Myrosinase

Broccoli does not contain sulforaphane. It contains glucoraphanin, and an enzyme stored in separate cells that can turn glucoraphanin into sulforaphane when the two meet. Whether that enzyme is still working when the food reaches you is the difference between a useful dose and almost none, and it has been measured in people often enough to say so with numbers rather than in principle.

What does myrosinase do?

Myrosinase hydrolyzes the bond that holds the sugar group onto a glucosinolate. In the case of glucoraphanin, the enzyme cleaves off glucose, a sulfate group leaves, and the remaining fragment rearranges into sulforaphane. Nothing else in the pathway is slow or uncertain. The hydrolysis is the step that either happens or does not.

In the intact plant it cannot happen, because myrosinase and its substrates are held in different cells. Chewing, chopping, blending or an insect's mandibles break both compartments at once, and the reaction runs where the damage is. That arrangement is a defense system with a trigger, and it is also why preparation method matters so much to anyone eating these plants on purpose.

How much does the enzyme change what you absorb?

Between about four and six fold in most of the trials that have asked, and higher in the one that compared raw broccoli against cooked. Six human trials have tested some version of the question. Three came from the same Johns Hopkins group, and the other three from three unrelated groups, one of which worked in Reading and another around Corvallis. They agree with each other.

Read the left column before the numbers, because the comparisons are not interchangeable. Shapiro 2006 and Shapiro 2001 both bypass the enzyme entirely by giving people sulforaphane that had already been formed in the laboratory, which sets an upper bound on what the conversion step costs. Okunade and Vermeulen are closer to a kitchen: broccoli cooked until the enzyme is dead, with and without a repair. Clarke is the supplement question asked directly, with the precursor dose matched to within 1 percent so that the enzyme was the only difference.

The agreement across those designs is the useful part. The same effect measured by unrelated groups, on different preparations, by several analytical methods, is harder to explain away than any single trial would be. The one figure that reaches 90 percent recovery is the one where no conversion was required at all, which is the clearest statement in this literature that the enzyme step, and not absorption, is where the dose is lost.

Do humans have myrosinase?

No. Human tissue does not produce the enzyme, so when the plant's own copy has been destroyed, there is one route left and it belongs to gut bacteria.

That route is real and it is not reliable. Shapiro's 2001 paper already described myrosinase as an enzyme of plants and bowel microflora, and Saha's 2012 trial showed directly that human colonic flora convert glucoraphanin, producing sulforaphane, erucin and the corresponding nitriles. The nitriles matter here: bacterial hydrolysis does not only run slower, it can send the fragment down a different path and produce a compound that is not sulforaphane at all.

How much any individual converts appears to depend on which bacteria they carry. The exploratory arm of a 2025 trial in prediabetes found that the adults who responded carried a higher abundance of a Bacteroides-encoded transcriptional regulator needed for the conversion, and that the abundance of its gene operon tracked their serum sulforaphane. That analysis is labeled exploratory in the paper and it is the first mechanistic account of a spread that has been visible in this literature since Fahey's group reported recoveries ranging from 1 to 40 percent of dose between individuals. It is an explanation offered, not an explanation established.

What destroys it?

Heat, mainly, and the thresholds have been measured. Peroxidase, lipoxygenase and myrosinase are all lost from broccoli at 86 °C and above. At 76 °C, 92 percent of lipoxygenase activity goes for an 18 percent loss of myrosinase-dependent sulforaphane formation, so a processor working at the lower temperature would keep most of the enzyme. That is laboratory work on frozen broccoli, never fed to anyone, and the practical reading of it is about what a factory could do rather than what a cook should.

Domestic cooking has been tested too, on 18 cabbage accessions, and the result is awkward in a way that single-answer advice tends to hide. Myrosinase was most stable after stir-frying, holding up to 65 percent of its activity, while steaming or microwaving destroyed over 90 percent of it in some accessions. But stir-frying also cost up to 70 percent of the glucosinolates, and steamed cabbage retained as much as 97 percent of them. The method that saves the enzyme throws away the substrate. The authors concluded that steaming produced the highest isothiocyanate concentrations, which is the outcome that matters, and that is the finding to take rather than the enzyme figure on its own. Note that this was cabbage, which Kushad's survey found to be sinigrin-dominant, so most of the isothiocyanate formed there is allyl isothiocyanate rather than sulforaphane.

Freezing on its own does little. Commercial blanching before freezing does a great deal, and the foods high in sulforaphane page covers the human trial behind that, where fresh broccoli soup delivered about ten times what frozen did.

Which foods are high in myrosinase?

Nobody can answer this from the published literature in a form worth putting in a table, and the sites that do answer it are not citing a measurement. Myrosinase activity in unextracted fresh tissue has been estimated for five broccoli and three kale cultivars, and that paper's abstract reports no per-cultivar figures, so this site carries none. The usual approach of isolating crude protein first, the authors note, may underestimate activity in a whole food.

The same paper settles a related question that gets answered wrongly. Broccoli myrosinase shows no selectivity between sinigrin and glucoraphanin: the products of the two correlated at r = 0.859, p = 0.006. The enzyme cuts whichever glucosinolate is in front of it. So a vegetable can be rich in myrosinase and yield almost no sulforaphane, because what the enzyme finds to work on is a different glucosinolate. Kale is the common example, and the foods page sets out why it keeps appearing on lists it does not belong on.

Two added sources have been tested rather than assumed. Brown mustard seed powder is the one tested in people, at 1 g with 200 g of cooked broccoli, and it quadrupled sulforaphane metabolite excretion in 12 adults. Daikon radish is the one tested in the laboratory, and its myrosinase holds up to heat far better: it survived 10 minutes at 125 °C, and 0.25 percent added to thawing frozen broccoli restored sulforaphane formation with no visible change to the vegetable. Nobody has fed the daikon version to a person and measured the result.

What this means for a capsule

Almost everything sold for sulforaphane contains glucoraphanin, because the precursor survives manufacturing and sulforaphane does not. The enzyme question is therefore the main thing separating two products with identical label figures, and Clarke's trial is the one that tested it with the dose matched: 5-fold lower excretion from the enzyme-free supplement than from sprouts carrying 218.4 µmol of the same precursor.

A label stating a myrosinase content is not stating a myrosinase activity. The enzyme is a protein, and heat or time can denature it while leaving it fully present to any assay that weighs it. What else a sulforaphane supplement label does and does not tell you is set out on the supplements page, and the doses human trials have actually used are collected on the how much sulforaphane per day page. For the figures behind the food side of this, with each row's assay beside it, see the sulforaphane content by food table.