, if a given cluster isn't completely ionized, it just won't be measured quantitatively. By that, I mean whatever *does* get burned off the outside of the particle and is therefore dissociated into single metal atoms and ionized, *would* get measured. But the remnants of the undissociated core wouldn't enter the machine properly: I'm not sure it would even make it past the cones, and would certainly be the wrong mass to make it through the quadrupole mass filter upstream of the TOF chamber. (Fluidigm or another mass spec guru, please correct me if I'm wrong).

Regarding the Qdots: commercially-available Qdots made for fluorescent flow cytometry are currently made from natural abundance Cadmium (in the form of CdS, CdSe, and/or CdTe). Therefore, they are a mixture of several different isotopes of Cd.

While the exact ratio depends slightly on the exact source (mine) for the Cd, typical natural abundance from sheets listed in the Resources section here on Cytoforum would be:
106Cd= 1.25%
108Cd= 0.89%
116Cd= 7.49%

As you can see, several of these Cd isotopes are the same mass as the isotopes of some other useful metals (106-110Pd, 113In, etc). This impacts your panel design options. This also means that it is common to monitor only one channel for a Qdot probe: 114 or 112 are good choices, as they don't share masses with any other reagents that we currently use, and are a high percentage of the total Cd mass. Yes, technically, by monitoring only the 114 channel you are throwing away ~70% of your signal, but since each Qdot is a nanoparticle made of hundreds of metal ions, you can "afford" the loss.


Re: Nanoparticles in CyTOF

PostPosted: Tue Dec 01, 2015 4:46 pm
by ErinSimonds
Kalle, since you're doing MRI, am I right in assuming that these are gadolinium nanoparticles? Keep in mind that like cadmium, Gd also has a bunch of stable isotopes (7 to be precise). As Mike said, if you focus on only the most abundant isotope, you'll be throwing away a lot of signal. You can sum the signal from all of those isotopes to get more sensitivity (though perhaps exclude Gd152, as it's only 0.2% of natural isotopes). This will increase your background noise a bit as well, but can be worth it to get more signal.

In terms of ionization, I think you're in safe territory with <10nm particles. Of course, we're assuming that your particles are comprised of readily ionizable elements. Some elements like gold and tungsten don't ionize efficiently. There's a recent thread on here about gold nanoparticles that you may want to read.

I agree with Mike that it's probably not possible to alter the mass windows on the CyTOF in the way you described. You can do a little bit of fiddling with mass windows, but I think you'll be restricted to the 90-210 Da range by the software.

To correlate the signal to numbers of nanoparticles, you will need to run a standard solution of nanoparticles. From there you can calculate the number of "counts" on the TOF that correspond to 1 nanoparticle. I'd recommend that you borrow Mike's trick of mixing the nanoparticle suspension with CyTOF tuning solution. The advantage of this approach is that the tuning solution itself has a known ion concentration of several elements. Check out the section titled "Determination of Metal Ions per Ab Molecule" in this article:

Sorry if my earlier post was confusing -- I was trying to say that the Qdots did emit the expected signal. The signal correlated very nicely with the anticipated nanoparticle volume, so I interpreted that to mean that they are being ionized completely.

Best of luck!


Re: Nanoparticles in CyTOF

PostPosted: Tue Dec 01, 2015 5:35 pm
by mleipold
Hi Kalle and Erin,

Erin's suggestion about using our published trick of using CyTOF Tuning Solution as a diluent/internal standard for metal counting does work well for certain metals (including lanthanides). However, remember, Tuning solution is a weak (2%) solution of Nitric acid. This helps keep the lanthanide metal salts happy in solution, but also should start "eating" your nanoparticles (thereby making a solution of the atoms rather than a suspension of intact particles).

Therefore, you would be checking whether you get the Solution Mode signal intensity to linearly increase with increasing (dissolved) nanoparticle concentration, not *intact* nanoparticle concentration. As such, you may want to *plan* to dissolve some of your particles in concentrated nitric acid overnight, then dilute *into* Tuning solution as an internal standard.


Re: Nanoparticles in CyTOF

PostPosted: Fri Dec 04, 2015 9:29 am
by antonio
Hi Mike,

nice trick, I noticed now in your manuscript.

Since the tuning solution contain 5 metals and the trasmission change over the mass range. Do you approximate your calculation to the nearest metal?


Re: Nanoparticles in CyTOF

PostPosted: Fri Dec 04, 2015 4:28 pm
by mleipold
Hi Antonio,

Yes, that's probably the best way to do it.

The Tuning Solution's 5 metals do a pretty good job of approximating transmission coefficients in a nearest-neighbor approach.

If you wanted to be really hard-core about it, you can use the Multielement Standard Solution from the Tricot et al paper. The metals in it are present in their natural-abundance form (eg, all 7 Gd isotopes at their mass fractions), so that covers all (or just about all) masses in the Lanthanide range, so you can calculate things directly.

The HIMC has used that to profile several machines, and it works really well. It's just that there are a lot of calculations needed the first time you set up the spreadsheet, due to some elements having isotopes of the same mass that contribute fractionally to the signal at a given mass (not to mention if you further
"complicate" things by assuming something like 2% M+16 oxide for each isotope). But further experiments become trivial once you set up that spreadsheet the first time.


Re: Nanoparticles in CyTOF

PostPosted: Fri Dec 04, 2015 5:16 pm
by mleipold
Hi Antonio,

I would also like to mention that if you want to do this sort of quantification with metals vs Standard Solutions, you will want to make sure that your measurement conditions are appropriate for the metal. Remember, not all metals are equally "happy" in all solutions (pH, counterion, etc).

As Erin pointed out, the Mei et al Cytometry A paper describes this for Pt and for Lanthanides. However, note that the Pt was diluted in HCl, not Tuning solution or nitric. When Henrik was doing all of these comparisons, we did find that the acid matrix did make a difference for Pt. We also were working on a similar quantification for Pd (we came up with the metal quantification idea *after* publishing Mei et al J. Immunol), and that was showing similar acid-dependent quantification.

Spex Certiprep makes a ton of ICP-MS standard solutions. As a rule of thumb, we used the acid matrix in the standard for that element, in order to quantify the amount of metal. So, that's a good starting place.

Hopefully Henrik will comment more on these trials.