Bristol-Myers Squibb locks into novel autoimmune strategy (Nature Biotechnology)

Nature Biotechnology, June 2016

Padlock Therapeutics, a startup with an innovative autoimmune R&D pipeline, recently caught the eye of big pharma Bristol-Myers Squibb (BMS), barely two years after the biotech was formed. The New York–based pharma agreed in March to buy the Cambridge, Massachusetts–based firm with $225 million in upfront and near-term milestone payments and another $375 mil­lion should drug development go according to plan. Padlock is developing small mol­ecules to block peptidylarginine deiminase (PAD) enzymes, which catalyze a protein modification long known to immunologists to be involved in certain types of autoimmu­nity. If the strategy proves successful, these PAD blockers could be a powerful alternative to blockbuster biologics like Humira (adali­mumab). But questions about side effects loom large because PADs play important roles in innate immunity and in gene regu­lation, which may be one reason few com­panies have ventured into this space, at least publicly.

BMS obviously feels the potential payoff justifies the risks and the steep price for Padlock. “Other companies have tried, but  none were able to get the kind of [chemi­cal] traction that we were able to get,” says Padlock co-founder Paul Thompson, a bio­chemist at the University of Massachusetts in Worcester. Thompson, then at the Scripps Research Institute, cofounded Padlock in late 2013 with Kerri Mowen, a Scripps molecu­lar immunologist who recently died at a tragically young age of a brain aneurysm, and Michael Gilman, CEO of Padlock and venture partner at Atlas. The company set up with seed funds from Atlas Venture fol­lowed by a $23-million series A financing, which included Atlas, Johnson & Johnson Innovation–JJDC (the venture arm of Johnson & Johnson), MS Ventures and Index Ventures (now Medici Ventures). Other biotechs in this space are 4SC Discovery in Planegg-Martinsried, Germany, with a PAD inhibitor program in the discovery stage, and ModiQuest in Oss, Netherlands, which is developing a monoclonal antibody against a PAD-modified substrate.

PADs are a novel target in autoimmu­nity. The enzymes catalyze citrullination, the conversion of the amino acid argi­nine to citrulline. As a post-translational modification, citrullination removes a posi­tive charge from the protein surface, a small change that can have major consequences by altering hydrogen bonding, and by modify­ing the protein’s structure and even function. In people with certain genetic backgrounds citrullinated proteins can be immunogenic.

The study of citrullination, however, was until recently “a backwater,” says Padlock’s Thompson. “Nobody thought it was inter­esting.” The rheumatology community was the exception. That’s because in 1998 Dutch researchers showed that 76% of rheumatoid arthritis patients produce antibodies that recognize citrullinated autoantigens (J. Clin Invest. 101, 273–281, 1998). A commercial assay for such anti­bodies is now the standard for diag­nosing the disease. And although no one knows what triggers the autoim­mune response in rheumatoid arthri­tis, the autoanti­gens are certainly citrullinated proteins, says rheumatologist Patrick Venables of Oxford University in the UK, “and citrulli­nated proteins are created by PAD enzymes.” Indeed, citrullinated antigens can be found in patients up to a decade before they show up in the doctor’s office with clinical symp­toms of rheumatoid arthritis. The hope for PAD inhibitors in rheumatoid arthritis is that removing the citrullinated antigens will resolve the disease, much like in celiac dis­ease where eliminating the antigen (gliadin protein in dietary gluten) cures the condi­tion.

Although this rationale has existed for at least a decade, two problems stood in the way of targeting these enzymes, says Thompson. The first was the poor under­standing of the physiological role of PADs, which made side effects hard to predict, and the second was the difficulty in making specific and potent inhibitors that penetrate cells and are also metabolically stable. Of the five PADs in the enzyme family, the favorite target is PAD4, which has been genetically linked to some rheumatoid arthritis cases and whose expression is mostly limited to neutrophils. Thompson’s laboratory, begin­ning in 2005, made a series of increasingly potent covalent pan-PAD inhibitors that he and others used over the last decade in vari­ous mouse models of autoimmune disease to demonstrate efficacy and safety. Also during this period, Brentford, UK–based GlaxoSmithKline (GSK), using a screening approach (Nat. Chem. Biol. 11, 189–192, 2015), developed a series of specific small-molecule PAD4 inhibitors. GSK licensed these compounds to Padlock last year, and they now are in the hands of BMS. These compounds, by showing specificity with noncovalent binding, which most pharma consider to be theoretically safer, “reinvigo­rated interest in the field,” says Thompson.

Padlock’s drugs block citrullination tak­ing place inside and outside cells. In the course of rheumatoid arthritis, neutrophils in inflamed joints release PAD4 into syno­vial fluid, where calcium concentrations are high. Because PAD enzymes are calcium-dependent, this adds fuel to the fire by sending PADs hunt­ing for proteins to citrullinate, creating more autoantigens in a feed-forward inflammatory loop. But PADs also citrullinate histones, transcription factors and other proteins inside cells, which may help drive disease.

Whether PAD inhibitors will still be effec­tive once rheumatoid arthritis is established remains unknown, because advanced disease is multifaceted. In animals PAD inhibitors are effective both in preventing disease and after disease onset, says Thompson. These results suggest PAD activity continues late in the disease process and is worth target­ing. Also, Venables recently used a pan-PAD inhibitor to treat mice with collagen-induced arthritis, a standard rheumatoid arthritis model, and at a high dose, the drug on its own completely halted the progression of the disease (the work has not yet been pub­lished).

But Venables hesitates to predict success for PAD inhibitors in rheumatoid arthri­tis. PADs are promiscuous and citrullinate many other proteins. “The worry is going to be side effects, isn’t it?” he says. “Because there are important physiological functions [of PADs].” In two 2004 papers (Cell 118, 545–553, 2004; Science 306, 279–283, 2004) researchers reported that PAD4 citrullinates histones and this could turn off gene expres­sion by antagonizing histone methylation. So inhibiting PAD4 might have unpredictable effects on the expression of multiple genes. Thompson says tumor formation is unlikely, because in cells and animals PADs generally promote oncogenesis, not the reverse.

On the plus side, inside cells, PAD activ­ity appears to be well controlled, and levels are generally very low, Thompson says. And blocking citrullination could lead to some beneficial gene regulatory effects. For exam­ple, work by Venables and others at Oxford shows that PAD4 citrullination of the tran­scription factor E2F1 promotes inflamma­tory cytokine gene expression, with a PAD inhibitor dampening the inflammation (Sci. Adv. 2, e1501257, 2016). This is a completely different PAD disease mechanism than auto­antigen citrullination—an unexpected bonus for PAD inhibitors, if it plays out in humans. “PAD inhibition in transcription is probably what made BMS much more interested” in these drugs, speculates Venables. It’s also why PAD inhibitors probably need to penetrate into cell nuclei to be effective.

For better or worse, PADs are also impor­tant in innate immunity. In fact, their best-established physiological role is to facilitate one form of neutrophil cell death. Bacterial exposure sometimes prompts neutrophils to explode spectacularly. They eject their chromatin, decorated with DNA, histones, enzymes and antibacterial peptides, forming neutrophil extracellular traps (NETs), which trap the pathogen for killing by phagocytes. PADs are essential for at least some forms of NETs, because citrullination of neutrophil histones decondenses the chromatin, releas­ing the proteins and nucleic acids needed to form the NET. (The neutrophil dies a noble death.) “It’s a complex process, but the PADs have been particularly implicated,” says Mariana Kaplan, a rheuma­tologist at the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS). As a consequence, a major worry about PAD inhibitors is that they will block this important mechanism of innate immu­nity and leave patients open to infection. Reassuringly, Thompson, who speculates that other immune mechanisms compensate for NET loss, says, “PAD4 knockouts don’t show an increased risk of developing infections.”

And inhibiting NETs may even have thera­peutic benefits, as NETs have been implicated in lupus pathogenesis. “There seems to be an imbalance between NET formation and NET clearance in lupus,” says Kaplan. Her group has identified a subset of neutrophils, pres­ent at varying levels in lupus patients, that in vitro form NETs spontaneously (without microbial stimuli) (J. Immunol. 187, 538–52, 2011). Other groups have reported NET dys­regulation in lupus patients who have severe disease and high titers of anti-DNA antibod­ies (a lupus hallmark). Such NETs may expose modified DNA and other autoantigens to immune attack. Kaplan’s laboratory has used PAD inhibitors in two animals models of lupus with positive results (Ann. Rheum. Dis. 74, 2199–2206, 2015) and NIAMS has a coop­erative R&D agreement with Padlock. Kaplan notes, however, that any future anti-PAD therapy may only work in a subset of lupus patients.

Rheumatoid arthritis is the obvious first indication for PAD inhibitors, because citrul­linated autoantigens are such a signature feature. “The presence of antibodies to citrul­linated antigens is quite specific to rheumatoid arthritis,” says Kaplan. Tackling PAD4 may not be enough; Venables thinks PAD2 must also be inhibited. “PAD2 and PAD4 have some shared redundancy,” he notes. This may be why GSK did not pursue a specific PAD4 inhibitor, Venables speculates. “They were develop­ing it for about ten years, but then they just totally dropped it,” he says. But dual PAD2– PAD4 inhibition raises the risk of side effects, because PAD2 is widely expressed. “There’s no predicting,” says Venables, who as a rheu­matology trainee saw the first patient treated with a tumor necrosis factor (TNF) inhibi­tor, in 1992. “There was a huge prejudice against anti-TNF when it was first tested,” he recalls. “PAD inhibition has gone through the same prejudice.”

That prejudice has helped limit the field of PAD inhibition to Padlock (now part of BMS) and a few known competitors. 4SC Discovery has not disclosed details on its early-stage PAD inhibitor program. ModiQuest, which has been issued a patent on PAD inhibi­tors together with a Dutch academic group, is instead developing a therapeutic anti­body against a citrullinated epitope of a his­tone protein, through a spinout company, Citryll. ModiQuest believes this downstream approach will be more specific and safer than inhibiting PADs. But neither this monoclo­nal antibody nor BMS’s PAD inhibitors have yet been tested in humans. Only with clinical results will researchers begin to know if target­ing an obscure post-translational modification will fulfill its promise in rheumatoid arthritis and beyond.

Ken Garber, Ann Arbor, Michigan