What salt actually does to a seedling鹽對一株苗到底做了什麼
Salt stress arrives in two phases on two clocks, and our three readouts report them differently. That single fact explains most of the places our own readouts disagree. 鹽逆境分兩個階段、走兩種時間尺度，而我們三個讀值分別反映不同階段。光是這件事，就能解釋我們讀值互相矛盾的大部分情況。
At day six we are mostly reading the osmotic phase, which is why root length is our cleanest readout. The purple leaves we photographed in sets 4 and 6 at day twelve and thirteen are the ionic phase showing up later. We never separated the two experimentally, and we should not claim we did. 第六天我們量到的主要是滲透階段，這就是根長最乾淨的原因。第4與第6組在第12、13天拍到的紫色葉片，是離子階段後來才出現。我們從未用實驗把兩個階段分開，也不該宣稱分開過。
ACC deaminase
Plants under stress make more ethylene, and above a certain level ethylene stops being a signal and starts being damage: it inhibits root elongation and accelerates senescence. ACC deaminase does not act on ethylene. It acts one step earlier, on the molecule ethylene is made from. 植物在逆境下會製造更多乙烯，超過一定程度之後，乙烯就不再是訊號而是傷害：它抑制根的伸長、加速老化。ACC 脫胺酶不是作用在乙烯上，而是作用在前一步——乙烯的原料。
Two consequences follow directly from that reaction, and both are things a judge may raise before you do. 這個反應直接帶出兩個結果，而且兩個都可能被評審先提出來。
- The nitrogen confound is built into the mechanism. The reaction releases ammonia. An enzyme preparation that works partly as a nitrogen source is not a far-fetched objection — it is a prediction of the chemistry. The control that settles it is a nitrogen-matched arm, and we do not have one. 氮的混淆是機制本身帶來的。這個反應會放出氨。說「酵素製劑可能部分是當氮源在用」不是刁難，而是化學本身的預測。能解決它的對照是等氮組，而我們沒有。
- Relief is not defence. Lowering stress ethylene removes an inhibition rather than funding a new defence, which is why an exogenous ACC deaminase could in principle give a benefit without the growth cost resource-allocation theory predicts. 解除不等於防禦。降低逆境乙烯是拿掉一個抑制，而不是出錢蓋一套新的防禦。這就是為什麼外源 ACC 脫胺酶原則上可以不用付出資源分配理論預測的生長代價。
The cell wall problem細胞壁的問題
This is the sharpest question in the advisor memo, and it has a clean answer if you move the problem instead of defending it. Our enzyme runs at about 41 kDa on the 24 August blot. An intact plant cell wall excludes particles far smaller than that, so the enzyme does not enter a cell, and we never claim it does. 這是老師備忘錄裡最銳利的一題，而只要把問題移開、不要硬守，答案其實很乾淨。我們的酵素在 8 月 24 日的墨點上約 41 kDa。完整的植物細胞壁連遠比它小的顆粒都擋得住，所以酵素不會進入細胞，我們也從不宣稱它進去。
The resolution is that the substrate is not inside either. ACC moves out of cells into the apoplast and is transported between tissues, and the conversion to ethylene happens where ACC oxidase is. An enzyme sitting in the apoplast, or on a wet leaf surface, can deplete that pool before the conversion. 解法是：受質本來也不在裡面。ACC 會流出細胞進入質外體，並在組織間運輸，而轉成乙烯的反應發生在 ACC 氧化酶所在的地方。待在質外體、或待在濕的葉面上的酵素，可以在轉換發生前把這個池子消耗掉。
Anchor it in the established case: a rhizosphere bacterium lowers root ethylene without entering a plant cell, by consuming ACC outside it. We moved the enzyme out of the bacterium; the location of the chemistry did not change. That reframing is why the exogenous route is worth trying at all. 用已確立的案例定錨：根圈細菌不進入植物細胞，也能靠在細胞外吃掉 ACC 來降低根部乙烯。我們只是把酵素從細菌裡搬出來，化學反應發生的位置沒有變。正是這個重新框定，讓外源施用這條路值得一試。
LEA14 · BoPep4
These are the two protectants we designed ourselves, and they act on a different problem from ACC deaminase. Ethylene is a signal; dehydration is physics. LEA-family proteins are intrinsically disordered and highly hydrophilic, and the working picture is that they hold water around proteins and membranes and act as a molecular shield as a cell dries, whether the drying comes from drought, cold or salt. 這兩個是我們自己設計的保護劑，處理的問題和 ACC 脫胺酶不同。乙烯是訊號，脫水是物理。LEA 家族蛋白本質上是無序且高度親水的，目前的理解是它們在細胞乾燥時把水留在蛋白與膜周圍，作為分子層次的屏障——不論乾燥來自乾旱、低溫或鹽。
On the published wiki, neither has ever been applied to a plant: module G failed, module H produced no clone, and the SamyQ fusions were only assembled at Level 1 at the end of August. Say that plainly. If a judge presses on newer bench work, the Red lines page has what exists and what it does not support. 在公開的 wiki 上，兩者都從未用在植物上：模組 G 失敗、模組 H 沒有菌落、SamyQ 融合直到八月底才完成 Level 1 組裝。這點要直說。如果評審追問更新的實驗，紅線頁有「存在什麼」與「它不能支持什麼」。
Trehalose, and why the textbook answer failed in our hands海藻糖，以及教科書答案為什麼在我們手上失敗
Trehalose is a compatible solute: a small molecule a cell can accumulate to high concentration without poisoning its own chemistry, which both lowers internal water potential and stabilises proteins and membranes. It is published as protective under salinity, and it is the reason we chose it as a reference. 海藻糖是相容性溶質：細胞可以累積到高濃度而不毒害自身化學反應的小分子，既能降低內部水勢，也能穩定蛋白與膜。文獻上它在鹽逆境下有保護作用，這也是我們選它當參考化合物的原因。
In our hands the window turned out to be narrow in both directions. At 1 mM it was toxic on its own — unstressed seedlings reached 14.6 mm against 51.7 untreated. At 0.5 mM it was harmless and rescued nothing. Between the two, two simultaneous trials of the same design pointed opposite ways. 在我們手上，這個窗口兩邊都很窄。1 mM 本身就有毒——無逆境苗只有 14.6 mm，未處理是 51.7 mm。0.5 mM 無害但什麼也沒救。而在兩者之間，同設計同時進行的兩個試驗結論相反。
CH peptide: why one product covers three stressesCH 胜肽：為什麼一個產品能跨三種逆境
The product was sold for cold and drought, and our clearest plate came from salinity. That sounds surprising until you say the one sentence that makes it expected: cold, drought and salinity all end in cellular dehydration. A molecule that protects against one is a reasonable candidate for the others. 這個產品是賣給低溫與乾旱用的，而我們最清楚的一盤來自鹽逆境。聽起來意外，直到你講出那句讓它變成理所當然的話：低溫、乾旱、鹽逆境最後都落在細胞層次的脫水。能保護其中一種的分子，本來就是其他兩種的合理候選。
What we cannot do is name the molecule. It is a finished product and its active components were not disclosed to us, so every statement has to stay at product level. The route to a molecule is already written in our Future work: fractionate the preparation and run the fractions through the same three-timing assay. 我們做不到的是指出分子。它是成品，活性成分沒有揭露給我們，所以每一句話都必須停在產品層級。要走到分子，方法已經寫在我們的 Future work：把製劑分離成不同分液，再用同一套三時機檢定跑一次。
Glutathione: two hypotheses, two different measurements穀胱甘肽：兩個假說，兩種不同的量測
The advisor's last question is whether our protectants work through glutathione. We have no redox measurement of any kind, so this is reasoning — but the two possibilities are cleanly distinguishable, and saying how is the answer. 老師最後一題問的是，我們的保護劑是不是透過穀胱甘肽作用。我們沒有任何氧化還原的量測，所以這是推理——但兩種可能可以乾淨地分開，而「怎麼分開」就是答案。
Direct substitution直接替代
The protectant does the antioxidant job itself. Prediction: an effect that does not depend on the plant's own pathway, and that should show in vitro.保護劑自己做抗氧化的工作。預測：效果不依賴植物自身路徑，且在試管中就該出現。
Upstream relief上游解除
The protectant lowers the stress, so the plant's own glutathione pool is less depleted. Prediction: a higher GSH to GSSG ratio under stress, and no change in an unstressed plant.保護劑降低逆境，所以植物自己的穀胱甘肽池消耗較少。預測：逆境下 GSH/GSSG 比值較高，無逆境時則無變化。
The measurement fits on the extracts we already take: a glutathione assay plus glutathione reductase activity, run alongside the chlorophyll extraction. And it connects to a gap we already published — Prof. Cheng suggested a ROS staining experiment in March and we did not do it. Pairing the two shows the hole was known, not discovered in the room. 這個量測可以掛在我們本來就會取的萃取液上：穀胱甘肽分析加穀胱甘肽還原酶活性，與葉綠素萃取一起做。而且它連回我們早已公開的缺口——鄭老師三月建議的 ROS 染色我們沒做。把兩者綁在一起，顯示這個洞是早就知道的，不是現場才發現的。
Where the whole chain goes dotted整條鏈在哪裡變成虛線
One picture for the end of a mechanism answer. It is the project's own summary, and saying it first is better than being walked through it. 機制類回答收尾用的一張圖。這是專案自己的總結，主動講比被帶著走好。