Fructose metabolism: KHK, ChREBP and hepatic lipid synthesis

Fructose provides both carbon for lipid synthesis and signals that can increase lipogenic enzyme expression. The important question is how much reaches each tissue, how quickly, and against which background of energy intake, insulin action and existing liver disease. A pathway diagram establishes biochemical possibilities; it does not establish that every fructose-containing food causes liver injury.

Evolution, food form and terminology

Humans and other apes lack functional urate oxidase, UOX. Ancestral-enzyme reconstruction supports a progressive decline of uricase activity during primate evolution, followed by gene inactivation. Uricase initially converts uric acid into 5-hydroxyisourate; subsequent reactions yield allantoin. The idea that higher urate promoted energy storage during periods of scarce fruit is an evolutionary hypothesis, not proof that a specific dietary dose causes modern metabolic disease. Neither historical daily fructose intakes nor a universal current intake can be inferred from this hypothesis (Kratzer, 2014).

Sucrose yields approximately equal molar amounts of glucose and fructose after digestion. HFCS-55 contains about 55% fructose and 42% glucose in its dry sugar fraction, with other saccharides making up the remainder; other commercial formulations differ. The relevant exposure includes total added sugars, food form, portion and energy balance. Whole fruit supplies water, fibre and an intact food matrix and usually takes longer to eat than a sweetened beverage. Juice and a rapid experimental fructose bolus are not interchangeable with whole fruit. Adding energy through sugary drinks differs from replacing another carbohydrate while holding energy constant.

MASLD, metabolic dysfunction-associated steatotic liver disease, requires hepatic steatosis and at least one defined cardiometabolic risk factor, with alcohol exposure and other causes assessed within the steatotic-liver-disease classification. Older trials use NAFLD terminology and should retain their original population descriptions. MASH additionally involves steatohepatitis; progression to fibrosis and cirrhosis varies and is not inevitable. Visceral adipose delivery, dietary fat, synthesis, oxidation and export all influence hepatic fat, as explained in the article on ectopic fat and the TOFI phenotype (Rinella, 2023).

Intestinal handling before hepatic exposure

Apical GLUT5, encoded by SLC2A5, transports fructose by facilitated diffusion; basolateral GLUT2 participates in exit from the enterocyte. Glucose uptake also uses the sodium-dependent SGLT1 system. Reports of apical GLUT2 recruitment depend on model and conditions and do not justify one universal human absorption scheme.

Enterocytes contain KHK and aldolase B and can convert fructose into glucose and organic acids. Oral isotope tracing in mice demonstrated substantial intestinal clearance at lower loads and greater delivery to liver and colon when intestinal capacity was exceeded (Jang, 2018). Those mouse doses are not a validated number of apples or a safe beverage threshold for humans. Eating speed, adaptation, co-ingested glucose and intestinal capacity affect exposure. The portal liver sees a different concentration profile from peripheral blood, but a fixed portal concentration or hepatic extraction percentage cannot be assigned to every meal.

KHK isoforms, kinetics and structure

The human KHK gene lies at 2p23.3. Mutually exclusive alternative exons generate KHK-A and KHK-C. The protein regions differ over residues 72–115 in the structural alignment, affecting the central fold and dimer interface. KHK-C is prominent in liver, kidney and intestine, whereas KHK-A is more widely distributed. These are expression preferences, not absolute tissue restrictions. Recombinant-enzyme measurements show higher fructose affinity for KHK-C; reported constants depend on assay conditions (Asipu, 2003; Diggle, 2009; Trinh, 2009).

The estimates use the simple Michaelis–Menten relationship, not measurements of portal concentrations or whole-organ flux:

$$\frac{v}{V_{\max}}=\frac{[S]}{K_m+[S]}.$$

Here [S] and Km have the same concentration units. At 2–5 mmol/L, the model does not predict complete saturation of KHK-C. Vmax and turnover values cannot be transferred between different preparations without enzyme abundance and assay conditions.

KHK belongs to the pfkB carbohydrate-kinase family and forms dimers. Each subunit has a central α/β domain and a smaller β-sheet domain contributing to the interface and active-site cleft. Trinh and colleagues resolved apo KHK-A, a KHK-A fructose/nucleotide-analogue complex and apo KHK-C. PDB 3B3L is the human KHK-C structure at 2.9 Å; PDB 2HLZ is KHK-A, not KHK-C. PDB 3NBD is a fungal lectin and is irrelevant to this enzyme. In the KHK-A complex, Asp15, Gly41, Asn42, Asn45 and Asp258 contact fructose; the conserved C-terminal motif organizes phosphate binding. Asp258 is the proposed catalytic base, not Asp16. A structure or model supports specific contacts; it does not prove every proposed inhibitor mechanism.

Isoform-switching mechanisms involving hnRNP proteins have been investigated in cancer models and should not be presented as a universal normal-liver response. In Ishimoto's mouse study, deleting KHK-A aggravated fructose-associated metabolic abnormalities, whereas deleting both A and C protected. That comparison is not a selective KHK-C-only knockout experiment (Ishimoto, 2012).

Phosphorylation, energy and urate

KHK catalyses an ATP-coupled reaction requiring magnesium:

$$\text{Fructose}+\mathrm{ATP}\longrightarrow\text{fructose-1-phosphate}+\mathrm{ADP}.$$

This biochemical shorthand suppresses ionization states and metal coordination. Fructolysis enters the triose-phosphate pool downstream of phosphofructokinase-1, bypassing that particular glycolytic control point. It still depends on substrate availability, ATP, downstream enzymes and tissue regulation. Fructose-1-phosphate also modulates glucokinase regulatory protein, promoting glucokinase release; the resulting glucose phosphorylation can support glycogen and other pathways. Glycogen synthesis branches through glucose-6-phosphate, not directly through pyruvate.

Rapid phosphorylation can accumulate fructose-1-phosphate and transiently lower ATP and available inorganic phosphate. KHK transfers ATP's phosphate; it does not directly consume free phosphate. ATP regeneration, phosphate trapping and adenine-nucleotide metabolism explain the energetic response:

$$2\mathrm{ADP}\rightleftharpoons\mathrm{ATP}+\mathrm{AMP}.$$ $$\mathrm{AMP}+\mathrm{H_2O}+\mathrm{H^+}\longrightarrow\mathrm{IMP}+\mathrm{NH_4^+}.$$

AMP availability and relief of phosphate-related inhibition can favour AMPD activity, including hepatic AMPD2. IMP is converted to inosine, then hypoxanthine; xanthine oxidoreductase, XOR, produces xanthine and uric acid. XDH and XO use different electron acceptors; oxidative products depend on enzyme state and conditions. Large ATP falls measured in particular experiments are not guaranteed after an ordinary fruit serving. The linked article on KHK, ATP depletion and urate examines this pathway and its renal and vascular context.

For the XO route, the following balanced net equations use neutral molecular forms and hydrogen peroxide as the oxygen-reduction product. Superoxide can also form; XDH uses a different electron acceptor.

$$\mathrm{C_5H_4N_4O}+\mathrm{H_2O}+\mathrm{O_2}\longrightarrow\mathrm{C_5H_4N_4O_2}+\mathrm{H_2O_2}.$$ $$\mathrm{C_5H_4N_4O_2}+\mathrm{H_2O}+\mathrm{O_2}\longrightarrow\mathrm{C_5H_4N_4O_3}+\mathrm{H_2O_2}.$$

The successive substrates are hypoxanthine, xanthine and uric acid.

Cell and animal experiments connect intracellular urate, mitochondrial oxidative stress, NOX4-associated processes and reduced aconitase activity with lipogenesis (Lanaspa, 2012). NOX4 predominantly generates hydrogen peroxide, and its regulation is not identical to p47phox-dependent oxidases. Oxidants can damage ACO2's iron–sulfur centre; this is different from claiming urate directly blocks the enzyme allosterically. Citrate export through SLC25A1 exchanges citrate and other substrates across the inner mitochondrial membrane, making carbon available for cytosolic synthesis. These findings do not establish complete cessation of the TCA cycle or a fixed loss of aconitase activity in every human liver.

Carbon routes from fructose to triacylglycerol

Aldolase B cleaves fructose-1-phosphate into DHAP and unphosphorylated glyceraldehyde. TKFC then phosphorylates glyceraldehyde to glyceraldehyde-3-phosphate, GA3P; triose-phosphate isomerase interconverts GA3P and DHAP. Downstream glycolysis generates pyruvate, including the phosphoenolpyruvate-to-pyruvate step. Mitochondrial pyruvate entry through MPC1/2 and pyruvate dehydrogenase supplies acetyl-CoA; citrate synthesis and export connect mitochondrial carbon with cytosolic DNL.

$$\text{Fructose-1-phosphate}\longrightarrow\mathrm{DHAP}+\text{glyceraldehyde}.$$ $$\text{Glyceraldehyde}+\mathrm{ATP}\longrightarrow\mathrm{GA3P}+\mathrm{ADP}.$$

The parallel GPD1 reaction supplies the glycerol backbone:

$$\mathrm{DHAP}+\mathrm{NADH}+\mathrm{H^+}\rightleftharpoons\text{glycerol-3-phosphate}+\mathrm{NAD^+}.$$

The simplified figure separates fatty-acid carbon from the glycerol backbone; intervening reactions are described in the text. Fructose carbon also becomes glucose, lactate and oxidized products. Microbial conversion to acetate can supply hepatic acetyl-CoA through ACSS2; Zhao's mouse experiments establish that route in the tested conditions, not an obligatory dominant source in all humans (Zhao, 2020).

ChREBP, SREBP-1c and the lipogenic machinery

ChREBP, encoded by MLXIPL, is a carbohydrate-responsive bHLH–leucine-zipper transcription factor. Its N-terminal glucose-sensing module contains LID, a low-glucose inhibitory domain, and GRACE, a glucose-response activation conserved element. Metabolite sensing, phosphorylation and protein interactions regulate activity. Classical models involve glucose-6-phosphate, xylulose-5-phosphate, PP2A, PKA/AMPK-dependent phosphorylation and 14-3-3-associated localization. They should not be reduced to a proven high-affinity fructose-1-phosphate ligand switch. ChREBP partners with MLX on carbohydrate-response elements, typically paired E-box motifs separated by five bases (Li, 2006).

ChREBP-α can promote transcription from the alternative ChREBP-β promoter. The shorter β isoform lacks much of the N-terminal inhibitory region; its initial characterization concerned adipose tissue (Herman, 2012). A feedback loop is not evidence that a single fructose bolus locks the human liver into lipogenesis for days. Insulin-responsive SREBP-1c provides an interacting transcriptional programme involving SCAP-mediated ER-to-Golgi trafficking and proteolytic activation. ChREBP and SREBP programmes cooperate but respond to different inputs.

ACLY releases cytosolic acetyl-CoA, biotin-dependent ACC1 supplies malonyl-CoA, and FASN uses acetyl/malonyl units and NADPH to make palmitate:

$$\text{Citrate}+\mathrm{CoA}+\mathrm{ATP}\longrightarrow\text{acetyl-CoA}+\text{oxaloacetate}+\mathrm{ADP}+\mathrm{P_i}.$$ $$\text{Acetyl-CoA}+\mathrm{HCO_3^-}+\mathrm{ATP}\longrightarrow\text{malonyl-CoA}+\mathrm{ADP}+\mathrm{P_i}.$$ $$\begin{aligned} \text{Acetyl-CoA}+7\text{ malonyl-CoA}+14\mathrm{NADPH}+14\mathrm{H^+}&\longrightarrow\\ \text{palmitate}+7\mathrm{CO_2}+8\mathrm{CoASH}+14\mathrm{NADP^+}+6\mathrm{H_2O}.& \end{aligned}$$

The first two equations use conventional biochemical shorthand; the third gives the usual net FASN reaction. SCD1 desaturates fatty acyl-CoA substrates; ELOVL6 elongates C16 chains toward C18. GPAT, AGPAT, lipin and DGAT subsequently build triacylglycerol through glycerolipid intermediates. Elongation is not esterification.

Malonyl-CoA inhibits CPT1A and can restrain mitochondrial long-chain fatty-acid entry. This is graded, compartment-sensitive regulation, not complete shutdown of oxidation. ACC2-associated local malonyl-CoA also matters. AMPK activation inhibits ACC through phosphorylation; it is incorrect to call berberine an AMPK inhibitor or inactive ACC a dephosphorylated enzyme.

Isotope measurements distinguish fractional DNL from absolute production and from total hepatic fat. Donnelly's NAFLD study attributed approximately 59% of measured hepatic triacylglycerol fatty acids to circulating nonesterified fatty acids, 26% to DNL and 15% to dietary fat. These are study-specific source fractions, not normal universal values or the proportion of every fructose dose converted into fat (Donnelly, 2005).

Storage, insulin signalling and injury

Neutral triacylglycerol stored in PLIN2-coated lipid droplets can buffer more reactive lipid species. Total liver fat therefore does not specify lipotoxicity. Membrane-associated sn-1,2-DAG can activate PKCε, a novel, calcium-independent but DAG-responsive PKC. Experimental phosphorylation of insulin-receptor Thr1160 in the human numbering scheme inhibits receptor kinase activity; the homologous mouse site is Thr1150. This is not an atypical PKC pathway. Impaired IRS–PI3K–Akt signalling can weaken FoxO1 inhibition and increase gluconeogenic gene expression. Continued lipogenesis alongside dysregulated glucose production is often termed selective hepatic insulin resistance, but multiple signalling and substrate mechanisms contribute (Petersen, 2016).

Ceramide effects depend on molecular species and location. CerS6 is associated with C16 species; CerS2 mainly generates very-long-chain ceramides. PP2A- and atypical-PKC-associated effects on Akt are model-dependent. Saturated acyl species can disturb ER membranes and calcium handling. The unfolded-protein response initially adapts through IRE1–XBP1, PERK–eIF2α–ATF4 and ATF6 processing; prolonged stress can engage JNK, CHOP and injury pathways. IRS1 Ser307 in rodents corresponds to human Ser312, and phosphorylation at this site is not invariably inhibitory.

SERCA function, ER calcium release and mitochondrial uptake through MCU can connect organelle stress. Permeability-transition-related mitochondrial dysfunction and BAX/BAK-mediated outer-membrane permeabilization are distinct processes. Cytochrome c, APAF1 and caspase-9/3/7 mediate intrinsic apoptosis; a permeability-transition pore is not required for every apoptotic event. These mechanisms explain possible injury pathways without making steatosis an irreversible progression to MASH.

VLDL export requires ApoB100, MTTP and phosphatidylcholine. Export may increase in MASLD and still fail to balance delivery and synthesis; universal ApoB deficiency is not the explanation. ATP, mitochondrial DNA and HMGB1 released from injured cells can engage macrophages. TLR4-related signalling can prime inflammatory responses, while inflammasome activation requires additional context. Palmitate should not be described as an established direct TLR4 ligand, nor urate crystals as obligatory in MASH. Kupffer cells and recruited immune cells contribute to inflammatory signalling.

TGF-β–SMAD2/3 and PDGF-associated pathways activate hepatic stellate cells, promote proliferation and matrix production, and alter retinoid-containing droplets. Collagen I/III, fibronectin and TIMP-regulated matrix turnover contribute to fibrosis; sinusoidal capillarization can impair exchange and favour hypoxia. PNPLA3 I148M, rs738409, and TM6SF2 E167K, rs58542926, modify susceptibility through lipid-droplet and export biology. Their effects interact with other exposures; neither variant supplies a fixed fructose-specific fibrosis multiplier for an individual (Romeo, 2008; Kozlitina, 2014).

What human interventions demonstrate

Lustig (2016) also examined nine-day sugar replacement with starch. The before/after design and observed weight loss limit attribution to fructose alone.

KHK inhibition is mechanistically informative; drug-discovery potency and animal effects cannot be converted into an approved clinical regimen. PF-06835919 and supplement mechanisms should not be conflated. Food reduction, energy balance, weight change and pharmacological enzyme blockade answer different questions.

Supplements and eating windows: mechanisms versus outcomes

EASL–EASD–EASO guidance does not recommend nutraceuticals as MASLD therapy because efficacy and safety evidence is insufficient. Selected trials of vitamin E or pioglitazone, such as PIVENS, do not validate the above combination or universal dosing.

A restricted eating window can alter intake and circadian exposure, but a 16:8 schedule does not guarantee glycogen depletion after a fixed number of hours. TFEB, autophagy, lipophagy and lysosomal acid lipase supply useful experimental mechanisms, not proof of extra clinical benefit. In the 12-month TREATY-FLD trial, 88 adults with obesity and NAFLD were randomized to an eight-hour eating window or daily calorie restriction with similar calorie goals; time restriction did not confer an additional liver-fat benefit (Wei, 2023). Adherence, food quality and achieved energy balance matter.

Interpreting follow-up measures

$$\mathrm{FIB\!\!\!-4}=\frac{\text{age}\times\mathrm{AST}}{\text{platelets}\times\sqrt{\mathrm{ALT}}}.$$

Age is in years, AST/ALT in U/L and platelets in 10⁹/L. In appropriate adults, values below 1.3 usually identify lower risk; values at or above 1.3 prompt further assessment, commonly elastography, and values above 2.67 suggest higher risk. For adults older than 65, a lower-risk threshold of 2.0 is used. These are risk pathways, not diagnoses, and clinical context can override a reassuring result. Fibrosis risk, cardiometabolic status and competing liver causes guide interpretation (EASL–EASD–EASO, 2024).

No fixed supplement course can promise enzyme normalization within 12–16 weeks or fibrosis regression by six months. Reducing sugary drinks, improving the overall dietary pattern, physical activity and individualized management of obesity, diabetes and cardiovascular risk are supported practical directions. Whole fruit need not be eliminated because a concentrated fructose experiment increased a biomarker. The biological detail is useful when it clarifies exposure and uncertainty, rather than substituting for assessment of an individual.