Osmolality, Tonicity, and Cellular Volume Regulation

A plasma sample can contain many dissolved particles without exerting a proportionately large sustained effect on cell volume. The distinction depends on which particles are present, which membranes separate them, and how rapidly they cross or disappear. Understanding these relationships connects laboratory measurements with water redistribution, cellular adaptation, and neurological vulnerability.

47.1 Osmolality Counts Particles in Water

Osmolality is the concentration of osmotically active particles per kilogram of solvent, usually reported as mOsm/kg of water. Osmolarity expresses particle concentration per liter of solution, in mOsm/L. The denominators differ: solvent mass is not solution volume. At physiological concentrations the numerical values are close enough for many clinical estimates, but the terms are not formally interchangeable.

Particle number matters more than molecular mass. One millimole of glucose contributes approximately one milliosmole if it remains as glucose molecules. One millimole of completely dissociated sodium chloride contributes approximately two milliosmoles because sodium and chloride are separate particles. Real solutions depart from ideal behavior through interactions between ions and other constituents; counting chemical formula units alone does not capture those interactions.

Electrical charge is another distinct quantity. Osmoles count particles, whereas equivalents count charge. Electroneutrality requires accompanying counterions, which also contribute to particle concentration. A sodium concentration therefore carries information about a much larger collection of extracellular particles than sodium ions alone.

Clinical laboratories commonly measure osmolality by freezing-point depression. Dissolved particles lower the freezing point of water; an instrument compares the sample's behavior with calibrated standards. It reports the collective effect of particles without identifying them individually. Glucose, urea, sodium salts, and an unmeasured small solute can all contribute to this physical measurement.

The concentration predicted from selected laboratory analytes is a separate result. A primary comparison of fourteen calculation formulas used freezing-point measurements as the measured comparator and found that performance depended on the chosen equation. The calculated value is consequently an estimate tied to its inputs and coefficients, rather than an alternative direct particle measurement. 1

A substantial lipid mass can occupy sample volume while contributing relatively few dissolved particles. Conversely, a small mass of a low-molecular-weight compound can contribute many particles. Solute mass and sample appearance do not reliably specify osmolality.

47.2 Tonicity Depends on a Membrane and a Time Interval

Water moves across a permeable membrane in response to differences in water chemical potential. An extracellular solute that remains outside a cell can lower extracellular water activity and draw water from the cell. A solute that equilibrates across that membrane may raise measured osmolality on both sides without sustaining the same cell-shrinking force.

Tonicity describes the effective osmotic influence of a solution on cell volume. It therefore requires a specified membrane and relevant period of observation. It is not an intrinsic property of a particle independent of its surroundings. Transporters, metabolic removal, membrane injury, and blood–tissue barriers can change the relationship between a plasma concentration and a cellular response.

Water permeability determines how quickly a volume response can occur. In a foundational experiment, expression of the red-cell CHIP28 protein in frog oocytes increased osmotic water permeability, supporting the water-channel function subsequently associated with aquaporin-1. This was a membrane experiment, not a measurement of whole-body water distribution. 2

For an ideal dilute solution, osmotic pressure can be related to concentration using the gas constant. Let Δπeffective be effective osmotic pressure difference in atmospheres, R the gas constant of 0.08206 L·atm/(mol·K), T absolute temperature in kelvin, and Δcᵢ a particle concentration difference in Osm/L. The reflection coefficient σᵢ describes how effectively that solute–membrane interaction generates osmotic water flow, with values approaching one for an effectively retained solute and zero for a freely penetrating solute under the model's conditions.

For a single ideally retained solute difference of 10 mOsm/L, Δc is 0.010 Osm/L. At 310 K, the ideal difference is 0.08206 × 310 × 0.010 = 0.2544 atm, or approximately 193 mmHg using 760 mmHg per atmosphere. A modest particle difference can therefore represent a large potential water-driving force.

This is not a prediction of an equivalent rise in blood pressure. Water movement changes concentrations, cell mechanics constrain expansion, and hydrostatic forces can oppose flow. The equation idealizes solution behavior and membrane transport. With a penetrating solute, the concentration difference itself also changes during equilibration; a single static calculation cannot describe the complete time course.

47.3 Compartment Mass Balance Explains Water Redistribution

Intracellular fluid and extracellular fluid contain different predominant ions but ordinarily approach osmotic equilibrium through water movement. Sodium salts dominate extracellular effective particles, while potassium salts and organic constituents contribute substantially inside cells. Sodium–potassium pumping and selective permeability maintain these chemical differences without allowing a large persistent separation of bulk electrical charge.

For a simple closed model, let nᵢ and nₑ be the amounts of retained effective particles in intracellular and extracellular compartments, in mOsm. Let Vᵢ and Vₑ be their water volumes in liters. The model approximates liters of compartment water as liters of fluid. If water equilibrates freely and the common effective particle concentration is C in mOsm/L, each compartment's volume is its retained particle amount divided by C.

Choose illustrative starting volumes of 28 L intracellularly and 14 L extracellularly, totaling 42 L. At a common concentration of 300 mOsm/L, the intracellular compartment contains 28 × 300 = 8,400 mOsm and the extracellular compartment 14 × 300 = 4,200 mOsm. These are selected model values, not obligatory fractions of body weight or universal human compartment sizes.

Now mathematically add 1 L of water without particles. Total water becomes 43 L while total retained particles remain 12,600 mOsm. The new common concentration is 12,600/43 = 293.0 mOsm/L. Intracellular volume becomes 8,400/293.0 ≈ 28.67 L, and extracellular volume becomes 4,200/293.0 ≈ 14.33 L. Approximately two-thirds of this addition enters cells because two-thirds of the model's original retained particles are intracellular.

The distribution follows the starting particle allocation. An individual with different body composition or pre-existing compartment abnormalities need not share those fractions. The model also excludes renal excretion, ongoing losses, metabolism, solute exchange, and regulatory volume responses.

Electroneutrality constrains how particles can be added or removed. Retaining a sodium salt adds both cation and anion; moving potassium out of cells requires counterion movement or another charge-balancing process. Water redistribution can change a measured concentration even when the total amount of the measured ion has not changed.

47.4 Clinical Calculations Require Explicit Unit Conventions

A useful simple estimate of plasma particle concentration combines sodium, glucose, and urea. Sodium is doubled to approximate sodium plus accompanying extracellular anions and associated constituents; it does not mean that one sodium ion itself counts as two particles. When sodium, glucose, and urea are all in mmol/L, the conventional estimate is:

The numerical result is approximately mOsm/L. This compact expression omits some measured and unmeasured contributors and simplifies nonideal solution behavior. Different validated formulas use additional terms or coefficients; changing equations changes the residual comparison with measured osmolality. 1

In a common mass-unit convention, sodium remains in mmol/L, but glucose and blood urea nitrogen are reported in mg/dL. Glucose is divided by approximately 18 to obtain mmol/L. Blood urea nitrogen, abbreviated BUN, is divided by approximately 2.8. BUN describes the mass of nitrogen contained in urea, not the mass of the entire urea molecule.

For sodium 140 mmol/L, glucose 90 mg/dL, and BUN 14 mg/dL, the calculation is 2 × 140 + 90/18 + 14/2.8 = 280 + 5 + 5 = 290 mOsm/L. The glucose and urea contributions are each approximately 5 mOsm/L in this example.

If the laboratory instead reports the mass concentration of whole urea in mg/dL, divide that value by approximately 6. Urea 30 mg/dL and BUN 14 mg/dL both correspond to approximately 5 mmol/L of urea. Dividing whole urea by 2.8 would overstate its particle contribution. If glucose and urea are already in mmol/L, neither mass conversion should be applied again.

These equations estimate total particle concentration, including urea. An estimate of systemic effective osmolarity often omits the urea term, giving approximately 2 × sodium plus glucose in mmol/L. That omission makes a permeability and equilibration assumption. It does not establish that urea can never exert an osmotic effect, or that another unmeasured effective solute is absent.

47.5 The Osmolal Gap Is a Residual, Not a Chemical Identification

The osmolal gap is commonly obtained by subtracting a calculated estimate from measured osmolality. Strictly, this often compares measured mOsm/kg with calculated mOsm/L. Clinical use accepts their approximate numerical comparability in plasma, but the subtraction does not make the underlying denominators identical.

The residual incorporates unmeasured particles, analytical error, equation bias, sample timing, and differences between the model and the real solution. A formula that systematically predicts a lower value generates a larger gap from the same blood sample. The local method and reference distribution therefore matter.

An unmeasured alcohol or other small compound may contribute to the gap. Its particle contribution depends on its molar concentration, not simply its mass concentration. Whether it also changes sustained tonicity depends on membrane transport. A substance can be readily measurable osmotically while exerting relatively little lasting cell-volume effect after equilibration.

Metabolism complicates interpretation further. The parent compound and its metabolites may contribute differently to measured osmolality and acid–base abnormalities at different times. A small gap late in an exposure does not reliably reconstruct the earlier concentration or exclude clinically important metabolites.

Lynd and colleagues evaluated the gap in a retrospective, selected population investigated for toxic alcohol exposure. Diagnostic performance varied with the calculation and decision boundary, and false-negative results occurred. This study supports using the gap within a broader assessment rather than treating one number as either toxin identification or a reliable isolated exclusion test. 3

The physiological task is to explain the residual while retaining uncertainty. Simultaneously obtained samples, the sodium measurement method, ethanol or other relevant measured substances, acid–base findings, and specific assays may alter that explanation. No single gap value establishes the identity, concentration, or cellular effect of an unknown solute.

47.6 Urea Illustrates Why Equilibration Time Matters

Urea contributes to freezing-point osmolality because its molecules are dissolved particles. Across many systemic cell membranes it can equilibrate sufficiently that an established rise in urea does not sustain the same transcellular water shift as an equivalent rise in retained extracellular sodium salts. This is why the usual estimate of systemic tonicity excludes urea.

Equilibration is a process, however, rather than instantaneous access to every water compartment. Transport properties differ between tissues and barriers. A plasma concentration changing faster than tissue urea can follow may produce a transient concentration difference, even when the solute eventually equilibrates.

Rapid removal from blood is particularly relevant to the brain. Water and urea need not cross the blood–brain interfaces at the same effective rate. When plasma urea falls while brain urea remains relatively elevated, water can move toward the more concentrated cerebral compartment. A steady-state classification of urea as ineffective therefore cannot simply be applied to an abrupt extracorporeal clearance transition.

In nephrectomized rats, Silver and colleagues compared dialysis conditions with and without a fall in plasma urea. Brain urea declined much less than plasma urea in the group developing increased brain water, supporting a reverse-urea contribution in that model. These were controlled animal experiments, not a validated prediction rule for an individual patient. 4

The mechanism should not be reduced to urea alone. Earlier experiments in uremic dogs identified cerebral edema and brain–blood osmotic differences during rapid dialysis that were not explained by measured brain urea or major electrolyte changes; the investigators implicated additional osmoles and acid–base changes. Different experimental preparations can reveal different contributions. 5

The response depends on the starting solute distribution, its rate of change, tissue permeability, and adaptation. Neither a high BUN nor its decline fully characterizes brain osmotic conditions or specifies a safe removal rate.

47.7 Hyperglycemia Redistributes Water and Changes Sodium Concentration

Glucose can become an important extracellular effective osmole when its extracellular accumulation outpaces cellular uptake and metabolism. Water then leaves cells, expanding extracellular water and diluting extracellular sodium. A reduced plasma sodium concentration in this setting does not by itself prove that sodium has been lost from the body.

The actual patient may simultaneously lose sodium, potassium, and water through urine or other routes. Osmotic diuresis can reduce total body water while transcellular redistribution lowers the sampled sodium concentration. A concentration therefore reflects the combined consequences of external balance and internal redistribution, which can act in different directions.

Insulin changes this system by modifying glucose uptake, metabolism, and associated ion movement. Glucose is consequently not an immutable extracellular particle. Its effective distribution depends on tissue, transport conditions, and the interval considered. The response to falling glucose cannot be represented as simple removal of a permanently extracellular salt.

A glucose-corrected sodium estimates what sodium might be after glucose returns toward a selected reference under specified assumptions. Empirical correction factors summarize observed relationships; they do not directly measure a hidden sodium concentration. In six healthy volunteers exposed to experimentally induced acute hyperglycemia, Hillier and colleagues found an average relationship different from the conventional factor and evidence of nonlinearity. Their preparation did not reproduce every aspect of a dehydrated critically ill patient with diabetes. 6

Consider an illustrative sodium of 128 mmol/L and glucose of 540 mg/dL. Relative to a chosen glucose reference of 100 mg/dL, the excess is 440 mg/dL, or 4.4 increments of 100 mg/dL. A factor of 1.6 mmol/L per increment gives 128 + 1.6 × 4.4 = 135.0 mmol/L. A factor of 2.4 gives 128 + 2.4 × 4.4 = 138.6 mmol/L.

The different answers demonstrate model sensitivity. Neither calculation proves the patient's original water deficit, total sodium content, or future concentration. Ongoing losses, intake, kidney function, and changing glucose distribution can alter the trajectory. The calculation is an interpretive aid rather than a fluid-selection or treatment-rate rule.

47.8 Pseudohyponatremia Is a Measurement Problem with Physiological Consequences

Within a plasma sample, sodium is dissolved in its aqueous phase. Lipids and proteins occupy part of the sample volume but contain little sodium in their nonaqueous fraction. When that fraction increases substantially, sodium concentration in plasma water can remain appropriate even though a method assuming the usual water fraction reports a reduced concentration per total plasma volume.

Indirect ion-selective electrode methods dilute a measured sample volume and use calibration assumptions about its composition. An unusually large nonaqueous fraction can violate those assumptions. Direct ion-selective electrodes examine an undiluted sample and avoid this particular dilution-related volume-exclusion effect, although all methods retain their own calibration and analytical limitations.

The underlying displacement was demonstrated long before current analyzers. Albrink and colleagues examined hyperlipemic human sera and showed that lipid-associated displacement of serum water could produce spuriously low concentrations of water-soluble constituents. Their laboratory study established the physical problem rather than a contemporary analyzer-specific correction formula. 7

In a later study of patient plasma across different protein concentrations, Dimeski and Barnett compared indirect and direct electrode measurements. Protein-dependent discrepancies included falsely low indirect sodium with high protein and falsely high values with low protein. The relationship was not sufficiently simple to justify one accurate universal protein correction. 8

Pure pseudohyponatremia does not represent the same extracellular hypotonicity as a genuine reduction in plasma-water sodium concentration. Measured osmolality can therefore help reveal a mismatch, but a normal or elevated result is not conclusive by itself. Hyperglycemia, uremia, unmeasured osmoles, and true sodium–water disorders can coexist with the analytical artifact.

A paired direct measurement can clarify the sodium result when sampling and analyzer conditions are compatible. Inserting an artifactually low sodium into an osmolarity equation also lowers the calculated value and can enlarge the apparent gap. Misreading that gap as an unidentified toxin would compound the original laboratory problem. Agreement among measurement method, effective particles, and the patient's physiological context is more informative than any one label.

47.9 Cell and Capillary Barriers Govern Different Distributions

An isotonic extracellular salt solution and water remaining after glucose metabolism have different compartment effects. The difference depends on where their particles remain, not simply on the volume introduced. A solution initially close to plasma osmolality can change its effective behavior when its solute enters cells or is metabolized.

Return to the closed model containing 28 L intracellular water, 14 L extracellular water, and a common effective concentration of 300 mOsm/L. Mathematically add 1 L of solution containing 300 mOsm of particles retained extracellularly. Total water becomes 43 L and total particles become 12,900 mOsm, so 12,900/43 = 300 mOsm/L. Intracellular volume remains 28 L; extracellular volume becomes 15 L.

The added volume stays extracellular in this idealization because it arrives with the particles needed to maintain extracellular concentration. Sodium chloride illustrates the relevant short-term extracellular behavior, although actual saline is not a perfectly ideal solution and physiological conditions determine its effective distribution.

Contrast this with the earlier addition of 1 L of particle-free water: intracellular volume increased by approximately 0.67 L and extracellular volume by approximately 0.33 L. Water remaining after glucose uptake and metabolism can approach that distribution if glucose no longer remains as an effective extracellular solute. Persistent hyperglycemia delays or changes that simplification.

Neither model means that extracellular fluid is entirely intravascular. Capillary endothelium separates plasma from interstitial fluid, whereas cell membranes separate extracellular from intracellular water. Small extracellular salts usually distribute across much of the capillary barrier more readily than across cell membranes. Plasma–interstitial partition also depends on hydrostatic forces, protein distribution, endothelial properties, and lymphatic return.

Real kidneys change retention during and after an input. In a crossover experiment in healthy volunteers, Lobo and colleagues found different dilution and excretion responses to saline and glucose-containing fluid. Those observations illustrate why a closed-compartment calculation is not a complete time-dependent account of an infusion, especially in critical illness. 9

These examples describe distribution mechanisms, not administered doses or a fixed intravascular fraction. Organ dysfunction, endothelial injury, ongoing losses, and solute metabolism can all move the patient away from the model.

47.10 Cells Regulate Their Volume by Moving Osmoles

Initial osmotic water movement can be followed by an active change in cellular particle content. A swollen cell may lose osmoles and water through regulatory volume decrease. A shrunken cell may accumulate osmoles and regain water through regulatory volume increase. These responses modify the fixed-particle assumption used in simple compartment calculations.

Volume decrease often involves potassium and anion exit, with water following the net loss of effective particles. In human lymphocyte experiments, Grinstein and colleagues demonstrated swelling-associated increases in anion permeability that supported salt loss and volume recovery. Coordinated ion movement permits substantial particle loss while preserving approximate bulk electroneutrality. 10

Volume increase can recruit pathways that accumulate sodium, chloride, or other solutes. Rat thymic lymphocyte experiments showed osmotic activation of sodium–hydrogen exchange during shrinkage. Its effect on retained osmoles depends on proton buffering and accompanying transport, rather than a literal exchange of equal numbers of freely dissolved particles. Sustained recovery requires a net effective osmole gain. 11

Different cells use different combinations of channels, exchangers, and cotransporters. Their response also depends on ATP supply, electrochemical gradients, cytoskeletal signaling, and the severity of stress. An injured cell may not regulate volume effectively even when its membrane initially permits water movement.

Longer adaptation can alter organic osmolytes, including amino acids, polyols, and methylamines. These compounds can support osmotic balance while limiting the disruption associated with maintaining very high intracellular inorganic salt concentrations. Bagnasco and colleagues identified substantial organic solute accumulation in rat and rabbit inner renal medulla, providing direct biochemical evidence of this strategy in a naturally hyperosmotic tissue. 12

Compatible does not mean biologically inert. Changing these solutes requires transport or metabolism, and some also participate in signaling or neurotransmission. Recovery can therefore be incomplete or delayed. Volume regulation changes how a cell responds to the next extracellular change, making exposure history a component of present physiology.

47.11 Brain Adaptation Creates a Memory of Prior Tonicity

Acute extracellular hypotonicity favors water entry into brain cells. The rigid skull limits accommodation of added intracranial volume, although displacement of cerebrospinal fluid and venous blood provides some initial buffering. Cellular swelling, extracellular water distribution, and the available intracranial reserve together determine the physiological consequences.

The brain can reduce its retained osmoles during sustained hypotonic exposure. Early electrolyte changes and later changes in organic solutes reduce the particle content drawing water inward. Conversely, sustained hypertonic exposure can promote osmole accumulation that limits persistent shrinkage. Adaptation changes cell composition rather than making the extracellular disturbance disappear.

Lien and colleagues measured brain electrolytes and organic osmolytes in rats with experimentally induced chronic hyponatremia and during its correction. Organic solute depletion persisted while some electrolyte changes occurred more rapidly. The study demonstrated that different components of osmotic adaptation recover on different schedules. Its experimental conditions do not define a universal safe human correction rate. 13

When extracellular tonicity changes again, the adapted brain starts with altered intracellular particle content. A brain depleted of osmoles can be vulnerable to shrinkage when extracellular tonicity rises; a brain that has accumulated osmoles can be vulnerable to swelling when extracellular tonicity falls. The same current plasma concentration can therefore accompany different risks after different exposure histories.

Brain injury cannot be predicted from cell volume alone. Barrier function, inflammation, regional cellular responses, systemic illness, and oxygen supply may modify the consequences. Clinical neurological syndromes should not be reduced to a single instantaneous osmotic-pressure calculation.

Acute water redistribution and slower adaptation explain why a laboratory value is insufficient without its trajectory. Reversing a biochemical abnormality does not immediately restore cellular composition, and one numerical correction boundary cannot establish safety for every patient.

47.12 Integrating Measurement, Effective Particles, and the Clinical State

Consider a paired illustrative sample with sodium 128 mmol/L, glucose 360 mg/dL, BUN 84 mg/dL, and measured osmolality 312 mOsm/kg. Before interpreting the sodium, identify the glucose and urea contributions and confirm that the measurements describe the same physiological interval.

Using the stated mass-unit equation, glucose contributes 360/18 = 20 mOsm/L and urea contributes 84/2.8 = 30 mOsm/L. Calculated total osmolarity is therefore 2 × 128 + 20 + 30 = 306 mOsm/L. The approximate measured-minus-calculated gap is 312 − 306 = 6, accepting the conventional numerical comparison between mOsm/kg and mOsm/L.

If urea has equilibrated systemically and no additional effective solute is present, the sodium-plus-glucose estimate of effective osmolarity is 256 + 20 = 276 mOsm/L. The measured particle concentration is substantially influenced by urea. It cannot simply be read as evidence that cells experience the full measured value as a sustained extracellular shrinking force.

This estimate does not directly measure intracellular water or prove the absence of effective unmeasured particles. An abrupt urea change could invalidate the steady-state assumption for the brain. Glucose uptake, changing kidney function, and ongoing external balances could be moving the patient through a changing rather than stable state.

Now suppose the sample is markedly lipemic and sodium was measured by an indirect electrode. A falsely low sodium would lower both calculated total and estimated effective concentrations. It could also enlarge the apparent gap. A compatible direct sodium measurement would help resolve that analytical question before attributing the calculated pattern entirely to water balance.

Finally, establish the trajectory and mechanism. Ask which particles are measured, which remain extracellular over the relevant interval, whether water or solute amounts have changed, and whether cells have had time to adapt. This sequence connects laboratory interpretation with compartment physiology while avoiding the assumption that every high measured osmolality means the same cellular state.

References

1. Martín-Calderón JL, Bustos F, Tuesta-Reina LR, Varona JM, Caballero L, Solano F. Choice of the best equation for plasma osmolality calculation: Comparison of fourteen formulae. Clinical Biochemistry. 2015;48:529–533. doi:10.1016/j.clinbiochem.2015.03.005

2. Preston GM, Carroll TP, Guggino WB, Agre P. Appearance of water channels in Xenopus oocytes expressing red cell CHIP28 protein. Science. 1992;256:385–387. doi:10.1126/science.256.5055.385

3. Lynd LD, Richardson KJ, Purssell RA, et al. An evaluation of the osmole gap as a screening test for toxic alcohol poisoning. BMC Emergency Medicine. 2008;8:5. doi:10.1186/1471-227X-8-5

4. Silver SM, DeSimone JA Jr, Smith DA, Sterns RH. Dialysis disequilibrium syndrome (DDS) in the rat: role of the “reverse urea effect.” Kidney International. 1992;42:161–166. doi:10.1038/ki.1992.273

5. Arieff AI, Massry SG, Barrientos A, Kleeman CR. Brain water and electrolyte metabolism in uremia: effects of slow and rapid hemodialysis. Kidney International. 1973;4:177–187. doi:10.1038/ki.1973.100

6. Hillier TA, Abbott RD, Barrett EJ. Hyponatremia: evaluating the correction factor for hyperglycemia. American Journal of Medicine. 1999;106:399–403. doi:10.1016/S0002-9343(99)00055-8

7. Albrink MJ, Hald PM, Man EB, Peters JP. The displacement of serum water by the lipids of hyperlipemic serum; a new method for the rapid determination of serum water. Journal of Clinical Investigation. 1955;34:1483–1488. doi:10.1172/JCI103199

8. Dimeski G, Barnett RJ. Effects of total plasma protein concentration on plasma sodium, potassium and chloride measurements by an indirect ion selective electrode measuring system. Critical Care and Resuscitation. 2005;7:12–15. PMID:16548813

9. Lobo DN, Stanga Z, Simpson JA, et al. Dilution and redistribution effects of rapid 2-litre infusions of 0.9% (w/v) saline and 5% (w/v) dextrose on haematological parameters and serum biochemistry in normal subjects: a double-blind crossover study. Clinical Science. 2001;101:173–179. doi:10.1042/CS20000316

10. Grinstein S, Clarke CA, DuPre A, Rothstein A. Volume-induced increase of anion permeability in human lymphocytes. Journal of General Physiology. 1982;80:801–823. doi:10.1085/jgp.80.6.801

11. Grinstein S, Rothstein A, Cohen S. Mechanism of osmotic activation of Na⁺/H⁺ exchange in rat thymic lymphocytes. Journal of General Physiology. 1985;85:765–787. doi:10.1085/jgp.85.5.765

12. Bagnasco S, Balaban R, Fales HM, Yang YM, Burg M. Predominant osmotically active organic solutes in rat and rabbit renal medullas. Journal of Biological Chemistry. 1986;261:5872–5877. doi:10.1016/S0021-9258(17)38464-8

13. Lien YHH, Shapiro JI, Chan L. Study of brain electrolytes and organic osmolytes during correction of chronic hyponatremia. Implications for pathogenesis of central pontine myelinolysis. Journal of Clinical Investigation. 1991;88:303–309. doi:10.1172/JCI115292