In the spring of 2020, doctors across the world prescribed azithromycin to COVID patients by the millions. The logic seemed sound: azithromycin had putative anti-inflammatory properties. A now-retracted study claimed it improved outcomes when combined with hydroxychloroquine. Even after the retraction, prescriptions continued. In some hospitals, it was given routinely.
Six years later, we know what that did. A Nature Microbiology study published this March — 1,164 hospitalized COVID patients, longitudinal metatranscriptomics — found that azithromycin altered the upper respiratory microbiome within a single day. It enriched pathobionts like Klebsiella and Staphylococcus. It increased macrolide resistance gene expression. Changes persisted for over a week. And the anti-inflammatory benefit that justified all those prescriptions? It didn't exist. No difference in host inflammatory gene expression. Zero.
But the damage went deeper than resistance genes. Among the commensals azithromycin destroyed was a bacterium most clinicians have never heard of — a gram-positive, lactic-acid-producing organism with a genome smaller than 2 megabases, no known virulence factors, and a name that sounds like a footnote: Dolosigranulum pigrum.
It turns out this footnote is a gatekeeper. And losing it may open the door to everything I've been writing about for the past 26 posts.
The First Chokepoint
Three independent research groups, working on three different continents, have converged on the same organism. Their findings, taken together, map a mechanism chain that stretches from the nose to the bone marrow.
Five-year longitudinal cohort. 156 patients — healthy controls, influenza, moderate COVID, severe COVID — sampled at acute phase and three-month follow-up. D. pigrum and Corynebacterium species were depleted in patients who developed PASC and abundant in those who recovered. The depletion was not subtle: LogFC of −3.98 for D. pigrum in PASC patients (q = 1.42 × 10−6). Antibiotic treatment during acute illness was associated with lower abundances of both protective taxa — and higher PASC frequency.
1,548 nasal swabs from a community-dwelling population in Washington, D.C. Two retrospective case-control studies plus 16S rRNA microbiome analysis. High D. pigrum density correlated with decreased nasal ACE2 and TMPRSS2 expression — the two receptors SARS-CoV-2 needs to enter cells. Conversely, pathobionts like S. aureus, H. influenzae, and M. catarrhalis increased receptor expression. The bottom line: elevated ACE2/TMPRSS2 meant a 3.6-fold increase in infection risk (95% CI: 1.71–7.47). Cases also had more unstable receptor expression in the days before testing positive.
In vitro challenge: Calu-3 human lung epithelial cells pre-treated with D. pigrum strain 040417, then exposed to live SARS-CoV-2. The bacterium upregulated IFN-β — the frontline antiviral interferon — and reduced pro-inflammatory chemokines (CXCL8, CCL5, CXCL10). At 48 and 72 hours post-infection, viral replication was significantly reduced. Cellular damage was reduced. But this was strain-specific: a different D. pigrum strain (030918) showed weak or no effect. The gatekeeper has variants, and not all keys fit the same lock.
Each study alone is suggestive. Together, they form a causal chain: D. pigrum suppresses the receptors that let virus in (Park), boosts the interferon response that fights virus off (Villena), and its absence predicts the chronic disease that follows (Ward).
Two Paths From the Same Infection
The gatekeeper thesis gives us something we've lacked: a mechanistic explanation for why two people can inhale the same virus and end up in utterly different places. One recovers in a week. The other is still sick two years later. The difference may begin in the nose — specifically, in whether D. pigrum was standing guard when the virus arrived.
The same virus, two trajectories. The fork begins in the nasal microbiome.
The right column isn't speculation. Each step links to specific evidence I've covered across 26 previous posts. Spike fragments functioning as xenoAMPs that kill plasmacytoid dendritic cells by membrane geometry — that's Post #26, drawing on Zhang & Wong's PNAS work. The resulting IFN-I collapse leading to monocyte reprogramming — that's Post #19 and Post #20, validated by Kumar et al. in Nature Immunology. The bone marrow epigenetic lock that makes it permanent — Post #21.
What the gatekeeper thesis adds is the upstream origin. The cascade doesn't start with spike fragments or monocyte reprogramming. It starts with whether the nose was defended.
The Antibiotic Catastrophe
This is where the story becomes tragic. During the first two years of the pandemic, empiric antibiotic prescribing for COVID patients was widespread. Azithromycin was the most common — prescribed both for suspected secondary infections and for its hypothesized anti-inflammatory properties. In some cohorts, over 70% of hospitalized COVID patients received antibiotics.
The Nature Microbiology data from Langelier et al. makes the damage quantifiable. Among 1,164 hospitalized patients:
| Metric | Azithromycin (n=366) | No Antibiotics (n=474) |
|---|---|---|
| Microbiome altered | Within 1 day | No |
| MLS resistance genes increased | Yes | No |
| Pathobionts enriched | Klebsiella, Staphylococcus | No |
| Changes persisted | >1 week | — |
| Anti-inflammatory benefit | None | — |
No anti-inflammatory benefit. No antiviral benefit. But it destroyed commensal bacteria — including the ones that Ward et al. showed were protective against PASC. We prescribed a drug that did nothing to fight the virus and everything to disarm the nose's own defenses.
This isn't hindsight bias. The RECOVERY trial had already shown by 2021 that azithromycin didn't reduce mortality or hospital stays in COVID. But prescribing habits lagged the evidence by years. And every prescription may have removed a gatekeeper.
The Ratchet Gets a Starting Point
In Post #25, I described the ratchet — the finding that each COVID reinfection tightens biological damage that never fully reverses. What I couldn't explain was where the ratchet begins. Now the nasal microbiome data offers a candidate.
SARS-CoV-2 itself depletes D. pigrum. Multiple studies have found that acute COVID infection reduces Dolosigranulum and Corynebacterium abundance, with the depletion proportional to disease severity. Severe and critical patients show the greatest loss of nasal commensals, accompanied by expansion of Staphylococcus, Prevotella, and Peptostreptococcus.
This creates a vicious cycle:
First infection depletes D. pigrum. Depleted D. pigrum means elevated ACE2/TMPRSS2. Elevated receptors mean higher viral load on reinfection. Higher viral load means more spike fragments, more immune damage, more depletion. Each turn of the ratchet makes the nose less defended.
This connects the nasal microbiome to the cumulative reinfection risk I mapped in Post #25 — the Quebec data showing 37% Long COVID after three infections, the RECOVER-EHR pediatric data showing PASC diagnosis doubling with each reinfection, the Singapore multi-organ hazard ratios that never attenuated over time. If each infection progressively depletes the gatekeeper, the ratchet has a microbial starting mechanism.
What D. pigrum Actually Is
A brief biological sketch, because this organism deserves to be more than an acronym in a findings table.
Dolosigranulum pigrum is the sole recognized species in its genus. Gram-positive. Firmicutes, family Carnobacteriaceae. It was first described in 1993, and for most of its known history, it was considered a benign commensal — present in about 41% of nasal samples, peaking in infancy and declining through adolescence. Its genome is small (under 2 megabases) and riddled with auxotrophies: it can't synthesize many amino acids, polyamines, or cofactors on its own. It depends on its neighbors.
That dependency is key. D. pigrum produces L-lactic acid in millimolar concentrations, which inhibits the pathobiont Moraxella catarrhalis — but to suppress Streptococcus pneumoniae, it needs Corynebacterium as a partner. The two together form a mutualistic pair that Ward et al. found co-depleted in PASC patients. This isn't a solo organism defending the nose; it's one node in a commensal network.
In animal models, intranasal D. pigrum before bacterial challenge reduced lung pneumococcal counts and accelerated innate immune recruitment. In RSV-infected mice, it altered respiratory and systemic cytokine profiles and lowered viral loads. It has CRISPR systems but no virulence factors across 34 sequenced genomes. This is a defensive specialist — a guard, not a weapon.
The Probiotic Frontier
The obvious question: can we put the gatekeeper back?
UCLouvain's team explicitly raised this possibility. They envision a nasal spray probiotic — administered before respiratory virus season — to restore D. pigrum in at-risk populations. The idea is biologically plausible. But reality imposes constraints.
D. pigrum's auxotrophies mean it may not survive alone as a probiotic formulation. It needs metabolic support from Corynebacterium and possibly other commensals. Any nasal probiotic might need to be a consortium — multiple organisms co-delivered to rebuild the commensal network, not just one species dropped into a depleted niche.
As of March 2026, no clinical trial has tested intranasal D. pigrum in humans for COVID prevention or Long COVID. The field is pre-clinical. A 2025 PNAS study from the Stubbendieck lab developed a computational framework showing that probiotic-based nasal decolonization of pathobionts outperforms antibiotics — and preserves microbiome diversity. A 2025 JCM study found that probiotic nasal rinses increased Dolosigranulum abundance in patients with chronic nasal inflammatory disease, though the change didn't reach statistical significance.
These are early signals, not proof of concept. But the therapeutic logic is coherent: if D. pigrum depletion opens the door to the entire Long COVID cascade, restoring it — before infection, not after — becomes upstream prevention rather than downstream damage control.
What This Changes
The gatekeeper thesis doesn't replace any of the mechanisms I've previously described. It adds the opening scene. The bone marrow trilogy (Posts #19–#21) explains how LC-Mo monocytes get reprogrammed and locked. Post #26 explains how spike fragments select which immune cells live and die. Post #25 explains how each infection compounds the damage.
What D. pigrum provides is the first domino. The question was always: why do some people generate enough spike fragments to trigger the xenoAMP mechanism while others don't? The answer may be that their nasal microbiome controlled how much virus got in, how many fragments were made, and whether IFN-β was present to fight back.
And the darkest implication: we may have knocked that domino over ourselves — with antibiotics prescribed in good faith, backed by a retracted study, continued out of habit, and paid for with the nasal defenses of millions of patients who trusted us to help them.
→ ACE2/TMPRSS2 elevated → higher viral load
→ Spike fragments generated → xenoAMPs kill pDCs (#26)
→ IFN-I collapse → monocyte reprogramming (#19)
→ CD38-NAD+ depletion loop (#20)
→ Autoantibody factories activated (#21)
→ NET-mediated microclots (#14) → mast cell neuropathy (#16)
→ Orexin neuron death (#17) → neurodegeneration (#22)
→ Cumulative with each reinfection (#25)
Twenty-seven posts. One cascade. And the gatekeeper sits at the top.
Sources
Ward B, et al. Association of nasopharyngeal Dolosigranulum pigrum and Corynebacterium species with post-acute sequelae of SARS-CoV-2 in a longitudinal cohort. Microbiology Spectrum, March 2026. DOI: 10.1128/spectrum.02313-25
Park J, et al. The nasal microbiome modulates risk for SARS-CoV-2 infection. eBioMedicine, 2025. DOI: 10.1016/S2352-3964(25)00104-5
Villena J, et al. Immunobiotic Dolosigranulum pigrum modulates innate-adaptive immune responses against SARS-CoV-2. Pathogens, 2021. PMC8224358
Langelier C, et al. Empiric azithromycin alters the upper respiratory microbiome and resistome without anti-inflammatory benefit in COVID-19. Nature Microbiology, March 2026. DOI: 10.1038/s41564-026-02285-8
Stubbendieck RM, et al. Computational and in vitro evaluation of probiotic treatments for nasal Staphylococcus aureus decolonization. PNAS, 2025. DOI: 10.1073/pnas.2412742122
Hurst JH, Stubbendieck RM, Kelly MS. Dolosigranulum pigrum: A promising nasal probiotic candidate. PLOS Pathogens, 2024. DOI: 10.1371/journal.ppat.1011955