Tag Archives: beauty

The Needle Issue #16

7 Oct
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

Commercial interest in targeted epigenetic therapies — agents that target specific genes without altering bases in their sequence or causing double-strand breaks or even single nicks in the DNA — continues to grow, as underscored by the latest financing announced by Epigenic Therapies. The unique selectivity and specificity of targeted epigenetic therapeutics offers compelling advantages over small-molecule epigenetic drugs, which target a specific epigenetic reader, writer or eraser, but affect genes across the genome and affect many diverse tissues, leading to narrow therapeutic windows that make them difficult to develop for conditions outside of cancer.

Today, Haystack is aware of at least eight private companies (nChroma Bio (resulting from a merger of Chroma and Nvelop), Encoded Therapeutics, Epigenic Therapeutics, Epitor Therapeutics, Moonwalk Bio, Navega Therapeutics, Regel Therapeutics, and Tune Therapeutics), and two public companies (Modalis Therapeutics and Sangamo Therapeutics) that are pursuing the targeted epigenetic approach against disease (let us know if you know of any others). Another company, Flagship Pioneering’s Omega Therapeutics, went out of business in August after filing for bankruptcy in February. A smaller set of companies are also pursuing targeted epigenetic therapies against RNA modifications.

All of these therapies are designed around an alluring set of simple principles: take a gene-specific DNA-binding domain — zinc-finger proteins (ZFPs), ‘dead’ Cas9 (dCas9) with mutations in its RuvC and HNH endonuclease domains, or transcription activator-like effectors (TALEs) — and tether it via an amino acid linker to an enzymatic effector module. This effector is either an enzyme that directly places or removes a specific epigenetic modification (e.g., TET, histone demethylases or the histone acetyltransferase p300) or a transcriptional activator (e.g., VP16) or repressor (e.g., KRAB).

A particularly compelling application for such treatments is genetic disorders of haploinsufficiency (like Dravet’s) or imprinting disorders (like Angelman’s or Prader Willi). There are also many of these diseases where the therapeutic genes would be too large (>4.0 kb) for a traditional AAV gene-therapy approach; in contrast, epigenetic editing machinery can be packaged into an AAV vector.

Currently, the diseases being pursued by companies include hepatitis B, hypercholesterolemia, epilepsies (SCN1A (Dravet syndrome) and SCN2A), chronic pain, and muscular dystrophies. Those with the most advanced programs are Encoded’s AAV-9 intrathecally delivered SCN1A-targeting zinc finger protein linked to a VP16 activation domain in phase 1 testing for Dravet and Sangamo’s AAV- STAC-BBB-delivered SCN9A-targeting zinc finger protein linked to a KRAB repressor domainin a phase 1/2 trial for patients with chronic pain. In this context, two papers published in the past couple of weeks represent important proofs of the efficacy of targeted epigenetic therapies.

In a first paper published in Nature, the groups of Kevin Bender and Nadav Ahituv at UCSF (scientific co-founders of Regel Therapeutics) sought to test a targeted epigenetic therapy in patients with SCN2A mutations that exhibit decreased NaV1.2 function. These individuals have impaired action potentials, synaptic transmission and manifest diverse neurological symptoms and seizures, with few therapeutic options, beyond symptomatic anti-seizure medications that have a dizzying range of debilitating side effects.

The UCSF teams leveraged conditional genetic knock-in technolgoy or CRISPRa technology — an AAV-delivered SCN2A-promoter-targeting dCas9 fused to a VP16 activator domain — to upregulate transcription of the SCN2A gene. Using either approach, they were able to boost transcript levels from the healthy SCN2A allele, ameliorating electrophsiological deficits and chemical-induced seizure activity in Scn2a+/− mouse models. Importantly, these effects were seen in adolescent mice, which conventionally have been thought to be too old to respond to treatment. This suggests that rescue of normal dendritic excitability with epigenetic agents at later stages of life might be capable of restoring neuronal function, with implications for patients.

In a separate set of experiments, the authors showed that their epigenetic approach was able to rescue neurophysiological activity in haploinsufficient neuron-like cells from SCN2A-knockout human embryonic stem cells. This cross-species reproducibility provides further confidence that CRISPRa-mediated upregulation could be translated into human treatments.

​In a second paper in Nature Biotechnology, a team from Epigenic Therapeutics (Shanghai, China) describes the design and validation of optimized epigenetic regulators (EpiRegs) to silence genes in a precise, durable way without altering genomic DNA. Epigen’s Shaoshai Mao and his collaborators at the Chinese Academy of Sciences and the First Affiliated Hospital of Anhui Medical University tested combinations of TALE- and dCas9-based systems, systematically optimizing effector domains and fusion architectures, looking for effective regulators of gene expression. The best-performing variant, EpiReg-T (a TALE-based system, which eliminates the need for a guide RNA), achieved 98% silencing of target genes in mice, substantially outperforming dCas9-based versions.

Using lipid nanoparticles (LNPs) for delivery, a single administration of EpiReg-T in macaques induced long-term repression of the PCSK9 gene, which encodes a validated target for the treatment of hypercholesterolemia. EpiReg-T reduced PCSK9 expression by >90% and LDL-cholesterol by about 60%, with effects persisting for nearly a year (343 days).

Mechanistically, the team used whole-genome bisulfite sequencing and cleavage under targets and tagmentation (CUT&Tag) to show that EpiReg-T induced stable DNA methylation and repressive histone marks at the PCSK9 promoter. The silencing persisted even after liver regeneration and could be reversed by targeted epigenetic activation. Multiomic analysis in mice, macaques and human hepatocytes confirmed high specificity of the manipulation and minimal off-target effects. Overall, these finding, as well as similar results reported in April by Chroma Medicine, establish epigenetic editing as a promising therapeutic platform for durable and reversible gene silencing.

Overall, targeted epigenetic therapies offer clear safety advantages over small molecules that indiscriminately target all genes under the control of an epigenetic eraser or writer enzymes. They avoid the potential risks associated with creating single- or double-strand DNA breaks associated with CRISPR/Cas9 gene, base or prime editing therapies. And they avoid the insertional mutagenesis risks associated with traditional viral gene therapies. What’s more, in applications requiring gene upregulation in haploinsufficient disease, these approaches maintain the endogenous regulatory context of the functional allele. This is in stark contrast to traditional gene-therapy replacement approaches, where overexpression of an introduced therapeutic gene can often lead to toxicities and immunogenecity.

Of course, questions still linger around the persistence of the changes elicited by these epigenetic agents. Will they persist in patients for long periods — for years or even decades? If they can, then epigenetic therapy may offer compliance advantages over small molecules, antibodies, ASOs or even siRNAs, which have treatment durations of six months or less.

Like all genetic medicines, though, delivery remains the key headache. Thus far, AAV vectors, lipid nanoparticles or ribonucleoproteins (RNP) have all been explored to deliver epigenetic therapies (with some evidence that RNPs might have advantages because they can result in higher dCas9 dosages within target cells). For AAV vectors, the fact that targeted epigenetic therapy might only need to be given once might be an advantage in terms of immunogenicity/neutralization concerns against the vector.

A broader point is that the safety profile of targeted epigenetic editors may offer advantages if AAV vectors are used as delivery vehicles: if the epigenetic agents themselves can be delivered at high dosage (given their intrinsic favorable safety profile and presumed maximal tolerated dose), perhaps AAV vector dosages could be lower than current practice. With many current gene therapies requiring dosages of 1013 or more viral particles/kg in patients, it is increasingly becoming clear that unacceptable liver toxicities arise from the virus at these levels in clinical studies. It will be interesting to follow this space as more agents enter human testing.

Tags: , , , , , , , , , , , , , , , , , , , ,

Finalist in the RESI Boston Innovator’s Pitch Challenge – Meet M6P Therapeutics

30 Sep

In this interview, Caitlin Dolegowski speaks with Cuong Do, Founder and Chairman of M6P Therapeutics, about the company’s groundbreaking lysosomal targeting platform, its applications in rare disease and oncology, and the experience of pitching at RESI Boston.

Cuong Do
 CaitiCaitlin Dolegowski

Caitlin Dolegowski (CD): M6P Therapeutics has achieved what was long thought impossible, delivering proteins to lysosomes. Can you explain the significance of this breakthrough?

Cuong Do (DO): An enzyme called GlcNac-1-phosphotransferase (PTase) is responsible for adding mannose 6-phosphate to the surface of lysosomal enzymes. People have tried and failed for decades to increase the expression of M6P, and everybody gave up. Our co-founder Stuart Kornfeld never gave up. He and his post-doc were able to engineer a variant of PTase that turned out to be 20X more effective than PTase itself in adding M6P to lysosomal enzymes. We built upon this breakthrough to create a platform that is able to create enzyme replacement therapies that have very high M6P content. Furthermore, our gene therapies are the only ones that result in M6P-containing enzymes being produced by the transduced cells.

We expanded upon the innovation and created chimeric antibodies that contain M6P as well. This allows these antibodies (after they bind to the targeted antigens) to be brought to lysosomes in virtually all cells in our bodies for degradation. This is a significant advantage over traditional antibodies relying on Fc clearance by only select immune cells.

CD: You have multiple rare pediatric drug designations and two programs nearing the clinic. What are the most exciting upcoming milestones for your pipeline?

DO: We are preparing to start an Investigator Initiated Trial in Australia for our M021 ERT for Pompe Disease in hopes of obtaining early human data demonstrating M021’s superiority over the standard of care.

CD: How does your lysosomal targeting platform extend beyond rare diseases, particularly in oncology with your chimeric PD-L1 and PD-1 antibodies?

DO: We figured out a way to add M6P to any protein, including antibodies. Our chimeric antibodies can be cleared by virtually all cells in the body since virtually all cells have receptors for M6P. This is especially effective for clearing surface antigens from cell surfaces. Our chimeric PD-L1 antibody is able to clear virtually all PD-L1 from the surface of tumor cells and thus activate T-cells and drive T-cell mediated tumor killing. Our chimeric version of Keytruda is able to remove PD-1 from the surface of T-cells and has shown to be more effective in inhibiting tumor growth in vivo than Keytruda itself.

CD: Can you walk us through your IP position and how it supports your growth strategy?

DO: We have invested heavily in IP that has created a portfolio of 9 patent families, 9 issued patents, and ~20 still in prosecution.

CD: Where are you in your fundraising journey, and what types of investors or partners are you looking to engage with?

DO: We have raised ~$40 million in our Seed and A rounds, which we invested to get our programs to where they are today. We are trying to raise a $5 million bridge now in anticipation of a $50+ million Series B next year. In addition to investors, we want to engage with potential partners who might be interested in our molecules.

CD: How did participating in the Innovator’s Pitch Challenge at RESI Boston help advance your business development or investor connections?

DO: We met a few companies who might be interested in partnering on some of our molecules. We’re continuing the conversations.

IPC Applications are now open for the next Innovator’s Pitch Challenge at RESI London 2025 and RESI JPM 2026, with spots filled on a rolling basis.

Tags: , , , , , , , , , , , , , , , , , ,

The Needle Issue #15

23 Sep
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

On September 11, the Lasker Foundation awarded the 2025 Lasker~DeBakey Clinical Medical Research Award to Michael Welsh, Jesús González and Paul Negulescu for discoveries that led to the development of Trikafta, a triple combination of cystic fibrosis transmembrane conductance regulator (CFTR) potentiators and correctors to treat cystic fibrosis. This award recognizes the contribution of Trikafta to improving the quality of life of ~90% of the 40,000 people living with this condition in the United States, reducing infection-related hospitalizations and lung transplants, among other benefits.

But what about the other 10% of patients who don’t respond to Trikafta, many of whom carry so-called Class I alleles that cannot be rescued by this drug combination? Although a lot of progress has been made, several obstacles lie in the path of effective medicines for people who produce no, or negligible amounts of, CFTR protein.

It should come as no surprise that the main therapeutic strategies for Class I alleles aim to put missing CFTR back into lung cells. Among these strategies, mRNA delivery is the most advanced. VX-522, an RNA therapeutic program from Vertex and Moderna currently in Phase 2, is an inhaled drug that aims to deliver full-length CFTR mRNA to the lung using lipid nanoparticles (LNPs). Two related, competing mRNA delivery programs are at a similar stage of clinical development: ARCT-032 by Arcturus Therapeutics using their LUNAR LNPs; and RCT-2100 by ReCode Therapeutics, which uses a lung-targeted SORT (selective organ-targeting) LNP.

A key feature of RNA-based therapies is that any therapeutic benefit would likely be transient, requiring periodic administration of the medicine to achieve sustained effects. Gene therapy and gene editing have the potential to be a curative, “one and done” procedure. Thus far, however, only gene therapy programs have advanced far enough to be in human testing.

Of these, 4D Molecular Therapeutics’ 4D-710 and Spirovants’ SP-101 use different AAV subtypes designed to optimize delivery to airway basal epithelial cells of a CFTR minigene that lacks the regulatory domain. Both projects are in Phase 1/2 of clinical development.

As the large size (6.2 kb) of the CFTR transgene exceeds the packaging capacity of AAV vectors, Krystal Biotech and Boehringer Ingelheim have launched Phase 1/2 clinical programs using viral vectors with a greater payload capacity: KB407 is a re-dosable herpes simplex virus (HSV)-1 vector with a cargo capacity >30 kb that delivers two copies of the CFTR gene to lung epithelial cells using a nebulizer. BI 3720931 is Boehringer’s inhaled lentiviral vector pseudotyped with Sendai virus F and HN envelope proteins (rSIV.F/HN) engineered to deliver a single copy of the CFTR gene. Further behind in the pipeline, Carbon Biosciences’ CGT-001 is a nebulized non-AAV parvovirus-based vector capable of delivering full-length CFTR gene. Thus far, it has been tested in nonhuman primates and in human bronchial cells in culture.

Companies are also pursuing oligonucleotide therapies to modify disease-causing mutations at the RNA level. SPL84 is an inhaled antisense oligonucleotide (ASO) addressing a splicing defect (cryptic exon; class V mutation) in the ~1,600 CF patients who carry the 3849+10kb C→T mutation. SpliSense has advanced the ASO into phase 2 testing, but it also has in preclinical development an exon-skipping ASO against the class I mutant W1282X. By masking the mutant premature termination codon in exon 23, SP23 induces the splicing machinery to skip exon 23 and stitch together exon 22 and exon 24, forming a partially functional CFTRΔex23 protein.

Gene editing is also beginning to appear on the therapeutic horizon. In July, Prime Medicine announced it had received $25 million in funding to advance prime editors, with a lead program focusing on G542X. Last year, Intellia Therapeutics and ReCode Therapeutics also announced a strategic collaboration to combine the CRISPR pioneer’s Cas9 DNA ‘writing’/insertion technology with Recode’s SORT LNPs. Academic groups have now shown that G542X correction is possible using inhaled LNP- or virus–like particle-delivered adenine base editors. And for RNA editing, at this year’s American Society of Gene & Cell Therapy Wave Life Sciences reported their oligo-based ADAR editors could achieve 21% correction (EC50 = 376nM) of CFTR W1282X nonsense mutations. This is likely a sliver of all the therapeutic activity underway; other programs are targeting mucus itself, which is much thicker than in healthy individuals. If we missed any drug-discovery projects in this space, please let us know!

Despite the plethora of programs, developing genetic therapies against cystic fibrosis patients with class I CFTR mutations faces some stiff translational challenges. For starters, targeted delivery of drugs to lung tissue remains a work in progress. The optimal cell type to be targeted by gene therapy/editing remains an open question, especially as the community continues to identify new cell types in the lung; is it enough to target the more prevalent epithelial cells (alveolar type 2 cells), or will it be necessary to target rarer stem cells (alveolar type 1 cells) to see a long-lasting therapeutic effect? What about the contribution of genetic modifiers and other ion channels known to affect airway dysfunction in CF airway epithelial cells? Also, how to figure out the pharmacokinetics and pharmacodynamics of these disease-modifying therapies in lungs and measure delivery in patients? Specifically, establishing protein expression levels after inhaling a DNA- or RNA-based product would likely require a bronchial biopsy, which is impractical particularly in this fragile patient population.

Last, not unlike most pathologies, new animal and in vitro models with predictive value need to be developed. The use of human bronchial epithelium culture is not as predictive of the efficacy of genetic therapies as it has been for small molecules. At present, the ferret is the gold standard disease model. But it is a time-consuming, challenging animal model, which is only supported by a few groups. All of which slows the path to clinical translation.

Six years after the approval of Trikafta, patient foundations like the CF Foundation​, Emily’s Entourage, and the Cystic Fibrosis Trust are devoting increasing resources to translational research to push forward treatments for patients with CFTR Class I mutations who do not respond to potentiators and correctors. The Lasker recognition of the science that led to Trikafta will surely inspire researchers working on those projects to overcome the remaining hurdles.

Tags: , , , , , , , , , , , , , , , , , , , ,

The Needle Issue #14

9 Sep
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

It is now nearly a decade since Dublin-based startup ​Inflazome burst onto the scene with the description of MCC950, the first nanomolar selective inhibitor of the inflammasome. Inflammasome-mediated low-grade inflammation has been associated with cancers, numerous chronic complex diseases—including inflammatory bowel disease, arthritis, metabolic dysfunction-associated steatohepatitis (MASH), atherosclerosis, Alzheimer’s, Parkinson’s and ALS—as well as rare autoinflammatory diseases, such as cryopyrin-associated periodic syndromes (CAPS). There is a wealth of compelling human genetic evidence from Muckle-Wells syndrome and other autosomal dominant familial genetic diseases showing gain-of-function mutations in NLRP3 are causative of autoinflammatory disease.

Given the ‘pipeline in a product’ potential of drugs targeting this pathway, big pharma has shown considerable interest, with Genentech/Roche snapping up Jecure Therapeutics for an undisclosed amount, and both Novartis and Roche splashing out hundreds of millions of dollars for pioneer companies IFM Tre and Inflazome, respectively. In 2022, Novo Nordisk licensed Ventus Therapeutics’ peripherally restricted NLRP3 inhibitor in a deal worth up to $703 million, lending weight to pharmacological inhibition of NLRP3 as a complement to glucagon-like peptide-1 agonists (GLP-1s) in cardiometabolic disease. And with several programs now entering the clinic, investment activity in the area has continued, with Enveda’s announcement last week of a $150 million series D round to fund a phase 1 trial for ENV-6946, an orally delivered gut-restricted small molecule targeting the NLRP3/tumor necrosis factor-like cytokine 1A (TL1A) pathway in inflammatory bowel disease.

Today, Haystack counts at least 17 independent companies pursuing inflammasome therapeutics (AC Immune, Azome Therapeutics, BioAge Labs, Cardiol Therapeutics, EpicentRx, Enveda, Halia Therapeutics, InflammX Therapeutics, Insilico Medicine, Neumora Therapeutics, NodThera, Olatec Therapeutics, Shaperon, Ventus Therapeutics, Ventyx Biosciences, ZyVersa Therapeutics and Zydus LifeSciences) and 8 programs now in clinical testing specifically targeting the key inflammasome component NLRP3.

While drugmakers have traditionally targeted downstream extracellular mediators of the inflammasome pathway (canakinumab or rilonacept against IL-1β or anakinra to block IL-1 receptor), NLRP3 represents a key upstream intracellular signaling hub, activated by innate immune pattern-recognition receptor (Toll like receptors 2/4) signaling via MyD88 and NFkappaB. Once activated, NLRP3 monomers unfold and associate into a massive 1.2 MDa oligomeric supracomplex with three other proteins: ASC, NEK7 and caspase 1. The mature complex then cleaves and activates proinflammatory cytokines interleukin (IL)-1β and IL-18 and primes gasdermin D to instigate cell pore formation and cell death via pyroptosis.

Discovering effective drugs against NLRP3 has proven challenging. The first NMR structure was obtained in 2016, but the structural basis for how NLRP3 ring-like oligomers associate with intracellular membranes and how its pyrin domains associate with ASC to orchestrate speck formation and caspase activation have only recently been elucidated. Thus far, the majority of small-molecule inhibitors (e.g., Inflazyme’s archetypal MCC950 and inzolemid; Zydus’s (ZYIL1), Olatec’s OLT117 and Jecure Therapeutics’ GDC-2394) form hydrogen bonds via a sulfonylurea group to NLRP3’s NACHT domain nucleotide-binding motifs, thereby obstructing ATP hydrolysis. Other companies are taking a different tack: thus, Halia Therapeutic’s small-molecule inhibitor orniflast and Monte Rosa Therapeutics’ MRT-8102 molecular glue target NEK7 rather than NLRP3.

But it has been less than straightforward to identify compounds with sufficient potency to target this pivotal innate immune signaling pathway without debilitating off-target effects. Indeed, several of the first wave of compounds entering the clinic have been dogged by serious toxicities, including liver problems (MCC950 and GDC-2394) and hypoglycemia (glyburide). Now, a team led by Rebecca Coll (Queen’s University Belfast) and Kevin Wilhelmsen (of BioAge Labs) reports in The Journal of Experimental Medicine the discovery and characterization of BAL-0028, a novel and selective small-molecule inhibitor of the human NLRP3 inflammasome.

Unlike previously studied inhibitors, BAL-0028 acts through a unique mechanism of action; it binds NLRP3’s NACHT domain at a site distinct from other inhibitors that act by directly interfering with ATPase activity. BAL-0028 has nanomolar potency against human and primate NLRP3 but, remarkably, has weak activity against the mouse target, highlighting species-specific differences.

As BAL-0028 showed very high plasma protein binding in mice, limiting its use in vivo, the team developed a derivative, BAL-0598, with improved pharmacokinetic properties. In a humanized NLRP3 mouse peritonitis model, BAL-0598 effectively reduced IL-1β and IL-6 production, confirming its anti-inflammatory activity in vivo. Importantly, both BAL-0028 and BAL-0598 inhibited hyperactive NLRP3 mutants associated with autoinflammatory diseases, in some cases more effectively than Vertex’s VX-765, a caspase 1 inhibitor, and compounds like MCC950, one of the best characterized NLRP3 inhibitors available.

The novel mechanism of action of BAL-0028 and BAL-0598 would suggest their off-target effects may be different from those associated with other NLRP3 inhibitors blocking ATP hydrolysis. The concern that such compounds might also bind other members of the NOD/NLR family (e.g., NLRP1, NLRP4 or AIM2 inflammasomes) is mitigated by most published studies indicating that NLRP3’s unique fold around the ATP binding site makes small-molecule binders selective for this family member alone. The most likely explanation from trials published to date is that the observed toxicities are associated with small molecule chemotype rather than any NLRP3 class-specific problem. In any case, the findings from this study support further investigation of these compounds as candidates for treating inflammatory and age-related diseases where NLRP3 plays a role. The race to develop a safe and effective NLRP3 inhibitor is on, with big pharma billion-dollar bets and startups jostling to create best-in-class assets across cancer, cardiovascular, neurodegenerative and metabolic disease.

​

Tags: , , , , , , , , , , , , , , , , , , , ,

The Needle Issue #13

3 Sep
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

While most parts of biotech early-stage financing have been in the doldrums in the past two or three years, so-called tech-bio startups have been thriving. Since the posterchild $1.0 billion mega series A round last April of Xaira Therapeutics, which was founded by scientists out of Nobel prize winner David Baker’s group at the University of Washington, several startups seeking to develop machine learning models for designing miniproteins or peptide binders of challenging or ‘undruggable’ targets have emerged, including Enlaza Therapeutics, Vilya, and UbiquiTx. All of these have been developing their own proprietary models based on Alphafold 3, Boltz-1 or Chai-1 for structure prediction and tools based off RFdiffusion, Bindcraft and ProteinMPNN for peptide design. Predicting CDR loops for de novo antibody design is a considerably more challenging task than for simple peptides, but Nabla Bio, founded last year by scientists out of George Church’s lab at Harvard, claims it is doing just that for GPCRs and ion channels. Earlier this month, Chai Discovery also launched with a $100 million series A from Menlo Ventures to optimize multimodal generative models such as Chai-2, which, according to the company, already “achieves a 16% hit rate in de novo antibody design.”

Designing peptides that can selectively bind to a protein target and show therapeutic activity remains a challenge, however, as it often depends on the availability of high-quality structural information about the target molecule, which is seldom available for many disease-relevant proteins that are unstructured or conformationally disordered. Similarly modeling protein-protein interactions like antibody-antigen interactions that are extremely dynamic and floppy also poses problems. All of which raises the question as to whether binders could be predicted simply using amino acid sequence information instead of structural data.

Now, a team led by Pranam Chatterjee from Duke University has addressed this question. In a recent paper in Nature Biotechnology, Chatterjee and his collaborators report the creation of PepMLM, a peptide binder design algorithm based on masked language modeling. A key feature of the algorithm is that it depends exclusively on protein sequence, not structure. Built upon the ESM-2 (Evolutionary Scale Modeling 2) protein language model, PepMLM masks and reconstructs entire peptide regions appended to target protein sequences. This design compels the model to generate context-specific binders. To train PepMLM, the team used high-quality curated datasets from PepNN and Propedia comprising ~10k putative peptide-protein sequence pairs. PepMLM output was consistently found to outperform RFDiffusion on held-out/structured targets, with a higher hit rate (38% to 29%) and low perplexities that closely matched real binders, with generated sequences showing target specificity, even in stringent permutation tests.

The model generated binders predicted to have higher binding scores than native and structure-based binders designed through other methods. Indeed, in vitro validation experiments confirmed the high affinity and specificity of PepMLM-generated binders.

Chatterjee and his colleagues went on to turn their binders into degraders by fusing them to E3 ubiquitin ligase domains, such as CHIP/STUB1. When tested in vitro, over 60% of these degraders knocked down their target proteins. PepMLM peptides achieved nanomolar binding affinity on the drug targets neural cell adhesion molecule 1 (NCAM1), a key marker of acute myeloid leukemia, and anti-Müllerian hormone type 2 receptor (AMHR2), a critical regulator of polycystic ovarian syndrome (where RFDiffusion-predicted peptides failed to bind). The authors also demonstrated that PepMLM-predicted peptides fused to E3 ubiquitin ligases not only degraded MSH3 but completely eliminated mutant huntingtin protein exon 1 containing 43 CAG repeats in Huntington disease patient-derived fibroblast cells. Similar results were obtained for a PepMLM-predicted peptide binder of MESH1, a protein controlling ferroptosis, in collaboration with Ashley Chi Jen-Tsan’s group at Duke University (RFDiffusion again gave no hits). And with Madelaine Dumas and Hector Aguilar-Carreno’s group, in collaboration with Matt Delisa’s group at Cornell University, PepMLM-derived peptides bound and reduced levels of viral phosphoproteins from Nipah, Hendra, and human metapneumovirus (HMPV); indeed, in live HMPV infection models, the PepMLM peptide mediated high levels of P protein clearance.

The ability of PepMLM to design binders purely on the basis of target-protein sequence is an important advance towards designing therapeutic peptides against hitherto inaccessible targets that lack structural data. Future work should explore how to incorporate chemical modifications such as cyclization or stapling to enhance stability of the binders, as well as the evaluation of the strongest candidates in vivo. Another challenge will be to ameliorate the immunogenicity of these foreign de novo proteins. The use of protein engineering approaches, such as incorporation of mirror amino acids that can cloak foreign peptides from the immune system, may offer solutions. But it is likely that candidates discovered using sequence or structure prediction tools will still require lengthy development programs to be turned into safe and effective drugs, despite the hype.

Tags: , , , , , , , , , , , , , , , , , , ,

The Needle Issue #12

12 Aug
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

The Summer BIO report “The State of Emerging Biotech Companies: Investment, Deal, and Pipeline Trends” highlights how much China-based programs have contributed to the drug pipeline over the past 10 years.

A couple of weeks ago, Bloomberg also summarized deal data showing how the share of global licensing by Chinese biotech companies has jumped over the past two years.

Judging by a report listing 16 ‘high-value’ currently unlicensed assets from China being hawked by longtime Phalanx Investment Partners analyst David Maris, there is more licensing to come.

In this context, we read with interest a recent Science Immunology paper describing a monoclonal antibody (mAb) program targeting a novel phagocytic checkpoint under development at yet another Chinese biotech: MedimScience, founded in Hangzhou City in 2021. MedimScience is one of a growing cadre of companies, including LTZ Therapeutics, Dren Bio, Chengdu Kanghong, Antengene and ImmuneOnco, looking to develop novel myeloid cell engagers/phagocytic checkpoint inhibitors.

Phagocytic checkpoint inhibitors are drugs that circumvent the molecular cloaks that tumors throw around themselves to avoid uptake and destruction by myeloid cells, such as macrophages, monocytes, and neutrophils. The strategy first came to the fore through pioneering work on the ‘don’t eat me’ signal CD47, work carried out by Ravi Majeti and Irv Weissman at Stanford. Results from their preclinical studies spurred the launch of startup Forty Seven (subsequently acquired in 2020 by Gilead) and the first-in-class anti-CD47 IgG4 magrolimabprogram.Phase 1b trial results of magrolimab combined with azacitidine in acute myeloid leukemia (AML) patients were so impressive that, by 2022, more than 20 different companies had anti-CD47 programs in clinical development. This blew up spectacularly when early trials failed to be reproduced in larger efficacy trials of combinations — failure that was largely attributed to intolerability/anemia issues related to the target, slow action/early disease progression, and a failure to account for patient heterogeneity with regard to P53 mutation status. But the strategy is compelling and the hunt for new phagocytic checkpoints has continued with new antibody formats seeking to avoid these pitfalls.

Now, Cheng Zhong and his colleagues at MedimScience report the identification of a new evasion actor — PSGL-1 — that suppresses macrophage-mediated phagocytosis in a variety of hematological malignancies. PSGL-1, which was previously known largely for its role in cell adhesion, is highly expressed in various hematologic cancers, including AML, T-acute lymphoblastic leukemia (T-ALL) and multiple myeloma (MM).

Moreover, high PSGL-1 expression has been found to correlate with poor patient survival in AML, T-ALL and MM.

Using several mouse models, the researchers found that tumors lacking PSGL-1 show slower progression, increased macrophage infiltration, and higher rates of phagocytosis by macrophages, effects that were independent of T cells or dendritic cells.

Mechanistically, the team found that PSGL-1 disrupts the interaction between the cell-adhesion molecule ICAM-1 on tumor cells and the integrin LFA-1 (CD11a/CD18) on macrophages. And when they tested Novartis’ lifitegrast, an inhibitor of ICAM-1/LFA-1 binding, they found this largely abrogates the phagocytosis of PSGL-1 knockout tumor cells, confirming PSGL1’s role in impairing prophagocytic signaling and cytoskeletal reorganization required for effective tumor-cell engulfment.

The authors went on to develop a humanized mAb against PSGL-1 and show its ability to induce phagocytosis of human tumor cells in vitro and to reduce tumor burden in mouse models of AML, T-ALL, and MM. The antibody showed a good safety profile in non-human primates with no significant toxicity at high doses. Additionally, PSGL-1 blockade synergized with chemotherapy (doxorubicin) and antibody-based therapies (anti-CD47 and anti-CD38), further underscoring the translational potential of this strategy, particularly in treatment-resistant settings.

Tags: , , , , , , , , , , , , , , , , ,

The Needle Issue #11

22 Jul
Juan-Carlos-Lopez
Juan Carlos Lopez		 Andy-Marshall
Andy Marshall

Haystack chat

Molecular glue degraders (MGDs) are currently having a bit of a moment. In the first half of 2025, the number of papers describing such compounds has doubled.

2025 has also witnessed a whole raft of MGD startups publish research related to their programs:

Startup (location) Scientific founders (location) 2025 paper
Ambagon Therapeutics (Eindhoven, The Netherlands) Michelle Arkin (UCSF, San Francisco, CA), Luc Brunsveld and Christian Ottman (Eindhoven University of Technology) Molecular glues of the regulatory ChREBP/14-3-3 complex protect beta cells from glucolipotoxicity
Cyrus Therapeutics (Seoul, South Korea) Keon Wook Kang (Seoul National University, Seoul, South Korea) High cereblon expression in neuroendocrine cancer confers vulnerability to GSPT1 molecular glue degrader
Matchpoint Therapeutics (Cambridge, MA) Nathanael Gray and Tinghu Zhang (Stanford University, Stanford, CA) and Edward Chouchani and Jianwei Che (Dana Farber, Boston, MA) Structure-guided design of a truncated heterobivalent chemical probe degrader of IRE1α
Monte Rosa Therapeutics (Boston, MA) Rajesh Chopra and Ian Collins (The Institute of Cancer Research and Cancer Research UK); Nicolas Thomä (Friedrich Miescher Institute, Basel, Switzerland) Structure-guided strategy for identifying human proteins predicted to be compatible with cereblon-based molecular glue degraders (see below for further details)
Oniria Therapeutics (Barcelona, Spain) Héctor G. Palmer, Esther Riambau, Isabel Puig, Josep Tabernero, Xavier Barril, and Carles Galdeano (Vall d’Hebron Institute of Oncology, University of Barcelona and ICREA) Cullin-RING ligase BioE3 reveals molecular-glue-induced neosubstrates and rewiring of the endogenous Cereblon ubiquitome
Proxygen (Vienna, Austria) Georg Winter (CeMM Research Center for Molecular Medicine, Vienna, Austria) Selective analysis of protein degradation by mass spectrometry enables degradome analysis and identification of direct protein substrates of molecular glues
Proteovant Therapeutics (King of Prussia, PA) Shaomeng Wang (University of Michigan, MI) Development of PVTX-405 as a potent and highly selective molecular glue degrader of IKZF2 for cancer immunotherapy
Sartar Therapeutics (Helsinki, Finland) Olli Kallioniemi and Harri Sihto (University of Helsinki, Finland) Pharmacokinetic profile and in vivo anticancer efficacy of anagrelide administered subcutaneously in rodents
SEED Therapeutics (King of Prussia, PA) Ning Zheng (University of Seattle, WA), Michele Pagano (New York University, NY) and Avram Hershko (Technion Institute of Technology, Haifa, Israel) UM171 glue co-opts CRL3 RING E3 ligase substrate coreceptor KBTBD4 as well as HDAC1/2, resulting in degradation of CoREST corepressors
Shenandoah Therapeutics (South San Francisco, CA) Jerry Crabtree and Nathanael Gray (Stanford University, Stanford, CA) A bivalent molecular glue linking lysine acetyltransferases to redirect p300 and CBP to activate programmed cell death genes normally repressed by the oncogenic driver, BCL6
Zenith Therapeutics (Basel, Switzerland) Daniel Nomura (UC Berkeley, CA); Nicolas Thomä (Friedrich Miescher Institute, Basel, Switzerland), and Martin Stahl (former Roche, LifeMine) Putative molecular glue niclosamide acts via ubiquitin E3 ligase CRL4AMBRA1-mediated degradation of cyclin D1 following mitochondrial membrane depolarization

On the commercial front, the march of startups receiving funding shows no sign of slowing down, with Trimtech Therapeutics and Booster Therapeutics raising substantive rounds. The first few months of the year have also seen the continuation of last year’s pharma MGD scramble to license programs from Triana Biomedicines and Neomorph, with deals based around molecular glues from Abbvie and Merck targeting Neomorph and Springworks, respectively.

In June, one of the flagship developers, Kymera Therapeutics, priced a $250.8 million follow-on offering (no mean feat in the present market) after announcing positive phase 1 safety data for KT-621, a novel MGD against STAT-6, and clinching a deal with Gilead Sciencesforanother small-molecule glue targeting cyclin-dependent kinase 2 (CDK2). All in all, we count 27 companies currently active in this preclinical space (Ambagon Therapeutics, Amphista Therapeutics, Booster Therapeutics, Captor Therapeutics, Cyrus Therapeutics, Degron Therapeutics, Dunad Therapeutics, F5 Therapeutics, Frontier Medicines, Lifemine Therapeutics, Magnet Biomedicine,Matchpoint Therapeutics, Montara Therapeutics, Monte Rosa Therapeutics, Neomorph, Oniria Therapeutics, Proxygen, Sartar Therapeutics, SEED Therapeutics, Shanghai Dage Biomedical Technology, Shenandoah Therapeutics, SK Biopharmaceuticals (Proteovant Therapeutics),Triana,Trimtech,Venquis Therapeutics, YDS Pharmatech, and Zenith Therapeutics). There are likely more.

Unlike their more recent cousins, the PROTACs (proteolysis targeting chimeras), MGDs have a long history. The archetypal MGD, thalidomide, was discovered back in the 1950s. From the late 1990s, a new generation of immunomodulatory imide drug (IMiD) derivatives of thalidomide were synthesized, culminating with the approvals of lenalidomide and pomalidomide for myeloma (which formed the basis for the Celgene (now BMS) franchise).

Unlike PROTACs, which use two ligands with a linker and tend to be rather unwieldy, MGDs are small, single compounds that induce conformational changes in E3 ubiquitin ligases and target proteins, reshaping both to enable binding. The vast majority of MGDs bind Cereblon (CRBN), leading to ubiquitination of the protein of interest and degradation in the 26S proteasome, although work is progressing to broaden MGD action to some of the other 600 or so E3 ubiquitin ligases (e.g., DCAF11,15 or 16, DDB1, SIAH, KEAP1, VHL, β-TrCP, Nedd1 and, just last week, TRIM21).

A key challenge in finding new MGDs has been a lack of understanding of the structural rules whereby MGDs turn their target proteins into CRBN ‘neosubstrates’, which has meant MGD ‘hit-finding’ is much more challenging, with fewer degrees of freedom than PROTACs.

What drug hunters have established is that many protein targets of glues contain a β-hairpin structural motif known as the ‘G-loop’. When a MGD brings a target together with CRBN, one end of the MGD interacts with a binding pocket in the C-terminal domain of CRBN, while the other end protrudes from the pocket and interacts with the G-loop (part of the so-called ‘degron’) in the neosubstrate. But how many proteins possess the β-hairpin G-loop or whether the loop is strictly necessary for MGD action have remained open questions. A recent study by Monte Rosa Therapeutics’ scientists starts to tackle these issues, disclosing a large cadre of potential new substrates for CRBN, some of which depart from the canonical β-hairpin G-loop, radically expanding MGD target space.

To map the full range of proteins potentially recruitable by CRBN through MGDs, the team led by John Castle and Sharon Townson developed computational algorithms to search for β-hairpin G-loop motifs in protein structures from two databases: Protein Data Bank and AlphaFold2. This approach resulted in 1424 candidate proteins, some of which were experimentally validated in MGD assays. The list included previously known neosubstrates, but also new proteins such as NEK7—a protein of interest as an autoimmunity target.

The researchers then wondered if the full β-hairpin structure of the G-loop is required for CRBN recognition and rescreened the structure databases looking for a minimal, structurally defined helical G-loop motif. This resulted in the identification of 184 additional potential neosubstrates, including mTOR, a well-established therapeutic target for drugs like rapamycin and sirolimus. Crystallographic data showed that the binding of this helical G-loop to CRBN is similar to that of the canonical β-hairpin G-loops.

As these protein–protein interactions have been well characterized, the team then tried to identify an even wider set of potential neosubstrates, looking now for proteins with sequences that might result in surfaces with electrostatic properties similar to known CRBN interactors, independently of secondary structure and the existence of G-loops. Using surface-matching algorithms, they identified and validated VAV1 (another autoimmune disease target) as a CRBN neosubstrate, providing compelling evidence that G-loops are not strictly necessary for the action of MGDs.

These findings show that CRBN recruitment through MGDs can be driven by a broader set of structural features than previously thought. The identification of a large number of neosubstrates potentially opens up a whole new set of previously ‘undruggable’ targets to MGDs (>1,600 proteins from many target classes, according to the Monte Rosa team).

The big questions, though, are still ahead. How will drug developers mitigate the risks of ‘off-tissue’ toxicity as this swathe of novel MGD compounds and new targets make their way into the clinic?One answer to the toxicity concern is molecular glue antibody conjugates (MACs), which can better localize glues to the tissue of interest. But that’s a subject for a whole other future Haystack Chat!

Tags: , , , , , , , , , , , , , , , ,

Newer Entries →