The Enzyme the Tumour Threw Away

Three articles about tumours winning metabolic exchanges because they had more escape routes than the immune cell. This is the one that runs the other way — and it works because of something the tumour discarded on purpose.

A pattern has been building across this series, and it has not been encouraging.

Alanine: the tumour runs the transaminase, the T cell barely expresses it. Glutamine: block the enzyme and the tumour reroutes five ways while the T cell has none. Methionine: the tumour overexpresses a transporter and simply outbids the T cell for the supply. Each time, an intervention aimed at a dependency the tumour genuinely has lands hardest on the cell you were trying to help.

So it is worth paying attention to the case that inverts.

Many tumours silence the enzyme argininosuccinate synthase 1, and they do it for a reason: it frees aspartate for nucleotide synthesis and they grow faster. The bill arrives later. Without ASS1 they cannot convert citrulline into arginine — and T cells can. For once the asymmetry runs the other way, and it runs that way because of a trade the tumour made on its own.

This follows Ammonia in Disguise, The Drug That Should Have Worked and The Amino Acid You Can Actually Move. It is the last of the four, and the only one with good news in it — qualified, preclinical good news with a serious catch, but good news.

Part I / The Rule the First Three Established

Selectivity Is About Escape Routes, Not Targets

The intuition behind metabolic oncology is that if a tumour depends heavily on something, removing that something should hurt the tumour most. Three articles' worth of evidence says the intuition is wrong, and the reason is simple once stated.

Every cell in the tumour microenvironment uses these pathways. Glutamine, alanine, methionine and the rest are not tumour-specific nutrients; they are general-purpose metabolic currency. So an intervention that removes one of them is not selective for the tumour. It is selective for whichever cell has the fewest alternatives — and that is almost never the tumour.

The tumour is the cell with options. It has a stroma secreting amino acids into it, a macropinocytic habit that lets it digest extracellular protein, isoform redundancy at most enzymatic steps, upregulated transporters, and signalling that reroutes within hours. The activated T cell beside it has none of those. It is the least adaptable cell in the neighbourhood, which makes it the one a blunt metabolic intervention starves first.

That is why glutaminase inhibition impaired the response to anti-PD-1 rather than enhancing it, and why every successful version of these strategies turned out to be the one that added something — clearance capacity, a wider block the T cell could adapt to — rather than removing a shared nutrient.

Which leaves an obvious question. Is there any metabolic intervention where the tumour is the one with no way out?

Part II / A Trade, Not an Accident

Why a Tumour Would Discard a Urea Cycle Enzyme

Argininosuccinate synthase 1 does one job: it condenses citrulline with aspartate to make argininosuccinate, which the next enzyme splits into arginine and fumarate. It is how a cell makes its own arginine, and it is the rate-limiting step of that synthesis.

A large number of tumours switch it off, usually by hypermethylating the promoter. On the face of it that looks like damage. It is better understood as a trade, and the accounting was worked out in 2015.

With ASS1 running, aspartate is consumed by the urea cycle. Silence ASS1 and cytosolic aspartate accumulates instead, where it feeds CAD — the rate-limiting complex of de novo pyrimidine synthesis. The effect is double: more substrate for CAD, and more CAD phosphorylation by S6K1. Pyrimidine output rises, and so does proliferation. In breast cancer and lung squamous carcinoma, low ASS1 tracks with worse survival.

The tumour did not lose ASS1. It spent it — and the currency it bought was nucleotides.

The bill is arginine auxotrophy. A cell without ASS1 cannot make arginine and must import it, which is the entire basis of arginine-deprivation therapy with pegylated arginine deiminase. That vulnerability has been pursued for years, and ASS1 status by immunohistochemistry is already used to select patients for it.

What has been pursued much less is the other side of the same coin.

Part III / The Inversion

A Nutrient Only One Side Can Use

If the tumour has thrown away the enzyme that converts citrulline into arginine, and the T cell has not, then citrulline is something close to a selective nutrient. Supply it, and only cells with ASS1 can do anything with it.

Currency	What the tumour has	What the T cell has	Who benefits
Alanine	ALT/GPT2, rising with grade	Barely any transaminase	Tumour
Glutamine	Transaminases, stroma, macropinocytosis, GLS2	No bypass; needs more when activated	Tumour
Methionine	SLC43A2, outcompetes for supply	Loses SAM, H3K79me2, STAT5	Tumour
Citrulline	ASS1 silenced — cannot convert it	ASS1 intact — converts it to arginine	T cell

Three rows of the tumour holding the enzyme the immune cell lacks, and one row where it is the other way round.

And the machinery on the T cell side is not hypothetical. Memory CD8 T cells actively run the urea and citrulline cycles to dispose of ammonia: they upregulate CPS1 through β-hydroxybutyrylation to fix free ammonia, mitochondrial arginase 2 cleaves arginine into urea and ornithine, and nitric oxide synthase converts arginine into NO and citrulline. Ammonia disposal is coupled to memory formation — which, given that the first article in this series was about tumours flooding their surroundings with nitrogen, is a connection worth sitting with.

The one in vivo test combining this with immunotherapy is small but unusually clean in its arms. In a murine lung adenocarcinoma model, mean tumour volume at day 28 ran 1,162 mm³ for control and 427 mm³ for anti-PD-1 alone. Adding oral L-arginine gave 452 mm³ — no improvement at all. Adding L-citrulline on top brought it to 198 mm³, with progression-free survival rising from 11 to 17 days and CD8 tumour-infiltrating lymphocytes roughly doubling.

The arginine arm is the control that matters. Supplementing the amino acid the T cells actually need did nothing. The effect appeared only when it was supplied in the form the tumour cannot intercept. That is the inversion stated as an experiment rather than a theory.

Part IV / The Catch

The Same Marker Cuts Both Ways

Here is where enthusiasm has to stop, because the selectivity argument contains its own boundary and that boundary is not a technicality.

Everything above depends on the tumour having silenced ASS1. Against an ASS1-competent tumour, citrulline feeds both compartments and the advantage simply disappears. That alone would make ASS1 stratification mandatory for any trial.

But it is worse than neutral in that group. ASS1-high breast cancers do not respond to immune checkpoint inhibitors, and the mechanism is metabolic: ASS1 drives purine synthesis, the resulting pyrimidine-to-purine imbalance suppresses immunoproteasome expression, and CD8 T cells become unresponsive to anti-PD-1. Inhibiting purine synthesis in those tumours restores the balance and the response.

So ASS1 status splits patients into two groups needing opposite interventions. ASS1-silenced: the tumour cannot use citrulline, the T cell can, and supplying it is rational. ASS1-high: the tumour uses citrulline perfectly well, is already resistant to checkpoint blockade through purine metabolism, and wants a purine synthesis inhibitor instead. A single intervention applied without knowing which group a patient is in would be expected to help one and do nothing useful for the other.

And Four Things Nobody Has Done

Stated plainly, because the gap between this argument and anything clinical is wide.

• No citrulline-alone arm. The lung cancer study gave arginine and citrulline together. Citrulline's isolated contribution is inferred from the arginine arm doing nothing, which is suggestive but not the same as testing it.
• No study stratified by ASS1. The entire selectivity case rests on a marker that no citrulline experiment has yet measured, despite it being routine immunohistochemistry.
• No supplementation data for the memory finding. The urea-cycle paper characterizes what T cells do endogenously. It did not administer citrulline, combine with immunotherapy, or run adoptive transfer.
• No mechanism confirmation. Nitric oxide was never measured in the one in vivo study, so the proposed vascular and signalling contributions remain unverified.

Add the usual translation problem: the mouse protocol used 2 g/kg daily by mouth, which does not convert into a human dose by arithmetic, and these were subcutaneous implants in a single model.

Part V / The Design Rule

Look for What the Tumour Discarded

The useful generalization is not about citrulline. It is about where to look.

Every failed intervention in this series tried to exploit something the tumour needs. That fails because need does not imply vulnerability when the needing cell has five ways to meet the need and its neighbours have none. The successful case exploits something the tumour gave up — and a discarded capability has no redundancy behind it, because the cell deleted it deliberately.

This shape is not unique to ASS1. MTAP is co-deleted with CDKN2A in a large number of cancers; losing it leaves those tumours unable to salvage methionine from methylthioadenosine, which is what makes them selectively sensitive to MAT2A and PRMT5 inhibition. Same logic: the deletion that conferred an advantage is the one that removes an escape route. The difference with ASS1 and citrulline is that the intervention is a nutrient rather than an inhibitor — you are feeding the system something only one side can metabolize, instead of blocking something both sides need.

It also puts the earlier conclusion of this series in a cleaner form. "Enhance clearance rather than block production" kept turning out to be the right answer — ammonia-clearing agents reactivating T cells and improving anti-PD-L1 where enzyme inhibitors hurt them, broad glutamine antagonism outperforming selective glutaminase inhibition, methionine supplementation rather than restriction restoring T cell function. Those are all the same rule. They work because they exploit an asymmetry the tumour created, rather than imposing a shortage that lands on whoever is least able to adapt.

Selected References

1. Diversion of aspartate in ASS1-deficient tumours fosters de novo pyrimidine synthesis.
   Rabinovich S, Adler L, Yizhak K, et al. Nature. (2015) — Nature. Why silencing ASS1 is a trade: cytosolic aspartate rises, CAD is activated by substrate availability and S6K1 phosphorylation, pyrimidine synthesis and proliferation follow.
2. Targeting purine synthesis in ASS1-expressing tumors enhances the response to immune checkpoint inhibitors.
   Keshet R, Lee JS, Adler L, et al. Nature Cancer. (2020) — Nat Cancer. The other half of the marker: ASS1-high tumours resist checkpoint blockade through purine synthesis and immunoproteasome suppression.
3. Ammonia detoxification promotes CD8⁺ T cell memory development by urea and citrulline cycles.
   Tang K, Zhang H, Deng J, et al. Nature Immunology. 24(1):162–173 (2023) — Nat Immunol. Characterizes the endogenous cycles; does not test supplementation.
4. Enhancement of anti-PD-1 immunotherapy in non-small cell lung cancer using arginine and citrulline supplementation.
   Miyamoto N, Yoshida M, Tsukumo S, et al. Journal of Thoracic Disease. 17(7):4814–4825 (2025) — Full text. The arm-by-arm volumes, and the authors' own stated limitations.
5. A pan-cancer analysis of the role of argininosuccinate synthase 1 in human tumors.
   Ding et al. Frontiers in Oncology. 13:1049147 (2023) — Full text. ASS1 silencing across tumour types, and its retention in normal tissue.
6. Cancer SLC43A2 alters T cell methionine metabolism and histone methylation.
   Bian Y, Li W, Zou W, et al. Nature. (2020) — Nature. The methionine row of the table.
7. Glutaminase inhibition impairs CD8 T cell activation in STK11-/Lkb1-deficient lung cancer.
   Best SA, Gubser PM, et al. Cell Metabolism. 34(6):874–887 (2022) — Full text. The glutamine row.
8. T cell activation depends on extracellular alanine.
   Ron-Harel N, Ghergurovich JM, Notarangelo G, et al. Cell Reports. 28(12):3011–3021.e4 (2019) — Record. The alanine row.
9. Microenvironmental ammonia enhances T cell exhaustion in colorectal cancer.
   Bell HN, Shah YM, et al. Cell Metabolism. (2023) — PMC. Clearance rather than blockade.
10. L-Citrulline: immunometabolic agent for cancer therapy.
    Synergies for Cancer Treatments — the longer case for citrulline, including the arginase-bypass and vascular arguments not covered here.