Ammonia

Ten molecular identities for one nitrogen atom. Only the centre is what a clinical ammonia assay reports — glutamate, the carriers, the incorporation products and the exits all read as zero. Glutamate sits between the free molecule and both arms because it is where assimilation happens: the instant nitrogen is fixed there, the measurement falls to zero while the atom is still in the tumour and still working. A tumour running its nitrogen economy well therefore looks, on the instrument, like a tumour with no nitrogen problem at all. Every card opens the slide behind it — hover for a preview, click to expand. The full set is below.

The argument

Nitrogen metabolism in tumours

Ammonia in Disguise

One nitrogen atom, ten molecular identities — and the mechanism behind each one.

October 2026

This deck unpacks the mechanism behind each card of the "Ammonia in disguise" figure. The spine of the argument: free ammonia is the only nitrogen species a clinical assay reports, and it is the form the nitrogen is least often in. Five colours run through the deck and match the figure: red for what free ammonia does while still free, violet for glutamate as the first fixed form, blue for the carriers, green for what the nitrogen is built into, amber for the exits.

Ammonia in Disguise

The argument

One measurement, ten fates

A clinical ammonia assay reports a single quantity: the sum of NH₃ and NH₄⁺ in plasma. Inside a tumour the same nitrogen also travels and acts as nine other molecules, and every one of them reads as zero.

Seen as itself

1

One molecule, two unmasked actions

Fixed as

1

Glutamate, where the nitrogen lands

Carried as

3

Glutamine, alanine, aspartate

Built into

3

Bases, glutathione, polyamines

Leaves as

2

Urea, and one that is installed

Free ammonia is the form it is measured in, and the form the nitrogen is least often in.

Set up the whole deck here. The five counts add to ten identities for one nitrogen atom. The first card reads one, not two, because the antacid and the immune effect are two actions of a single molecule rather than two identities. Stress that this is not a claim that the assay is wrong — it measures exactly what it claims to. The claim is that the quantity it measures is a poor proxy for the thing we actually care about, which is nitrogen flux through a tumour.

One measurement, ten fates

The chemistry and the supply

The chemistry

One equilibrium, two behaviours

NH₃ + H⁺ ⇌ NH₄⁺ · pKa 9.25

At pH 7.4 roughly 98–99 % of the pool sits as NH₄⁺. The two species share a pool but not a behaviour.

NH₃

Uncharged, lipid-soluble, crosses membranes on its own, and moves through aquaporins and the Rh glycoproteins RHAG, RHBG and RHCG.

NH₄⁺

A cation of almost the ionic radius of K⁺. It rides K⁺ channels, NKCC1 and NHE. It does not diffuse.

The consequence: diffusion trapping

NH₃ crosses into any acidic compartment, picks up a proton there, and becomes an ion that cannot cross back out. The compartment loses a proton and gains a trapped cation.

---

This one property generates two of the mechanisms in this deck: the alkaline tumour interior, and the loss of mature perforin from cytotoxic granules.

Same chemistry as any lysosomotropic weak base — chloroquine works on the identical principle.

Everything downstream rests on this slide. Two points worth labouring. First, the pKa is far above physiological pH, so the pool is overwhelmingly the ion; but it is the small uncharged fraction that does the moving, and it is continuously replenished by the equilibrium. Second, trapping is directional: a compartment only has to be more acidic than its surroundings to accumulate ammonium indefinitely. That is why the effect lands hardest on lysosomes and secretory granules.

One equilibrium, two behaviours

The supply

Where the free ammonia comes from

Route	Enzyme	What is released
Glutaminolysis	GLS1, GLS2	The glutamine amide nitrogen, as free NH₄⁺
Glutamate oxidation	GLUD1	NH₄⁺, with the enzyme run oxidatively
Purine nucleotide cycle	AMPD2, AMPD3	NH₃ from the deamination of AMP to IMP
Glycine cleavage	GLDC, AMT	NH₃ and CO₂, beside the one-carbon unit
Asparagine hydrolysis	ASRGL1, ASNase	NH₃ from the asparagine amide
Gut microbiota	Bacterial urease	NH₃ in the lumen, adjacent to colorectal tumours

In the colorectal model of Bell and colleagues, ammonia accumulated because the tumour had lost the capacity to detoxify it, not because it produced more. That moves the therapeutic target from supply to clearance, and it is why the two exit slides carry most of the weight.

The list is deliberately broad: ammonia is not a single-pathway by-product, it is the common residue of nitrogen turnover. Glutaminolysis dominates in most proliferating tumours, but the purine nucleotide cycle matters in muscle-adjacent and highly glycolytic tissue, and the microbiota contribution is specific to colorectal disease. The Bell finding is the one to dwell on: if accumulation is a clearance failure rather than an overproduction problem, then restricting glutamine is the wrong lever and installing an exit is the right one.

Where the free ammonia comes from

Seen as itself

Seen as itself · action one

The intracellular antacid

The step

NH₃ diffuses in down its gradient. Inside, it binds a free proton, and the ion it becomes cannot diffuse back out. One proton is removed from the cytosol per molecule, and the removal is not reversed.

Tumours hold a reversed pH gradient — an interior near 7.3–7.6 against an exterior near 6.7–7.1 — normally built by NHE1, CA IX and XII, MCT1 and MCT4 and the V-ATPase. Every one of those costs ATP or expression. Ammonia supplies the same alkalinity as a by-product of nitrogen turnover.

What the alkaline interior buys

• Higher glycolytic flux: PFK-1 and several partners sit on a steep pH response curve
• Resistance to apoptosis: caspases and the apoptotic nucleases have acidic-to-neutral optima
• Cytoskeletal remodelling and motility, downstream of the same gradient
• Ion trapping of weak-base chemotherapeutics away from their targets

Running the exit would also consume base. Keeping the nitrogen and keeping the alkalinity are one decision — see the urea slide.

The point to make here is economic rather than toxicological. The textbook framing treats intracellular ammonia as a hazard the cell must survive. In a tumour it is closer to a subsidy: the cell gets a pH advantage it would otherwise have to pay for with ATP-driven transport. Hold the footer line until the urea slide, then come back to it — the bicarbonate accounting is what makes the two red cards and the amber card a single strategy.

The intracellular antacid

Seen as itself · action two

Perforin never reaches maturity

01

Ammonia accumulates in tumour interstitial fluid, where the lymphocyte actually sits

02

NH₃ crosses into the secretory lysosome and protonates there, raising granule pH

03

At the raised pH, perforin dissociates from the proteoglycan that holds it

04

Once dissociated it is open to inactivation and proteolytic degradation

05

Less mature perforin is delivered at the synapse; serial killing falls

What it reaches

Natural NK cytotoxicity, antibody-dependent killing with rituximab and daratumumab, and engineered CAR-NK and CAR-T cells alike. The defect is in the effector, so arming the effector better does not rescue it.

---

Note the symmetry with the previous slide: the same proton capture that alkalinises a tumour cell alkalinises a granule. One mechanism, two beneficiaries pointing opposite ways.

Domagala et al., Cancer Research 2025

This is the cleanest mechanistic chain in the deck, and worth presenting slowly. The authors tie the effect explicitly to ammonia's lysosomotropic character: it is the weak-base pH shift, not a receptor or a signalling pathway. Two proposed counters from that paper are worth mentioning in discussion — equipping effector cells with an ammonia-metabolising enzyme, or engineering perforin to hold its association at higher pH.

Perforin never reaches maturity

Seen as itself · action two

T cells reprogrammed, then exhausted

The observation

In an autochthonous metastatic colorectal model, ammonia accumulates in the microenvironment. T cell metabolism and redox signalling shift, proliferation falls and exhaustion markers rise.

Patients carry the correlate: raised serum ammonia, and an ammonia-related gene signature that tracks with blunted T cell responses, worse outcomes and failure of checkpoint blockade.

The accumulation traced back to lost disposal capacity, with HNF4α governing urea-cycle gene expression — not to higher production.

The intervention

Enhancing clearance with an approved hyperammonemia agent reactivated T cells, reduced tumour size and extended survival. Lowering tumour-associated ammonia restored the response to anti-PD-L1.

---

This is the load-bearing result for the whole figure: an intervention aimed at the nitrogen economy, not at the tumour cell, changed immunotherapy response.

Bell et al., Cell Metabolism 2023

Two mechanisms, two compartments. The perforin effect is intragranular and physicochemical; this one is metabolic and transcriptional, acting on the T cell's redox state. They are not alternatives — both can run in the same microenvironment, and neither shows on an ammonia assay as anything other than a modestly raised number. Worth flagging in discussion that the clearance agent used clinically for hyperammonemia is the same chemistry as the phenylbutyrate slide later in this deck, which is why the two exits section matters practically rather than only conceptually.

T cells reprogrammed, then exhausted

Fixed as

Identity two · fixed as glutamate

The step that empties the assay

GLUD1 · reductive amination  α-ketoglutarate + NH₄⁺ + NADPH → glutamate

GLUL · glutamine synthetase  glutamate + NH₃ + ATP → glutamine

What the tracing showed

Spinelli and colleagues followed ¹⁵N-ammonia into more than two hundred metabolites in breast cancer cells and in xenografts. Reductive amination at GDH was the primary assimilation route; secondary transamination carried the label onward into proline and aspartate. Ammonia accelerated proliferation rather than poisoning it.

Why glutamate is its own identity

Either reaction converts a species the assay counts into one it does not. The nitrogen has not left the tumour and has not stopped working. It has only changed the molecule it is attached to — and the measurement falls to zero at that instant.

Every carrier and every structure in the rest of this deck runs through this one molecule.

Spinelli et al., Science 2017

This is the slide that converts the figure from a taxonomy into an argument. Before this reaction, ammonia is a measurable waste product. After it, the same atom is biomass. The assay cannot distinguish a tumour that never made ammonia from one that made a great deal and assimilated all of it — and those two tumours have opposite phenotypes. If asked about the direction of GDH: the enzyme is reversible and normally runs oxidatively, releasing ammonia. It is the accumulation of ammonia itself that pushes it into reductive amination, which is a nice example of a product becoming a substrate once its concentration rises.

The step that empties the assay

Carried as

Carrier one

Glutamine

Two nitrogens per molecule: the α-amino nitrogen and the amide nitrogen. The amide is the labile one. GLS hydrolyses it to free NH₄⁺; GLUL re-fixes it at the cost of ATP. The two enzymes run as a cycle.

The cycle looks futile. It is a buffer. It lets a cell hold nitrogen in a neutral, transportable, non-toxic form and release it only where and when it is wanted.

Glutamine is also the dominant interorgan nitrogen shuttle in the body and the most consumed amino acid in most proliferating tumours, which is why it is the figure's main spoke.

The normal analogue

The liver already separates these jobs by zone. Periportal hepatocytes run the urea cycle and dispose of nitrogen. Perivenous, glutamine-synthetase-positive hepatocytes scavenge whatever escaped.

---

Tumours reproduce the scavenging half of that arrangement without the disposal half. Glutamine-synthetase-positive, β-catenin-mutant hepatocellular carcinoma is the clearest case.

The GS/GLS cycle is the single most useful idea in this section. If someone objects that a cycle consuming ATP to undo its own product makes no sense, the answer is that it buys control: the cell decides where the amide nitrogen is released. A tumour with high GS is not detoxifying ammonia in the hepatic sense — it is warehousing nitrogen in a form that will not show on an assay and will not harm it.

Glutamine

Carrier two

Alanine

ALT / GPT  glutamate + pyruvate ⇌ α-ketoglutarate + alanine

Why this carrier exists

A transamination moves the amino group between two carbon skeletons. Nothing is released and nothing is measured. Alanine is the vehicle of choice when a cell wants to shed nitrogen and carbon in the same molecule.

The systemic version is the glucose–alanine cycle: muscle ships nitrogen to the liver as alanine, and gets glucose back.

The tumour version runs locally

In pancreatic cancer the stroma supplies it: stellate cells secrete alanine that the tumour cell takes up, sparing glucose and glutamine for biosynthesis rather than fuel.

---

Nitrogen crosses between two cell types, in a traceable direction, and the free ammonia concentration never moves at any point on the route.

The pancreatic example is the best one for the figure's thesis, because it is heterocellular. Any sampling method that homogenises a tumour loses the direction of the transfer, and an assay that reads only free ammonia never saw the transfer at all. If someone asks what would detect it: labelled alanine, or a co-culture with compartment-resolved tracing.

Alanine

Carrier three

Aspartate

GOT1 / GOT2  glutamate + oxaloacetate ⇌ α-ketoglutarate + aspartate

It is the limiting one

Aspartate synthesis needs an electron acceptor. That is why inhibiting the electron transport chain is antiproliferative: the cell runs out of aspartate before it runs out of ATP.

It builds both rings

It donates N1 of the purine ring, and the ring nitrogen of pyrimidines through CAD. Citrin, SLC25A13, is the carrier that moves it between mitochondrion and cytosol.

It is also the exit

Aspartate supplies the second nitrogen of urea. Every molecule is therefore a decision between ring and exit — and silencing ASS1 makes that decision permanent.

The third card is the hinge of the whole deck. It is unpacked on the urea slide.

Aspartate is where the carriers section and the exits section meet, so flag it here and pay it off later. The competition is literal and chemical: cytosolic aspartate is a substrate for both ASS1, which sends the nitrogen out as urea, and CAD, which builds it into a pyrimidine ring. There is one pool and two consumers, and expression of ASS1 sets the split.

Aspartate

Built into

Built into · one

The nucleotide ledger, atom by atom

Purine ring · four nitrogens

N1

aspartate

N3, N9

glutamine amide

N7

glycine

---

Carbons for completeness: C4 and C5 from glycine, C2 and C8 from 10-formyl-THF, C6 from CO₂.

Pyrimidine ring · two nitrogens

N1

aspartate

N3

carbamoyl phosphate, from the glutamine amide through CPS2 within CAD

---

Both ring nitrogens arrive from the carrier pool of the previous section. Neither has existed as free ammonia at any point.

¹⁵N from labelled ammonia or amide-labelled glutamine appears here. On an ammonia assay, all of it reads as zero.

Worth saying out loud: a dividing cell commits four nitrogens per purine and two per pyrimidine, permanently, and a tumour doubling its genome is making this commitment at scale. That is the real size of the nitrogen demand, and none of it is visible in the measured pool. This slide also sets up the urea slide. If aspartate is supplying N1 of every purine and the ring nitrogen of every pyrimidine, then anything that diverts aspartate away from the urea cycle increases the supply for exactly this ledger.

The nucleotide ledger, atom by atom

Built into · two

Glutathione — nitrogen as redox capacity

glutamate + cysteine →[GCLC/GCLM]→ γ-glutamylcysteine →[GSS]→ glutathione

Glutamate is spent twice

Once as the backbone of the tripeptide, straight out of the assimilation step. And again as the export currency of the xCT antiporter, SLC7A11, which trades intracellular glutamate for the extracellular cystine that supplies the cysteine.

Cysteine is usually the rate-limiting input. Glutamate decides how much of it can be brought in.

The corollary

The nitrogen economy sets the ceiling on glutathione synthesis. Glutathione sets tolerance of reactive oxygen species and resistance to ferroptosis.

---

So a nitrogen constraint is a redox constraint, by a route that no ammonia measurement passes through.

This is the slide where the figure stops being about ammonia and starts being about vulnerability. If glutamate supply is what gates both the GSH backbone and cystine import, then nitrogen availability sits upstream of the whole ferroptosis-sensitivity question — which is usually discussed as if it were purely a lipid and iron story. A fair caveat if challenged: the quantitative contribution of recycled ammonia nitrogen to the GSH pool specifically has not been measured as cleanly as it has for the amino acids in the Spinelli tracing. The pathway is certain; the flux share is an open number.

Glutathione — nitrogen as redox capacity

Built into · three

Polyamines — nitrogen as structure

ornithine →[ODC]→ putrescine →[SRM]→ spermidine →[SMS]→ spermine

The entry

Ornithine arrives from arginine through ARG1 or ARG2, or from glutamate through OAT. ODC is short-lived, MYC-driven and rate-limiting, which makes it the control point for the branch.

The methionine link

Each elongation adds an aminopropyl group from decarboxylated SAM, through AMD1. The nitrogen economy is therefore coupled directly to the methionine and SAM cycle.

What they do

They are polycations that bind DNA, RNA and chromatin as a counter-charge pool. Spermidine is also the sole substrate for hypusination of eIF5A, which a subset of proteins needs in order to be translated.

DFMO is the pharmacological probe for the whole branch, through irreversible inhibition of ODC.

The methionine coupling is the point most worth making, because it means this branch cannot be reasoned about as a nitrogen pathway alone. Every spermidine costs one decarboxylated SAM, so polyamine demand pulls on the methionine cycle, and anything that restricts methionine constrains polyamine elongation whether or not ornithine is plentiful. Note also the shared substrate with the urea cycle: ARG1 produces ornithine both for polyamine synthesis and as the regenerating intermediate of ureagenesis. Another case of one pool, two consumers.

Polyamines — nitrogen as structure

Leaves as

Exit one

Urea — the exit most tumours close

NH₄⁺ + HCO₃⁻ →[CPS1]→ carbamoyl-P →[OTC]→ citrulline →[ASS1 + aspartate]→ argininosuccinate →[ASL]→ arginine →[ARG1]→ urea

Per molecule of urea: two nitrogens leave — one from free NH₄⁺ at CPS1, one from aspartate at ASS1 — and two bicarbonate are consumed, one built into the carbonyl and one protonated to CO₂ and water. The second has been measured directly in perfused liver.

1 · The nitrogen stays

Nothing leaves. The whole supply from the earlier slide stays available to the carriers and the structures.

2 · Aspartate is freed

With ASS1 silenced, its substrate goes to CAD instead, and pyrimidine synthesis rises. Acidosis and hypoxia push ASS1 down further.

3 · The base is spared

No bicarbonate is consumed. The alkaline interior and the closed exit turn out to be the same decision.

Rabinovich et al., Nature 2015 · urea-cycle dysregulation as a general phenomenon, Cell 2018

This is the slide to spend the most time on. Three independent advantages arrive from a single act of silencing, which is what makes it look like design rather than damage. On the bicarbonate stoichiometry, since it sometimes gets challenged: the molecular accounting gives one bicarbonate incorporated by CPS1, and the acid-base accounting gives two consumed overall, because the two ammonium ions releasing protons require a second bicarbonate to neutralise them. The net reaction is 2 NH4+ + 2 HCO3- to urea, CO2 and water. Atkinson and Camien made the argument on stoichiometric grounds; the perfused rat liver experiment confirmed it by measuring extracellular acidification during stimulated ureagenesis. The downstream consequence, from the 2018 work: urea cycle dysregulation shifts the pyrimidine-to-purine ratio enough to leave a detectable mutational signature. The nitrogen decision is legible in the genome even when it is invisible in the plasma.

Urea — the exit most tumours close

Exit two

An exit that can be installed

phenylbutyrate →[β-oxidation]→ phenylacetate →[ACSM]→ phenylacetyl-CoA →[+ glutamine]→ phenylacetylglutamine → urine

Why it works where urea cannot

Two nitrogens leave per molecule — the glutamine amide and its α-amino nitrogen. Identical nitrogen content to urea, by a route that never touches the urea cycle.

That independence is the whole point. A tissue that has silenced ASS1 still conjugates glutamine, so the exit can be opened in exactly the tumours that closed the other one.

Conjugation happens in liver and kidney, and the product is cleared renally — so disposal is systemic, driven by drawing down the circulating glutamine pool.

Open questions worth naming

Pairing ornithine with phenylbutyrate is argued to work sequentially: ornithine replenishes the cycle intermediate and feeds glutamine synthesis, phenylbutyrate then carries that glutamine out. The rationale is coherent; the sequencing is not yet settled by trial data.

---

Phenylbutyrate is also an HDAC inhibitor. Any antitumour effect it shows cannot be attributed to nitrogen disposal without a control that separates the two.

Bell et al. used an approved hyperammonemia agent on this principle and restored checkpoint-blockade response.

Keep the caveat card visible while presenting. The HDAC confound is the first objection any oncologist will raise, and naming it first is stronger than defending against it later. The clean experiment would be a nitrogen-scavenging agent with no chromatin activity, or an HDAC inhibitor with no conjugation chemistry, run head to head. On the ornithine pairing: present it as a mechanistic argument with a clear rationale rather than as an established protocol. What would settle it is a measurement of PAGN excretion with and without ornithine loading, which is a tractable study.

An exit that can be installed

The measurement

The blind spot

A tumour running its nitrogen economy well reads, on the instrument, as a tumour with no nitrogen problem at all.

A pool is not a flux

A small pool turning over quickly reads the same as a small pool sitting still.

Plasma is not interstitium

The concentration that matters is the one where the lymphocyte sits, not the one in the vein.

One species is not nine

Nine of the ten identities in this deck are chemically invisible to the assay.

Pause here. This is the thesis of the figure stated as plainly as it can be put, and the three cards are the three distinct failures stacked underneath it — kinetic, spatial, and chemical. They compound: a tumour can be high-flux, high-interstitial and fully assimilated at once, and the number still comes back unremarkable. Add the pre-analytical point verbally if the audience is clinical: the plasma ammonia assay is fragile. It needs the sample on ice and separated quickly, and haemolysis inflates it. So the measurement is not only narrow, it is noisy in the narrow range it does cover.

The blind spot

The repair

What would actually measure it

Approach	What it reads	What it would settle
¹⁵N flux tracing	Label spreading across the nitrogen metabolome	Which of the ten fates actually carries the flux
Interstitial fluid sampling	The concentration where lymphocytes sit	Whether plasma under-reads the microenvironment
Urea-cycle expression	ASS1, CPS1, OTC, SLC25A13 levels	How far the exit has been closed, and by what
Pyrimidine : purine ratio	Nucleotide imbalance and transversion bias	Whether diverted aspartate reached the genome
Urinary PAGN and urea	Nitrogen disposal actually achieved	Whether an installed exit is carrying real flux
pH imaging, inside and out	The gradient across the tumour cell membrane	How much of the alkalinity ammonia contributes

The nitrogen is readable. It is simply not readable on an ammonia assay.

Close the argument constructively. The figure is not a complaint about an assay; it is a claim that the field has an instrument-shaped hole in its evidence, and that several of the things that would fill it already exist and are tractable. If asked which one to do first: the expression signature and the urinary disposal measurements are both cheap and available in existing cohorts, so they are the fastest way to test whether nitrogen retention tracks with anything clinical. The tracing work is the definitive answer but needs dedicated material.

What would actually measure it

The summary

The ten identities, side by side

Identity	Route	What it does	NH₃ assay
Free NH₃ / NH₄⁺	—	Buffers protons; raises granule pH in lymphocytes	Reads it
Glutamate	GLUD1, GLUL	The first fixed form; hub for all the rest	Zero
Glutamine	GLUL, GLS	Main shuttle; holds nitrogen in a neutral form	Zero
Alanine	ALT, GPT	Moves nitrogen and carbon between cell types	Zero
Aspartate	GOT1, GOT2	Donates purine N1 and the pyrimidine ring nitrogen	Zero
Nucleotide bases	CAD, PPAT	Four nitrogens per purine, two per pyrimidine	Zero
Glutathione	GCLC, GSS	Sets ROS tolerance and ferroptosis resistance	Zero
Polyamines	ODC, SRM, SMS	Bind chromatin; hypusinate eIF5A for translation	Zero
Urea	CPS1 to ARG1	Removes two nitrogens and two bicarbonate; usually silenced	Zero
Phenylacetyl‑glutamine	PAA conjugation	Removes two nitrogens into urine; can be installed	Zero

Use this as the recap slide and as the handout. The colour in the first column maps back to the original figure, so anyone who saw the figure can locate themselves immediately. The last column is the argument in one strip: one row says "reads it", nine say "zero". That asymmetry is the entire claim, and nothing in the deck depends on anything more contentious than it.

The ten identities, side by side

Sources

Where each mechanism comes from

Assimilation

Spinelli JB et al. Metabolic recycling of ammonia via glutamate dehydrogenase supports breast cancer biomass. Science 2017;358(6365):941–946. doi:10.1126/science.aam9305

T cells

Bell HN et al. Microenvironmental ammonia enhances T cell exhaustion in colorectal cancer. Cell Metabolism 2023. doi:10.1016/j.cmet.2022.11.013

Perforin

Domagala J et al. Ammonia suppresses the antitumor activity of natural killer cells and T cells by decreasing mature perforin. Cancer Research 2025. doi:10.1158/0008-5472.CAN-24-0749

Aspartate

Rabinovich S et al. Diversion of aspartate in ASS1-deficient tumours fosters de novo pyrimidine synthesis. Nature 2015.

Urea cycle

Urea cycle dysregulation generates clinically relevant genomic and biochemical signatures. Cell 2018.

Bicarbonate

Atkinson & Camien's stoichiometric account of ureagenesis as bicarbonate disposal, confirmed by measuring extracellular acidification in isolated perfused rat liver (Biochem J).

Volume and page numbers are given only where verified. Check the last three before publication.

Three citations on this slide are deliberately incomplete. The Rabinovich and Cell 2018 entries are correct as to title, journal and year but the volume and pages were not verified, and the bicarbonate entry names the argument and the confirming experiment without a full citation for either. Complete them from the DOIs before this deck goes anywhere public.

Where each mechanism comes from

✕

← Previous Next →