Dan Fulop

🚨🐄 With growing alarm over the renewed spread of 𝙱𝟹.𝟷𝟹 H5N1 in US dairy herds, the recent spillover into a Utah mink farm, and ongoing human cases, the #1 question remains: 𝗜𝘀 𝘁𝗵𝗶𝘀 𝘃𝗶𝗿𝘂𝘀 𝗼𝗻 𝗮 𝗽𝗮𝘁𝗵 𝘁𝗼 𝗲𝘃𝗼𝗹𝘃𝗲 𝗵𝘂𝗺𝗮𝗻-𝘁𝘆𝗽𝗲 (α𝟮–𝟲) 𝗿𝗲𝗰𝗲𝗽𝘁𝗼𝗿 𝗯𝗶𝗻𝗱𝗶𝗻𝗴? Yet after 2 years and thousands of mammal-to-mammal infections across cows, cats, mink, and wildlife, the main 𝙱𝟹.𝟷𝟹 lineage still hasn’t budged toward α𝟸–𝟼. Here’s my current thinking on why — and why 𝙱𝟹.𝟷𝟹’s unprecedented mammalian host breadth might actually be pushing it in the opposite direction, 𝗹𝗼𝗰𝗸𝗶𝗻𝗴 𝗶𝘁 𝗱𝗲𝗲𝗽𝗲𝗿 𝗶𝗻𝘁𝗼 𝗯𝗶𝗿𝗱-𝘁𝘆𝗽𝗲 (α𝟮–𝟯) 𝗯𝗶𝗻𝗱𝗶𝗻𝗴: ──────────────────── 🔑 𝗤𝘂𝗶𝗰𝗸 𝗥𝗲𝗰𝗲𝗽𝘁𝗼𝗿 𝗥𝗲𝗳𝗿𝗲𝘀𝗵𝗲𝗿 Flu enters cells by grabbing sialic acid sugars (𝙷𝙰) and exits by cutting them (𝙽𝙰): • 🦆🐄 α𝟸–𝟹 (𝗯𝗶𝗿𝗱-𝘁𝘆𝗽𝗲): Dominant in bird guts/airways, deep human lung, eye — and the 𝗰𝗼𝘄 𝗺𝗮𝗺𝗺𝗮𝗿𝘆 𝗴𝗹𝗮𝗻𝗱. • 👃🧑 α𝟸–𝟼 (𝗵𝘂𝗺𝗮𝗻-𝘁𝘆𝗽𝗲): Dominant in the human nose & upper throat (where cough/sneeze transmission happens). Historically, avian flu needed to flip 𝙷𝙰 from α𝟸–𝟹 to α𝟸–𝟼 (often via 𝚀𝟸𝟸𝟼𝙻) to spread between humans. Why hasn’t mammal-to-mammal spread in cattle nudged 𝙱𝟹.𝟷𝟹 toward α𝟸–𝟼? ──────────────────── 🔒 𝟭. 𝗧𝗵𝗲 𝗖𝗼𝗺𝗺𝗼𝗻-𝗗𝗲𝗻𝗼𝗺𝗶𝗻𝗮𝘁𝗼𝗿 𝗧𝗿𝗮𝗽 𝗼𝗳 𝗮 𝗠𝘂𝗹𝘁𝗶-𝗛𝗼𝘀𝘁 𝗚𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝘁 𝙱𝟹.𝟷𝟹 isn’t evolving in an isolated upper-airway niche. In the tree of >5,600 genomes (thanks @nextstrain): • 🔀 𝟮𝟳𝟵+ 𝘀𝗽𝗶𝗹𝗹𝗯𝗮𝗰𝗸𝘀 from cattle into wild birds, poultry, barn cats, raccoons, skunks, rodents & mink • ↩️ 𝟴𝟳 𝗯𝗮𝗰𝗸-𝘀𝗽𝗶𝗹𝗹𝗼𝘃𝗲𝗿𝘀 from birds back into cattle Across 𝗮𝗹𝗹 of those niches — cow udders, bird guts, poultry, and carnivore/mink lungs — α𝟸–𝟹 is the shared entry receptor! In lab studies on 𝙱𝟹.𝟷𝟹 𝙷𝙰, forcing in the human-type 𝚀𝟸𝟸𝟼𝙻 mutation collapses α𝟸–𝟹 binding and destabilizes the 𝙷𝙰 trimer. In a virus cycling between cow udders and farm wildlife, any mutant that trades α𝟸–𝟹 for α𝟸–𝟼 immediately loses fitness in the very tissues keeping the chain alive. 🧬 𝟮. 𝗕𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝗮𝗻 α𝟮–𝟯 𝗦𝘂𝗽𝗲𝗿-𝗚𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝘁 𝗜𝗻𝘀𝘁𝗲𝗮𝗱 Rather than shifting toward α𝟸–𝟼, clade 𝟸.𝟹.𝟺.𝟺𝚋 expanded its host range by making its binding pocket more flexible: 𝙺𝟸𝟸𝟸𝚀 → 𝚃𝟷𝟿𝟿𝙸 → 𝙳𝟷𝟶𝟺𝙶 → 𝚅𝟷𝟺𝟽𝙼 This lets 𝙷𝙰 grab a wider variety of α𝟸–𝟹 sugars (fucosylated, branched, and cattle NeuGc variants) while keeping α𝟸–𝟼 affinity near zero. ✂️ 𝟯. 𝗖𝗼-𝗘𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝗮𝗿𝘆 𝗟𝗼𝗰𝗸-𝗜𝗻 𝗪𝗶𝘁𝗵 𝗡𝗔 As 𝙱𝟹.𝟷𝟹 𝙷𝙰 tightened its grip on α𝟸–𝟹 sugars in cattle, the viral scissor protein (𝙽𝙰) co-evolved matching mutations: 𝙽𝙰: 𝙺𝟷𝟷𝟷𝚁 | 𝚃𝟽𝟸𝙰 | 𝙽𝟺𝟹𝟺𝚂 | 𝙺𝟺𝟹𝟸𝙴 (𝙽𝟺𝟹𝟺𝚂 & 𝙺𝟺𝟹𝟸𝙴 sit right in 𝙽𝙰’s secondary helper pocket that grabs α𝟸–𝟹 sugars). Once 𝙷𝙰 and 𝙽𝙰 are co-tuned for α𝟸–𝟹, switching 𝙷𝙰 alone to α𝟸–𝟼 breaks their balance. ──────────────────── ⚠️ 𝗧𝗵𝗲 𝗖𝗼𝘂𝗻𝘁𝗲𝗿-𝗔𝗿𝗴𝘂𝗺𝗲𝗻𝘁: 𝗪𝗵𝘆 𝗪𝗲 𝗖𝗮𝗻’𝘁 𝗥𝗲𝗹𝗮𝘅 Even if the cow–bird–mammal cycle locks the 𝗺𝗮𝗶𝗻 𝘁𝗿𝘂𝗻𝗸 into α𝟸–𝟹, three things make the resurgence, mink farm outbreak & human cases genuinely concerning: ⚙️ 𝟭. 𝗧𝗵𝗲 𝗶𝗻𝘁𝗲𝗿𝗻𝗮𝗹 𝗺𝗮𝗺𝗺𝗮𝗹𝗶𝗮𝗻 𝗲𝗻𝗴𝗶𝗻𝗲 𝗶𝘀 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝗯𝘂𝗶𝗹𝘁: Two years in cows fixed mammalian polymerase adaptations (𝙿𝙱𝟸:𝙼𝟼𝟹𝟷𝙻, 𝙿𝙰:𝙺𝟺𝟿𝟽𝚁) and a 4-mutation interferon-stealth stack in 𝙽𝚂𝟷 (𝙰𝟼𝟶𝚃 + 𝚀𝟸𝟻𝚁/𝙸𝟻𝟺𝙻/𝙴𝟽𝟶𝙰/𝙼𝟷𝟸𝟺𝚃) — and in Utah mink it immediately stacked 4 more (𝙿𝙱𝟸:𝚀𝟷𝟹𝚁, 𝙿𝙱𝟷:𝙺𝟹𝟾𝟾𝚁, 𝙿𝙰:𝙸𝟷𝟸𝟿𝚃, 𝙼𝟸:𝙶𝟻𝟾𝙳/𝚂𝟼𝟺𝙵). Spillovers no longer start from scratch. 🪜 𝟮. 𝗖𝗮𝘁𝘁𝗹𝗲 𝗲𝘃𝗼𝗹𝘂𝘁𝗶𝗼𝗻 𝗸𝗲𝗲𝗽𝘀 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝘀𝘁𝗲𝗽𝗽𝗶𝗻𝗴-𝘀𝘁𝗼𝗻𝗲 𝗛𝗔 𝘃𝗮𝗿𝗶𝗮𝗻𝘁𝘀: In 𝙱𝟹.𝟷𝟹 𝙷𝙰, 𝚀𝟸𝟸𝟼𝙻 works poorly alone, but gives clean human α𝟸–𝟼 binding when paired with 𝙰𝟷𝟽𝟸𝚃 (𝚃𝟷𝟼𝟶𝙰, which deletes a nearby sugar shield). In cattle, 𝙰𝟷𝟽𝟸𝚃 has already arisen ~𝟭𝟬 𝗶𝗻𝗱𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 𝘁𝗶𝗺𝗲𝘀 (#2 selected site in 𝙷𝙰). 🌿 𝟯. 𝗗𝗲𝗮𝗱-𝗲𝗻𝗱 𝘀𝗽𝗶𝗹𝗹𝗼𝘃𝗲𝗿𝘀 & 𝗿𝗲𝗮𝘀𝘀𝗼𝗿𝘁𝗺𝗲𝗻𝘁 𝗱𝗼𝗻’𝘁 𝗰𝗮𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝘂𝗱𝗱𝗲𝗿 𝗳𝗶𝘁𝗻𝗲𝘀𝘀: A virus that spills into a human or pig — or reassorts with seasonal/swine flu — is in an α𝟸–𝟼 upper airway and never has to cycle back into a cow udder or a bird. ──────────────────── 💡 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆 Expecting the 𝗰𝗮𝘁𝘁𝗹𝗲 𝗿𝗲𝘀𝗲𝗿𝘃𝗼𝗶𝗿 𝗶𝘁𝘀𝗲𝗹𝗳 to gradually drift toward α𝟸–𝟼 probably misreads what selection in cows and farm wildlife rewards — which is locking in α𝟸–𝟹. The real risk isn’t where the main cattle trunk is heading, but that expanding mammal circulation keeps feeding a pre-adapted internal engine + permissive 𝙷𝙰 variants into side-branch exposures. Key papers in reply 👇