I’m using an Agilent Bravo NGS B to automate an RNA library construction method where biotinylated RNA is bound to strepavidin magnetic beads (Thermo Dynabeads MyOne C1 streptavidin) and serial enzymatic steps are performed on-bead with washes in between them. Libraries built on the Bravo give 1/4 the yield of the same starting material prepared manually.
I suspect the clumping accounts for the lower automation yield. The beads start in good (i.e. ~monodisperse) suspension but clump early, and progressively, over 6-8h of handling. I routinely do this method manually in 1.5mL tubes without issues.
I’ve tweaked my Bravo protocols to minimize the time beads are dry between washes and increased how vigorously I resuspend the beads between steps (currently 2000rpm shake x 30sec followed by 500uL/s pipet mix x 30 cycles). This helped somewhat but clumps remain.
I wonder if anyone has had a similar issue with Dynabeads C1 streptavidin beads specifically or streptavidin beads generally? It would be great to hear any suggestions you might have for troubleshooting the clumping.
I’m happy to provide more details on any of the above but thought I’d keep things concise to start.
I guess this is tricky because we know nothing about your manual process but I can share some general experience notes.
For all magnetic beads, the mix is super important both before and during the protocol. A lot of people will use rotators and let the beads mix around for 30+ minutes if clumping is seen to be that bad but it’s also to mix thoroughly upon re-suspension.
The drying steps can also matter so make sure you’re not leaving too much EtOH behind in the wells.
Try to mimic the shake as much as possible but note that not all shakers are going to be able to mimic each other. Some are better for some jobs than others.
The magnet you use can matter, depends on the protocol. What kind of magnet are you utilizing?
Understated but labware could also have an impact.
All of these are obvious points and it varies from system to system and setup to setup but these are the most critical steps IMO.
Thanks for the reply. I realize there was a lot of detail I glossed over in my initial post. Here are some of the relevant details
Manual vs robot
The washes are a series of tris buffered salt/detergent solutions ranging from high salt buffer (50mM Tris pH 7.4, 2M NaCl, 0.5% Triton X100, 1mM EDTA) to no salt with 0.1% triton.
Manual protocol: 400uL wash volumes in 1.5mL low binding Eppendorf tubes. Enzyme reaction volumes are 20uL and tubes are rotated during incubations. I resuspend either by pipetting or moderate vortexing on a standard lab vortexer.
Automation: 100uL washes in Brooks FrameStar or Eppendorf twin.tecs plates. Enzyme reaction volumes 20-30uL. Bead resuspension by on-deck shaker pad (Teleshake from Thermo) and pipet/mix. Incubations are off deck on a rotator by inversion.
Drying
No ethanol is used in the protocol but your point about drying is well taken. The beads are kept in aqueous suspension as much of the time as I can manage. The time the beads spend dry between washes is ~15-20sec.
Magnet
I have an Alpaqua 96S ring pull. Perhaps not ideal as the beads are tend to spread out along well walls a bit. The plates sit on the magnet for ~30sec before supernatents are removed. I’ve been wondering whether an alternative magnet might help.
I’m just going to throw random thoughts your way, sorry but it’s boiling right now so I’m distracted.
Have you considered doing partial runs? First half automated, second half manual and the other way around… It’ll be interesting to see for example if the yields are marginally improved by some critical step.
What’s the drying time in the manual process vs automated? Is the aqueous suspension volume the same for both processes?
I use that magnet a lot and I like it, it allows me to get to the very bottom of the well and pipette as much out as I need to but then re-suspension becomes a big concern because as you noted, the beads are along the edges. Do you pipette while it’s mixing?
That’s a good thought. My manual runs have always used different volumes and magnets than the automated runs. I can adjust my manual protocol into PCR strips to better parallel the why the robot does things. Might give me more insight into the difference.
An alternative magnet is an interesting idea too. A side pull with a more concentrated bead pellet might reduce the potential for drying. I’m a bit restricted at the moment to using a universal magnet due to some other protocols I have that use deep well blocks on the magnet. That could, of course, be tweaked and it may be worth doing if it resolves my bead issue here. I had been thinking of testing out Alpaqua’s Magnum FLX (Magnum FLX® - Alpaqua) which isn’t side pull but claims lower elution volumes are possible.
I feel that’s the best place to start, remove or reduce the number of variables (in this instance, differences) between your manual vs your auto process.
Wash with same volumes, verify that intermediate steps leave the beads in similar aqueous volumes, and perhaps even trying using a plate that’s perhaps better or more similar to the wells in your manual process. Resuspension is critical as are drying times if needed. In the past, shaking has also been a deal breaker so I’ve learned to mix while shaking to get that thorough level of resuspension.
The Magnum FLX is a great magnet, always had good results with it. Another point to consider is bead volume viscosity: a good wash that is properly mixed should result in beads that spread evenly across the magnet ring. This allows you to pull your eluted resuspension from the center of the well and reduce bead carryover and clumping if you set the aspirate speed to low and slow.
The elution resuspension, as others have said, is a critical step. Likely need to mix far more than you normally would, and yeah shaking usually doesn’t cut it. I would have your Bravo mix from the 4 corners of the wells (manually move the head in x and y a few millimeters and run a mix cycle, repeat for each corner) to avoid a clump forming in the corner and capturing your sample from being resuspended. A good resuspension like a good wash will result in freely flowing beads that form a nice ring.
Also agree with @luisvillaautomata, I would break up the automation into its constituent steps (bind, wash, elute) along with manual steps to better pinpoint where optimization is needed.
This is great, I think I’ve got some idea about sensible next steps. It definitely makes sense to make a manual version of the protocol that’s closer to my automation version to help figure out what the difference is that’s causing clumping.
The tips on bead resuspension are also really helpful. I hadn’t considered that reducing the wash volume for automation might change solution properties as function of bead concentration. The manual testing should let me know if that’s an issue. If so, scaling down my bead amounts might be worth testing too. I’m pretty sure I have excess binding capacity.
Thanks all for taking the time to offer advice. I’ve been looking online for this kind of forum for a long time and am very glad to have found one.
With process development, I feel that it’s always important to have next steps. If you have an elite group of scientists they can help provide any guard banding results on their end to make some of the decision making simpler.
As far as next steps are concerned, I always like to have a few labware options, tip type options, accessory options (different magnets, shakers) and method variability (shaking or mixing or drying variability.)
With that said, you could plan it all well in advance as best as you can and still wind up with something that isn’t as great as the manual process. BUT at least you find out sooner.
It does seem like a better magnet might help. I use both PCR plates and 1.3mL round bottom, deep well plates in my protocol (https://www.thermofisher.com/order/catalog/product/260251). Would a post magnet of this sort work for both deep well and pcr plates?
In your work with the streptavidin beads, has clumping ever been an issue? I’ve been using dynabeads C1 which are 1um and hydrophilic. They give excellent yields in my manual protocol as compared to larger, hydrophobic beads (dynabeads M280) though I don’t know if that’s a function of bead size or surface. Could I ask which beads you’ve been using and a bit more about your application?
Thanks for checking back. Unfortunately I got a big batch of unrelated samples to process just when I was planning to follow through on some of the suggested testing so I’ve hardly had time. I did try manually wrangling the plate on deck to get the bead ring as low as the Alpaqua 96S allows (i.e. pause the robot and lift the plate briefly to place the bottoms of the wells on the edge of the magnets). Didn’t seem to help but probably could have done it more cleanly.
I definitely maxed out the pipetting and vortexing speeds possible on my deck and that alone is unable to keep the beads nice over a long wash cycle.
I also talked to our Agilent FAS and it turns out one of their SureSelect kits uses the same beads (Dynabead C1) in a similarly long series of washes without issue. That suggests maybe buffer composition is a factor (though it remains mysterious why it only matters on the robot). I’m trying to find out their buffer compositions for comparison.
Next steps are a version of my manual protocol done more like the robot and testing T1 beads in parallel. I’ll let you know how it goes.
@sgoldman101 , I’m familiar with all the beads you mentioned. I think you and others may be on the right track that it could be a function of your re-suspension volume. Try 200 uL of resuspension buffer instead of 100 uL.
Something else I just noticed. You mention you shake at 2000 RPM on a Teleshake shaker? I’m surprised you can hit 2000 RPM given your plate and volume. Does the plate fit in the shaker firmly? Are you sure it’s giving a good shake, and that the solution has a good vortex?
Yes, I’m wondering if volume could be a factor too and that’s definitely worth testing. I am indeed vortexing at 2000 rpm but not with my full 100uL wash. I dispense 30uL to the beads, vortex, then dispense and pipet mix the remaining 70uL. 30uL at 2000rpm doesn’t overtop my wells which I’ve checked by cutting a window in one edge of the plate skirt and using a free strobe tachometer app for my phone to see the liquid level during shaking.
Wow! It sounds like you’ve vetted this process. So you mix at a very low volume at high speeds and then pipette mix? Personally I think there is a lot going on there. But I also just realized the Twintec plate your using holds 250 uL max, so I see why you went in that direction.
Let us know how larger volumes work, and good luck!
Sorry not to have been better about updating this but I didn’t have much new info.
I had a chance to perform some additional testing on my bead issue, though I haven’t yet tested the scaling up the wash volumes or putting the manual protocol into strips or plates. Both are good ideas and are definitely on my list to try. What I have done is:
Washing the beads in a plate using the 96S deck magnet but with a manual pipet. This preserves bead quality and suggests the issue is not the magnet per se.
I tried handling the beads more gently: no vortexing and slower pipetting. Surprisingly, this reduced the clumping. It’s still not monodisperse like the stock solution but at least the clumps are smaller than my previous methods. No idea why it helps though.
Agilent uses Dynabead MyOne T1 beads in one of their NGS protocols with no apparent problems (albeit at 50C which I can’t do). I use the Dynabead MyOne C1 which differs in surface chemistry (T1 is hydrophobic, C1 hydrophilic). I’ve got some T1s on order to compare. I’ll let you know how it goes.