Phycocyanin's Antioxidant Mechanism: Where the Biliverdin Reductase Story Is Established, and Where It Stops
The phycocyanobilin-to-phycocyanorubin pathway is real enzymology. The leap from that pathway to an antioxidant claim on a finished product crosses four steps, and only the first is settled. A dose-arithmetic check on what the rodent data would actually require.
Executive Summary
The mechanism usually cited for phycocyanin's antioxidant activity runs like this. Phycocyanobilin — the chromophore that makes the protein blue — is a structural near-relative of biliverdin. Mammalian biliverdin reductase reduces it to phycocyanorubin, a homolog of bilirubin. Bilirubin inhibits NADPH oxidase. Therefore phycocyanin is an antioxidant.
Each link in that chain is a different quality of evidence, and the chain is usually presented as though it were one claim.
Link 1 — the enzymology — is settled. Purified rat liver biliverdin reductase converts phycocyanobilin to a rubinoid product, with Km and Vmax "very similar to those of the natural substrate, biliverdin IX α"[1]. That is a clean 1993 result and nothing has overturned it.
Link 4 — the human outcome — is not. The most recent meta-analysis of controlled clinical trials found no significant effect of spirulina on malondialdehyde (SMD −0.42; 95% CI −0.98 to 0.14; P = .14), and only marginal, non-significant trends on IL-6 and TBARS[7]. Eleven studies, 465 subjects — all of whole spirulina, none of isolated C-phycocyanin.
Between those two points sits the question a formulator actually has to answer: how much phycocyanobilin reaches plasma, and at what dose. In 2007, the hypothesis paper that launched this whole line of reasoning stated its own load-bearing assumption in plain text — that the extrapolation holds "assuming that humans and rodents digest and absorb Spirulina-bound PCB in a comparable manner"[2]. Nineteen years later that assumption still has not been tested in humans.
What follows is the chain, link by link, and the dose arithmetic that falls out of it. The arithmetic is the part that changes purchasing decisions.
1. The chromophore is not free, and that matters first
In C-phycocyanin, the pigment is not dissolved in the protein. It is covalently bonded to it, and that single structural fact governs everything downstream.
Each αβ monomer of C-phycocyanin carries three phycocyanobilin chromophores, each attached through a thioether bond to a cysteine residue — one on the α subunit at Cys-84, two on the β subunit at Cys-82 and Cys-153 (residue numbering varies slightly by organism)[8]. In the cell, dedicated lyases are required to make those bonds.
This is a real distinction from most botanical actives. Curcuminoids, carotenoids and polyphenols are non-covalently associated with their matrix and are released by solvent. Phycocyanobilin is not released by solvent. It requires either proteolysis or a chemical cleavage step — methanolysis is the standard laboratory route.
So the first question in the chain is not "is phycocyanobilin absorbed?" It is "does digestion liberate phycocyanobilin from the apoprotein at all, and how efficiently?" Gastric and pancreatic proteases will degrade the apoprotein. Whether that degradation cleaves the peptide bonds adjacent to the chromophore- bearing cysteines cleanly enough to release free or near-free phycocyanobilin, and in what yield, is not something the supplement literature has quantified.
The prodrug step is assumed, not measured.
2. The enzymology: the one link that is actually nailed down
Terry, Maines and Lagarias ran this in 1993 with purified rat liver biliverdin reductase and three bilin substrates[1].
Phytochromobilin and phycocyanobilin were "particularly good substrates" — kinetics comparable to biliverdin IX α itself. Phycoerythrobilin was the least preferred. ¹H NMR of the product confirmed that the enzyme selectively reduces the C10 methine bridge without touching the A-ring ethylidene substituent, which is exactly the transformation that turns a biliverdin-type molecule into a bilirubin-type one.
Two honest qualifications that the citing literature almost never carries.
First, this was purified enzyme in vitro, in rat. It establishes that the reaction is chemically and kinetically feasible. It is not a demonstration that phycocyanorubin is formed in human tissue after an oral dose, at a meaningful concentration, in the compartment where you would want it.
Second, the paper is titled "Inactivation of..." The authors were plant photobiologists. From their angle, biliverdin reductase destroys the chromophore: the rubinoid products "do not form photoactive adducts with recombinant apophytochrome." The reaction the nutraceutical hypothesis depends on was first characterised as a degradation pathway. Both framings are correct. It is a good reminder that a citation supports the sentence it was written for, not the sentence you want.
3. What "inhibits NADPH oxidase" was measured on
The animal work is real and it is consistent in direction. It is also narrower than the summaries suggest.
| Study | Model | What was dosed | What was measured |
|---|---|---|---|
| Riss 2007[6] | Hamster, atherogenic diet, 12 wk | Phycocyanin and Se-enriched phycocyanin by gavage | Aortic fatty streak area, cardiac superoxide, p22phox expression |
| Zheng 2013[3] | db/db mouse, 10 wk | C-phycocyanin 300 mg/kg; PCB 15 mg/kg | Albuminuria, mesangial expansion, TGF-β, NAD(P)H oxidase component expression |
| Zheng 2013 (in vitro)[3] | Cultured renal mesangial cells | PCB, bilirubin, biliverdin | NADPH-dependent superoxide production |
| García-Pliego 2021[5] | Mouse, HgCl₂ acute kidney injury | Purified PCB, 0.75 / 1.5 / 3 mg/kg/d | Oxidative stress markers, antioxidant enzymes, caspase activity |
Three things to notice.
Expression is not inhibition. Riss measured a 34% decrease in p22phox subunit expression[6] — the enzyme complex being built less, not the existing enzyme being blocked. The McCarty hypothesis is about direct inhibition by a bilirubin homolog[2]. These are different mechanisms with different dose-response behaviour, and the literature routinely merges them into "reduces NADPH oxidase."
The cleanest NADPH-oxidase result in Riss came from the selenium arm. That 34% figure was for Se-enriched phycocyanin specifically, not plain phycocyanin. Plain PC did produce the large fatty-streak reduction (82%), so the vascular result is not selenium-dependent — but the mechanistic marker that the hypothesis predicts was strongest in the arm carrying a second known antioxidant cofactor[6].
The prodrug question has been tested once, properly, and it came out in favour of the hypothesis. García-Pliego 2021 purified PCB from C-phycocyanin, characterised it by MS, and dosed it directly in a mercury AKI model. All doses prevented alterations in oxidative stress markers and caspase-9 activity; only 3 mg/kg/d prevented the full damage picture[5]. That is the right experiment, and it supports PCB as the active moiety — in mouse.
4. The dose arithmetic, which is where this becomes a purchasing question
The working is shown so it can be checked.
Step 1 — how much phycocyanobilin is in C-phycocyanin. Three PCB per αβ monomer[8], PCB free acid at 586.7 g/mol, monomer mass roughly 37 kDa. That gives ≈4.8% PCB by mass of pure C-phycocyanin.
Cross-check: McCarty put 15 g of whole spirulina at about 100 mg PCB — 0.67%[2]. At a typical 14% C-phycocyanin content, 0.14 × 4.8% = 0.67%. The two numbers, derived independently, agree. The 14% is a typical trade figure, not a cited value, so treat this as a consistency check.
Step 2 — convert rodent doses to human equivalents. Standard FDA body-surface-area allometry: divide a mouse mg/kg dose by 12.3[10], then multiply by 70 kg.
Step 3 — express the result as material you would actually buy. Viridia's published E18 specification is colour value ≥180 CU and pure phycocyanin ≥24%.
| Source dose | Species | Human-equivalent PCB/day | As pure C-PC | As E18 powder (24% C-PC) |
|---|---|---|---|---|
| PCB 3 mg/kg/d (García-Pliego, AKI)[5] | mouse | 17 mg | 0.36 g | ~1.5 g |
| C-PC 300 mg/kg/d (Zheng)[3] | mouse | (82 mg equiv.) | 1.7 g | ~7.1 g |
| PCB 15 mg/kg/d (Zheng)[3] | mouse | 85 mg | 1.8 g | ~7.4 g |
| 2 tbsp spirulina/d (McCarty's recommendation)[2] | human extrapolation | ~200 mg | 4.2 g | ~17 g |
| PCB 50 mg/kg single dose (PK study)[4] | mouse | 285 mg | 5.9 g | ~25 g |
Two observations a formulator can use immediately.
The internal consistency check passes. Zheng's two arms — 300 mg/kg of the holoprotein and 15 mg/kg of the free chromophore — land at 1.7 g and 1.8 g of C-phycocyanin equivalent. Those were run as independent interventions, and they converge. That is a point for the prodrug model and for my 4.8% figure.
The range is seventeen-fold, and every figure is well above a typical supplement serving. 17 mg to 285 mg of PCB per day. The low end needs about 1.5 g of E18 powder daily; the high end needs 25 g. A capsule typically holds a few hundred milligrams of powder, so even the lowest figure from the rodent data needs several capsules of E18-grade material a day.
These conversions are my arithmetic, not values reported in the cited papers. Body-surface-area allometry is a scaling heuristic for first-in-human dose selection, not a bioequivalence method, and it assumes comparable absorption — the assumption Section 5 shows is unsupported. Treat the table as an order-of- magnitude sanity check on dose plausibility, not as a dosing recommendation. Do not put these numbers on a label.
5. The pharmacokinetics, finally measured — and what they imply
In November 2025, a group at Instituto Politécnico Nacional published the first validated LC-MS/MS method for quantifying phycocyanobilin in plasma, and applied it[4]. This is the missing link, and it is worth reading carefully.
Their own framing is the important part: C-phycocyanin "is a prodrug catabolized to phycocyanobilin when it is administered orally," and "the PCB pharmacokinetic and pharmacodynamic profile remains unknown." That was the state of the field in late 2025.
After a single oral dose of 50 mg/kg PCB in mice:
A 50 mg/kg oral dose producing a peak plasma concentration of 40 ng/mL is a very small number. And the apparent volume of distribution and apparent clearance are the tell: both are apparent values, divided by bioavailability F. Figures of 5,709 L/kg and 1,168 L/h/kg are not physiological volumes or flows. They are what you get when F is very small and the model has nowhere else to put the discrepancy. Extensive tissue distribution contributes — PCB binds human serum albumin with high affinity — but the magnitude here points at low oral bioavailability as the dominant term.
A 2.5-minute Tmax is also unusual and deserves a flag rather than an explanation from me. It suggests near-immediate appearance in plasma, which is hard to square with slow intestinal uptake. Note this study dosed purified PCB, not C-phycocyanin — so it characterises the chromophore's own kinetics and leaves the release step from the holoprotein still unmeasured.
This is a mouse study, in one laboratory, at one dose, single administration. It is a methodological advance and the correct next step. It is not a human bioavailability figure, and nobody should present it as one.
6. The bilirubin arm, which is stronger than the phycocyanin arm
The part of the hypothesis with real human support is the part that is not about phycocyanin at all.
People with Gilbert syndrome carry chronically mildly elevated unconjugated bilirubin, and epidemiological studies have associated that state with lower coronary artery disease risk and higher serum antioxidant capacity. Bilirubin has also been reported to inhibit NADPH oxidase. That evidence concerns the body's own molecule.
It is also not evidence about phycocyanin. Mapping it across requires that oral C-phycocyanin produces circulating phycocyanorubin at concentrations approaching the bilirubin levels seen in Gilbert syndrome — commonly in the region of 1 to 3 mg/dL, which is 10,000 to 30,000 ng/mL. Set that against a 40 ng/mL PCB Cmax in mice at 50 mg/kg[4], and the gap is not a detail to be closed by a better formulation.
I am not arguing the comparison is fatal. Phycocyanorubin's potency at NADPH oxidase could differ from bilirubin's; local tissue concentration is not plasma concentration; and PCB, not PCR, was what was measured. But anyone invoking Gilbert syndrome in support of a phycocyanin claim should be asked to state that concentration gap out loud.
7. What you can defensibly say, and what you cannot
The single largest gap in this whole literature: there is no completed human trial of isolated C-phycocyanin with oxidative-stress endpoints. Every entry in the Mohiti meta-analysis is whole spirulina[7]. One properly designed trial of a purified C-phycocyanin preparation exists — PROPERTY (NCT05025826), 110 patients across seven French hospitals, testing phycocyanin against oxaliplatin-induced peripheral neuropathy[9]. The protocol was published in 2023. We did not find published results as of September 2026. When they do, they will be the most informative single data point in this field, in either direction.
| Statement | Status | Basis |
|---|---|---|
| "Phycocyanobilin is a substrate for mammalian biliverdin reductase" | Established | Purified enzyme kinetics, comparable to biliverdin IX α[1] |
| "Phycocyanobilin inhibits NADPH-dependent superoxide production in cell culture" | Established, in vitro | Cultured renal mesangial cells[3] |
| "PCB is the active moiety of orally administered C-phycocyanin" | Supported in rodents | Purified-PCB dosing reproduces the holoprotein effect[5] |
| "Oral C-phycocyanin delivers phycocyanobilin to human plasma" | Unmeasured | Mouse PK only, purified PCB, single dose[4] |
| "Phycocyanin reduces oxidative stress in humans" | Not demonstrated | Null on MDA; marginal on IL-6/TBARS; whole spirulina only[7] |
| "Phycocyanin is an antioxidant" (unqualified, on-pack) | Not supportable | Chain breaks at absorption and at human outcome |
For a B2B buyer, the practical consequence is narrow and specific: structure your substantiation file around the mechanism as a mechanism, and keep outcome language out of it. A dossier that says "phycocyanobilin is a biliverdin reductase substrate, with the following in vitro and rodent evidence, and the following stated gaps" will survive a regulatory review. A dossier that compresses that into "clinically proven antioxidant" will not. The chain traces back to a 2007 paper that presented itself as a hypothesis.
Where Viridia sits on this
We sell phycocyanin in E6 to E40 grades, and most of it goes into colour applications, where the performance question is measurable and the evidence is our own batch data.
A claim that cannot survive the buyer's own regulatory review creates problems for both sides later, and the mechanism is interesting enough without inflating it.
Our E-grade specifications state the pure-phycocyanin percentage, so the arithmetic in Section 4 can be run against the actual material rather than a nominal grade name. Phycocyanobilin content is not part of a standard phycocyanin CoA. If your substantiation depends on it, it needs a contract-laboratory methanolysis-and-HPLC assay, commissioned for that purpose.
When the PROPERTY results publish, they will change this picture in one direction or the other.
See the phycocyanin grades and specifications →
Request batch CoAs or the pure-phycocyanin percentage for a specific grade →
References
[1] Terry MJ et al. (1993). Inactivation of phytochrome- and phycobiliprotein-chromophore precursors by rat liver biliverdin reductase. Journal of Biological Chemistry 268(35):26099–26106; PMID 8253726; erratum J Biol Chem 1994;269(14):10965. Retrieved from https://pubmed.ncbi.nlm.nih.gov/8253726/
[2] McCarty MF (2007). Clinical potential of Spirulina as a source of phycocyanobilin. Journal of Medicinal Food 10(4):566–570; PMID 18158824 (review / hypothesis paper). Retrieved from https://pubmed.ncbi.nlm.nih.gov/18158824/
[3] Zheng J et al. (2013). Phycocyanin and phycocyanobilin from Spirulina platensis protect against diabetic nephropathy by inhibiting oxidative stress. Am J Physiol Regul Integr Comp Physiol 304(2):R110–R120; doi 10.1152/ajpregu.00648.2011; PMID 23115122. Retrieved from https://pubmed.ncbi.nlm.nih.gov/23115122/
[4] García-Pliego E et al. (2026). Pharmacokinetic profile of Arthrospira maxima-derived phycocyanobilin. Development and application of a novel bioanalytical method. Journal of Applied Phycology 38:315–324; doi 10.1007/s10811-025-03728-1. Retrieved from https://link.springer.com/article/10.1007/s10811-025-03728-1
[5] García-Pliego E et al. (2021). Phycocyanobilin is the molecule responsible for the nephroprotective action of phycocyanin in acute kidney injury caused by mercury. Food & Function 12:2985–2994; doi 10.1039/D0FO03294H; PMID 33704296. Retrieved from https://pubmed.ncbi.nlm.nih.gov/33704296/
[6] Riss J & et al. (2007). Phycobiliprotein C-phycocyanin from Spirulina platensis is powerfully responsible for reducing oxidative stress and NADPH oxidase expression induced by an atherogenic diet in hamsters. Journal of Agricultural and Food Chemistry 55(19):7962–7967; PMID 17696484. Retrieved from https://pubmed.ncbi.nlm.nih.gov/17696484/
[7] Mohiti S & et al. (2021). Spirulina supplementation and oxidative stress and pro-inflammatory biomarkers: A systematic review and meta-analysis of controlled clinical trials. Clinical and Experimental Pharmacology and Physiology 48(8):1059–1069; doi 10.1111/1440-1681.13510. Retrieved from https://onlinelibrary.wiley.com/doi/10.1111/1440-1681.13510
[8] Zhou W & et al. (2018). Molecular cloning of cpcU and heterodimeric bilin lyase activity analysis of CpcU and CpcS for attachment of phycocyanobilin to Cys-82 on the β-subunit of phycocyanin in Arthrospira platensis. PMC6273044. Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6273044/
[9] PROPERTY study group (2023). PROPERTY: study protocol for a randomized, double-blind, multicenter placebo-controlled trial assessing neurotoxicity in patients with metastatic gastrointestinal cancer taking PHYCOCARE® during oxaliplatin-based chemotherapy. Trials 24:47; doi 10.1186/s13063-023-07071-z; ClinicalTrials.gov NCT05025826. Retrieved from https://pmc.ncbi.nlm.nih.gov/articles/PMC9854012/
[10] US Food and Drug Administration (2005). Guidance for Industry: Estimating the Maximum Safe Starting Dose in Initial Clinical Trials for Therapeutics in Adult Healthy Volunteers (body-surface-area conversion; mouse Km factor 12.3). FDA CDER, July 2005. Retrieved from https://www.fda.gov/media/72309/download
More from Viridia
Keep exploring
Wild-Harvested Blue-Green Algae Is Often Not Spirulina
The main wild-harvested blue-green algae product, from Upper Klamath Lake, is a different genus, harvested from a lake where a microcystin producer blooms. Cultivated Limnospira has no genetic capacity to make microcystins. Any microcystin in a spirulina lot comes from a contaminant, and 'cultivated' is not a specification on its own.
Erinacines, Hericenones, and Which Lion's Mane the Trials Actually Used
The two headline Lion's Mane compound families sit in different tissues. Hericenones failed a human-cell NGF assay. What a CoA should carry.
What "Ganoderma lucidum" Actually Means on a Reishi CoA
Almost no commercial reishi is Ganoderma lucidum. Three binomials, an unresolved nomenclature dispute, a measured chemical difference, and what to specify instead.