The Needle Issue #26
|
Welcome to The Needle, a newsletter from Haystack Science to help you navigate translational research and the world of preclinical biotech startups from around the world. The news is in: April was the biggest month for new biotech initial public offering (IPO) filings in five years, with investors continuing to reward later-stage companies with assets in clinical development (all in all, six biopharma IPOs have raised $1.8 billion so far this year). Trade sales also continue to be healthy, surpassing total proceeds for all of 2025. By the start of May, biotech deals reached $84 billion; that is almost twice the $44.4 billion in deals a year earlier. But the year’s banner dealmaker prize must go to Gilead, which has already splashed out on three biotech enterprises. In February, it swallowed publicly traded chimeric antigen receptor (CAR)-T company Arcellx; a month later, it moved for T-cell engager-focused Ouro Medicines ($1.68 billion upfront); and, in recent weeks, it announced the marquee deal of the year so far, snapping up antibody-drug conjugate (ADC) developer Tubulis for $3.15 billion up front (and another up to $1.85 billion in contingent “biobucks”). One other preclinical deal really took our eye: LLM maker Anthropic’s $400 million acquisition of eight-months old Correlation Bio. The latter, one of numerous plays combining AI-assisted de novo protein design, foundational models of biology, and a high-throughput experimental validation backend was founded in New York City by former-Genentech computational scientists Samuel Stanton and Nathan Frey in September 2025. With closing of the deal, Correlation’s investor, Dimension, received a return of ~$200 million+ on a single-digit million-dollar investment. Not a bad return for a 24-week investment! All this dealmaking means money is moving and being recycled, which is good news for preclinical biotechs early in their gestation. The bad news for CEOs of preclinical startups is that most of the venture world continues its aversion to seed financing deals. But this issue’s bumper round of preclinical financings shows deals by biotech VC specialists are still getting done; no surprise perhaps that the tech-bio diet remains flavor of the month, with machine learning an essential ingredient in no fewer than 12 of the 31 funding announcements we uncovered (Unnatural, PerturbAI, Ternary, Clockwork Bio, Pinnacle, Ambrosia, Scala, Syneron, TippingPoint, Neomorph and DeepCyte and Helical). Among the companies active in translational research, several preclinical startups presented their preclinical programs at last week’s AACR in San Diego. As usual, anything we missed in the biotech startup world, let us know (info@haystacksci.com).
Haystack chat An old adage in drug development states that any successful program for an advanced medicine must overcome three central challenges: first, delivery; second, delivery, and third … delivery! Lipid nanoparticle (LNP) technology and N-acetyl galactosamine-(GalNAc) conjugates have opened the liver to a wide range of genetic medicines, and transferrin 1 receptor (TfR1) conjugates are beginning to access the CNS via intravenous delivery with brain-shuttle technology. But tissues like the lung, kidney, muscle and heart remain very much a work in progress. In the pulmonary space, a small cadre of companies are pursuing inhaled LNP delivery technologies. Recode Therapeutics, Vertex Pharmaceuticals and Arcturus are the main players, while other firms such as 4DMT and Krystal Biotech are focusing on viral gene therapies for lung delivery. Just a few days ago, one of these LNP programs got the chop. The Vertex/Moderna phase 1/2 study of VX-522, an aerosolized LNP to deliver mRNA encoding full-length cystic fibrosis transmembrane conductance regulator (CFTR) to the lungs of cystic fibrosis patients, which had been paused due to tolerability issues, is now permanently discontinued. According to reports, the Moderna LNP was the culprit, leading to lung inflammation. That leaves Recode and Arcturus as the frontrunners, a rather small field, given the entire market opportunity for a pulmonary delivery solution. All told, in 2023, there were 569.2 million cases of chronic respiratory diseases and 4.2 million deaths from respiratory disease. Recode now is enrolling patients into the phase 2 trial of its Selective Organ Targeting (SORT), LNP platform (RCT2100) that delivers an mRNA encoding CFTR in combination with the small-molecule CFTR potentiator ivacaftor (the SORT technology was originally licensed out of Daniel Siegwart’s group at UT Southwestern). The other LNP platform, Arcturus’ LUNAR LNP technology, also has encouraging interim data from its phase 2 trial in cystic fibrosis patients and from its program delivering ornithine transcarbamylase mRNA. These LNPs (and most other LNP delivery platforms) are built around the same four common components: an amino ionizable lipid, a helper lipid, a polyethylene glycol lipid and cholesterol. The formulations follow this scheme but with different combinations of proprietary lipid forms; thus, in Arcturus’ LUNAR LNP, distearoylphosphatidylcholine (DSPC) performs the helper lipid function, whereas in Recode’s SORT LNP, it is 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE). Overall, however, just a handful of novel lipid components have gone into humans so far. According to Siegwart, the field is in dire need of developing a broader palette of cationic lipids that are both efficient and non-toxic for the pulmonary epithelium; ultimately, the goal would be a delivery technology capable of targeting specific cell types in the lung (with many new cell subtypes continuing to be identified). In a recent article in Nature Biomedical Engineering, Siegwart and his group at UT Southwestern introduce the design and evaluation of a new class of lung-targeting (LuT) lipids that enable the highly efficient and selective delivery of mRNA and CRISPR–Cas9 gene-editing systems to the lungs. They synthesized and screened a library of 444 lipids using a combinatorial approach, systematically varying amine head groups and hydrophobic tails. Through in vivo testing and structure–activity relationship analysis, they identified key features in the lipids that most effectively targeted the lung: a distinctive ‘tripod-like’ structure, consisting of a quaternary amine head, three long alkyl chains and a short fourth chain. Compared to benchmark formulations, the best-performing LuT-containing LNPs achieved up to a 25.5-fold increase in mRNA delivery and a 9.2-fold improvement in gene-editing efficiency, with >90% of delivery localized to the lungs. These LuT-LNPs successfully transfected multiple lung cell types, including endothelial, epithelial and immune cells, with some formulations showing preferences for specific cell populations. Mechanistically, the improved performance was attributable to two main factors. First, the tripod-like structure of lipids promoted endosomal escape by facilitating membrane fusion and LNP disassembly, allowing efficient release of genetic cargo into cells. Second, LuT LNPs formed distinct protein coronas in the bloodstream, particularly enriching for vitronectin, a protein that enhances targeting to lung cells via receptor-mediated uptake. Siegwart and his team went on to show the therapeutic potential of LuT LNPs. The lead formulation, 1A7B13, enabled effective delivery of IL-10 mRNA in a mouse model of acute lung injury and achieved robust CRISPR–Cas9 gene editing in lung tissue. The LNPs showed minimal toxicity and no significant adverse effects in vivo. This research establishes clear design principles for lung-targeting LNPs and markedly expands the available toolkit for pulmonary gene delivery. It is just the beginning of the translational path, however. The Siegwart LuT-LNPs must home through the vasculature to the lungs after being delivered intravenously. This is very different from the aerosolized LNP delivery approaches of Recode and Arcturus currently in clinical testing. There may be a case to be made that some pulmonary vascular disease, lung endothelial targets, lung fibrosis, immune-cell or vascular-compartment targets might warrant the intravenous route, but aerosolized LNP delivery provides lower systemic exposure (and thus higher therapeutic index), is more patient-friendly, and rapidly/directly reaches the airway lumen. Regardless of the route of administration, the translational challenges associated with targeting the lung remain very difficult. In terms of testing formulations in different models, anatomical differences between mouse, ferret and human airways, including physiological size and branching complexity, impact LNP design and aerosol physics.The formulations used for mice may simply not work for people because of differences in cell composition, and lung epithelial and endothelial membranes and “surfaceomes”. As humans age and develop disease, cell protein and lipid composition may also change, requiring further optimization of LNP formulations. Mice have more narrow airways and faster breathing rates than humans, requiring smaller diameter aerosol droplets (often <2 µm) to ensure particles bypass the upper respiratory tract and reach the alveolar regions. Moreover, humans have ~23 branches in their airways, whereas mice have only 13, meaning an aerosol optimized for a ‘deep’ reach in a mouse might only reach mid-level bronchi in a human. Furthermore, ferrets are not a widely available model system to study the biodistribution and efficacy of LNPs. Indeed, there are just a few labs in the United States that upkeep ferret colonies. Last, a human lung's surface area (~70 m²) is nearly 8.500 times larger than a mouse's (~82 cm²), and human tidal volume is roughly 6,000 times greater. This requires significant dose scaling and affects how ‘diluted’ the LNPs become once they deposit. Designing in vitro and in vivo systems representative of human biology and capable of predicting LNP biodistribution is also a tall order (especially with such a small cadre of companies working on the problem). For small molecules, the measurement of efficacy in human basal epithelium-derived patient cells carrying a mutation of interest by and large will translate into what you see in the clinic. The pharmaceutical industry has amassed a lot of data to bolster pharmacology. Unfortunately, that correlation doesn’t necessarily hold for genetic modalities like mRNA or CRISPR/Cas9 constructs. For these medicines, it is very hard to figure out PK/PD. And so, the translation from preclinical work to the clinic can be tricky for an inhaled LNP technology delivering mRNA. It is difficult to really know the degree of protein expression from an inhaled LNP genetic medicine intracellularly without doing a bronchial biopsy (which is of course highly intrusive). And if you need to test your LNP in patients via biopsy, clinicians historically have been very resistant to carrying out such procedures, particularly in very sick patients like some of people with cystic fibrosis who carry nonsense mutations in CFTR. Thus, there is a need for alternative approaches. Certainly, there is an opportunity for more work on organoids or simpler patient cell-derived assays: 2D or 3D alternatives to large animal models like the ferret. What is clear is that there are enough patients worldwide living with lung disease that further research in this area needs to be encouraged. In this respect, the findings from Siegwart’s group are a step in the right direction, with broad implications for treating lung diseases by enabling safer and more precise delivery of RNA-based therapeutics and genome-editing technologies. Translational papers: Best of the rest We came across several intriguing small molecules in the preclinical space over the past few weeks: Also, several papers on biologics and other advanced therapeutics caught our eye: Last, in the genome-editing world, new technologies reported for kilobase-scale integration and template-independent editing: Last, several startups presented data at the American Association for Cancer Research annual meeting in San Diego, CA this April: Preclinical financings (March 16 to March 24) Preclinical financings (March 26 to April 5) Preclinical financings (April 6 to April 14) Preclinical deals (March 24 to April 1) Preclinical deals (April 6 to April 15) Stay in touch We hope you enjoyed this issue of The Needle and hit the button below to receive forthcoming issues into your inbox
If you’re interested in commercializing your science, get in touch. We can help you figure out the next steps for your startup’s translational research program and connect you with the right investor. Follow us on X, BlueSky and LinkedIn. Please send feedback; we’d love to hear from you (info@haystacksci.com). Until next week, Juan Carlos and Andy |