0:00Welcome to Base by Base, the paper cast that brings genomics to you wherever you are. Thanks for listening and don't forget to follow and rate us in your podcast app. So what if we told you that the key to unlocking new treatments for, say, aggressive cancers or really stubborn chronic infections lies hidden, not in some future discovery, but in dusty archives, sometimes, sometimes a decade old or more?
0:26Sounds a bit like science fiction, doesn't it? Retrieving secrets from the past, but this is a very real challenge facing clinical labs all over the world. For years and years, patient tissue samples have been preserved using this one method.
0:38It's called formal and fixed paraffin embedded, FFPE for short. In these FFPE blocks, they're kind of the backbone of our medical history. They're stable, they're cheap to store, and when you put them all together, I mean, they represent this massive, invaluable library.
0:51A global biobank, really. It holds the molecular story of how diseases progress. We just we desperately need a way to tap into that history. But there's a catch. A huge one. A huge one. The problem is that the very chemical process that preserves the tissue so beautifully, that formal fixation.
1:08It also just, um, brutally damages the nucleic acids inside, especially the RNA. It just shreds. It causes severe degradation, severe fragmentation. Yeah, it shreds it. And when your RNA is in pieces, getting high quality spatial data, and by that, we mean mapping every single molecule back to its exact spot in that tissue, that's been, well, basically impossible.
1:31The old methods just couldn't handle it. Not with that level of degradation, and certainly not if you wanted to see the whole transcript. Exactly. So the challenge has always been, how do you capture that complete molecular portrait, including all those critical non-coding RNAs, and math it at a, you know, near single cell level?
1:48And that's what this deep dive is all about, a technology that finally cracked it. And today we're celebrating the work of a really large collaborative team. We're talking about researchers from BGI research, the Chinese Academy of Medical Sciences, and Peaking Union Medical College, and Tungji University.
2:03They've really pushed our understanding of what's possible with these clinically archived samples. Okay, so let's unpack this FFPE hurdle a bit more because it really is the fundamental problem they had to solve.
2:14Why is Formalin so brutal on RNA? What's it actually doing? Well, the key ingredient is formaldehyde. It forms these cross links. So it's basically binding the nucleic acids to proteins and other molecules.
2:28Okay, so it's tangling everything up. It's tangling everything up, and at the same time, it's causing the RNA strands to break, to cleave. Imagine a fragile string of lights. Formal and not only tangles the string up, but it also snips it into tiny, useless little fragments.
2:42So you've got fragmented, tangled RNA, and when researchers try to use their existing spatial transcriptomics tool kits on this. It was a non-starter. Why? Because all those existing ST techniques, they were built for fresh frozen samples, you know, samples with beautiful, high quality intact RNA.
3:01The ideal scenario. The ideal scenario. When they tried to adapt them for FFPE, they just, they hit a wall. And what were the main designs that just couldn't make the leap? There were 2 main groups. The 1st were the FEHH-based techniques.
3:14These use predesign probes to find specific genes. Right. you're only looking for what you already know to look for. Exactly. The scope is just too small. You might capture a 200, maybe a couple of 1000 genes, but you're not getting the whole transcriptome.
3:29You're not getting the whole story. You need the whole picture, not just a handful of suspects. Precisely. And then the 2nd group, the sequencing based techniques, they had a different problem, a flawed capture strategy.
3:39How so? They almost all use what's called poly-a capture. So they use these little poly T primers that are designed to grab onto the polyate tail at the 3 prime end of a messenger RNA. And if the RNA is already in tiny pieces, that fragile little tail might be completely gone.
3:55Or, the rest of the message is gone. So even if you grab the tail, you got this tiny, useless fragment of data. It's like finding the last page of a book with no other pages. So that reliance on the 3 prime end meant the results were always biased, always incomplete.
4:09Always, mapping the whole transcriptum on FFPE sections with high resolution was, yeah, it was the Achilles heel of the field until now. And this is where Stereo Sec V2 comes in. So if the problem is chasing that fragile 3 prime tail, The solution must be to stop chasing it.
4:27How did they do that? That's the core innovation. V2 moves completely away from those poly T primers. Instead, it uses what are called random primers. specifically 6N random primers. So instead of trying to catch the tail with a hook, they're just casting a huge net designed to grab any piece of the RNA string.
4:45That is a perfect analogy, the random priming strategy is, well, it's revolutionary for this problem because it captures everything without bias. It doesn't care if a fragment is from the 5 prime end, the middle, or the 3 prime end.
4:56It just grabs what's there. And the result of grabbing any fragment is you get everything. You get uniform coverage across the entire gene body. This is so important because it just eliminates that 3 prime bias.
5:07And on top of that, it's way more efficient at profiling all the non-polyodonilated stuff. Like those really important non-coding RNAs. Okay, so a huge sequencing innovation solves the chemical degradation problem.
5:21But I imagine the whole workflow had to be incredibly gentle with these old, delicate samples. Oh, absolutely. The whole process was built for FFPE. It involves really critical steps like, um, deep paraffinization, rehydration, and then a very specialized chemical decross-linking step to untangle that RNA.
5:39And somehow, through all of that, they still maintain the incredible spatial resolution that Stereosec is famous for. Remind us how they do that. Right. They inherited that from V1. It's the use of DNA nana ball arrays, DNBs.
5:51And what's just mind blowing is the spacing. The distance between each of these D&Bs is only 500 nanometers. 500 animator. Okay, for you listening, put that in perspective. How small is that compared to a cell?
6:02Well, a typical mammalian cell can be anywhere from 10 to 100,000 centimeters wide. So 500 nanimeters. You're getting theoretical subcellular resolution. You're seeing context inside the cell. Wow. So you get this huge field of view and incredible resolution at the same time, which is just, It's a holy real, really.
6:22That combination is what unlocks these archives. Okay, so let's get to the proof. How did V2 actually perform when they threw some really terrible samples at it? This is where it gets really impressive.
6:32They tested on 10 triple negative breast cancer FFPE blocks. Now, some of these samples were up to 9 years old. And the RNA quality, measured by a metric called DV200, was as low as 18. Okay, wait, a BV 200 of 18 for a researcher, what does that mean?
6:49Is that basically unusable, just garbage RNA? That's exactly right. A good sample might have a DB 200 of 70 or 80? A score of 18 means less than a 5th of your RNA fragments or even 200 nucleotides long.
7:00It is scientifically speaking trash. And V2 worked on the trash samples. It showed incredible tolerance. The gene capture efficiency was consistent. Even those nine-year-old super low quality samples. I mean, that finding alone proves his potential for tapping into these massive global biobanks.
7:17And because of the random priming, they got that total RNA mapping. How did it stack up against the older pro-based methods in terms of gene detectors? It wasn't even close. They detected significantly more genes.
7:28They found over 23,000 James that were completely missed by a popular pro-based technique. 23,000. And a lot of those were the non-coding RNAs. Huge number of them were. We're increasingly realizing that these non-coding RNAs are central to, you know, how diseases work.
7:44Losing them is losing half the story. For example, V2 successfully mapped things like small nuclear or RNA, Snorinays, which are classic non-polly ARNAs. It's definitive proof that V2 profiles the total RNA.
7:58Let's talk about the clinical power. Unraveling cancer heterogeneity is so critical. In those TNBC samples, how did this new spatial map change their understanding of the tumor? Well, by analyzing the spatial transcriptum, they could infer something called copy number alterations, or CNAs.
8:14And this analysis actually spatially divided one of the tumors into two distinct subtypes, right there on the slide. It drew molecular boundary, and what did that boundary tell them? This was fascinating.
8:26One of those subtypes, they called it subtype two, it had this big amplification around the ESR one gene locus. The estrogen receptor gene. Exactly. A huge therapeutic target in other types of breast cancer.
8:38So finding this specific alteration, spatially located within a tumor that's defined by not having that receptor. It just opens up new ideas for really targeted localized treatment. And they went deeper than CNAs, didn't they?
8:51They could actually see changes in the gene structure itself. That's right. V2 could spatially identify alternative splicing events. They found a significant skipped Exxon event in the GIPC one gene, a known breast cancer biomarker.
9:05So it's not just about how much RNA is there, but how that RNA is built and where that's happening. So we've seen it unlock cancer secrets. The paper then takes this really cool turn into chronic infection.
9:17How does this power apply to something as complex as tuberculosis? Yeah, this application was brilliant. They use V2 on a mouse lung model infected with mycobacterium, tuberculosis, or MTB, and they watched it over an 8 week period.
9:31This is where that simultaneous host pathogen mapping really shines, right? Perfect for it. V2 successfully tracked the gene expression of both the host, so the mouse and the pathogen, the MTV bacteria, simultaneously, on the same piece of tissue.
9:46So you can watch the battle unfold spatially. You can. They saw the bacterial RNA peak at 4 weeks and then decrease by 8 weeks. And when they looked at the host's immune system at that 8 week mark, they saw a strong enrichment in the adaptive immune response right where the infection was being controlled.
10:02But the real kicker was the immune repertoires, the B cells. How did they manage to assemble those? The key information for that? Is it the 5 prime end, which is usually the 1st thing to go into graded samples?
10:12It all comes back to the random priming. That unbiased coverage means it captures enough fragments from the 5 Primand to let them assemble B cell receptor clones. BCRs. This is the part that is just so visually and scientifically powerful.
10:26What did they actually see these BCR clones doing in the mouse lung? They could see the clones spreading in and around the MTB infection sites, but then they found this incredible spatial gradient, a pattern.
10:40What was the pattern? The closer the BCL clones were to the infected areas? The higher their clone diversity was, and the higher their mutation frequency. Wow. So that's that's B cell affinity maturation, the process where B cells get better and better at recognizing the enemy, they're seeing it happen in real time.
10:57Not just in real time, but in situ. They visualized a complex dynamic immune process unfolding on a static archive slide. That completely changes how we can read these archives. And they confirm this in humans.
11:09They did. They took surgically respected human lung samples from 3 different tuberculosis patients. And they found a similar pattern. Even better. V2 found 16 recurrent BCR clones that appeared across all 3 patients.
11:21That's a huge deal. It strongly suggests these shared clones are MTV specific. which makes them incredibly valuable targets for designing future immunotherapies. So looking at the big picture, stereo sick V2 just blows the doors open on what's possible.
11:37We're not just limited to pristine fresh frozen tissue anymore. Exactly. The ability to robustly use these long-term stored FFP samples could just dramatically accelerate research. Think about rare diseases where all you have is historical cohort data.
11:51It unlocks global vital banks that we thought were you know, molecularly useless. And for personalized medicine, being able to see tumor heterogeneity, alternative splicing, the immune response, all at this resolution.
12:03It's a huge step forward for diagnostics and guiding treatment. It is. Now, this does raise an important question about limitations. The methodology is incredibly powerful. But while it covers the whole gene body, the individual reads are still relatively short.
12:17So if you're hunting for one specific point mutation, a single nucleotide variant, V2 might not be the best tool for that one job, why is that? Right. Because you're spreading your sequencing power so wide with random priming, you don't get super deep coverage at anyone's specific spot.
12:33So for that kind of high confidence variant analysis, the team still recommends you complement this with long read sequencing. And any caveats on the pathogen profiling, the MTB work? Just that it requires really rigorous quality control.
12:46These are old slides from non-sterile workflows, so you have to be careful about potential environmental contamination. Okay, so let's bring this deep dive home. The central insight here is that stereo sick V2's use of random priming effectively captures total, even heavily degraded RNA from these archived FFPE samples.
13:04And that one change delivered this unprecedented whole transcripts and coverage at subcellular resolution. It allows researchers to map gene expression, track non-coding RNAs, and spatially analyzed dynamic immune responses like those maturing BCR clones, all within these invaluable clinical archives.
13:21So what does this all mean for you? If this technology can uncover MTB specific immune responses in nine-year-old tissue blocks, What other ancient molecular secrets are just sitting there waiting to be revealed in the world's pathological archives? And how fast can we use them to find new treatments?
13:38This episode was based on an open access article under the CCBY 4.0 license? You can find a direct link to the paper and the license in our deep dive description. If you enjoyed this, follow or subscribe in your podcast app and leave a five-star rating.
13:53If you'd like to support our work, use the donation link in the description. Now stay with us for an original track created especially for this deep dive and inspired by the article you've just heard about.
14:02Thanks for listening and join us next time as we explore more science base by base.
