Place · Level 3 · Macro
The Dark Matter of Nutrition · why your liver is overbuilt
营养成分表是缺乏症投下的影子 · 酶的手笨是未知分子的指纹 · 次生代谢物是植物的化学武器 · 好处来自轻微的毒 · 提纯加大剂量把 U 型推过峰顶 · 问错的问题与问对的问题
Last updated
Story path
- 1The label is a shadowThe label is a shadow
- 2Why your liver is overbuiltWhy your liver is overbuilt
- 3Plants didn't make it for youPlants didn't make it for you
- 4A mild poison wakes your defenseA mild poison wakes your defense
- 5Why the pill reversesWhy the pill reverses
- 6What to do with not knowingWhat to do with not knowing
Chapter 1
The label is a shadow
The label is a shadow
Not one of the molecules that make garlic garlic appears on the nutrition label.
The USDA database lists 69 entries for raw garlic. Ascorbic acid (vitamin C) is there; so are amino acids like alanine. What isn't there is allicin — the molecule that makes garlic pungent and antimicrobial. Nor ajoene. Nor p-coumaric acid. Meanwhile researchers recently counted 6,802 small molecules in raw garlic (Menichetti 2024).
Those 69 rows weren't picked at random. To see why the list looks like this, start with a question about the body: what earns a molecule the word essential?
Essential is the gap between two rates
The answer isn't in the food. It's in you, and it's two measurable rates:
how fast you build ithow fast you spend it
Build faster than you spend, and you needn't get it from food. Build slower, and the shortfall has to be eaten. Build at zero while demand isn't zero, and it's essential: cut off the supply and your stores fall at the rate you spend them, and below some line the step that runs on it stops.
What stopping looks like is visible. Without vitamin C you can't build stable collagen, so gums bleed and wounds split open — scurvy. Without thiamine your nerves and heart can't get at the energy in sugar — beriberi. Niacin and vitamin D each collapse in their own way.
So the label is a medical history
Look at the shape of the sieve: to get on the list, a molecule's absence must cause trouble within weeks to months.
That condition selects a very particular class — molecules your body cannot build, without which some step seizes. Twentieth-century nutrition science found them by following the diseases: first the disease, then the molecule.
Which means the sieve necessarily misses an entire class: molecules whose absence causes no trouble in weeks. They have no deficiency disease, so nobody traced a disease back to them, so they aren't on the list.
Not on the list doesn't mean not in the food, and it certainly doesn't mean nothing happens in you. Allicin and ajoene live in that class. Skipping garlic gives you no disease — which is exactly why they're absent from the label, and exactly why they're easy to treat as nonexistent.
The nutrition label is a shadow cast by acute deficiency disease. The analogy stops there: nobody shrank the list on purpose. It was shaped by a question, and the question was what kills people when missing, not what is actually in here.
The USDA database lists 69 entries for raw garlic. Ascorbic acid (vitamin C) is there; so are amino acids like alanine. What isn't there is allicin — the molecule that makes garlic pungent and antimicrobial. Nor ajoene. Nor p-coumaric acid. Meanwhile researchers recently counted 6,802 small molecules in raw garlic (Menichetti 2024).
Those 69 rows weren't picked at random. To see why the list looks like this, start with a question about the body: what earns a molecule the word essential?
Essential is the gap between two rates
The answer isn't in the food. It's in you, and it's two measurable rates:
how fast you build ithow fast you spend it
Build faster than you spend, and you needn't get it from food. Build slower, and the shortfall has to be eaten. Build at zero while demand isn't zero, and it's essential: cut off the supply and your stores fall at the rate you spend them, and below some line the step that runs on it stops.
What stopping looks like is visible. Without vitamin C you can't build stable collagen, so gums bleed and wounds split open — scurvy. Without thiamine your nerves and heart can't get at the energy in sugar — beriberi. Niacin and vitamin D each collapse in their own way.
So the label is a medical history
Look at the shape of the sieve: to get on the list, a molecule's absence must cause trouble within weeks to months.
That condition selects a very particular class — molecules your body cannot build, without which some step seizes. Twentieth-century nutrition science found them by following the diseases: first the disease, then the molecule.
Which means the sieve necessarily misses an entire class: molecules whose absence causes no trouble in weeks. They have no deficiency disease, so nobody traced a disease back to them, so they aren't on the list.
Not on the list doesn't mean not in the food, and it certainly doesn't mean nothing happens in you. Allicin and ajoene live in that class. Skipping garlic gives you no disease — which is exactly why they're absent from the label, and exactly why they're easy to treat as nonexistent.
The nutrition label is a shadow cast by acute deficiency disease. The analogy stops there: nobody shrank the list on purpose. It was shaped by a question, and the question was what kills people when missing, not what is actually in here.
清单 · 表上有谁, 表上没有谁
美国农业部的数据库给生蒜列了 69 行。里面有抗坏血酸, 也就是维生素 C; 有丙氨酸这样的氨基酸。表上没有的是蒜素 (allicin), 那个让大蒜辣、让它抗菌的分子; 也没有蒜烯, 没有对香豆酸。而研究者最近在生蒜里数出了 6,802 个小分子 (Menichetti 2024)。那 69 行不是随手挑的。蒜素和蒜烯就在那一类不上表的分子里 —— 你不吃大蒜不会得任何病, 这正是它们不上表的原因, 也正是它们容易被当成不存在的原因。
营养成分表是急性缺乏症投下的一道影子。比喻到此为止: 这张表不是被谁故意做小的, 它是被一个问题问出来的, 而那个问题是缺了会不会出人命, 不是里面到底有什么。
机制 · 缺了会怎样, 所以这张表是一份病历
停了会怎样, 是看得见的。维生素 C 断供, 你造不出稳定的胶原, 于是牙龈出血、伤口裂开, 这叫坏血病。硫胺素断供, 你的神经和心脏拿不到糖里的能量, 这叫脚气病。烟酸和维生素 D 各有各的塌法。所以这张表是一份病历
这个条件筛出来的是很特别的一类分子 —— 身体自己造不出来, 而且某一道工序离了它就转不动。二十世纪的营养学正是沿着这些病, 一个一个把它们找出来的: 先有病, 后有分子。
于是反过来, 这个筛子必然漏掉一整类东西: 那些不吃也不会在几周内出事的分子。它们没有对应的缺乏症, 所以没人循着病找到它们, 所以它们不在表上。
Numbers · how much is tracked
Pull the camera back and the shadow's width is measurable.The USDA's long-tracked core panel is 150 essential micro- and macronutrients, organized mainly around energy metabolism and deficiency: fatty acids, amino acids, sugars, fiber, minerals, vitamins. Since 2003 it has also reported flavonoid content for selected foods, extending the main panel to 188 components (Menichetti 2024).
Set that against the whole of food chemistry:
In 2019 the food compound database FooDB held 26,625 compounds. Those 150 are about 0.5% of it. More than 99% of the biochemicals in food go untracked by any national database — and many of them have documented roles in health and disease (Barabási 2020).By 2024, integrating literature, mass-spectrometry repositories, experiments, composition databases and pathway predictions, that library became 139,443 molecules: 92,612 detected, 46,831 inferred (Menichetti 2024).
But those two numbers can't just be laid side by side — you have to say what each one is first. 26,625 is what FooDB held in 2019. 139,443 is not that same database grown up; it's a separate library the authors assembled (FooDB itself held about 71,000 by 2023). And it matters that 46,831 of them are inferred — computed from pathways and species relatedness, not measured in food.
So using these two to say scientists estimate between twenty thousand and a hundred forty thousand compounds in food is reading two different rulers as one error bar. The honest sentence is just this: the 150-odd we track are a rounding error on every ruler anyone has picked up.
Back to raw garlic: 69 rows against 6,802 molecules — of which 1,984 were actually detected and 4,818 inferred. Count only the detected column and it's still 69 against 1,984. FooDB also records how many bioactivities each molecule has — the alanine on the label has 3, ascorbic acid 105; while the absent allicin has 64, ajoene 46, p-coumaric acid 24 (Menichetti 2024). The missing ones are not nobodies.
A prior offense
This has happened once before. In the 1980s, detractors of the Human Genome Project insisted only the coding regions were worth the cost of decoding — 1.4% of all base pairs — labelling the remaining 98.6% junk DNA. Today it is estimated that 66% of disease-carrying variants sit in exactly those non-coding regions (Barabási 2020).
The analogy stops there — nothing here predicts how many treasures hide among those hundred-odd thousand molecules. It makes one point: a list filtered by a purpose is easily mistaken for the whole. And the purpose that filtered the nutrition label was finding what drops people within weeks.
Chapter 2
Why your liver is overbuilt
Why your liver is overbuilt
You would not build a clumsy enzyme for a molecule you already know.
Start with an enzyme that isn't clumsy, for contrast.
What specificity looks like
Your intestinal wall carries a glucose transporter (SGLT1). It is a severe fusspot: fastest with glucose, a fraction of that with galactose, and it barely moves xylose at all (Wright 2011).
Why can it afford to be so picky? Because it needn't be otherwise. Glucose has exactly one shape, and that shape hasn't changed in hundreds of millions of years. The body knows precisely who it's waiting for, so it built a slot that recognizes that one face. Specificity pays in speed and thrift.
Now look at your liver
Liver cells hold a large family of enzymes, cytochrome P450, whose job is drilling a handle onto foreign molecules: press an oxygen atom on, creating a socket other things can attach to (Guengerich 2008). The most famous member is CYP3A4.
It works the opposite way to that transporter. CYP3A4 is the most promiscuous of the human CYP enzymes: structural work found its active site changes shape around whatever binds, with the pocket volume expanding by more than 80% (Ekroos 2006).
It isn't a lock committed to one set of teeth. It's a hand that reshapes itself to grip things. It grips imprecisely, so it's slow and expensive — but it can grip what it has never met.
Two more properties of this family are just as odd:
Humans carry 57 CYP genes — and the family splits into two populations, which says more than the total does. Only about a dozen handle foreign molecules, clustered in the CYP1, CYP2 and CYP3 families; five of those cover roughly 95% of drug metabolism (Guengerich 2008). The other forty-odd mostly have dedicated endogenous substrates: building steroid hormones and bile acids, making eicosanoids, activating vitamin D (Zanger 2013).And it is exactly the dozen that meet foreign molecules whose substrates overlap and which can be induced — the forty-odd on endogenous duty are specific (Zanger 2013). Inside a single gene family, the body builds specific enzymes for molecules it knows and promiscuous ones for molecules it doesn't. The glucose transporter and the clumsy hand, shrunk to a contrast within one family.They can be induced. When a kind of molecule keeps arriving, the body builds more of the matching enzyme. Capacity can be expanded on demand.
After the handle comes step two: another set of enzymes welds on a big, water-loving, charged tag — glucuronic acid, sulfate, or glutathione (Jancova 2010). Charged things can't slip back through a membrane, so once the kidney filters it, it can never sneak back.
Now set the two side by side
On one side, a glucose transporter, specific to the point of severity. On the other, an enzyme family of dozens of members with overlapping substrates and expandable capacity, far less precise.
Same body. Why two designs?
Because they face different kinds of problem. Glucose's list has one entry and never changes; you can build to order. The list P450 faces is not known to the body in advance, and it differs at every meal.
Against a list whose contents you don't know, specificity is meaningless — you cannot machine a lock for a molecule you haven't met. The only workable design is breadth: build a hand that changes shape, let it grip anything a little and accept that it grips poorly; give it dozens of overlapping colleagues as backup; let it scale on demand.
Which yields the inference this whole story rests on
You would not build something this slow, this expensive, and this imprecise for a known, finite list of molecules. Specificity always pays better. There is only one reason a body accepts the cost of clumsiness: it expects to meet molecules it has never seen, and to meet them often.
The non-specificity of these enzymes is itself the fingerprint of unknown molecules.
And the fingerprint has a provenance. A classic hypothesis for why the P450 superfamily diversified into so many members: a continuing chemical arms race between animals and plants — plants keep producing new defensive molecules, animals keep producing new enzymes to take them apart (Gonzalez & Nebert 1990).
Put differently: your liver was shaped into its present form by somebody else's chemical weapons.
The next scene goes to see who made those weapons, and why. For the full picture of the liver's two steps, see the hepatic island.
Start with an enzyme that isn't clumsy, for contrast.
What specificity looks like
Your intestinal wall carries a glucose transporter (SGLT1). It is a severe fusspot: fastest with glucose, a fraction of that with galactose, and it barely moves xylose at all (Wright 2011).
Why can it afford to be so picky? Because it needn't be otherwise. Glucose has exactly one shape, and that shape hasn't changed in hundreds of millions of years. The body knows precisely who it's waiting for, so it built a slot that recognizes that one face. Specificity pays in speed and thrift.
Now look at your liver
Liver cells hold a large family of enzymes, cytochrome P450, whose job is drilling a handle onto foreign molecules: press an oxygen atom on, creating a socket other things can attach to (Guengerich 2008). The most famous member is CYP3A4.
It works the opposite way to that transporter. CYP3A4 is the most promiscuous of the human CYP enzymes: structural work found its active site changes shape around whatever binds, with the pocket volume expanding by more than 80% (Ekroos 2006).
It isn't a lock committed to one set of teeth. It's a hand that reshapes itself to grip things. It grips imprecisely, so it's slow and expensive — but it can grip what it has never met.
Two more properties of this family are just as odd:
Humans carry 57 CYP genes — and the family splits into two populations, which says more than the total does. Only about a dozen handle foreign molecules, clustered in the CYP1, CYP2 and CYP3 families; five of those cover roughly 95% of drug metabolism (Guengerich 2008). The other forty-odd mostly have dedicated endogenous substrates: building steroid hormones and bile acids, making eicosanoids, activating vitamin D (Zanger 2013).And it is exactly the dozen that meet foreign molecules whose substrates overlap and which can be induced — the forty-odd on endogenous duty are specific (Zanger 2013). Inside a single gene family, the body builds specific enzymes for molecules it knows and promiscuous ones for molecules it doesn't. The glucose transporter and the clumsy hand, shrunk to a contrast within one family.They can be induced. When a kind of molecule keeps arriving, the body builds more of the matching enzyme. Capacity can be expanded on demand.
After the handle comes step two: another set of enzymes welds on a big, water-loving, charged tag — glucuronic acid, sulfate, or glutathione (Jancova 2010). Charged things can't slip back through a membrane, so once the kidney filters it, it can never sneak back.
Now set the two side by side
On one side, a glucose transporter, specific to the point of severity. On the other, an enzyme family of dozens of members with overlapping substrates and expandable capacity, far less precise.
Same body. Why two designs?
Because they face different kinds of problem. Glucose's list has one entry and never changes; you can build to order. The list P450 faces is not known to the body in advance, and it differs at every meal.
Against a list whose contents you don't know, specificity is meaningless — you cannot machine a lock for a molecule you haven't met. The only workable design is breadth: build a hand that changes shape, let it grip anything a little and accept that it grips poorly; give it dozens of overlapping colleagues as backup; let it scale on demand.
Which yields the inference this whole story rests on
You would not build something this slow, this expensive, and this imprecise for a known, finite list of molecules. Specificity always pays better. There is only one reason a body accepts the cost of clumsiness: it expects to meet molecules it has never seen, and to meet them often.
The non-specificity of these enzymes is itself the fingerprint of unknown molecules.
And the fingerprint has a provenance. A classic hypothesis for why the P450 superfamily diversified into so many members: a continuing chemical arms race between animals and plants — plants keep producing new defensive molecules, animals keep producing new enzymes to take them apart (Gonzalez & Nebert 1990).
Put differently: your liver was shaped into its present form by somebody else's chemical weapons.
The next scene goes to see who made those weapons, and why. For the full picture of the liver's two steps, see the hepatic island.
对照 · 专一长什么样
先看一个手不笨的例子, 好有个对照。你的小肠壁上有一种搬葡萄糖的转运体 (SGLT1)。它挑食挑得厉害: 搬葡萄糖最快, 换成半乳糖速度掉到几分之一, 换成木糖几乎不动 (Wright 2011)。
它凭什么这么挑? 因为它不需要不挑。葡萄糖只有一种长相, 而且这个长相几亿年没变过。身体清楚地知道自己在等谁, 于是造了一个只认这一副面孔的口子。专一的回报是又快又省。
机制 · 它攥得有多不准
这家子里名气最大的是 CYP3A4。它的做事方式和那个转运体正好相反。CYP3A4 是人类这些酶里底物最杂的一个: 结构研究发现, 它的活性口袋会随着结合的分子改变形状, 体积能撑大八成以上 (Ekroos 2006)。
两个种群 · 同一个家族里的两种造法
这个家族还有两个同样反常的性质:人有 57 个 CYP 基因 —— 而这个家族内部就分成两个种群, 这一点比总数更能说明问题。真正管外来分子的只有十几个, 集中在 CYP1、CYP2、CYP3 三个家族; 其中 5 个就承担了约 95% 的药物代谢 (Guengerich 2008)。剩下四十来个多数有自己专门的内源底物: 造甾体激素、造胆汁酸、做类二十烷酸、活化维 D (Zanger 2013)。而恰恰是碰外来分子的那十几个, 底物彼此重叠、还能被诱导; 那四十来个干内源活的, 是专一的 (Zanger 2013)。同一个基因家族里, 身体对知道的分子造专一的酶, 对不知道的分子造宽泛的酶 —— 刚才那个葡萄糖转运体和这只手笨的手, 在这里缩成了一个家族内部的对照。
产能 · 能诱导, 而且还有第二道工序
它们可以被诱导。 同一类分子来得多了, 身体就把对应的酶造得更多。产能是能临时扩的。凿完把手还有第二步: 另一组酶往把手上焊一个又大又亲水、还带电的标签, 可能是葡萄糖醛酸、硫酸根、或者谷胱甘肽 (Jancova 2010)。带电的东西钻不回细胞膜, 于是肾一滤, 它就再也溜不回来。
推论 · 为什么这个设计只能是宽泛的
对一份你不知道内容的名单, 专一是没有意义的。唯一可行的设计是宽泛: 造一只能改形状的手, 让它谁都能攥一点, 攥不好也认了; 再配十几个底物互相重叠的同事一起兜; 再让它按需求扩产。你不会为一个已知的、数量有限的分子清单, 造这么一套又慢又费又不精准的东西。专一永远更划算。身体愿意付出手笨的代价, 只有一个理由: 它预期自己会遇到没见过的分子, 而且经常遇到。
出处 · 一场化学军备竞赛
而这个指纹有出处。P450 超家族为什么会分化出这么多成员, 一个经典假说是: 这是动物和植物之间一场持续的化学军备竞赛 —— 植物不断造出新的防御分子, 动物不断造出新的酶去拆它们 (Gonzalez & Nebert 1990)。换句话说, 你的肝是被别人的化学武器塑造成今天这个样子的。
下一幕去看那些武器是谁造的、为什么造。肝这两步的全貌, 去肝那一岛。
Chapter 3
Plants didn't make it for you
Plants didn't make it for you
When a plant built those molecules, it wasn't thinking about you. It was thinking about killing the insect chewing on it.
Detonate on bite
Cut a head of garlic, chew a mouthful of broccoli, and the smell that hits your nose has only just appeared. It isn't there in the intact plant cell.
The brassica design is explicit: a molecule called a glucosinolate and an enzyme called myrosinase sit packed in separate compartments of the cell, each quiet. Break the cell wall, the two mix, the enzyme cleaves the glucosinolate, and isothiocyanates are produced — that pungency is them (Fahey 2001).
It's a chemical landmine wired to detonate on bite: no energy spent while idle, triggered only at the moment of chewing, and going off inside the mouth doing the chewing.
These molecules do no work for the plant
Collectively they're called secondary metabolites. Secondary means they take no part in the plant's own growth and energy production — that's primary metabolism. Secondary metabolites are for dealing with the outside.
Of the several thousand garlic molecules counted in the previous scene, the researchers' own words: many are secondary metabolites acting as the plant's chemical defense against stressors such as predators and extreme weather conditions (Menichetti 2024).
Defense against whom? Insects, fungi, bacteria, the sun. Not you.
The pungency in your mouth is a stray round
To state it plainly: you and the caterpillar on the broccoli are biochemical relatives. A molecule that can jam its enzymes can often reach yours too — you're simply far bigger, so the same mouthful lands a far smaller dose on you.
So when you eat plants, you're eating a poison prepared for someone else. An accident, not a design.
This scene welds the first two together
Return to scene two's question: why did your liver build an entire infrastructure for molecules it has never met?
Here's the answer. Your ancestors ate other species' chemical weapons every day, for hundreds of millions of years. The weapons kept updating, because the plants were evolving too. So the disassembly tools couldn't be bespoke; they had to be general, expandable, and broad to the point of clumsiness. The arms-race hypothesis from the previous scene (Gonzalez & Nebert 1990) is about exactly these two things driving each other.
It also sets up the next scene
If these molecules' day job is to mildly interfere with your biochemistry, then their benefit to you is unlikely to be doing your work for you. Far more likely: they poke you gently, and your response to being poked is where the benefit comes from.
That idea has a name — xenohormesis: many dietary phytochemicals are toxins the plant uses against insects and stress, but on our side, at the low doses humans actually eat, they activate adaptive cellular stress responses and thereby confer stress resistance and other benefits (Surh 2011).
The next scene takes that sentence apart, down to which atom of which molecule the poke lands on.
Detonate on bite
Cut a head of garlic, chew a mouthful of broccoli, and the smell that hits your nose has only just appeared. It isn't there in the intact plant cell.
The brassica design is explicit: a molecule called a glucosinolate and an enzyme called myrosinase sit packed in separate compartments of the cell, each quiet. Break the cell wall, the two mix, the enzyme cleaves the glucosinolate, and isothiocyanates are produced — that pungency is them (Fahey 2001).
It's a chemical landmine wired to detonate on bite: no energy spent while idle, triggered only at the moment of chewing, and going off inside the mouth doing the chewing.
These molecules do no work for the plant
Collectively they're called secondary metabolites. Secondary means they take no part in the plant's own growth and energy production — that's primary metabolism. Secondary metabolites are for dealing with the outside.
Of the several thousand garlic molecules counted in the previous scene, the researchers' own words: many are secondary metabolites acting as the plant's chemical defense against stressors such as predators and extreme weather conditions (Menichetti 2024).
Defense against whom? Insects, fungi, bacteria, the sun. Not you.
The pungency in your mouth is a stray round
To state it plainly: you and the caterpillar on the broccoli are biochemical relatives. A molecule that can jam its enzymes can often reach yours too — you're simply far bigger, so the same mouthful lands a far smaller dose on you.
So when you eat plants, you're eating a poison prepared for someone else. An accident, not a design.
This scene welds the first two together
Return to scene two's question: why did your liver build an entire infrastructure for molecules it has never met?
Here's the answer. Your ancestors ate other species' chemical weapons every day, for hundreds of millions of years. The weapons kept updating, because the plants were evolving too. So the disassembly tools couldn't be bespoke; they had to be general, expandable, and broad to the point of clumsiness. The arms-race hypothesis from the previous scene (Gonzalez & Nebert 1990) is about exactly these two things driving each other.
It also sets up the next scene
If these molecules' day job is to mildly interfere with your biochemistry, then their benefit to you is unlikely to be doing your work for you. Far more likely: they poke you gently, and your response to being poked is where the benefit comes from.
That idea has a name — xenohormesis: many dietary phytochemicals are toxins the plant uses against insects and stress, but on our side, at the low doses humans actually eat, they activate adaptive cellular stress responses and thereby confer stress resistance and other benefits (Surh 2011).
The next scene takes that sentence apart, down to which atom of which molecule the poke lands on.
机制 · 这些分子在植物身上不干活
它们有个统称, 叫次生代谢物。次生的意思是, 它们不参与植物自己的生长和产能, 那些叫初生代谢。次生代谢物是拿来对外的。上一幕数过的那几千个大蒜分子, 研究者的原话是: 其中许多是次生代谢物, 是这株植物对付捕食者和极端天气的化学防御 (Menichetti 2024)。
防的是谁? 虫、真菌、细菌、太阳。不是你。
接线 · 前两幕在这里焊上
现在回头看第二幕那个问题: 你的肝为什么要建一整套为未知分子准备的基础设施?答案就在这儿。因为你的祖先每天都在吃别的物种的化学武器, 吃了几亿年。武器年年更新, 因为植物那边也在进化; 所以拆解工具不能是定制的, 只能是通用的、可扩产的、宽泛到手笨的。上一幕那个军备竞赛的假说 (Gonzalez & Nebert 1990), 说的就是这两件事互相催出来的。
预告 · 好处的方向是反的
如果这些分子的本职是轻微地干扰你的生化, 那它们对你的好处就不太可能是替你干活。更可能的是: 它们轻轻捅你一下, 而你被捅之后的反应, 才是好处的来源。这个想法有名字, 叫 xenohormesis, 中文可以叫异源刺激效应: 很多膳食植物化学物本来是植物用来对付虫子和逆境的毒素, 但到了人这边、在人吃到的那种低剂量下, 它们激活的是适应性的细胞应激反应, 于是带来抗压能力和其它好处 (Surh 2011)。
下一幕就把这句话拆开, 看那一下捅具体捅在哪个分子的哪个原子上。
Chapter 4
A mild poison wakes your defense
A mild poison wakes your defense
Broccoli is good for you not because it cleared away free radicals on your behalf. It's because it mildly poisoned you, forcing you to turn your own defenses up.
This scene takes that sentence apart down to the atoms.
Step one · a carbon short of electrons
You chew broccoli, myrosinase cleaves a glucosinolate into sulforaphane, and the landmine from the last scene has just gone off (Fahey 2001).
Sulforaphane carries an isothiocyanate group, written —N=C=S. The carbon in the middle has electrons pulled away by the nitrogen and sulfur flanking it, leaving it electron-poor — chemically, electrophilic. An electrophilic carbon has a fixed temperament: it hunts for electron-rich things to crash into, then sticks covalently, unbreakably.
The most electron-rich and most easily struck thing in a cell is the sulfur on the amino acid cysteine.
Step two · what it hits is a sensor
Cells hold a protein called KEAP1, carrying a row of cysteines. Its everyday job is concrete: it is a bridge.
One end of the bridge is the Cul3 ubiquitin ligase; the other is a transcription factor called Nrf2. KEAP1 keeps handing Nrf2 to the ligase, and the ligase tags Nrf2 for destruction. So Nrf2 is dismantled as fast as it's built, its concentration stays low, and the genes it governs stay low too (Hu 2011).
Note that this is the resting state. Your cells build Nrf2 every moment and destroy it every moment. That isn't waste — it's a loaded spring, held down by a hand.
The electrophilic carbon of sulforaphane crashes into exactly that hand. It sticks covalently onto KEAP1's cysteines; number 151 (Cys151) is among the most readily modified of them, and modifying it is required for sulforaphane's activity (Hu 2011).
This scene takes that sentence apart down to the atoms.
Step one · a carbon short of electrons
You chew broccoli, myrosinase cleaves a glucosinolate into sulforaphane, and the landmine from the last scene has just gone off (Fahey 2001).
Sulforaphane carries an isothiocyanate group, written —N=C=S. The carbon in the middle has electrons pulled away by the nitrogen and sulfur flanking it, leaving it electron-poor — chemically, electrophilic. An electrophilic carbon has a fixed temperament: it hunts for electron-rich things to crash into, then sticks covalently, unbreakably.
The most electron-rich and most easily struck thing in a cell is the sulfur on the amino acid cysteine.
Step two · what it hits is a sensor
Cells hold a protein called KEAP1, carrying a row of cysteines. Its everyday job is concrete: it is a bridge.
One end of the bridge is the Cul3 ubiquitin ligase; the other is a transcription factor called Nrf2. KEAP1 keeps handing Nrf2 to the ligase, and the ligase tags Nrf2 for destruction. So Nrf2 is dismantled as fast as it's built, its concentration stays low, and the genes it governs stay low too (Hu 2011).
Note that this is the resting state. Your cells build Nrf2 every moment and destroy it every moment. That isn't waste — it's a loaded spring, held down by a hand.
The electrophilic carbon of sulforaphane crashes into exactly that hand. It sticks covalently onto KEAP1's cysteines; number 151 (Cys151) is among the most readily modified of them, and modifying it is required for sulforaphane's activity (Hu 2011).
化学 · 那个碳为什么缺电子, 那座桥怎么搭的
萝卜硫素身上挂着一个异硫氰酸基团, 写作 —N=C=S。中间那个碳原子被两边的氮和硫拽走电子, 自己就缺电子, 化学上叫亲电。那座桥的两头是这样的: 一头是 Cul3 泛素连接酶, 另一头是一个叫 Nrf2 的转录因子。KEAP1 一直把 Nrf2 递给连接酶, 连接酶就给 Nrf2 挂上降解标签。于是 Nrf2 一造出来就被拆掉, 浓度一直很低; Nrf2 管的那批基因也就一直低着 (Hu 2011)。
萝卜硫素共价粘到 KEAP1 的半胱氨酸上; 其中第 151 号那个 (Cys151) 是最容易被它改的几个之一, 而且改到它是萝卜硫素起效的必要条件 (Hu 2011)。
Step three · the hand lets go
Step three · the hand lets goThe bridge breaks, Nrf2 stops being tagged, so it accumulates, enters the nucleus, pairs with a small Maf protein, binds a switch sequence in the genes called the ARE, and starts transcribing cytoprotective enzymes (Hu 2011).
And who gets transcribed? Here's the elegant loop:
Glutathione S-transferase — recognize it? That's the weld-a-water-balloon-onto-the-handle step from scene two. What this plant toxin does is turn up the very production line that clears it.NAD(P)H quinone oxidoreductase 1 and heme oxygenase-1, two more cytoprotective tools (Hu 2011).And one that matters more: through the ARE, Nrf2 governs the rate-limiting enzyme of glutathione synthesis, glutamate cysteine ligase (GCL, built from the GCLC and GCLM subunits) (Lu 2013).
That last one is this scene's landing point. Glutathione is your cells' principal reducing agent — and one of the water balloons welded on in scene two. Sulforaphane cleared no free radical for you. It isn't even a scavenger. What it did was widen the rate-limiting valve on the machine that builds scavengers.
So the benefit runs the other way
Not: you swallow a scavenger and it cleans for you.
But: you swallow a mild poison; your cell notices something is covalently attacking its cysteines, releases the hand that was holding the spring down, and raises its own defensive capacity.
The benefit is built by you. The plant only pressed a button.
The button's existence makes the point too: your cells carry a sensor dedicated to noticing that an electrophile has arrived, and that sensor is wired to the master switch for defensive genes. Bodies don't install sensors for events that never happen — the same argument as scene two. The Keap1-Nrf2 pathway is regarded as the most important one underlying the health benefits of dietary phytochemicals (Surh 2011).
The analogy stops here
If a metaphor is wanted: this is more like a fire drill than a delivery of fire extinguishers. The drill is a nuisance, but afterwards your capacity to put out fires is genuinely higher.
The analogy stops there — in the real event there is no fire and no alarm bell. There is an electron-poor carbon hitting a sulfur atom, and a bridge coming apart.
Chapter 5
Why the pill reverses
Why the pill reverses
The same molecule is good for you from a vegetable, and can harm you once purified, dosed up, and packed into a capsule.
The previous scene's mechanism carries this consequence built in.
The shape is a U
If the benefit comes from a mild poisoning, then dose and effect cannot be a straight rising line.
Too low: the sensor is never touched and nothing happens.About right: the sensor gets touched, defense is turned up, net result good.Too high: the electrophile is no longer touching only KEAP1's cysteines. Cysteines are everywhere in a cell, and so is everything else it can stick to covalently. At that point it is simply what it always was — a poison.
So there's a peak in the middle. Purifying and dosing up pushes you off the rising limb and straight over that peak. Food struggles to do this, because you'd have to eat a mountain of broccoli to reach the amount in one capsule.
And a second route, more insidious
High-dose antioxidants do something else: they erase the signal itself.
The Ristow 2009 trial is explicit. People exercised; half also took vitamin C 1000 mg/day plus vitamin E 400 IU/day, half took none.
The result: the improvement in insulin sensitivity appeared only in the group not taking antioxidants.
Worse is what follows. Exercise normally induces the body's own reactive-oxygen defenses (superoxide dismutases 1 and 2, glutathione peroxidase) to be built in greater quantity; with antioxidant supplementation, that induction was blocked too (Ristow 2009).
Look carefully at what happened. The reactive oxygen produced in exercising muscle isn't only damage — it's a signal. The body reads it and learns to raise its defenses and metabolic capacity. Swallow high-dose antioxidants, neutralize the signal before it lands, and the body never gets the memo, so it never adjusts.
You removed the alarm. The room is cleaner, at the price that the fire brigade never comes again.
This is the mirror image of scene four: there, a little stress turns defense up; here, a big dose flattens the stress, so defense never turns up at all.
The previous scene's mechanism carries this consequence built in.
The shape is a U
If the benefit comes from a mild poisoning, then dose and effect cannot be a straight rising line.
Too low: the sensor is never touched and nothing happens.About right: the sensor gets touched, defense is turned up, net result good.Too high: the electrophile is no longer touching only KEAP1's cysteines. Cysteines are everywhere in a cell, and so is everything else it can stick to covalently. At that point it is simply what it always was — a poison.
So there's a peak in the middle. Purifying and dosing up pushes you off the rising limb and straight over that peak. Food struggles to do this, because you'd have to eat a mountain of broccoli to reach the amount in one capsule.
And a second route, more insidious
High-dose antioxidants do something else: they erase the signal itself.
The Ristow 2009 trial is explicit. People exercised; half also took vitamin C 1000 mg/day plus vitamin E 400 IU/day, half took none.
The result: the improvement in insulin sensitivity appeared only in the group not taking antioxidants.
Worse is what follows. Exercise normally induces the body's own reactive-oxygen defenses (superoxide dismutases 1 and 2, glutathione peroxidase) to be built in greater quantity; with antioxidant supplementation, that induction was blocked too (Ristow 2009).
Look carefully at what happened. The reactive oxygen produced in exercising muscle isn't only damage — it's a signal. The body reads it and learns to raise its defenses and metabolic capacity. Swallow high-dose antioxidants, neutralize the signal before it lands, and the body never gets the memo, so it never adjusts.
You removed the alarm. The room is cleaner, at the price that the fire brigade never comes again.
This is the mirror image of scene four: there, a little stress turns defense up; here, a big dose flattens the stress, so defense never turns up at all.
研究 · 把信号本身抹掉的那个试验
Ristow 2009 那个试验讲得很清楚。让人去运动, 一半人同时吃维生素 C 每天 1000 mg 加维生素 E 每天 400 IU, 另一半不吃。结果: 胰岛素敏感性的改善只出现在不吃抗氧化剂的那组。
更要命的是下面这条。运动本来会诱导身体自己那套活性氧防御 (超氧化物歧化酶 1 和 2、谷胱甘肽过氧化物酶) 造得更多; 吃了抗氧化剂之后, 这个诱导也被摁掉了 (Ristow 2009)。
看清楚这里发生了什么。运动时肌肉里产生的活性氧不只是损伤, 它还是一个信号: 身体读到它, 才知道该把防御和代谢能力往上调。你吃下大剂量抗氧化剂, 把这个信号提前中和掉, 身体收不到通知, 于是不调了。
你把报警器拆了。 房间是干净了, 代价是消防队再也不来。
这正是第四幕的镜像: 第四幕靠一点应激把防御开大; 这里靠大剂量把应激抹平, 于是防御也就不开了。
The beta-carotene affair, stated precisely
The beta-carotene affair, stated preciselyThe most famous reversal usually gets waved through in a sentence: antioxidants turned out to cause cancer. That sentence is lazy. The real mechanism needs three conditions at once.
Start with the chemistry. Burton and Ingold found in 1984 that beta-carotene is an antioxidant that watches the oxygen pressure: only at partial pressures of oxygen significantly below 150 torr — the pressure in normal air — is it a good radical-trapping antioxidant. Raise the oxygen pressure and it not only loses antioxidant activity but shows an autocatalytic pro-oxidant effect, especially at relatively high concentrations (Burton 1984).
Most tissues in the body sit at the low end of that range, which is why it behaves itself under ordinary conditions.
But a smoker's lung isn't at the low end. It's among the highest-oxygen places in the body, and it's flooded with oxidants from smoke.
So the three conditions line up:
Purified — stripped of the hundreds of molecules that arrive with it inside a carrot.High dose — far beyond what anyone eats from vegetables.A specific environment — that lung, with high oxygen pressure plus smoke oxidants.
Hence the two trials stopped early. In ATBC (1994, male smokers, 20 mg/day beta-carotene) lung cancers rose about 18% (ATBC 1994). In CARET (Omenn 1996, beta-carotene plus retinol, smokers and asbestos-exposed workers) lung cancers rose about 28% (Omenn 1996).
Here is the part that must be exact: one word, hormesis, does not explain this. Slapping a U-curve on it and calling it finished uses a handsome word to cover three specific conditions. What is actually operating is the combination of those three — remove any one and the conclusion might change.
Don't fuse three trials into one
Don't fuse three trials into oneWhile we're here, separate the standard confusion:
ATBC (1994) — vitamin E and beta-carotene, male smokers, lung cancer up about 18%.CARET (Omenn 1996) — beta-carotene plus retinol, smokers and asbestos-exposed, lung cancer up about 28%.SELECT (Klein 2011) — selenium and vitamin E, healthy men, prostate cancer up about 17%. Not beta-carotene, not smokers, not lungs.
The three point at one lesson: purified high-dose antioxidants aren't free. But their mechanisms can't be borrowed across. The vitamin E in SELECT doesn't travel beta-carotene's oxygen-pressure road.
And one number, retired
That antioxidant capacity score once printed on packaging (ORAC) had its entire database withdrawn by the USDA in 2012. The stated reason is direct: in-vitro antioxidant capacity does not predict in-vivo health effects (USDA 2012).
Why it can't predict, scene four already answered. What matters was never how many radicals the molecule traps in a test tube. It's whether it can reach KEAP1's row of cysteines inside your cells. Those are entirely different things, and a molecule can excel at the first while being incapable of the second.
Chapter 6
What to do with not knowing
What to do with not knowing
What nutrients does this food have is the wrong question.
Where it goes wrong, the first five scenes have answered: the question assumes the answer lives on a label, and that label was sieved by whether absence causes trouble within weeks. Most of what you want to know isn't on this side of the sieve.
So what should you ask? Two things:
Does this mouthful bring a wide enough variety of unfamiliar molecules?Is any one of them outrunning the speed at which I clear it?
Both land on something concrete in the body. The first lands on KEAP1's row of cysteines: somebody has to touch it. The second lands on the capacity of the P450-plus-phase-two line: don't touch too hard.
The division of labor with the additives story
The zero-added story says: the body reads only three things — which molecule, how much, and what it arrived with; and once the clearance line saturates, dose versus harm turns from a slope into a hinge.
This story adds a premise to those three: the list of which molecule is far longer than you thought, and you don't hold the full copy.
Together the conclusion changes shape. Since you know neither the complete list nor how to compute each dose, the strategy cannot be to evaluate them one by one. Only two remain:
Keep that infrastructure built for unknown molecules busy, and busy with variety.Don't let any single molecule sit at the capacity ceiling for long.
Now dismantle two opposite errors
Not knowing gets used in two opposite directions. Neither holds.
First: unknown equals dangerous.
The instinct is natural — most of these molecules were never studied, so avoid them, pick the shortest ingredient list.
But it defeats itself. Among those hundred-odd thousand untracked molecules, the most vicious are precisely the natural ones: tetrodotoxin, aflatoxin — made by living things to kill other living things. Meanwhile you successfully process a pile of unstudied plant molecules every single day, using exactly the machinery of scene two. Your liver was built for this, and it wins every day.
So fearing the unknown means fearing something you succeed at daily. If you must rank by danger, be wary of mold and natural toxins, not the long chemical names of legal additives — that ranking is exactly inverted.
Second: unknown equals magical.
This one costs more. The whole superfood pitch is built on it: since science tracks only a fraction, my as-yet-undiscovered miracle active compound cannot be falsified.
Scene four supplies the test: for a molecule to act on you, it has to actually touch something in your body — hit a cysteine, wedge into a pocket, plug a gate. No contact, no event.
And the argument treats all plants equally. Tens of thousands of unknown molecules is not one berry's selling point. It is every plant's default configuration, including the cheapest cabbage at the market. Anyone selling it as exclusive is charging you for a universal fact.
So what can you take away
Not a list of what to eat. The entire content of this story is that nobody has that list.
Three judgments:
Variety beats betting. You don't know the list, so you can't pick the winner. Rotating is admitting you don't know — the only honest response.Whole food beats the purified capsule. The reason is scene five: purified plus high dose is the fastest way over the peak of the U.When you see the word unknown, ask whether it's being used to frighten you or to sell to you. Both exploit the same fact, and that fact only says your body was ready for it long ago.
Know that it is so, and know why it is so. This story's why is unusual: the infrastructure you carry for the unknown, by simply existing, is telling you that food is far more than what's on the label. You don't need to know those molecules. You only need to know your body has been handling them for you all along.
Where it goes wrong, the first five scenes have answered: the question assumes the answer lives on a label, and that label was sieved by whether absence causes trouble within weeks. Most of what you want to know isn't on this side of the sieve.
So what should you ask? Two things:
Does this mouthful bring a wide enough variety of unfamiliar molecules?Is any one of them outrunning the speed at which I clear it?
Both land on something concrete in the body. The first lands on KEAP1's row of cysteines: somebody has to touch it. The second lands on the capacity of the P450-plus-phase-two line: don't touch too hard.
The division of labor with the additives story
The zero-added story says: the body reads only three things — which molecule, how much, and what it arrived with; and once the clearance line saturates, dose versus harm turns from a slope into a hinge.
This story adds a premise to those three: the list of which molecule is far longer than you thought, and you don't hold the full copy.
Together the conclusion changes shape. Since you know neither the complete list nor how to compute each dose, the strategy cannot be to evaluate them one by one. Only two remain:
Keep that infrastructure built for unknown molecules busy, and busy with variety.Don't let any single molecule sit at the capacity ceiling for long.
Now dismantle two opposite errors
Not knowing gets used in two opposite directions. Neither holds.
First: unknown equals dangerous.
The instinct is natural — most of these molecules were never studied, so avoid them, pick the shortest ingredient list.
But it defeats itself. Among those hundred-odd thousand untracked molecules, the most vicious are precisely the natural ones: tetrodotoxin, aflatoxin — made by living things to kill other living things. Meanwhile you successfully process a pile of unstudied plant molecules every single day, using exactly the machinery of scene two. Your liver was built for this, and it wins every day.
So fearing the unknown means fearing something you succeed at daily. If you must rank by danger, be wary of mold and natural toxins, not the long chemical names of legal additives — that ranking is exactly inverted.
Second: unknown equals magical.
This one costs more. The whole superfood pitch is built on it: since science tracks only a fraction, my as-yet-undiscovered miracle active compound cannot be falsified.
Scene four supplies the test: for a molecule to act on you, it has to actually touch something in your body — hit a cysteine, wedge into a pocket, plug a gate. No contact, no event.
And the argument treats all plants equally. Tens of thousands of unknown molecules is not one berry's selling point. It is every plant's default configuration, including the cheapest cabbage at the market. Anyone selling it as exclusive is charging you for a universal fact.
So what can you take away
Not a list of what to eat. The entire content of this story is that nobody has that list.
Three judgments:
Variety beats betting. You don't know the list, so you can't pick the winner. Rotating is admitting you don't know — the only honest response.Whole food beats the purified capsule. The reason is scene five: purified plus high dose is the fastest way over the peak of the U.When you see the word unknown, ask whether it's being used to frighten you or to sell to you. Both exploit the same fact, and that fact only says your body was ready for it long ago.
Know that it is so, and know why it is so. This story's why is unusual: the infrastructure you carry for the unknown, by simply existing, is telling you that food is far more than what's on the label. You don't need to know those molecules. You only need to know your body has been handling them for you all along.
分工 · 和零添加那一篇的关系
零添加那一篇讲的是: 身体只读三件事, 是什么分子、多少、跟什么一起来; 而且清除产线一饱和, 剂量和伤害的关系就从斜线变成转折。这一篇给那三件事补了一个前提: 是什么分子那份名单, 比你以为的长得多, 而且你手上没有全本。
两篇合起来, 结论就换了形状。既然你既不知道完整名单、又没法逐个算剂量, 那策略就不可能是逐个评估。只剩两条:
让那套为未知分子准备的基础设施有事干, 而且换着花样干。别让任何一个分子长期顶在产能上限。
误区一 · 未知等于危险
未知这件事, 被两拨人朝相反方向用, 两边都不成立。这个念头很自然 —— 既然大多数分子没被研究过, 那还是躲开吧, 挑配料表最短的那个。
但它自己打自己。那十几万个未追踪的分子里, 最凶的几个恰恰是纯天然的: 河豚毒素、黄曲霉毒素, 都是生物为了杀死别的生物造出来的。而你每天都在成功处理掉一大堆没被研究过的植物分子, 靠的正是第二幕那套东西。你的肝就是为这件事建的, 而且它每天都在赢。
所以恐惧未知, 等于恐惧一件你每天都在做成的事。真要按危险排序, 该警惕的是霉变和天然毒素, 而不是配料表上名字长的合法添加剂 —— 那正好排反了。
误区二 · 未知等于神奇
这个更贵。超级食物的整套话术就架在这上面: 既然科学界只追踪了一小部分, 那我这个尚未被发现的神奇活性成分, 你就无法证伪。第四幕给了判据: 一个分子要在你身上起作用, 它得真的碰到你身体里的某个东西 —— 撞上一个半胱氨酸、卡进一个口袋、堵住一扇门。碰不到, 就是没发生。
而且这个论证对所有植物一视同仁。几万个未知分子不是某一种莓果的卖点, 它是每一棵植物的默认配置, 包括菜市场里最便宜的那棵白菜。谁拿它当独家卖点, 谁就是在把一个普遍事实标价卖给你。
落点 · 能带走的是三个判断
能带走的是三个判断:多样性打败押注。 你不知道名单, 所以你压不中宝。轮着吃, 是在承认自己不知道 —— 这是唯一诚实的对策。完整食物打败提纯胶囊。 理由在第五幕: 提纯加大剂量, 是把 U 型推过峰顶最快的办法。看到未知两个字, 先问它被用来吓你还是用来卖你。 两种都在利用同一件事实, 而那件事实本身只是在说: 你的身体早就为它准备好了。
知其然, 亦知其所以然。这一篇的所以然有点特别: 你身上那套为未知而建的基础设施, 它的存在本身就在告诉你, 食物远不止表上那些。 你不需要认识那些分子。你只需要知道, 你的身体一直在替你处理它们。
Red flags · and one thing for people on medication
This story is about mechanism. It is not medical advice and does not replace a doctor's judgment.One thing if you take medication
Scene two noted that P450 enzymes can be induced — use them more and the body builds more. They can also be suppressed.
And what suppresses them is often a plant. Grapefruit is the famous case: a class of molecules it carries inactivates CYP3A4 in the intestinal wall, so the fraction of a drug that should have been dismantled before reaching the blood isn't, and the same tablet can deliver considerably more drug into circulation (Bailey 2013). Dozens of medications are affected, from lipid-lowering drugs to certain blood-pressure drugs and immunosuppressants.
Two lessons run in opposite directions:
It confirms this whole story: your drug and the plant's secondary metabolites are competing for the same production line. The body can't tell which is medicine and which is fruit. It only reads molecules.It also warns that eating more plants isn't a zero-cost move for someone on medication. If you want to change things, ask a doctor or pharmacist rather than experimenting.
Don't handle these yourself — see a doctor
Unexplained weight lossPersistent fatigue with pallorBlack or bloody stools, or food sticking when you swallowRecurrent fever or night sweats
One more deserves its own line: a rash, swelling of the lips or tongue, or difficulty breathing after eating something. That is what an acute allergic reaction looks like, and it needs immediate care rather than watchful waiting.
bailey-2013-grapefruit-drug
References · 18
- Menichetti, G., Barabási, A.-L., & Loscalzo, J. (2024). Decoding the Foodome: Molecular Networks Connecting Diet and Health. Annual Review of Nutrition, 44, 257-288. 10.1146/annurev-nutr-062322-030557
- Barabási, A.-L., Menichetti, G., & Loscalzo, J. (2020). The unmapped chemical complexity of our diet. Nature Food, 1(1), 33-37. 10.1038/s43016-019-0005-1
- Wright, E. M., Loo, D. D. F., & Hirayama, B. A. (2011). Biology of human sodium glucose transporters. Physiological Reviews, 91(2), 733-794. 10.1152/physrev.00055.2009
- Guengerich, F. P. (2008). Cytochrome P450 and chemical toxicology. Chemical Research in Toxicology, 21(1), 70–83. 10.1021/tx700079z
- Ekroos, M., & Sjögren, T. (2006). Structural basis for ligand promiscuity in cytochrome P450 3A4. Proceedings of the National Academy of Sciences, 103(37), 13682-13687. 10.1073/pnas.0603236103
- Jancova, P., Anzenbacher, P., & Anzenbacherova, E. (2010). Phase II drug metabolizing enzymes. Biomedical Papers, 154(2), 103–116. 10.5507/bp.2010.017
- Gonzalez, F. J., & Nebert, D. W. (1990). Evolution of the P450 gene superfamily: animal-plant 'warfare', molecular drive and human genetic differences in drug oxidation. Trends in Genetics, 6(6), 182-186. 10.1016/0168-9525(90)90174-5
- Zanger, U. M., & Schwab, M. (2013). Cytochrome P450 enzymes in drug metabolism: regulation of gene expression, enzyme activities, and impact of genetic variation. Pharmacology & Therapeutics, 138(1), 103-141. 10.1016/j.pharmthera.2012.12.007
- Fahey, J. W., Zalcmann, A. T., & Talalay, P. (2001). The chemical diversity and distribution of glucosinolates and isothiocyanates among plants. Phytochemistry, 56(1), 5-51. Foundational chemistry of glucoraphanin conversion to sulforaphane by the enzyme myrosinase; myrosinase is heat-sensitive. 10.1016/S0031-9422(00)00316-2
- Surh, Y.-J. (2011). Xenohormesis mechanisms underlying chemopreventive effects of some dietary phytochemicals. Annals of the New York Academy of Sciences, 1229, 1-6. 10.1111/j.1749-6632.2011.06097.x
- Hu, C., Eggler, A. L., Mesecar, A. D., & van Breemen, R. B. (2011). Modification of Keap1 cysteine residues by sulforaphane. Chemical Research in Toxicology, 24(4), 515-521. 10.1021/tx100389r
- Lu, S. C. (2013). Glutathione synthesis. Biochimica et Biophysica Acta (BBA) - General Subjects, 1830(5), 3143-3153. 10.1016/j.bbagen.2012.09.008
- Ristow, M., Zarse, K., Oberbach, A., Klöting, N., Birringer, M., Kiehntopf, M., Stumvoll, M., Kahn, C. R., & Blüher, M. (2009). Antioxidants prevent health-promoting effects of physical exercise in humans. Proceedings of the National Academy of Sciences, 106(21), 8665-8670. 10.1073/pnas.0903485106
- Burton, G. W., & Ingold, K. U. (1984). Beta-carotene: an unusual type of lipid antioxidant. Science, 224(4649), 569-573. 10.1126/science.6710156
- The ATBC Cancer Prevention Study Group. (1994). The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. NEJM, 330(15), 1029–1035. 10.1056/NEJM199404143301501
- Omenn, G. S., et al. (1996). Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease (CARET). NEJM, 334(18), 1150–1155. 10.1056/NEJM199605023341802
- Klein, E. A., Thompson, I. M., Tangen, C. M., Crowley, J. J., Lucia, M. S., Goodman, P. J., et al. (2011). Vitamin E and the risk of prostate cancer: the Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA, 306(14), 1549-1556. Among 35,533 men, vitamin E 400 IU/day significantly increased prostate cancer risk (about 17% relative increase; 76 vs 65 cases per 1,000). 10.1001/jama.2011.1437
- U.S. Department of Agriculture, Agricultural Research Service. (2012). Withdrawal of the USDA ORAC (Oxygen Radical Absorbance Capacity) database for selected foods. ARS stated ORAC values are biologically misleading because in-vitro antioxidant capacity does not predict in-vivo health effects. www.ars.usda.gov