Spark UV 387: more interactions and a solution

Today’s post follows up on this previous report of an interaction between Spark UV 387 and Real Blue 670. Thanks to everyone who wrote in or commented that they’ve seen the same effects—it’s great to get the confirmation. I didn’t discover this effect anyway. It was brought to my attention by Vera Dermesrobian at KU Leuven.

Today we have two things: 1) additional interactions with other dyes (boo!) and 2) an apparent solution (yay!).

Let’s start with the bad news: Spark UV 387 also interacts with at least two other dyes emitting around 660nm. There may be more. Manuel Trebo at the Medical University of Innsbruck reached out to me with findings showing that Spark UV 387 interacts with BUV661. I’ve since tested and also found interactions with BV650. Like the Real Blue 670:Spark UV 387 interaction, these effects are not mitigated by Brilliant Stain Buffer or True-Stain MultiFluor Buffer, but can be dealt with by staining sequentially. You have to stain with the Spark UV 387 after the other dye(s). That’s the first odd bit, as before.

You also need to wash quite a bit in between, which I may not have emphasised adequately. We’ve been using the RB670 for surface staining and the Spark UV 387 for intracellular/post-fix staining, so there are always about 5 buffer changes between the staining steps. You can see in Manuel’s data that with sequential surface staining, you may have to wash a lot, so you may prefer to drop one of the dyes (see the previous post for tips on this).

Three wash steps were needed here.

Examples from Manuel Trebo

In these examples we see a strong interaction between CD45RA Spark UV 387 and CD4 RB670, CCR5 RB670 and a moderate interaction with CD19 BUV661. All of these are reduced with CD62L Spark UV 387, perhaps partly due to the lower expression level of CD62L.

The second odd bit: there are some indications that these interactions are conjugate-specific, or perhaps manufacturer-specific. This probably needs to be investigated further. I’m not sure whether every company gets their SIRIGEN polymers from the same source, or if there is in-house manufacturing with, perhaps, some differences on the external part of the structure for stability/conjugation. Anyway, what it means is that while I saw strong interactions with two BV650 conjugates, but I don’t see interactions with BV650 in our panels, and Manuel hasn’t seen interactions with BV650. So, odd and inconsistent. For now, be careful with BV, BUV and Super Bright tandem dyes.

I ran a couple of tests looking at conjugates of the same dye from different vendors, but couldn't find any difference in interactions. There is probably a minor interaction with BV711 (the CD4 T cells really shouldn't be staining for T-bet without permeabilisation since T-bet is in the nucleus).

Here are some tests I’ve run on human PBMCs, using CD8 in Spark UV 387. There’s the strong interaction with RB670, none with SBB675 (some non-specific staining from using too much antibody), no problem with PE-Cy5, nothing with PerCP, maybe something with BUV661, and definitely a problem with BV650.

Human PBMCs (no mouse IL-10 present)

I tested PE-Cy5 and saw no interaction there. Nor have I see interactions there in our panels. I tested PE-Cy5 because my best guess, on essentially no information, is that these dyes emitting around 660nm may all contain a Cy5-derived FRET acceptor as part of the dye. The structures of Real Blue 670, BUV661 and BV650 are all proprietary and trade secrets, so I have no idea if there is actually Cy5 in any of those, and PE-Cy5 is not a very good comparison anyway because it is protein-based.

One thing to point out here is that this is not just an interaction creating false positives between the pair of dyes. We’re also getting unmixing errors throughout the data, suggesting the that dyes are interacting closely enough to transfer energy between them, basically creating new fluorophore FRET pairs. This manifests as skewing (biasing) errors versus other fluorophores with similar emission wavelengths. Note: there is no staining for NKp46 or CD11b in the example below, we're just unmixing the channels.

Here we’re looking at empty channels, gating on the CD8 Spark UV 387 expressing cells. We expect nice round negatives like on the left, with PerCP, a non-interacting dye. With the samples co-stained for RB670 or TCRgd, we’re getting false signals in these channels where there should be nothing.

Anyway, on to the better bit: a likely solution.

Since I wrote to them a year or so ago when this issue first cropped up, BioLegend has been looking into it. While they don’t have any direct solutions to deal with the interactions with Spark UV 387, they do have a new version of the dye: Spark PLUS UV 387. Like the other “PLUS” dyes, this is brighter (think more dye molecules jammed together). And that fixes the problem.

In the interest of full disclosure, BioLegend has sent me this vial of Spark PLUS UV 387 for free to test. Otherwise, I don’t have any ongoing relationship with BioLegend except that of a customer.

Here are the examples from my tests. I’ve run the regular and PLUS versions side-by-side, with the conjugates that gave me problems.

human PBMCs, all y-axes should show no signal except CD14

The PLUS version is brighter. There’s no interaction visible with either RB670 or BV650. We do have RB670+Spark PLUS UV 387+ cells when we stain for CD14, but that’s correct in this case because it’s CD14+CD4low monocytes, and the CD4-bright T cells do not stain now.

The Spark PLUS version of UV 387 is a good solution to the problem. It appears to stop the interactions with other dyes (probably more testing will be required, but so far, so good), it allows you to use the conjugate whenever (no sequential staining), and it’s brighter. Brighter isn’t always better—it’s nice to have clean dim dyes for lineage markers—but in this case, Spark UV 387 is dim enough that even molecules like CD45 will still have no risk of going offscale with the PLUS version. So, I doubt there are many cases of existing panels where Spark UV 387 cannot be simply swapped to Spark PLUS UV 387.

Anyway, this is a nice resolution because Spark UV 387 is a very good dye in that it is quite clean (limited spillover). I like to use it for lineage markers like CD45, CD4, CD8, where every cell I care about analysing will be staining with the dye, because this way, it minimises the spread on those cells, maximising the amount of detector space available for other markers.

That’s all for today. Hope this helps.

Conjugates used in this post by Manuel Trebo (first two figures and data not shown):

BioLegend:

• SparkUV387 - CD45RA HI300 REF 304179 LOT B435885

• SparkUV387 - CD4 SK3 REF 344685 LOT B389947 

• SparkUV387 - CD14 S18004B REF 399215 LOT B476226 

• SparkUV387 - CD62L S21001C REF 285502 LOT B492468 

• BV650 - CX3CR1 2A9-1 REF 341625 LOT B485702 - this one only in larger panels combined with UV387 never in a direct comparison

Waters (BD):

• RB670 - CD195 3A9 REF 771684 LOT 5353867 

• RB670 - CD4 SK3 REF 571786 LOT 4261296 

• BUV661 - CD19 SJ25C1 REF 750536 LOT 87308 (bit of an older one)

Cytek (Testing Kit)

• cFluor B675 - CD4 SK3 REF R7-20152 LOT C0047071524R72 

Bio-Rad:

• StarBright B675 - CD8 LT8 REF MCA1226SBB675 LOT 100006890

Conjugates used in this post by Oliver Burton (other figures and data not shown):

BioLegend:

• Spark UV387 - CD4 SK3 344685

• Spark PLUS UV387 - CD4 SK3 (prototype)

• Spark UV387 - CD8 SK1 344776

• PE-Cy5 - NKp46 9E2 331952

• PerCP - CD14 63D3 367152

• CXCR3 - BV650 CXCR3-173 126531

Waters (BD):

• RB670 - CD14 63D3.rMAb 770847

• BV650 - TCRgd B1 564156

• BV650 - TCRgd 11F2 569510

• BV650 - IL-10 JES5-16A3 564083

• BUV661 - CD19 1D3 612971

• BUV661 - CD11b M1/70 612977

ThermoFisher:

• BUV661 - CD19 1D3 376-0193-82

• BUV661 - CD11b M1/70 376-0112-82

Bio-Rad:

• StarBright B675 - CD14 TÜK4 MCA1568SBB675

Green-fronted Lancebill, Costa Rica