Tremoço Salad with Queijo Minas

What it is

Tremoço, the lupin bean, bought the easy way: cooked, debittered and waiting in a jar of brine. You drain it, rinse it, taste one and soak it for half an hour to pull some salt back out. Then it meets a sharp dressing of olive oil, red wine vinegar, garlic and oregano, and the vinagrete crowd follows: tomato, red onion, green pepper, parsley and scallions. Cubes of queijo minas frescal on top, a crusty pão francês on the side. You won't light a single burner.

Tremoço usually gets filed under snacks. Lusíadas Saúde, a Portuguese health group, calls it the snack of Portuguese summer afternoons, and in Brazil the consumer association PROTESTE says it's still little explored in the kitchen. So I'm giving it a plate. It's Protein Champion number eight because, gram for gram, cooked lupins carry about 1.8 times the protein of number one's black beans, and its teammate is the humble bread roll.

One plain safety line before you shop. Lupin is a legume in the same plant family as peanuts, and the FDA says people who are allergic to peanuts may also react to lupin, sometimes severely. Some people react to it without a peanut allergy. If either could be you, skip this one and talk to your doctor or allergist.

Why you should eat it

The short version, before tremoço gets filed under snacks again. USDA FoodData Central lists 15.6 g of protein per 100 g of cooked lupins (record 172424), about 1.8 times cooked black beans' 8.86 g (record 173735), and each plate gets 100 g. That same 100 g carries just 9.88 g of carbohydrate and 2.92 g of fat. Fiber, by the database: 2.8 g. Full honesty, I think that figure runs low, because it doesn't line up with USDA's own raw lupin record, so let the vinagrete and the roll top the plate up. The fat comes mostly from the cheese and the olive oil, the saturated part mostly from the cheese. The watch-out is salt: USDA's salted record lists 240 mg of sodium per 100 g, and jarred tremoço sits in salty brine, so the method rinses and soaks. If you're managing your blood pressure, read the jar's label and take it to your doctor. Why a bread roll belongs here comes while the beans soak.

Tips

• Taste before anything. Salty and mild is right. A bitter taste can mean the toxic alkaloids weren't fully removed, and the German Federal Institute for Risk Assessment (BfR) says not to eat them.
• Buy tremoço ready to eat. Raw dried lupins need days of careful debittering: BfR advises buying lupins already debittered unless you really know the process, and one case in a 2025 review of lupin poisonings describes soaking with a 10-minute boil every day for at least 5 days. If you ever do debitter, never drink the bitter soaking water or cook with it (BfR).
• Skins on or off, your call. Lusíadas Saúde says the fiber is higher in the skin, so keeping it on is the fiber-friendly choice. Slipping the skins off gives a softer salad, and it's a lovely job to hand to whoever turns up in the kitchen.
• Short on time? Skip the soak and rinse twice instead. The plate will be saltier, so leave the extra salt out entirely.
• The pão francês is a teammate, not a garnish. Check your bakery's ingredient list before you buy, and if the rolls list lard, pick another plain crusty bread.

Substitutions

• Queijo minas frescal: Any fresh white cheese made with non-animal rennet, or leave it off Leaving it off takes some protein and saturated fat with it, and the lysine top-up from 'While it soaks' goes too. A different cheese changes the card.
• Pão francês: A wholemeal roll or any plain crusty bread Still a grain teammate for the sulfur pair, though I haven't checked wholemeal's amino acid shares here. The card is built on USDA's hard roll, so it won't match.
• Tremoço, for anyone who can't eat lupin: Drained cooked chickpeas The vinagrete, cheese and roll still work, with less protein on each plate, and it becomes a different champion's plate. If you avoid lupin because of an allergy, ask your allergist before you swap in another legume: no source here checked whether a lupin allergy carries over to chickpeas.

Storage

Opened jar: keep leftover tremoço in the fridge, still covered by its own brine (Lusíadas Saúde). Dressed salad: into the fridge within 2 hours and eat within 3 to 4 days (Nebraska Extension), though in my kitchen judgment it's at its best on day one, before the tomatoes go soft.

While it soaks: the scarcest ingredient sets the batch

Tremoço in the water, dressing whisked, thirteen minutes with nothing to do. Pull up a chair.

Quick refresher from number one: protein is built from amino acids, and nine of them have to come from food. Every plant protein carries all nine. The mix is what changes.

So here's lupin's mix. USDA FoodData Central's record for cooked lupins lists 15.6 g of protein per 100 g, a generous load for a bean. Of every 100 g of that protein, 5.33 g is lysine, the amino acid grains run short on, and 1.94 g is the sulfur pair, methionine plus cysteine.

Now the yardstick. The FAO, the UN's food agency, publishes how much of each essential amino acid a gram of protein should carry for older children, teenagers and adults. I held cooked lupin up against it: eight of the nine clear the line. The sulfur pair doesn't, at 84% of the reference. A 2022 review of white lupin says the same thing in words: low in the sulfur pair, high in lysine.

Why does one short bar matter? Because on paper, a protein is scored by its scarcest essential amino acid. Here's the picture I'd draw on the board: brigadeiros. You can only roll as many as your condensed milk allows, and three more bags of chocolate sprinkles won't get you a single extra sweet. Lupin has sprinkles to spare. Its condensed milk is the sulfur pair.

So the bread roll brings a second tin. It carries 3.96 g of the sulfur pair per 100 g of its protein, about twice lupin's share, while lupin carries about twice the roll's lysine. Four rolls beside 400 g of tremoço put the pair at 108% of the reference, by my arithmetic.

One wrinkle, because grains will be grains: the roll thins the lysine a touch, to 95% of the reference. That's the queijo minas's job. USDA lists no amino acids for its fresh cheese, so I borrowed its cottage cheese record as a stand-in, and by that estimate the whole plate clears every line, lysine and the sulfur pair both around 113 to 115%.

Where the picture stops. These are shares of the protein sitting in the food, before your gut has its say: the FAO's real score also needs digestibility, and nobody here measured it, so I won't hand you a score. Your body doesn't roll one batch per meal, either: Mariotti and Gardner, the reviewers from number one, found no evidence yet that plant proteins need a partner in the same meal. And lupin brings so much protein that its thin share still adds up, 0.30 g of methionine plus cysteine in 100 g of cooked lupins.

So, in one line: lupin brings the protein and the lysine, the bread brings the sulfur pair, the cheese tops the lysine back up, and together the plate clears the line.

Case one: the review that counted the lupin trials

Three researchers searched three big medical databases for controlled trials of lupin in adults, published from 2000 to September 2021. They found 21 papers and 998 people: about a quarter of them healthy, the rest living with type 2 diabetes, high cholesterol or overweight. The trials ran from a single meal to 12 months, mostly in Australia, with others in Germany, Italy and Ecuador. Twelve tested whole lupin, four lupin protein and five lupin fiber, and most of it was Australian sweet lupin flour baked into bread, pasta or biscuits, or stirred into drinks.

The trials were too different to pool into one number, so the team counted instead. Feeling full went up with lupin in 5 of the 6 trials that measured it. Blood pressure improved in 5 of 7. Blood fats improved in at least one way in 7 of 11, and blood sugar control in 6 of 13. No trial saw a change in markers of inflammation. Whole lupin did the most: 9 of its 12 trials found a difference in at least one marker, and it went 4 for 4 on fullness.

For your plate: the steadiest signal was fullness, and it came from whole lupin.

Where it stops: almost none of this was a jar of tremoço. It was mostly flour in baked goods, so stretching it to your salad is my reading, not theirs. No pooled number, small trials, English-language papers only, and the authors say more research is needed before anyone makes health claims. No outside funding, but the review was supported by the Grains & Legumes Nutrition Council, a not-for-profit charity that employed one of the authors while the search ran.

Case two: lupin against whey after a pasta lunch

Twelve healthy young adults in Cologne, Germany, five men and seven women aged 23 to 33, came in for three test lunches at least a day apart, in a balanced order. Each lunch was the same ready-made spaghetti with meat sauce, the study's standard meal, sized to body weight. It came plain, or with whey protein isolate stirred in, or with sweet white lupin flour dosed to match the whey's protein. Then blood tests, out to three hours.

In the first hour, the blood sugar curve came out 46% smaller with whey and 54% smaller with lupin than with the plain pasta, with no real difference between the two. Over three hours, both sat 35% below the plain meal. First-hour insulin ran 39% higher after whey than after lupin, though neither differed from the plain pasta.

For your plate: in this test, lupin flour, dosed to match the whey's protein, softened the first-hour sugar rise from a carb-heavy lunch about as well as whey, the gym's favorite powder, with a smaller insulin bump than whey.

Where it stops: twelve healthy people, one lunch per condition, lupin flour in a sauce rather than whole beans in a salad, and a trial registered after the fact. The lupin meal also carried a little more fat and energy than the whey meal, and the authors say the mechanism is unknown. For a 70 kg person the lupin dose was about 26 g of protein by my arithmetic, roughly what 166 g of cooked lupins hold; one plate here has 100 g. It came from the German Aerospace Center, with a PhD scholarship from the Helmholtz Association, and the authors declare no conflicts. If you manage your blood sugar, take this to your doctor before you change your plate around it.

Case three: the leg in the brace

Same lab as case two, a different trial, and four of the same authors. Thirteen healthy young men, average age 26, wore a brace called Hephaistos on one leg for 60 days, 8 to 16 hours a day. It takes the load off the calf while you walk. A coin toss split them: six got nothing extra, and seven got electrical stimulation of the calf, 20 minutes twice a day, plus 19 g a day of a 'lupin crunchy' mixed into their food. That's about 7.8 g of lupin protein a day by my arithmetic, roughly what 50 g of cooked lupins holds.

After 58 days the calf had shrunk 11.9% in the men with nothing extra and 3.5% in the men with stimulation and lupin, a difference the statistics cleared (p = 0.044 in the results). Strength with the knee bent dropped 22.3% against 9.7%. With the knee straight, both groups lost strength and the gap disappeared.

For your plate: lupin protein rode along with a muscle-saving routine that worked in a tiny trial. It can't take the credit alone.

Where it stops: the authors say plainly they can't tell how much came from the stimulation and how much from the lupin, and that separating the two is the next study. Thirteen men, one leg each, no dummy powder and no fake stimulation, and the trial was registered after the men had already enrolled. The German Aerospace Center paid for it from its own budget, the stimulator came from its maker, and the authors declare no competing interests. A hint about a combination, not a verdict on lupin.