0:02
Hi, everyone.
0:02
Thanks for coming.
0:03
I know I am your last stop before lunch, so we'll make it quick and let's see if I can make this work.
0:13
So just going to do a really brief introduction, then I'm going to talk about our FFPE tissue dissociation kit and then I'll spend a couple minutes at the end on our spatial biology solution for both protein and RNA analysis on the same slide, of course, again on FFPE tissue.
0:31
So a quick introduction to Miltenyi Biotec.
0:34
So if you guys are not familiar with the company, we do have a pretty big portfolio diversity.
0:40
So what this crowd is probably familiar with is our research portfolio.
0:45
So that ranges from cell separation to our imaging portfolio.
0:50
We also have a really large clinical and GMP portfolio that's employed in cell manufacturing, like in the manufacturing of CAR T cells.
1:00
But hopefully you also know us from having the best swag.
1:04
So if you haven't stopped by the Miltenyi booth yet, we are giving out spatial biology puzzles and our Happy Cell plush toys.
1:11
So if you have any questions after the talk or just haven't gotten your swag yet, definitely stop by.
1:17
OK, so today, as I mentioned, I'm just going to be focusing on our sample preparation and our imaging products.
1:27
So let's talk about FFPE.
1:28
So I don't need to tell this crowd the value of FFPE.
1:32
So of course, there's like massive FFPE archives that can be used for retrospective analysis and it can also be used for actual patient outcomes since it's very commonly available for solid tumours.
1:44
However, there's a lot of challenges with FFPE.
1:47
While, it's really great at maintaining tissue morphology and protein biomarkers.
1:52
It just wreaks havoc on the RNA and DNA.
1:56
So while there have been a lot of advances earlier for bulk RNA and DNA sequencing of FFPE tissue, single cell sequencing had been lagging behind.
2:07
So initially it was more just RNA sequencing, but in the past couple years we've actually also gotten around to the, I think I said RNA already, but there was DNA sequencing then we've gotten around to the RNA sequencing anyway.
2:22
So if I did a really quick PubMed search of just single cell sequencing plus FFPE and as you can see not many publications up until 2023 where you see a big jump.
2:37
So what happened, well there was a big development in 2022.
2:40
So the biggest one is that the 10X genomics released their gene expression flex assay, which works on fixed RNA.
2:49
But we don't want to give them all the credit.
2:52
There is a lot of academic development too.
2:54
So that same year another lab released a paper on doing FFPE was just common on Poly A tail single cell sequencing chemistry.
3:04
And while the data is sparse when you use that Poly A tail, they did use some clever bioinformatic tools to make it useful.
3:14
And then the year after that, there was another publication where they basically used DNA blocking and random primers to get some really good results out of FFPE tissue.
3:27
So one thing I want to point out is all of these different publications up here use different methods to prepare their FFPE tissue, so to get it into a single nuclei solution.
3:39
So we also have our own method.
3:42
So we've came out with our first FFPE tissue dissociation kit in 2018.
3:48
So this was well before, well, single cell sequencing was around, but it was still really new at the time and way before anyone was even trying to do single cell RNA seq on FFPE tissue.
4:01
So what we actually released this kit for bulk DNA sequencing, which I'll get into a little more detail on the next slide.
4:10
However, what happened when 10X first released their protocol for FFPE analysis on their assay, they did use the Miltenyi kit.
4:21
The good news is it worked, it was compatible.
4:24
The bad news is we actually did not make that kit for RNA.
4:28
It was not optimised for RNA and it turns out there are better methods.
4:33
Of course, we want to make our own kit and we want to make it the best kit with the best specs.
4:38
So that's what we did.
4:40
Last year we released our FFPE kit for RNA profiling.
4:46
So just a little more detail about the difference between the two kits, because the first kit is actually really cool, a little underappreciated.
4:54
Basically what this kit is made for is the idea is to dissociate your FFPE tissue but maintain the cytoskeleton around the nuclei.
5:12
So we actually call this cells instead of nuclei coming out of the FFPE tissue.
5:16
And the reason why we wanted to maintain that cytoskeleton is because you can then separate your carcinoma cells from the surrounding stroma based on that cytoskeleton.
5:26
So cytokeratin versus vimentin.
5:28
And when you do that cell separation and you can do bulk DNA sequencing on just the carcinoma cells, you can really improve your detection of those cancer associated allele variants.
5:42
However, with the RNA technology, people prefer the nuclei and also looking forward for when we start to get into the technologies like attack and DNA sequencing, we also really just want the nuclei.
5:54
So that was our goal for this kit.
5:57
The other big difference here, so we use different enzymes, RNase free enzymes, but the other big difference is we skip the antigen retrieval step.
6:07
So the antigen retrieval step is not necessary for the 10X probe base chemistry.
6:12
And of course we want to avoid degrading the RNA if we can or any more than it already is.
6:20
And so with this kit, we do get just the pure nuclei.
6:25
So to give you a little bit more detail of the workflow, it's really simple.
6:29
You have your FFPE curls, it can be 25 to 50 Micron thick is what we recommend.
6:36
You just put it on our general Max instrument with the enzymes in the kit.
6:40
It's about a 40 minute protocol on the instrument hands free.
6:45
Then you can filter it and do whatever you want with those nuclei.
6:50
So most people are using it for the flex assay.
6:54
So if we just before we go into the sequencing, I didn't just want to show what the nuclei look like, they're beautiful.
7:02
So this is tonsil tissue 10X magnification with PI overlaid on the bright field.
7:09
And then if we zoom in, you can see that the nuclei have really smooth well defined borders, which is what you want to see for good quality nuclei.
7:19
We didn't stop just there to demonstrate that we are getting nuclei and not the rest of the cell.
7:27
So we did do an experiment where we took that same tonsil tissue and we either dissociated it, then did a bulk RNA extraction or we left the scroll intact and then again did a bulk RNA extraction and we did three prime RNA sequencing.
7:45
So first of all what we discovered is we wanted to show that we're not degrading the RNA.
7:51
So you can see our we have 85% uniquely mapped reads, so really good and there is no difference between the intact versus dissociated RNA.
8:02
The other thing we saw was that we were enriching for introns, which is exactly what you would expect if you're isolating the nuclei and removing the cytoplasmic elements.
8:15
So of course, we did take this into sequencing.
8:19
For sequencing, we did a different comparison.
8:22
So the method that 10X ended up going with after the initial use of our FFPE kit was a dissociation using Liberase.
8:32
So that was our benchmark.
8:35
So we wanted to make sure whatever we produce is going to outperform Liberase.
8:38
So here we did two different tonsil tissues or FFPE tonsil samples and we either dissociated with our Miltenyi FFPE kit or with the Liberase.
8:50
And first, I just want to point out that yeah, all the data looked beautiful, high-quality data.
8:56
And both whether you're using Librease or Miltenyi, you are able to isolate all the expected cell types.
9:04
So there's no bias in that way.
9:07
Where we did see the difference was with our QC metrics and just the overall amount of data you get from your experiment.
9:17
So if you look at the QC metrics, the most stark one is probably tonsil 2.
9:24
We loaded the same number of cells and got almost double the number of cells from the using the Miltenyi dissociation.
9:34
So we start, we have here we got 5,000 cells and here we got 10,000.
9:38
And for tonsil one we got about 5,000 versus about 8,000.
9:44
And I mixed up those numbers, but you can read them.
9:50
So the reason why we got more cells, it's because you can see the knee here on the barcode read plot is much more defined.
9:56
So that means that you are filtering out fewer cells.
10:01
The other thing you'll notice is we have UMI counts per cell.
10:05
We're also significantly increased as well as the median genes per cell.
10:11
Now someone who is really used to looking at single cell QC metrics will also go, hey, wait a minute, your mean reads per cell is also a lot higher in the Miltenyi sample versus the Liberase sample.
10:25
That was not intentional.
10:26
And at this point in time, I want to point out we did not perform this experiment in the house.
10:32
This experiment was outsourced to the Cedars Sinai's genomics core and they did everything and they loaded all the cells equally just like they did every other experiment because this was a Multiplex experiment too.
10:46
It wasn't just our samples in the run and this is just the way it came out.
10:55
We have more data where we saw the same effects that we generated in house where we just happened to get more read with our data.
11:03
But overall, my hypothesis is that for whatever reason, we are in fact recovering more library complexity.
11:12
And so the reads just reflect the fact that we have more RNA there to analyse.
11:20
So moving on, I was also a little concerned that you guys would look at the data and be like, oh, you just floated the libraries on balance.
11:33
So I did have them do the analysis where you extrapolate the mean genes per cell against the mean reads per cell.
11:41
And you can see we still have much higher library complexity.
11:46
The other thing we have is a much lower percent of mitochondrial reads.
11:50
So you definitely don't want a lot of mitochondrial reads in your RNA.
11:54
It's junk and an indicator of poor quality.
11:57
So we have that lower.
12:00
And then so to move on, I did want to talk about one of the common questions we get asked is can I instead of using curls, use sections on a slide?
12:13
So we wanted to test this out.
12:16
Is our kit basically efficient enough that we could actually use sections on a slide rather than the FFPE scrolls?
12:26
So what we found out is yes, it works, but it's not really optimised yet.
12:34
So we're getting there.
12:36
So what we did, we took the same tissue, this was tumour tissue and we either did 225 Micron FFPE scrolls or 25 Micron FFPE sections on a slide.
12:49
And even though we end up using double the amount of tissue in the slides compared to the scrolls, you can see we got fewer cells recovered with the slide method.
13:01
So this is pretty intuitive.
13:03
If you're doing thin sections, you're literally ending cutting the cells in half and fractionating the cells and the nuclei.
13:11
So we are.
13:13
But even that being said, we are working on ways to make this even more efficient so you can still get some usable data out of those sections.
13:21
Of course, the other reason why we want to do the slide method is becauseif you want to get not only single cell sequencing, you want to get spatial analysis out of the same exact tissue section.
13:34
I can say we're not quite there yet, but we're moving in the direction.
13:37
So in theory, you could go and do some spatial analysis of your tissue as long as it doesn't destroy the tissue and then take it into single cell sequencing.
13:49
So for this, I wanted to introduce our MACSima solution.
13:55
So the MACSima is instrument for high content protein imaging as well as targeted RNA profiling.
14:04
The way that it works is that it's a cyclic, so pretty intuitive.
14:09
You stain.
14:12
So for protein with antibodies, you image, you erase, you repeat and pretty much the same thing for probes.
14:19
For probes though, we do have probe the probe based chemistry.
14:24
So this is non decoding.
14:26
We just do the probe hybridization, ligation and amplification off the instrument to amplify our signal and then we'll put it on the instrument to actually do the imaging.
14:39
Once again, just labelling 1 probe at a time.
14:42
So it's 11 colour to 1 probe act during one imaging session.
14:49
The other thing that's kind of cool is I'm not going to go into this, but they did spend time on the sample preparation to optimise getting both RNA and protein out of the same section.
15:06
So as you saw before that antigen retrieval step really important for getting your antibodies to bind the protein, not so great for RNA.
15:17
So it's definitely a balance.
15:19
So we do have an optimised protocol for this on our website.
15:24
So just to show you some of the example of the tonsil tissue that we analysed both with protein and RNA.
15:33
So what you can see here is just beautiful results, nice clear segmentation.
15:40
So the yellow here is dividing cells, dividing B cells in the germinal centre.
15:45
And if we zoom in on this, you can see I have Ki 67 in yellow punctate and MK167 as proliferation marker as yellow punctate.
16:00
And you can see that they overlap and correlate really well.
16:04
So one of the things that we are working on was the status set.
16:08
So we already had the MACSima data and the sequencing data is newer.
16:14
So we haven't had enough time to really dig into it and make a correlation.
16:19
However, what we did do actually last year already is we used publicly available tonsil data for both single cell sequencing as well as bulk RNA sequencing and correlated it to our MACSima RNAsky data.
16:34
You can see even though these are totally different samples processed in different labs that we're seeing really great correlation.
16:43
So we're still going to finish this up and we're hoping to see even better correlation when we do it with samples from the same tissue, right.
16:54
So that's it, short and sweet.
16:57
So just to summarise everything, we've talked about garbage in garbage out a couple times I've heard at this conference.
17:06
So FFPE is no different.
17:08
Sample preparation method has a big impact on your data quality.
17:13
We have a new FFPE kit for RNA profiling that outperforms Liberase and we have our MACSima imaging platform where we saw tight correlation between our RNA sky probes and RNA sequencing data.
17:28
All right, questions.
17:36
Thank you, Carina.
