Acid-Base Balance

Adult human arterial plasma at rest holds about 24 millimoles of bicarbonate per litre and about 40 nanomoles of free hydrogen ions per litre. Metabolism in a US adult on a typical diet adds roughly 50 mEq of non-volatile acid a day, yet pH stays near 7.40, held on three clocks: buffers in seconds, lungs in minutes, kidneys over days. Water, pH and weak bonds gave you pH = pKa + log([A-]/[HA]), and the kidney the nephron; here both get numbers.

A buffer that would be unremarkable in a bottle

A volatile acid leaves as a gas: carbon dioxide, removed by the lungs. A non-volatile acid, such as sulfuric acid from sulfur-containing amino acids or an organic acid such as lactic acid or a ketoacid, cannot be breathed off; it is buffered, and the kidney excretes the net fixed-acid load. Lactate and ketoanions can also be metabolised, which gives bicarbonate back. A disturbance that starts with pCO2 is respiratory; one that starts with non-volatile acid or base is metabolic.

For human plasma at 37 C, with pCO2 in mmHg and bicarbonate in mmol/L:

pH = 6.1 + log10( [HCO3-] / (0.03 x pCO2) )

The 0.03 is the solubility of CO2 in plasma, in mmol/L per mmHg. The 6.1 is an apparent pK for the combined reaction CO2 + H2O = HCO3- + H+ in plasma at 37 C, hydration and ionization together; carbonic acid by itself is a much stronger acid. Henderson set out the mass-action relation in 1908, and in 1916 Hasselbalch rewrote it in Sorensen’s pH notation (carrying oxygen and carbon dioxide tells his work with Bohr and Krogh). At a pCO2 of 40 mmHg, dissolved CO2 is 0.03 x 40 = 1.2 mmol/L, the ratio is 24 / 1.2 = 20, and pH = 6.1 + 1.30 = 7.40. pH follows the ratio, not either number alone.

Now the puzzle. A pK of 6.1 sits 1.3 units below 7.40, and in a closed container the pair buffers no better than any other weak acid pair. Its power comes from being open to CO2: added acid turns bicarbonate into CO2, the lungs remove it, and the denominator does not climb. Add 10 mmol/L of strong acid to a made-up plasma with no other buffers. Sealed in a syringe, bicarbonate falls to 14 and dissolved CO2 rises to 11.2 mmol/L, so pH falls to about 6.20. With pCO2 held at 40 mmHg, dissolved CO2 stays at 1.2 and pH falls only to about 7.17. Alveolar ventilation holds that denominator (ventilation and perfusion).

pCO₂ before the load 40 mmHg Bicarbonate before the load 24 mmol/L Non-volatile acid load 10 mmol/L

The sample is

open to CO₂ closed to CO₂

pH

[H⁺], nmol/L

HCO₃⁻ to dissolved CO₂ ratio

Bicarbonate, mmol/L

Dissolved CO₂, mmol/L

pCO₂ (dissolved CO₂ ÷ 0.03), mmHg

Same load, the other way

In this sketch: one made-up compartment of human plasma at 37 C with no hemoglobin, protein, phosphate, cell or bone buffering and no renal response, so it shows direction and rough size and no more. pH = 6.1 + log10(HCO₃⁻ / (0.03 × pCO₂)) with the lesson’s pK of 6.1 and CO₂ solubility of 0.03 mmol/L per mmHg, and [H⁺] = 10^(9 − pH) nmol/L. The acid load takes bicarbonate away mmol for mmol. Open: the gas phase holds pCO₂, so dissolved CO₂ stays where it was. Closed: total CO₂ in the sample (bicarbonate plus dissolved CO₂) stays constant, so dissolved CO₂ rises by the amount of bicarbonate lost. Made up for the sketch: the slider ranges, the starting values you choose, and a floor of 1 mmol/L of bicarbonate (the logarithm fails at zero), so a load is capped at 1 mmol/L below the starting bicarbonate. The equation is applied across the whole slider range without checking that plasma could sit there, and the sketch gives no interpretation, classification or reference range.

Rule to keep: pH follows the ratio of bicarbonate to dissolved CO₂, not either number alone; added acid lowers the numerator either way, but only a system open to CO₂ keeps the denominator from climbing.

Uncatalysed interconversion of CO2 and bicarbonate is slow, with a half-time of seconds to tens of seconds, while blood spends about one second in a pulmonary capillary; carbonic anhydrase closes that gap.

Two limits follow. Plasma bicarbonate cannot be set independently of pCO2, so it mixes respiratory and metabolic effects, and the equation says nothing about other acids. Base excess, introduced in the late 1950s by Siggaard-Andersen with Astrup’s equilibration method, measures the metabolic part of a sample: the acid or base needed to bring whole blood in vitro to pH 7.40 at a pCO2 of 40 mmHg. Standard base excess adjusts it for the whole extracellular space, because in the body base excess still moves with pCO2. Bicarbonate read with pCO2, standard base excess and Stewart’s strong-ion difference give nearly the same answer for a given sample but explain it differently, and comparisons of Stewart’s approach with the traditional one have been inconsistent.

Where the protons actually go

Follow an acute load of non-volatile acid. Outside cells, bicarbonate takes up protons and becomes CO2 for the lungs, with plasma proteins and phosphate taking a smaller share; over hours cells and bone take up a further share. In a tissue capillary, CO2 entering a red cell is hydrated by carbonic anhydrase, most of the bicarbonate leaves on the Cl-/HCO3- exchanger AE1 as chloride enters, and hemoglobin buffers nearly all of the H+ formed. The Bohr and Haldane effects that couple this to oxygen are in carrying oxygen and carbon dioxide.

Intracellular buffering only limits how far and how fast intracellular pH moves; returning it to its set point takes acid-base transporters (membrane transport). In cultured neonatal mouse calvariae (skull bones) in vitro, acute metabolic acidosis led bone to take up hydrogen ions into its mineral phase, with surface carbonate depleted and sodium and potassium leaving.

Buffering is not removal. Every proton a buffer holds is still in the body, and bicarbonate spent on a fixed-acid load is gone until the kidney makes more, or until a retained lactate or ketoanion is metabolised. In a validation study that enrolled 1,082 US adults aged 50 to 74, the 765 who completed both 24-hour urine collections had a mean net endogenous acid production, estimated from urine, of about 47 to 48 mEq per day (interquartile range about 36 to 64); ketoacids add to such loads (insulin, glucagon and glucose). Whether lactate formation itself releases protons in working muscle is contested, and exercise physiology takes up that debate.

The proximal tubule gets the bicarbonate back

Human kidneys make about 160 to 170 L of ultrafiltrate a day (glomerular filtration). At 24 mmol/L, 165 L carries about 4,000 mmol of bicarbonate, roughly eighty times the day’s non-volatile acid; losing 3 percent would cost more base than that acid consumes. The proximal tubule takes back 60 to 70 percent of filtered water and NaCl and a greater proportion of the NaHCO3.

Reclaiming is indirect. Apical NHE3 exchanges Na+ in for H+ out, and a V-type H+-ATPase secretes more H+. In the lumen, carbonic anhydrase IV, anchored on the luminal membrane, turns H+ and filtered HCO3- into CO2 and water, fast enough to keep up with the fluid moving past the cells. CO2 enters the cell, cytosolic carbonic anhydrase II rehydrates it, the H+ is secreted again, and the HCO3- leaves across the basolateral membrane on the Na+-dependent cotransporter NBCe1. One bicarbonate vanishes from the lumen and one appears in the blood; no new base is made. The segment table and carbonic anhydrase inhibition as a probe are in tubular transport and diuretics.

NHE3 knockout mice survive mildly acidotic; their in situ microperfused proximal tubules absorbed 54 percent less bicarbonate and 63 percent less fluid, and a significant part of the remainder was carried by a bafilomycin-sensitive H+-ATPase. In humans, William Sly and colleagues identified carbonic anhydrase II deficiency (1983) as the defect in an inherited syndrome that includes failed renal acidification. Boron’s group has reported that two Na+-coupled “bicarbonate” transporters, studied in Xenopus oocytes, carry carbonate (CO3 2-), with the quantitative test made on NBCe1.

That the tubule secretes acid at all was once in doubt. An older view held that the tubule acidified urine by reabsorbing buffer base, filtered bicarbonate and dibasic phosphate, leaving the acid forms behind; then titratable acid excreted could not exceed what that reabsorption allows. Robert F. Pitts and colleagues tested this in 1945 to 1949, in clearance studies of dogs and of people (as summarised in 1956 by Orloff), setting filtered loads against excretion. In metabolic acidosis, titratable acid excretion exceeded what reabsorbing filtered bicarbonate or dibasic phosphate could explain. H+ was being added to tubular fluid, and Pitts proposed it was extruded in exchange for filtered sodium. Orloff stressed there was no direct evidence yet for ion exchange in tubule cells; these were whole-kidney measurements, and the molecular exchanger came decades later.

New bicarbonate: ammonium and titratable acid

Replacing bicarbonate that acid consumed takes new bicarbonate, made in proportion to acid excreted:

net acid excretion = titratable acid + NH4+ - HCO3-

About one-third to one-half is normally titratable acid, protons carried on urinary buffers, mainly phosphate. The other half to two-thirds is ammonium: proximal cells metabolise each glutamine to 2 NH4+ and 2 HCO3-, the bicarbonate going to the blood. Only excreted ammonium counts, because ammonium returned to the blood is handled by the liver in a bicarbonate-consuming step (protein turnover); here ammonium matters mainly as an acid-base carrier rather than as the nitrogen route of osmoregulation, though it does carry some nitrogen. People on a typical Western diet excrete 30 to 40 mmol of ammonium a day. In acidosis the proximal tubule takes up more glutamine and citrate and recovers less phosphate, raising both terms.

In the collecting duct, type-A intercalated cells pump H+ into the lumen through a V-type H+-ATPase and return new HCO3- to the blood on basolateral AE1. Type-B cells, the mirror image, secrete bicarbonate through luminal pendrin and keep the H+-ATPase on the basolateral side; isolated cortical collecting ducts from alkali-loaded pendrin-knockout mice failed to secrete bicarbonate (Royaux and colleagues, 2001). A third, non-A/non-B type has both apically, and what it does on balance is unsettled. Urine pH bottoms out near 4.5, a 1000-to-1 H+ gradient from plasma, so buffers must carry the acid. In acute acidosis distal H+ secretion rises at once while ammonium production lags; in chronic acidosis ammonium excretion is high and urine pH can rise to about 6. Urine pH is a poor index of ammonium excretion, and no master renal pH sensor has been identified.

The textbook account was diffusion and trapping: NH3 crosses membranes freely and is trapped as NH4+ in acid fluid, so deleting an ammonia transporter should matter little. Sophie Biver and colleagues (2008) acid-loaded Rhcg knockout mice beside wild-type controls and microperfused their collecting ducts. The knockouts could not acidify urine normally because ammonium excretion was impaired, and their ducts showed reduced apical NH3 permeability and impaired transepithelial NH3 transport. Protein-mediated transport is required in the mouse collecting duct. It did not show whether Rhcg carries NH3 or NH4+, says nothing about humans, and, tested under acid loading, does not show what sets baseline excretion.

Human genetics reaches the same segment through a different protein, the proton pump’s B1 subunit, whose gene Karet and colleagues found mutated in families with distal renal tubular acidosis and hearing loss. Predict the chain. Apical H+ pumping falls, so the duct builds a smaller gradient and urine pH stays higher; less phosphate is titrated and less ammonium held in the lumen, so net acid excretion, new bicarbonate, plasma bicarbonate and pH fall. Rising: retained acid buffered elsewhere, bone included; ventilation, so pCO2 falls; and plasma chloride, replacing lost bicarbonate, so the anion gap is unchanged. In carbonic anhydrase II deficiency, by contrast, the acidosis is proximal, distal or combined, so proximal reclamation can fail as well.

Reading the numbers, and what the rules are worth

Worked problem. In this made-up example for human plasma at 37 C, take pK 6.1, solubility 0.03 and steady state, and compare against bicarbonate 24 mmol/L, pCO2 40 mmHg, pH 7.40 and albumin 4.0 g/dL. The sample: pCO2 27 mmHg, HCO3- 12, Na+ 140 and Cl- 104 mmol/L, albumin 2.0 g/dL.

1. Dissolved CO2 = 0.03 x 27 = 0.81 mmol/L; 12 / 0.81 = 14.8; pH = 6.1 + 1.17 = 7.27. [H+] = 10^-7.27 mol/L, which is 10^(9 - 7.27) = about 54 nmol/L, against 40 at pH 7.40.
2. pH is low with a low bicarbonate. A primary rise in pCO2 would raise bicarbonate, so this is a metabolic acidosis: acid gained or base lost. The low pCO2 is moving pH back toward 7.40, and the anion gap below narrows the cause.
3. An empirical fit gives expected pCO2 = 1.5 x 12 + 8 = 26 mmHg; the measured 27 lies within the scatter one review attaches to it. A pCO2 of 36 would mean ventilation below the fit, pointing to a second, respiratory process.
4. Anion gap = [Na+] - [Cl-] - [HCO3-] = 140 - 104 - 12 = 24 mmol/L. An acid whose anion the panel does not measure, such as a ketoacid or lactate, consumes bicarbonate and leaves its anion, so the gap rises; base lost directly is replaced by chloride, leaving the gap unchanged.
5. Albumin is the main unmeasured anion. Corrected gap = 24 + 2.5 x (4.0 - 2.0) = 29 mmol/L, the same as 24 + 0.25 x (40 - 20) with albumin in g/L.
6. Against a made-up comparison gap of 12 (140 - 104 - 24), the gap rose 12 (17 corrected) as bicarbonate fell 12, ratios of 1.0 and 1.4. That ratio genuinely varies, larger for lactate than for ketoacid anions, so alone it cannot establish a second process.

The gap’s reference range is not fixed either: it shifted down in one laboratory when ion-selective electrodes raised measured chloride (Winter and colleagues, 1990), so each laboratory’s own range applies.

Compensation follows mechanism. In metabolic acidosis pCO2 falls. In respiratory disturbances the kidney moves bicarbonate in the same direction as pCO2, slowly: for a primary rise in pCO2, chemical buffering raises bicarbonate by about 1 mmol/L per 10 mmHg at once, and renal adaptation over days adds more, with one prospective study of chronic hypercapnia giving about 5 mmol/L per 10 mmHg (Gonzalez and colleagues, 2018), and altitude’s falling bicarbonate is told in control of breathing. Compensation usually blunts the pH change rather than erasing it, and in chronic hypercapnia pH can sit inside the normal range, so compare the measured values against an expected response and the clinical setting. The fits are regressions on observed people, not laws.

Design problem

In the chicken, blood from a leg vein drains directly into the peritubular region of the same-side kidney, so an agent infused into one leg reaches one kidney’s tubule cells while the other kidney is a control in the same bird. Orloff’s 1956 review reported, as unpublished observations, that acetic acid added to blood perfusing one side raised local CO2 tension (it reacts with plasma bicarbonate) and, on that side only, lowered potassium excretion (already raised by a potassium sulfate infusion) and raised urine H+ concentration; it credits the method to Sperber (1946). (a) What does the paired design buy that an intravenous infusion into a mammal cannot? (b) What shows whole-body acid-base state did not change? (c) Which rival explanation remains, and how would you exclude it?

One way to do it

(a) Each bird is its own control: systemic pH and pCO2, blood pressure and anaesthesia are shared, and the infused side meets the agent first in its peritubular region. Flow, filtration and delivery on each side still have to be measured rather than assumed equal.

(b) Arterial pH and pCO2 before, during and after, with urine collected simultaneously and separately from each ureter, and flow and electrolytes measured on each side.

(c) The acid also lowers peritubular bicarbonate and may change sodium delivery and local blood flow. An acid carrying no CO2 still makes CO2 when it meets plasma bicarbonate, so these infusions cannot clamp peritubular pCO2, bicarbonate and pH separately. Measure local pH, pCO2, bicarbonate, filtration and flow on each side, with matched vehicle and salt controls, and then set the bath directly in an isolated perfused tubule. Even then the potassium result only shows that local acid-base conditions shift potassium and proton handling together, without identifying a shared driving force or transporter, and one bird species cannot be generalised to mammals. The modern form of the logic is the isolated perfused tubule, with the bath set directly.