Roasted Vegetables with Green Pepita Sauce

What it is

Cauliflower, sweet potato, zucchini, red peppers and red onion, roasted hot on two trays until the edges brown. Underneath them goes a thick green sauce: toasted pumpkin seeds blended with cilantro, jalapeño, garlic, lime juice, a little olive oil and cold water. On top, crumbled queso fresco, the toasted seeds you kept back and a few cilantro leaves.

Pepitas are pumpkin seeds with the shell already off. They usually get sprinkled on top of things. Here they're the base of the plate, and most of them disappear into the sauce. I'm lacto-vegetarian, so this is a platter I'd happily put in the middle of my own table, and the only thing on it that comes from an animal is the cheese, made with non-animal rennet.

It's also Protein Champion number thirteen. Pumpkin seed protein runs short on one amino acid, lysine, and this whole plate is built to cover it. One of the helpers is a vegetable you'd never suspect.

Why you should eat it

Here's the short version, before anyone files pumpkin seeds under salad decoration. USDA FoodData Central lists 30.2 g of protein per 100 g of raw pepitas (record 170556), with 6 g of fiber in that same 100 g. Each plate gets about 40 g of seeds, so roughly 12 g of protein from them alone, plus about 5.5 g from the queso fresco (18.1 g per 100 g, same database). The trays carry most of the fiber, with the sweet potato and cauliflower bringing about half of it. The cheese brings salt: USDA lists 751 mg of sodium per 100 g of queso fresco. The fine print: pepitas are almost half fat, 49 g per 100 g in that record, so with the olive oil this is a rich plate. And the seed's protein has one soft spot, an amino acid called lysine. The cauliflower and the cheese are here to help with it, and I'll show you how while the trays roast.

Tips

• Stay with the pan while the seeds toast. Pull them the moment they puff, start to pop and turn golden at the edges.
• Two trays, one layer each, and a stir every 10 minutes, the rhythm UNH Extension gives for roasting. Swapping the trays at each stir gives both of them a turn near the top.
• Like heat? Leave a few jalapeño seeds in the sauce.
• The sauce can be made a day ahead and kept covered in the fridge (see Storage). Blend it on Sunday and Monday's dinner is just the trays.
• If you're pregnant, 65 or older, or have a weakened immune system: the CDC advises avoiding unheated queso fresco-type cheeses, even when they're made with pasteurized milk. Use feta made with pasteurized milk instead (non-animal rennet, check the label), which the CDC lists as a safer choice. Everyone else: pick a queso fresco made with pasteurized milk.
• Want a plant partner for the lysine as well? Champion number one's black beans carry 6.86 g of lysine per 100 g of their protein (USDA FoodData Central 173735), so a ladle of them beside this plate would add more. That's a sum on paper, not a tested pairing: no study here put pepitas and beans together, and the card doesn't count them.

Substitutions

• Queso fresco: Feta made with pasteurized milk and non-animal rennet (check the label) The CDC lists pasteurized feta among its safer choices for people at higher risk of Listeria. It's saltier: USDA lists 1,140 mg of sodium per 100 g for feta (173420) against 751 mg for queso fresco (172223), so go easy on the salt. The card is built on queso fresco, so it won't match.
• Raw pepitas: Roasted unsalted pepitas Skip the toasting step. USDA's roasted unsalted record (170557) carries the same amino acid shares (4.09 g of lysine per 100 g of protein). Salted roasted pepitas (169415) list 256 mg of sodium per 100 g against 7 mg raw, so cut the sauce salt if that's what you have. The card is built on raw.
• Sweet potatoes: Peeled butternut squash, in the same size cubes UNH Extension lists peeled winter squash such as butternut among the vegetables for the same roast. The card is built on sweet potato, so it won't match.
• Cilantro: Flat-leaf parsley A milder green sauce for anyone who finds cilantro soapy. No record was run for parsley here; the card is built on cilantro.

Storage

Cool leftovers quickly and get them into the fridge within 2 hours of cooking. Keep the vegetables, the sauce and the cheese in separate covered containers, and eat them within 3 to 4 days (USDA FSIS, Leftovers and Food Safety). Reheat the vegetables until they reach 165°F (74°C), as FSIS advises; the green sauce is served cold. If you're in one of the higher-risk groups, read the CDC line in the tips before eating the queso fresco cold.

While it cooks: the ingredient that runs out first

Sauce blended, trays roasting, six minutes with nothing to do. Pull up a stool.

Your body builds protein to a fixed recipe of amino acids, and nine of them have to come from food. A bag of seeds tells you how many grams of protein are inside. It doesn't tell you which amino acid runs out first, and on paper that's the one that sets the limit for the whole batch.

Picture a batch of pão de queijo. You can have a mountain of grated cheese, but once the polvilho runs out, that was the last tray. The short ingredient decides how many trays go in, and the extra cheese just waits on the counter.

For pepita protein, the short ingredient is lysine. USDA FoodData Central's record for raw pepitas works out to 4.11 g of lysine per 100 g of protein. The review in case one quotes an adult requirement of 4.5 from WHO, FAO and UNU, and its own pumpkin seed sample came in lower still, at 2.7. Two sources, two numbers, same side of the line.

Plot twist: methionine, the amino acid case one finds short in lentil, chickpea, pea and soy protein, is the comfortable one here. USDA's record gives 2.00 g per 100 g of protein, case one gives 1.9, and the requirement it quotes is 1.6.

So the plate brings help. Counting the seeds and the vegetables only, by my arithmetic the tray's protein comes to 5.26 g of lysine per 100 g, over that 4.5 line, and the cauliflower supplies about a third of that lysine. USDA lists 11.3 g of lysine per 100 g of cauliflower protein. There isn't much protein in a cauliflower, but what's there is lysine-rich.

The cheese isn't in that sum, because USDA has no amino acid values for queso fresco. Case one gives animal proteins 5.3 to 9.0% lysine, and the fresh cheeses USDA does measure, cottage cheese and feta, work out to about 8.4 and 8.6, roughly twice the pepita's share. My reading: the cheese helps too, and I can't put a number on how much.

The picture stops in two places. In the kitchen, every gram of polvilho ends up in a tray. In your body, digestion takes its share first, and every number above is counted in the food, before that happens. The review in case one says it plainly: what your muscles do with a meal can't be predicted from its amino acid makeup alone. And paper isn't blood: in case two, a pea and pumpkin protein blend built to match whey on paper still peaked lower in the blood.

So, in one line: pepita protein runs short on lysine and is fine on methionine, and the cauliflower and the cheese are on this plate to cover the gap.

Case one: where pumpkin seed protein sits in the lineup

Four researchers at Maastricht University, in the Netherlands, wrote this review from a talk one of them gave to the Gatorade Sports Science Institute's expert panel. Nobody ate anything for it: they lined up the amino acid makeup of plant and animal proteins against the adult requirements from WHO, FAO and UNU, and gathered the few human studies on plant protein and muscle.

The lysine lineup: plant proteins ran from 1.4 to 6% of their protein, animal proteins from 5.3 to 9.0%. Pumpkin seed sat at 2.7%, under the 4.5% requirement, next to oat (2.1), brown rice (2.4) and quinoa (3.3). On methionine it changes sides: 1.9% against a 1.6% requirement, while lentil, chickpea, pea and soy protein sit at 0.3 to 0.4%. Whole plant foods also gave up less of their protein, about 50 to 75% absorbed for chickpeas, mung beans and yellow peas against 85 to 95% for eggs and chicken, which the authors put down to fiber and tannins. Their three fixes for a weak spot: eat more of that protein, blend it with one that has what it lacks, or fortify it with the missing amino acid.

For your plate: this platter is fix number two, seeds blended with vegetables and cheese.

Where it stops: it's a narrative review, so the authors chose the evidence, and pumpkin seed appears only in the composition comparison, never in a human test. The product behind that 2.7% is named only in a figure I couldn't read as text, and USDA says 4.11, so the two disagree on how short. The Gatorade Sports Science Institute, a division of PepsiCo, paid an honorarium for the talk and the article, and two authors have had grants or fees from FrieslandCampina, PepsiCo, Tereos and Top Institute Food and Nutrition.

Case two: a pea and pumpkin blend against whey, in the blood

Twelve healthy adults aged 65 and over signed up in Dublin and Padova, and nine made it to the analysis: five women and four men, average age 74. On four fasted mornings, each ate a soup carrying a protein mix plus 10 g of pea fiber: whey, a plant blend of 67% pea and 33% pumpkin protein, or one of two other blends. All were matched for leucine, the amino acid most tied to the muscle-building signal, and balanced to the FAO's reference, so the pea and pumpkin soup needed 41.3 g of protein against 30.9 g for whey. Blood every 30 minutes, for three hours.

On paper, the pea and pumpkin soup even beat whey on methionine, 1.18 g against 0.65. In the blood, the story turned. An hour in, leucine was 74% higher after whey, and whey's essential amino acids peaked higher too. Over three hours the totals showed no significant difference, pea and pumpkin only trending lower. Methionine and cysteine dropped below their fasting levels after every plant soup; the authors' best guess is that older guts and livers keep more of them back.

For your plate: matching a protein on paper didn't match it in the blood, so I'd rather lean on a mix of foods and a dairy partner than on seeds alone.

Where it stops: nine people, no power calculation, and pumpkin was a third of a lab powder, its own part impossible to pull out. The authors say the results aren't comparable to whole foods, pumpkin dishes included, and blood levels stand in for digestion without measuring muscle building. European public funders paid, plus an Irish Research Council scholarship with Kerry Group. Staff of the Fraunhofer Institute, which made the powders, are co-authors; the authors declare no conflict of interest.

Case three: a shake with pumpkin seed in it, against whey

Fifteen healthy adults at the University of Surrey, in England: nine women and six men, average age 26, none exercising more than three times a week. On three mornings, after a 12-hour fast, each drank 300 mL of water carrying 20 g of protein from whey isolate or from an unflavored plant blend of pea, brown rice and pumpkin seed protein, or plain water. For three hours they rated their hunger and fullness every hour, then sat down to as much pasta in tomato and cheese sauce as they wanted.

An hour in, both shakes had cut hunger and raised fullness compared with water. By two and three hours, the effect was gone. Desire to eat dropped below water's only after whey, and the two shakes never differed from each other on appetite. Lunch came to 1,275 kcal after the plant shake, 1,309 after whey and 1,425 after water, with no significant difference between them, and the rest of the day's eating didn't differ either. Whey did push insulin higher at 30 and 60 minutes.

For your plate: a shake with pumpkin seed in its blend held hunger off for about an hour, the same as whey.

Where it stops: one 20 g dose, drunk alone on an empty stomach, by young, lean, lightly active people. The paper never says how much of the blend was pumpkin. The budget didn't stretch to measuring amino acids in the blood, taste wasn't matched between the drinks, and the menstrual cycle wasn't controlled beyond avoiding periods. A UK research council doctoral program paid for the work. One author consults for FormNutrition, the company that supplied the plant powder; the authors say the company had no part in the design, the analysis or the writing.

Case four: 65 g of pumpkin seed and blood sugar, in Viçosa

This one comes from my home state: researchers at the Universidade Federal de Viçosa, in Minas Gerais, ran the only test here that used the seed itself rather than an extracted protein. Their volunteers were adults with normal blood sugar, 25 for the food records and 15 for the blood sugar readings. On three separate days each ate a high-carbohydrate mixed meal: with no seed, with 65 g of pumpkin seed, or with 65 g of flaxseed, all three with a similar nutritional makeup, the authors say. A finger-prick test checked blood sugar seven times, from just before the meal to two hours after.

Across those two hours, the rise in blood sugar was about 35% smaller after the pumpkin seed meal than after the meal with no seed (P = .025). Flaxseed made no significant difference. At any single checkpoint, the three meals didn't differ, and appetite and what people ate that day didn't differ either. The authors call pumpkin seed a food with potential that now needs confirming in long-term studies.

For your plate: one plate here carries about 40 g of pepitas, less than the 65 g tested, inside a very different meal.

Where it stops: full honesty, the paper sits behind a paywall and a login I didn't get past, so I've read only its summary. I can't tell you the volunteers' ages, sexes or weights, whether the seeds were whole or ground, what 'pooled' in the title means, how the no-seed meal was matched, or who paid for the work. It was one meal each in 15 people with normal blood sugar. If you're managing your blood sugar, take this to your doctor before you build meals around it.