Absorption, the Liver and Enterohepatic Circulation

About 2 litres of drink enter an adult gut each day, the body adds roughly 7 to 9 litres of its own secretions, and about 200 grams of stool comes out the far end. Stool water is a small difference between two large flows, which makes it a sensitive quantity. This lesson is about the transporters that set it, the route fat takes out of the gut, and the organ that receives everything else first.

Digestive systems walks a meal down the tube, digestion has the enzyme chemistry, and gut motility and secretion covers the secretions themselves. Start from the products of hydrolysis in the lumen.

Eleven litres in, two hundred grams out

Take an adult human at steady state, with no change in the fluid held in the gut. Reviews give intake as about 2 L a day and secretions (saliva, gastric juice, bile, pancreatic juice, intestinal fluid) as roughly 7 to 9 L a day, so the load entering the lumen is about 9 to 11 L a day. About 99 percent of the total is reabsorbed. Daily stool mass varies with diet and population; about 200 g a day is a round illustrative figure, not a fixed adult average. The colon can absorb up to 4 to 5 L a day, reserve capacity rather than routine load.

stool water = load entering the lumen x (1 - fractional reabsorption)

Worked problem: the water balance. Round illustrative numbers: intake 2.0 L a day plus secretions 9.0 L a day gives a load of 11.0 L a day.

1. At 99 percent reabsorption. Stool water = 11.0 x 0.01 = 0.11 L a day, consistent with a stool of about 200 g a day.
2. At 96 percent. Stool water = 11.0 x 0.04 = 0.44 L a day, four times as much for a change of three percentage points. Check: 10.56 + 0.44 = 11.0 L a day.
3. Read it. The reabsorbed fraction sits close to one, so what you observe is the residue, which magnifies any shift in transport.

Circular folds, villi and microvilli give the area that makes 99 percent possible, and secretions lost rather than reabsorbed take their bicarbonate with them, which is how the route of a loss sets the consequence in acid-base balance.

Crypts secrete, villi absorb, and the stool is the difference

Along the crypt-villus axis of mammalian small intestine, secretion predominates in the crypts and absorption on the villi. In the classic model crypt cells carry the secretory machinery and villus cells the absorptive machinery, though reviews note that both processes can run in the same cell (epithelia and membrane transport have the routes, pumps and carriers).

In a crypt cell, the basolateral sodium pump and NKCC1, which brings sodium, potassium and chloride in together, load chloride above its electrochemical gradient. Apical CFTR, a chloride channel regulated mainly by cyclic AMP- and cyclic GMP-dependent protein kinases, is the exit. Open CFTR and chloride leaves, sodium follows paracellularly to balance charge, and water follows the osmotic force of both. On the villus, the apical sodium/hydrogen exchanger NHE3 works with apical chloride/bicarbonate exchange, mainly DRA, to absorb sodium chloride electroneutrally, and SGLT1 carries sodium with sugar. In the distal colon sodium also enters electrogenically through ENaC. Net fluid movement is the sum of secretion and absorption along the small intestine and the colon, together with the luminal solute load and transit time.

The classic experiment. Into the 1960s the dominant explanation of cholera diarrhea was a nineteenth-century model from Virchow: fluid leaking from the circulation through a damaged lining. In cholera, clearance of small molecules such as mannitol, a paracellular probe, into the stool was found to be very low, which already made filtration untenable.

Michael Field and colleagues took the question to an isolated sheet: isolated rabbit ileal mucosa in an Ussing-type chamber, voltage clamped at zero so the current supplied measures net electrogenic ion transport, with radioisotopes giving each direction separately. On that picture a sheet with no circulation behind it should not be able to drive ions toward the lumen either; the report was the opposite. Toxin added on the luminal side raised short-circuit current slowly over hours, and at 3 to 4 hours the net chloride flux was secretory and net sodium absorption had been reduced to zero, not reversed. Adding glucose to the lumen then restored a large net sodium absorption, so that route was intact. The authors read this as a cyclic AMP-dependent active secretory process of the epithelial cells, which points to secretion by the epithelium rather than filtration. Separate measurements in living rabbit ileal loops tied the same electrical pattern to fluid movement. Scope: one species, one segment, an isolated sheet with no blood flow or normal motility, with only whatever neural elements survive stripping, at a voltage held fixed rather than free to vary as it does in vivo.

Why glucose in the lumen changes the sodium flux

A messenger that opens CFTR and inhibits NHE3 cuts one absorptive route. It leaves the other working. SGLT1 runs on the inward sodium electrochemical gradient maintained by the basolateral pump, which is why poisoning that pump stalls it (digestion works that through in one cell). Add a transported sugar on the luminal side and sodium absorption through SGLT1 proceeds whether or not the cyclic-nucleotide-sensitive route works.

Human SGLT1 couples two sodium ions to each sugar, a 2:1 stoichiometry measured in transport studies, with structural work and mutagenesis identifying two likely sodium sites. The 2:1 ratio is the fixed coupling per transport cycle, not a cap on how much is absorbed. Take a sealed litre holding 75 mmol of a transported sugar and 75 mmol of sodium, the figures of the reduced-osmolarity rehydration solution: with no sodium coming in, 75 mmol of sodium can carry only 37.5 mmol of sugar through SGLT1. That is an accounting exercise about a closed compartment. In the gut, where luminal sodium is high and continuously resupplied, sodium is not the limiting partner.

Fat takes the other door

Sugars and amino acids leave the enterocyte basolaterally into capillaries draining to the portal vein. Most long-chain fat does not. Bile acids combine with the products of lipolysis into mixed micelles about 4 to 5 nm across, small enough to reach the spaces between microvilli. Westergaard and Dietschy, in rabbit jejunum in vitro, corrected for the unstirred layer of water held against the brush border and found the micelle’s main role is to beat that resistance, with absorption proceeding from individual lipid molecules in equilibrium with the micelles.

Uptake is partly protein-assisted: in mice, NPC1L1 lies in the ezetimibe-sensitive sterol pathway, and CD36-null mice with lymph fistulae secrete less lipoprotein into lymph. Inside the cell, products are re-esterified and microsomal triglyceride transfer protein (MTP) assembles lipid with newly made apoB-48 into chylomicrons; MTP activity was absent in intestinal biopsies from four people with abetalipoproteinemia. Chylomicrons pass into the lamina propria, enter the villus lacteal and travel the mesenteric lymphatics to the thoracic duct (capillary exchange and lymph). Short- and medium-chain fatty acids enter portal blood directly, and some long-chain fatty acids do too, but most dietary long-chain lipid leaves in chylomicrons and bypasses the liver’s first pass.

A pool used five times a day

Bile acids are made from cholesterol in the liver, held in the gallbladder between meals, delivered to the duodenum, used, and recovered, mostly actively and mostly in one place: ASBT in the terminal ileum. Returning bile acids are removed efficiently in their first passage through the sinusoids, NTCP carrying most conjugated bile acid uptake, with BSEP the major canalicular export pump. CYP7A1 is the rate-limiting enzyme of the classic synthesis pathway, and gut bacteria make the secondary bile acids deoxycholic and lithocholic acid. Human conjugation is mostly to glycine, about 3:1 over taurine; mice are about 95 percent taurine, so pool composition does not transfer.

Worked problem: the bile acid budget. Human review figures: pool 3.5 g, about 5 cycles a day, about 95 percent reabsorbed per cycle, fecal loss 200 to 400 mg a day. Assume steady state and that the whole pool recirculates each cycle.

1. Delivery to the intestine. 3.5 g x 5 a day = 17.5 g a day.
2. Loss predicted at 95 percent. 17.5 x 0.05 = 0.875 g a day, two to four times the measured 200 to 400 mg a day, so the textbook numbers do not close.
3. Solve for the implied efficiency. At a mid-range loss of 0.30 g a day, fractional escape = 0.30 / 17.5 = 0.017, so reabsorption is about 98.3 percent per cycle. The 95 percent figure is rounded; use one self-consistent set at a time.
4. Synthesis. It replaces 0.30 g a day, 8.6 percent of the pool; the rest of the delivery is recycled, not made.
5. Sensitivity. At 90 percent reabsorption, loss = 17.5 x 0.10 = 1.75 g a day, six times as much, with synthesis rising to match.

Pool size 3.5 g Cycles per day 5 Fractional reabsorption per cycle 98.3 percent

Ileal bile acid uptake

ileal uptake intact ileal uptake blocked

Delivery to the ileum = pool × cycles

Fecal loss = delivery × (1 − f)

Required synthesis, as a share of the pool per day

Reabsorption per cycle used by the sketch

In this sketch: steady state, and the whole pool recirculates each cycle, which is the lesson’s own assumption. Pool size, cycles per day and reabsorption per cycle are set independently, so some combinations may not occur in a person. The reference values are the lesson’s human review figures: pool 3.5 g, about 5 cycles a day, about 95 percent reabsorbed per cycle, fecal loss 200 to 400 mg a day. Made up for the sketch: the slider limits (pool 2.0 to 5.0 g, 3 to 8 cycles a day, 85.0 to 99.5 percent), the 90.0 percent cap used when ileal uptake is blocked (the lesson’s sensitivity case uses 90 percent; the ASBT-deleted mice excreted 10- to 20-fold more bile acid in stool, which this sketch does not reproduce), the logarithmic loss scale and the arrow thickness. Values are rounded for display: at the default the loss is 0.2975 g per day, so synthesis reads 8.5 percent of the pool, where the lesson, rounding the loss to 0.30 g first, gets 8.6 percent.

Rule to keep: at steady state synthesis equals fecal loss, and fecal loss is the delivered flux times the fraction that escapes reabsorption, so a small change in a number close to one moves the loss several-fold. The three textbook figures (a 3.5 g pool, about 5 cycles a day, about 95 percent reabsorbed) are rounded and do not close with a measured loss of 200 to 400 mg a day; that mismatch is the point, and the readouts are arithmetic on assumed inputs, not a measurement.

The brake on synthesis runs through the gut. Bile acids taken up by ileal ASBT activate FXR, a nuclear receptor (nuclear receptors), which induces FGF19; FGF19 reaches the hepatocyte and acts through FGFR4 with beta-klotho to lower CYP7A1. Mice lacking the ortholog FGF15 have more hepatic CYP7A1 and more fecal bile acid.

The modern experiment. Dawson and colleagues deleted Slc10a2, the gene for ASBT, in mice. If other routes could take over, the pool would stay roughly intact. Instead fecal bile acid excretion rose 10- to 20-fold, and the pool fell by about 80 percent and was enriched in cholic acid despite increased synthesis, with no steatorrhea on a low-fat diet. The authors concluded that ASBT is needed for efficient intestinal bile acid absorption and that the alternative routes cannot compensate for its loss, even with synthesis increased. It does not show the same consequences in people, or behaviour at a high fat load.

Reasoning from the lesion. Interrupt ileal uptake and the loop reads forwards. Delivery to the colon rises, and bile acids there have been shown to induce fluid secretion and raise mucosal permeability, so colonic net fluid absorption falls. Ileal FXR activation and FGF19 fall, the brake on CYP7A1 is released, and synthesis rises. With mild malabsorption the extra synthesis can hold the pool; with severe loss, as in these knockout mice, the pool shrinks anyway, leaving fewer bile acids to form micelles at a meal. Fat absorption need not change: those mice had no steatorrhea at a small fat load, though their cholesterol absorption fell about 20 percent. People with SLC10A2 mutations do present with diarrhea and steatorrhea, so the mouse result is not the human one.

The liver as the first organ downstream

Textbooks give about 75 percent of hepatic blood flow by the portal vein and about 25 percent by the hepatic artery, in a fasting adult at rest (measured human values vary, and portal flow rises after a meal), so most of what the gut absorbs, long-chain fat excepted, meets hepatocytes first. For a compound completely absorbed with no gut-wall metabolism, systemic availability F = 1 - E_H, where E_H is the fraction removed in one hepatic passage; a measured F of 0.10 to 0.20 gives hepatic extraction of 0.80 to 0.90 if these conditions hold; if they are unverified, these values are only upper bounds.

F = 1 - E_H, for a completely absorbed compound with no metabolism in the gut wall

Between meals the liver is a glucose source, by glycogenolysis and gluconeogenesis (integration of metabolism, gluconeogenesis and insulin, glucagon and glucose hold the pathways and their control). They do not run in sequence. Rothman and colleagues followed liver glycogen in healthy humans by carbon-13 NMR through a 68-hour fast with a labelled tracer for whole-body glucose production: gluconeogenesis already supplied 64 percent of production in the first 22 hours, then 82 and 96 percent in the following 14 and 18 hours.

The liver makes about 10.5 g of albumin a day in healthy people, a turnover of about 25 days, and the evidence that it readily doubles this on demand is thin: in one controlled study of a volunteer under daily plasmapheresis, synthesis rose about 25 percent.

Hepatic bilirubin handling takes three steps: uptake from blood, conjugation and export into bile. Heme breakdown is in blood cells and hemostasis; begin with unconjugated bilirubin arriving bound to albumin. UGT1A1 conjugates it with glucuronic acid, transporters including MRP2 export the conjugate into bile, and gut bacteria convert it to urobilinogen and urobilin. If heme breakdown outruns conjugating capacity, unconjugated bilirubin rises and does not appear in urine, because albumin-bound bilirubin is not filtered in free form. If conjugating capacity is low, as with the UGT1A1 promoter variant of Gilbert’s syndrome, unconjugated bilirubin rises with export intact. If canalicular export fails, through loss of MRP2 or obstruction of bile flow, conjugated bilirubin rises and appears in urine because it is water soluble. When bile flow is blocked, less conjugate reaches the gut, so urobilinogen and stool pigment fall. One caution concerns assays rather than physiology: delta-bilirubin is conjugated bilirubin bound covalently to albumin, so it is part of the conjugated pool but is not freely excretable, and the diazo method also reads a little unconjugated pigment. “Direct” bilirubin is therefore not the same as conjugate the liver can still clear into bile.

Design problem

You have isolated rabbit ileal mucosa in an Ussing-type chamber, sodium and chloride radioisotopes, glucose, mannitol as a paracellular probe, and a cell-free bacterial filtrate that produces secretion. Design measurements that separate three mechanisms for fluid appearing in the luminal bath: active electrogenic anion secretion, loss of barrier integrity, and reversal of the villus absorptive pumps. Then design the one further test that asks whether an absorptive route survives the filtrate.

One way to do it

Start electrically. Clamp the voltage at zero and record short-circuit current, which reports net electrogenic ion transport, with conductance alongside it. A rise in current with little change in conductance points to electrogenic transport; a large rise in conductance can mean either open channels or a junctional permeability change, so read it with the tracer measurement.

Then resolve the net fluxes. Measure unidirectional sodium and chloride fluxes, one isotope per bath, so that net secretion separates into which flux moved. A rise in serosa-to-mucosa chloride rules in active anion secretion; a fall in net sodium absorption without reversal is a reduced absorptive flux rather than evidence of a reversed pump, and transepithelial fluxes cannot settle the direction of any one membrane pump.

Test the barrier with mannitol clearance, which should not rise if the junctions held, and with resistance over time. Test the messenger with bumetanide serosally, for whether the current depends on NKCC1 chloride loading, and with a membrane-permeant cyclic AMP analogue, for whether the messenger alone reproduces the response.

The further test: add glucose to the mucosal bath and ask whether sodium-coupled absorptive flux still rises, with a non-transported sugar at the same concentration as the control separating an osmotic effect from cotransport. Limits: one species, one segment, an isolated sheet with no blood flow or normal motility, with only whatever neural elements survive stripping, at a voltage held fixed rather than free to vary.