An alternative view of the role(s) of surfactant and the alveolar model

Brian A. Hills


Currently, the study of surfactant proteins is much in vogue, but, in the early days, the physics underlying surfactant function was treated somewhat superficially, leaving assumptions that have become culturally embedded, such as the “bubble” model of the alveolus. This review selectively reexamines these assumptions, comparing each combination of alveolar model and role of surfactant for compatibility with the major features of pulmonary mechanics and alveolar stability, morphology, and fluid balance.

  • alveolar edema
  • alveolar fluid balance
  • alveolar models
  • pulmonary mechanics

the epochal experiment of von Neergaard (150) clearly demonstrated the major role, if not dominance, of surface forces in pulmonary mechanics, as witnessed by the reduction by 75% or more (87) in lung recoil when an aqueous fluid is substituted for air in reinflating excised lungs to functional residual capacity (FRC). Furthermore, compliance hysteresis is almost eliminated upon liquid filling (120).

Basic Physics

Surfaces possess additional energy (surface energy) simply by virtue of the imbalance of forces acting on molecules located at the boundary of any phase, including gases, liquids, and solids. Surface energy is manifest in many ways, such as the surface tension of water (69.9 mN/m at 37°C); the separation of two immiscible liquids after shaking; or the wettability of solid surfaces where, in equivalent units, values can reach 500–1,000 ergs/cm2for most metals, including platinum (10).

In accordance with the second law of thermodynamics, all surfaces tend to minimize their surface energy, as demonstrated by the tendency for a one-sided bubble to collapse, thereby generating a pressure difference (ΔP) related to surface tension (γ) and radius of curvature (r) in Fig.1, A andB, by the Laplace equationΔP=2γ/r Equation 1Because solids are rigid, surface energy is manifest in less obvious ways, such as the contact angle (θ) resulting when a droplet of water is placed on the surface (see Fig.1 C) where, by resolving forces horizontally at the triple point, one obtains the Young equation (5) asγSAγSL=γcosθ Equation 2where γSA is the interfacial energy between solid and air and γSL is that between solid and liquid. Although the contact angle is a simple and convenient means of quantifying surface hydrophobicity, it has its limitations (55).

Fig. 1.

Diagram depicting how any liquid-air interface tends to collapse inward whether air (A) or fluid (B) is the convex medium, with difference in pressure (ΔP) upon traversing surface being related to surface tension and geometric parameters by the Laplace equation (seeEq. 1 ). Thus concave surfaces tend to suck in more fluid, whereas convex surfaces tend to resolve that fluid. At rigid solid surfaces (C), relationship between surface parameters is described quantitatively by the Young equation (see Eq. 2 ). Note how surface becomes more hydrophobic [contact angle (θ↑)] as energy difference between the same surface dry and wet (γSA − γSL), where γSA is interfacial energy between solid and air and γSL is that between solid and liquid, is reduced, thus discouraging wetting. Note that, at alveolar level, air would be saturated with water vapor. γ, surface tension.


Many substances are attracted to liquid or solid surfaces, but their presence is not obvious, largely because they do not change surface energy significantly. Substances that do so are termed “surfactants,” each molecule consisting of two parts of widely differing “personality” covalently, i.e., irreversibly, bound to each other. One end is polar and seeks aqueous fluids or hydrophilic solid surfaces, whereas the other is nonpolar, seeking oil, air, or hydrophobic surfaces. Surfactants that are readily recruited to one type of interface will generally adsorb to others as their molecules orient to place each moiety in contact with a compatible phase, thereby reducing interfacial energy.

Solid Surfaces

Surfactants in general bind reversibly to solid surfaces by a process known as adsorption or chemisorption for stronger ionic binding, more characteristic of cationic surfactants (64, 90), especially those incorporating the strongly positively charged terminal quaternary ammonium (QA) ion. Equilibrium between the adsorbed monolayer and adjacent fluid is characterized by the Langmuir isotherm (2), as determined for surface-active phospholipid (SAPL) by Gershfeld (45). When binding to hydrophilic surfaces via their polar groups, the nonpolar groups of surfactant molecules are oriented outward (Fig.2 C), rendering the surface more compatible with air (γSA↓), less compatible with water (γSL↑) and, therefore, more hydrophobic (θ↑), as predicted byEq. 2 . In excess, surfactants often build up more layers parallel to the adsorbed monolayer as an oligolamellar coating (59, 100), those with QA ions often forming structures alternating lipid bilayers with water layers invoking chelate binding (88), as seen in vivo (see Fig. 12).

Fig. 2.

Diagram depicting how dipalmitoylphosphatidylcholine (DPPC) is surface active because it consists of a nonpolar end comprising 2 straight fatty acid chains and a polar end where it has long been argued whether the phosphate (−) and quaternary ammonium moieties (+) are planar, vertical, or can undergo a conformational change (see Ref. 154) from one to the other (A); how these molecules orient themselves at a water-air interface to reduce surface tension (B); and how they can adsorb (bind reversibly) to hydrophilic surfaces to render them more hydrophobic (less wettable) and form a barrier of close-packed alkyl chains in doing so (C). Surfactants that act at one interface usually act at others, although efficiencies in reducing surface energy may vary depending on thermodynamics of the whole system including mobile ions (see Fig. 3).

Oligolamellar layers, or largely intact monolayers, of surfactants can impart many highly desirable properties to a surface, which are of immense industrial importance, since they include water repellency, bactericidal barriers, boundary lubrication, corrosion inhibition, release (2, 60, 63, 71, 88, 90, 111), etc. Unfortunately, publication from the large research base supporting this vast industry, producing 5 million tons/yr worldwide and 75,000 tons/yr of just one QA surfactant (88), is often restricted by commercial considerations.

Tissue Surfaces

All mammalian biosurfaces are negatively charged (158), including alveolar epithelium where membrane-bound carbohydrates introduce carboxyl and sulfonyl groups (122) and are, therefore, most conducive to adsorption of cationic surfactants. Phosphatidylcholines, which are perfectly zwitterionic at physiological pH (5), can become effectively cationic when mobile cations, present in any body fluid, effectively neutralize the phosphate ion, as depicted in Fig.3. The phosphatidylcholine molecules can be attached by the same highly charged QA ion used to “anchor” many industrial surfactants (64, 90). Interspersion of cations within a plane of phosphate ions is a century-old method of generating a cohesive protective barrier known industrially as “phosphating” (39). Concomitant close-packing of the nonpolar chains is a further barrier, also placing the glycerol backbone of SAPL molecules out of reach of digesting enzymes, as demonstrated experimentally (11).

Fig. 3.

Diagram depicting how adsorption of DPPC to a tissue surface would render a wettable hydrophilic surface less wettable and how cohesion of the adsorbed layer can be improved by the time-honored mechanism of “phosphating” (see Ref. 39), whereby interspersed mobile cations pull phosphate ions into a tight matrix, thus enhancing barrier properties. Binding to the surface by strongly positive quaternary ammonium group is also enhanced by neutralization of phosphate ions. Note how straight (saturated) chains fit more easily into the matrix. An outermost hydrocarbon surface would also be conducive to deposition of neutral lipids in pulmonary surfactant, just as any oil present enhances corrosion protection imparted to nonbiological materials by adsorbed surfactants (59). [Adapted from Hills (80).]

Much evidence has been reviewed elsewhere for SAPL providing the gastric mucosal barrier to the back-diffusion of hydrogen ions (70,80) and the boundary lubricant providing effortless sliding of pleura and articular cartilage (73, 78, 82). Essentially, the same barrier might also mask bronchial irritant receptors (79) as one manifestation of an airway mucosal barrier.

Pulmonary Surfactant

These days, in respiratory physiology, the word “surfactant” has assumed the meaning of the whole complex of phospholipids (∼67%), proteins (∼8%), neutral lipids (∼21%), carbohydrates (∼2%), and minor constituents (∼2%) (105), the exact composition of which has been reviewed in detail elsewhere (40). At least, this is the approximate composition of lamellar bodies (LBs) that retain the surface-active state in which surfactant is produced in type II pneumocytes located in septal corners of the alveolus from which it is secreted onto the alveolar surface. The composition and structure of LBs in the lung have been reviewed in detail elsewhere (127, 140); although the LBs have been reported in many other tissues of the body, some pathologists are tending to dismiss them as “the membranous remains of dead cells” (46). This dismissal is debatable, since the LBs are found in nonscavenger cells such as type B synoviocytes (37, 133) and parietal cells (72, 148).

Although not identified in the lung until 1946 (144), the surface activity of dipalmitoyl phosphatidylcholine (DPPC) was first demonstrated by Leathes (98) in 1923 when he showed on the Langmuir-Adam trough how just a monolayer of “hydro-lecithine” could withstand a very high “surface compressive force,” reaching 200 atmospheres. According to the definition (1) of surface pressure (π) in Eq. 3 , this would mean a very low value for surface tension γ, assuming equilibriumπ=γoγ Equation 3where γo is the surface tension of water at the same temperature. If equilibrium conditions do not apply, we should really refer to an apparent surface tension (γ'), especially in discussing compressed monolayers.

The Liquid-Air Interface

The fascination of surfactants located on the surface of water is how, in the two dimensions of the surface, their molecules behave in a manner closely resembling gas molecules in a three-dimensional cylinder (2). When compressed by the barrier of the Langmuir-Adam trough and/or when temperature is lowered, they can be transformed into the equivalent of a liquid phase in the cylinder or even liquid crystal or a solid of one or more crystalline forms. The liquid and solid equivalents are known as the “expanded” and “condensed” states, respectively, as depicted in Fig.4 A for DPPC. Around the turn of the century, Langmuir (97) and Adam (1) studied these fascinating phase changes of many surfactants using a Langmuir-Adam trough (Fig.4 B), employing surface compression ratios of 5:1, or even 10:1.

Fig. 4.

A: diagram outlining transition zone and the principal phases for a monolayer of surfactant on a surface pressure-area (π/A) phase diagram for temperatures above and below transition temperature (Tc).B: standard Langmuir-Adam trough using a platinum dipping plate to measure surface tension γ. Note how the vertical pull, monitored by force transducer, is reduced by an appreciable contact angle θ, which, if above 90°, has been known to be misinterpreted as a negative surface tension. The Wilhelmy method is inappropriate for measuring surface tension of liquid-air interfaces when masked by solid DPPC.

Clements (24, 25) and Brown (18) appreciated how movement of the barrier of the Langmuir-Adam trough could be used to simulate respiratory area changes but retained the physicists' compression ratios of 5:1, as reproduced for lung surfactant in Fig.5.

Fig. 5.

Diagram depicting the 3rd γ/A loop of lung surfactant applied to the pool of a Langmuir-Adam trough for different methods of measurement of surface tension, including the Wilhelmy method ignoring contact angle (29). Note how very low (<10 mN/m) apparent values for γ can be attained by using compression ratios adopted from physical sciences, but reach only the equilibrium γ value for a maximum physiological compression of 25% (70, 118). Data for the stirrup and Langmuir methods are taken from Brown et al. (20) and Hills (69), respectively.


Phase changes for l-α-DPPC, the predominant and most surface-active component of lung surfactant (19, 105), have been studied in much detail at the liquid-air interface only by Bangham (7-9), Watkins (154), Gaines (42), Nag et al. (110), Chapman and Jones (21, 89), McConnell et al. (101), Christova et al. (23), Scarpelli and Mautone (126), Denizot et al. (32), Tchoreloff et al. (142), and many others. Although these studies describe the most academically fascinating crystal structures of DPPC, all are based on the liquid-air interface and few go on to relate their findings to physiological features of the lung.

It can be seen in Fig. 4 A how the surface area per molecule of the condensed (solid) state is reduced under compression to 40Å2, as determined in the original experiment of Leathes (98). This cross-sectional area applies to both polar and nonpolar ends, enhancing the surface activity of DPPC, because both sets of moieties are able to close pack with those of neighboring molecules without one inhibiting the other. This feature promotes monolayer cohesion, as discussed above.

Although the extremely high value for π recorded by Leathes (98) might indicate extreme surface activity, Watkins (154) points out how it is incorrect to refer to the surface tension of a monolayer that is condensed or hypercompressed. Gaines (42) and Bangham (9) pursue the theme that Eq. 3 is only valid under equilibrium conditions, casting doubt on claims of “near-zero surface tension” made by Clements and Tierney (29) based on their alveolar “bubble” model evaluated later.

Solid “shells” or monolayers of condensed DPPC have a well-defined gel to liquid-crystal transition temperature (Tc), i.e., melting point, of 41°C (12), which is above normal body temperature if the DPPC is pure. When adsorbed to solids, melting of close-packed hydrocarbon chains occurs at a higher temperature, i.e., the Kraft temperature (100), which is 46°C for DPPC (21). It could be most significant that this almost coincides with the temperature (47°C) at which lung mechanics change significantly on warming (87). Returning to the liquid-air interface, only in the expanded phase above Tc isl-α-DPPC free to spread. Otherwise, if expanded rapidly from a condensed state, the solid rafts of surfactant do not disperse but crack, as though opening up fissures in pack ice, often leaving “icebergs”, i.e., undispersed aggregates (50), of solid surfactant floating on the exposed liquid surface (see Fig.6 A). On the other hand, Tchoreloff et al. (142) describe how “piled amorphous aggregates” of lung surfactant can respread rapidly upon expansion of the surface film.

Fig. 6.

A: micrograph demonstrating how a condensed monolayer of DPPC below Tc fractures on expansion from a condensed state to expose liquid surface with floating “icebergs” (arrows). This cleavage pattern is demonstrated by blowing talc onto surface of pool, as described by Shah and Schulman (137).B: micrograph demonstrating a monolayer at surface of the aqueous hypophase (AH) in ovine lung, lamellar bodies expanding into tubular myelin from which surfactant is then recruited to the liquid-air interface and, probably, to the alveolar wall also (70). Note how surfactant monolayer (arrow) in this case is oligolamellar, rather than a true monomolecular, layer. Bar represents 50 nm.

Aqueous Hypophase

When equilibrium is established between the aqueous hypophase (AH) and its supernatant monolayer, DPPC reduces γ to an equilibrium surface tension of 24–26 mN/m (7). Cations in the AH reduce the area per molecule, with the effect depending on their charge Al3+ > Ca2+ > Na+ and concentration (137), an effect also highly relevant to monolayers chemisorbed to solids (Fig.3) and their cohesion (70), particularly applicable to cohesive planes of phosphate ions discussed above. However, nonequilibrium conditions apply to compressed monolayers where continuous compression is needed to maintain γ' <4 mN/m (126) as derecruitment occurs. Reverting to kinetics of a compressed DPPC monolayer, foreign molecules can be “squeezed out” into the AH (154), whereas collapse of the “refined” monolayer has been studied by Goerke and Gonzalez (53).

The reverse process of DPPC recruitment to the liquid-air interface is currently attracting much attention for its relevance to “surfactant rescue” of newborns with the respiratory distress syndrome (RDS) unequivocally caused by surfactant deficiency (4). Approaching the interface from the air, “dry” surfactant in which Tc has been reduced below 37°C by admixture with other phospholipids, e.g., phosphatidylglycerol (9), spreads very rapidly to form a monolayer (Fig.7 A). However, if approached from the fluid side or as a “wet” (e.g., nebulized) formulation, recruitment is much slower, and assistance is required in the form of Ca2+ (94), Mg2+, or, more effectively (23), surfactant proteins (62, 131) (see Fig.7 B). LBs are unique in that they can retain the surface activity of dry surfactant while dispersed in water, as indicated structurally by nuclear magnetic resonance spectroscopy (54).

Fig. 7.

Approaching liquid-air interface from the air phase with “dry surfactant,” showing how oligolamellar rafts above Tc rapidly spread to form a monolayer (A); and from the liquid phase, when recruitment to the interface is facilitated by surfactant proteins, especially the combination of SP-A and SP-B (B).A is reproduced from Morley et al. (106) with permission; B depicts data for dry surfactant from Bangham (12) and wet surfactant from Haagsman and Hawgood (62).

Surfactant Proteins

The recent burgeoning of research into the many roles proposed for surfactant proteins (SP-A, SP-B, SP-C, and SP-D) has been reviewed in detail elsewhere (149, 161). Contrary to original belief (92), surfactant proteins would not appear to be unique to the lung, with SP-A having been identified in the gut (124), synovium (35), and peritoneum (35), while our current research is indicating SP-B only in other organs having contact with air, such as Eustachian tube and skin. However, the concept of nonpulmonary surfactant proteins remains controversial. Physically, the insoluble proteolipid SP-B is the protein that, alone, is most effective in recruiting DPPC to the liquid-air interface, thus reducing γ (62, 113). However, it is the combination of SP-B with the water-soluble SP-A that really enhances DPPC recruitment to the interface (Fig.7 B), suggesting a carrier mechanism for transporting the otherwise highly insoluble DPPC, the critical micelle concentration of which is only 4.7 × 10−10 M (141). Such a mechanism, invoking reversible phospholipoprotein binding (103), has been proposed for transporting lubricating DPPC by the macromolecular glycoprotein lubricin found in the joint (134), so it is interesting to find SP-A closely associated with the monolayer—or multilayer (131)—at the liquid-air interface (110). Binding coefficients between amino-acid groups and phospholipids confirm appreciable bonding of SAPL to water-soluble proteins (103).

Conventional Approach

Each role of surfactant that has been proposed needs its own alveolar model, which can range from the conventional continuous liquid lining (25, 61) to one totally free of fluid under normal conditions (31), or the intermediate case of a discontinuous liquid lining (65, 70), often misquoted (14, 151) as dry. In 1954, Macklin (102) postulated a mucous lining, with later workers demonstrating an alveolar glycocalyx (122), i.e., “structured gel” (40). This is interesting because squeezing such a gel during expiration might cause a supernatant liquid layer to form by a reversible phenomenon known as “syneresis” (51), disappearing on inspiration.

Impressed by the stability of foam expelled from excised lungs, Pattle (115, 116) proposed a continuous liquid hypophase, the surface tension of which was greatly reduced, if not eliminated, by lipoprotein. The conventional model as it stands today (61) arose when Clements (25) realized that the surface tension was lowered by phospholipid rather than lipoprotein. However, the resulting surface monolayer theory goes much further than stating the obvious propensity for any surfactant to locate at a liquid-air interface, since the AH is depicted as a continuous fluid layer lining a concave alveolar wall, as though it were a one-sided bubble (61). Reduction of surface tension reduces lung recoil, “making breathing easy” (27).

This one-sided bubble model (Fig.8 A) has provided the basis for formulating exogenous surfactant for treating RDS and, rightly or wrongly, the success of these formulations in reducing infant mortality (135) has been taken by many as tacit confirmation of its validity.

Fig. 8.

A: diagram depicting the conventional “bubble” model of the alveolus (25, 61), in which surfactant is assumed to locate only at the liquid-air (LA) interface of a continuous AH. Note how the concave nature of interface tends to suck fluid into air space, especially at the more curved (r↓ in Eq.1 ) septal corners, unless γ is very low.B: “morphological” model (see Refs. 70, 77) reflecting the “pits” and “pools” of fluid found in morphological studies where surfactant adsorbs to both LA and tissue surfaces, rendering tissue less wettable to explain the apparently fluid-free areas. Note how pits and pools are normally concave but can become convex with excess fluid when the LA interface will now assist physiological water pumps in resolving edema (Eq. 4 ). Thus fluid control is self-regulating in B but not inA, unless some other mechanism is involved.

Substituting values of 5 cm water gauge for transmural pressure at FRC and 180 μm for an alveolar diameter (2r) for an adult human, the Laplace equation gives γ = 23 mN/m, the equilibrium value for DPPC at 37°C, as discussed above. Thus, superficially, this simple one-sided bubble model appears justified.

Septal Corners

In most alveolar models, air is displayed (perhaps incorrectly) on the concave side of the interface, such that, on passing from air to fluid (Fig. 1 A), the pressure will decrease by the recoil contributed by that interface (ΔP inEq. 1 ). Thus the interface per se induces a negative pressure (ΔP) sucking fluid into the alveolus, which is then balanced by the physiological water pumps (protein-oncotic and ion-channel) for resolving excess alveolar fluid. These pumps are discussed later, but their net effect has been estimated as −5.8 mmHg from the negative interstitial pressure demonstrated in Guyton's classic ping-pong ball experiment (57), the relevance of which to the lung was confirmed by various “wick” methods (129), or −7 mmHg considering Starling forces (58). Guyton (57) argued that, unless γ were very low (<5 mN/m), this pumping pressure would be inadequate to balance the collapsing pressure at the septal corners of alveoli for which the radius of curvaturer is only 10 μm (86).

Wilson (160) pointed out how fluid pressure induced at points of greater concavity (r↓ inEq. 1 ) will be even more negative than that at points of lesser concavity (r↑ in Eq.1 ), causing fluid to shift along the alveolar surface, thus reducing ΔP by “rounding off” septal corners.

Bat and Shrew

It was further argued (66, 70), however, that even if rounding off were taken to its limit, i.e., to a spherical interface, it is even more difficult to explain the apparently normal pulmonary physiology of mammals, such as the shrew and the bat, for which alveolar diameters (2r) have been quoted as 32 and 29 μm, respectively (143).

Such creatures would appear to need very low alveolar surface tensions to satisfy the conventional model (see Table1). This argument raises the current issue of “near-zero surface tension,” which can be traced to Pattle's statement (115), derived primarily from his studies of lung foams, that “the surface tension of lung bubbles is therefore zero,” a theme later adopted by Clements and Tierney (29) for phospholipid as the surface-active agent.

View this table:
Table 1.

Fluid pressures generated by surface forces

Near-Zero Surface Tension?

In response to the apparent shortcoming arising from septal corners, proponents (25, 61) of the conventional bubble model argued that, when the apparent surface tension of DPPC or lung surfactant is measured on a Langmuir-Adam trough using a Wilhelmy dipping plate (Fig. 4 B), the vertical “pull” on this plate does, indeed, reach very low values, especially at maximum compression (see Fig. 5). They justify their emphasis on maximal compression on the grounds that, if alveolar collapse or atelectasis were to occur during the respiratory cycle, it would do so at end expiration. However, continual compression is needed to maintain γ' <4 mN/m (126).

From a physical standpoint, Bangham (7) is particularly critical of the concept of near-zero surface tension describing it as “absurd,” since “0 surface tension implies no interface.”

Measurement of Surface Tension

The validity of the conventional bubble model is clearly very dependent on near-zero surface tension (<5 mN/m). The original Wilhelmy dipping platinum-flag method is unsatisfactory for measuring γ because DPPC adsorption to the flag induces a contact angle θ that varies throughout the simulated respiratory cycle (13), producing artifactually low values if θ is ignored, and so we are really measuring an apparent surface tension γ'. This error can be reduced by using a paper flag (43), whereas, in another attempt to overcome contact-angle artifact, the platinum flag was coated in tracheal epithelium rinsed free of its mucous lining (81). It was serendipitous that this failed to eliminate the θ that reached 60° and varied in a manner resembling the platinum flag, because this provided the first indication of adsorption of DPPC directly onto pulmonary epithelium.

Another method of eliminating contact-angle artifact altogether was the reintroduction of the American Society for Testing Materials' “fast-bubble technique” (3) in the form of a bubble surfactometer (38) oscillating at respiratory frequencies. This has the enormous advantage of requiring small samples.

Schürch et al. (132) have described a new version minimizing bubble distortion by employing a “captive bubble,” the surface area of which is now changed by alternate compression and decompression, but other shortcomings are introduced (126), e.g., adiabatic heating/cooling (77). However, results again confirm γ' <5 mN/m for compressions >30%.

Reverting to the well-controlled conditions of the Langmuir-Adam trough and yet avoiding contact-angle artifacts, Hills (69) employed the original Langmuir method (97) of measuring the horizontal force acting on a floating boom to measure γ, using Wilhelmy plates placed behind the boom and barrier only to detect any leaks. The minimum value of γ recorded for compression of a DPPC monolayer was 6.9 mN/m, which is close to the lowest value recorded by Brown et al. (20) (∼10 mN/m) in 1959 using a “stirrup” (also θ independent) to measure γ for lung rinsings on the Langmuir-Adam trough.

In summary, most studies were able to demonstrate a near-zero (<5 mN/m) surface contractile force, but only for surface compressions of at least 30% and, mostly, in excess of 50%. This raises the next issue concerning whether area cycles of 5:1 or more, originally adopted from physicists' studies of other surfactants, have any physiological relevance to respiratory movements in the lung.

Respiratory Area Changes

This topic has been reviewed elsewhere (56, 70), but, for a typical healthy adult with an FRC of 2.5 liters and tidal volume of 500 ml, the isometric change in alveolar surface area should be no more than 6%. Even when subjects are breathing between maximal inspiratory and forced-expiratory lung volumes, the change calculated on this basis (70) does not quite reach Pattle's originally estimated maximum of 25% (118). Studies of dog lungs frozen at various stages of inspiration show that most of the volume change occurs in distal bronchioles rather than in alveoli (52), whereas Whimster (159) compares the alveolus to a concertina that changes volume without changing surface area at all.

It is, therefore, difficult to see how alveolar surface area and, hence, that of a continuous AH, can change by the amount (>30%) needed to reduce apparent surface tension to “near zero,” again casting serious doubt on the validity of the conventional bubble model.

Alveolar Stability

When proponents of the conventional model put forward the concept of a continuous liquid lining to the alveolus, they introduce two aspects of instability: one inherent in single bubbles and another related to interconnected bubbles.

Historically, the importance of interalveolar stability was raised by Pattle (117), who emphasized the extreme stability of foam expressed from excised lungs. However, foam has very limited relevance, since each bubble can have a different internal pressure (see Fig.9 A). Once interconnected, however, smaller bubbles (r↓) would have higher internal pressures (ΔP↑ in Eq. 1 ) and, hence, the air would flow into larger bubbles, thus enlarging them further (ΔP↓) and increasing flow until the smaller ones collapsed (see Fig. 9 B). Aware of this, proponents of the bubble model argued (28) that, as smaller bubbles shrank, the area reduction would reduce γ (Fig. 5) until the collapsing pressure (ΔP in Eq. 1 ) equaled that in larger bubbles in which, by the reverse sequence, increasing bubble size decreases ΔP (see Fig.9 C). Superficially, this offers a satisfactory explanation until it is pointed out (70) how any continuous AH postulated for one alveolus is also continuous, allowing surfactant to flow between adjacent alveoli via both the pores of Kohn and terminal bronchioles (70).

Fig. 9.

Alveolar stability, depicting how 2 adjacent bubbles in a foam can have same surface tension and different internal pressures (A); how, when interconnected, the smaller bubble will tend to collapse forcing air into the larger to create instability (B); how, as the smaller bubble collapses, the reduction in surface area will decrease γ (see Fig. 5) while the increase in the larger bubble will increase γ until pressure in both bubbles is the same and stability is restored (C); and how the higher surface concentration of surfactant in the diminished smaller bubble will now be greater than that of enlarged larger bubble, causing surfactant (unless solid) to flow along the surface from the smaller to the larger bubble (D). This movement will tend to equalize surface tensions, thus reintroducing the instability problem depicted in B.

Thus the increased surface pressure of the smaller bubble will cause surfactant to flow from the smaller to the larger bubble (Fig. 9), at least during the inspiratory phase when the monolayer is spreadable. Thus the instability problem is reintroduced, yet again casting serious doubt on the validity of the bubble model and the continuous AH in general.

Alveolar Fluid Balance

A single open bubble at constant surface tension is inherently unstable. If internal pressure (ΔP) happens to fall slightly below 2γ/r, then it will continue to deflate at an ever increasing rate (asr↓). By the same token, a transient elevation of ΔP above 2γ/r will cause it to keep enlarging. Applied to the alveolus, this means that it will keep inflating by reducing the interfacial pressure (ΔP) opposing suction applied by the natural physiological water pumps (p) by continually increasing the net force (Δp) pumping fluid out of the alveolus (ΔP↑↑ as r↓)Δp=pΔP=p2γ/r Equation 4This raises the obvious question, particularly relevant at birth (84): Why expansion of the one-sided bubble should not continue until it reaches the alveolar wall. Alternatively, What physiological mechanism, if any, is in place to control the thickness of the AH and maintain the proposed continuous liquid lining?

Control of the AH

Despite the challenge (17), no satisfactory mechanism appears to have been forthcoming from proponents of the bubble model.

At least, ion channels, which normally pump Cl and, hence, water into the potential air spaces in the fetal lung, can also pump Na+ and, hence, water out of the alveolus after birth (139, 153), with the direction of flow being influenced by β-adrenergic stimulation (152, 153). Although this offers a possible mechanism for pumping water in both directions across the alveolar wall, it still falls short of explaining how the walls could sense the physical thickness of a liquid layer and then respond.

Morphology of the Aqueous Lining

Proponents (14) of the conventional bubble model have responded with a study designed specifically to address the issue of the liquid lining in which they first freeze the water to lock it in place and then pursue its profile by electron microscopy (EM). Bastacky et al. (14) claim that the liquid lining is continuous, with a mean thickness varying from 0.14 μm over flat sections up to 0.89 μm in areas of greater concavity, as anticipated from earlier discussion of rounding off. However, their EMs can be regarded as totally artifactual because, for some unknown reason, they inflated the lungs to 15 cmH2O before freezing and fixation, thus squeezing all blood out of the septal walls to render the surface totally concave. In normal blood-filled lungs, however, it can be seen that 60–80% of the alveolar surface is convex, as red blood cells grossly distend the wall in bulging their way through alveolar capillaries (see Fig. 10). Just as concave surfaces accumulate water by generating a negative pressure within the fluid, so convex surfaces (Fig.1 B) generate a positive pressure, now supporting (70, 155) rather than opposing, the physiological pumps (Δp↑ as r < 0 inEq. 4 ). With a predominantly convex alveolar surface, a higher surface tension, e.g., equilibrium γ or an alveolar value of ∼18 mN/m determined by alveolar micropuncture (121) would be preferable to γ < 5 mN/m (see Table 1).

Fig. 10.

A: alveolar wall of a normal blood-filled human lung demonstrating how 70%, or more, of alveolar surface is convex with respect to air at the microscopic level on account of red blood cells “bulging” their way through underlying capillaries. Thus surface forces would be “pumping” any excess fluid in the desired direction (see Table 1).B: depiction of the same convexities by transmission electron microscopy. Note how it is a moot point whether there is any fluid lining those convexities. [From Weibel and Bachofen (156) with permission].

In normal frozen blood-filled lungs, classic studies of the alveolar surface performed by EM in the early 1970s (48, 49, 157) essentially demonstrate any alveolar fluid as confined to “pools” at the septal corners and “pits” elsewhere along the alveolar surface. Fluid is seen in particular in the reentrant crevices formed between two capillaries inflated by blood (156). It is virtually impossible to determine whether the intervening space between pools is covered with an exceedingly thin liquid layer.

A simple physical indication of whether a surface is wetted is provided by any evidence of an “edge” to the pools and pits, as manifest by a contact angle (Fig. 1 C). Gil (48) disputes the concept of a fluid-free surface on the grounds that one does not observe a contact angle with fluid bulging into the air space. In his example of a concave region of a normal air-filled lung, however, his EM does display a contact angle but one formed by the first pleat, folding into the pool of a normal concave region where the negative interstitial pressure is balanced by a concave (negative) surface-contractile pressure. In frozen edematous lungs, however, fluid has been demonstrated as highly convex (34, 93) and, therefore, conducive to resolution by both physiological pumps and surface forces (see Figs. 1 B and8 B).

Alveolar Wall Tensioning

In a liquid-filled lung, alveolar epithelium is seen as a “pleated” surface (49) resembling soaking wrinkles. In the normal air-filled lung, by contrast, the apparently fluid-free flaccid wall is pulled tight into the same contour as the liquid-air interface of the pits and pools (Figs. 8 B and11,A-C). For such tensioning to occur, it can then be argued that there must be an edge to the pools, whereby the fluid can “grip” the surface at the triple point (Fig. 1 C).

Fig. 11.

Diagram depicting how, in the morphological model (see Fig.8 B) alveolar epithelium rendered less wettable by surfactant adsorption is more conducive to pool formation and how an edge to these pools, as manifest in cross section by triple points (see Fig. 1 C), enables the fluid to “grip” and, hence, tension the flaccid epithelium into the same contour (A); how edema formation can swell these pools until they become convex, but they will still tension the apparently fluid-free areas (72) (B); and how resolution of this edema reduces curvature (r↓ inEq. 1 ), thus reducing interfacial force (ΔP), which supports that derived from physiological water pumps (p) until it returns to an equilibrium contour shown inA, where the surface is concave; thus the system is self-regulating (C).D: depiction of how expansion of alveolar surface on inspiration expands the pools rather than fluid-free surface, thus offering a much larger compression ratio for the liquid-air interface. This does not apply to the bubble model.

Morphological Studies of Surfactant

Whereas there is no doubt that surfactant will locate at any liquid-air interface in the alveolus, the real issue concerns the nature of the epithelial surface. In most EM, aldehydes, especially glutaraladehyde (125), are used as fixatives because these fix proteins and, therefore, produce pictures with excellent detail. However, if one wishes to visualize lipid structures, including SAPL, it is necessary to use a lipid fixative, e.g., tannic acid (36). Using tannic acid Ueda and co-workers (145, 146) have produced many EMs of the alveolar surface, demonstrating an oligolamellar lining of SAPL immediately adjacent to the lipid bilayer of the epithelial membrane. The number of layers can range from a single adsorbed monolayer (145) to oligolamellar structures (146). We have confirmed such structures (83) (see Fig.12) including the continuity of the SAPL layer between alveolar epithelial cells, spanning intercellular junctions (147). This “bridging” of tight junctions by PL has also been demonstrated in other membranes such as the epithelium of the renal tubule (46), even when aldehyde fixation was used.

Fig. 12.

An electron micrograph of human lung produced by using lipid fixatives following the method of Ueda et al. (146) and reproducing their demonstration of surface-active phospholipid (SAPL) directly adsorbed to alveolar epithelium. However, in this case, note how oligolamellar layer of adsorbed (tissue-bound) SAPL spans intercellular junctions (arrow), as though “caulking” them (74) to prevent protein leakage (84), whereas, adjacent to the alveolar membrane, it offers an ideal semipermeable membrane needed for ion channels to pump water (84). [From Hills and Masters (84)].

Some proponents of the conventional bubble model argue that EMs such as that displayed in Fig. 12 are artifactς because tannic acid causes SAPL migration, as demonstrated by Shrijvers et al. (130). However, Shrijvers et al. also go on to show how, with aldehydes, radiolabeled SAPL is washed out in the fixation procedure, so that those structures would not be visualized even if they were authentic (Fig. 12), concluding that the issue is “open to debate.”

There are a few ultrastructural studies based on aldehyde fixation (91,95) and horseradish peroxidase (47) in which fragments of the oligolamellar structures displayed in Fig. 12 are shown; but one author refers to it as “artifact” because it conflicts with the bubble model, so culturally embedded is that concept.

Ueda et al. (147) also performed epifluorescence microscopy on alveolar tissue, obtaining the green-gold color characteristic of oligolamellar SAPL by using Phosphin E as the hydrophobic probe. No tannic acid was involved. We have repeated that procedure and confirmed their findings on fetal alveolar epithelium and peritoneal mesothelium, comparing color spectra with those of DPPC controls (83). Although it has been pointed out above how the morphological picture of an oligolamellar lining of SAPL at an epithelial surface is consistent with chemisorption, it is also feasible that these structures could have arisen at the air-liquid interface of adjacent pools. They could be deposited onto the epithelial surface by the respiratory “tide” or formed as Blodgett structures (16) by tidal flow of fluid in adjacent pools. Such structures formed at liquid-air interfaces in vitro are retained upon deposition onto a solid subphase (101).

If this lining layer is real, it raises some fascinating possibilities for its role in the lung and other organs.

Possible Roles for Epithelium-Bound Surfactant

Possible roles for surfactant in the active immune system have appeared from time to time, implicating surfactant proteins (149, 161) and demonstrating how SAPL (113) enhances the engulfment of pathogens by macrophages. The fascination to this writer is why an area of the magnitude of the lungs (90 m2) is not overwhelmed by the vast number of pathogens present in alveolar air, considering that burns over only 20% of the skin (2 m2) are life threatening. Surely, there must be an innate barrier, an “airway mucosal barrier,” in which case the lining of oligolamellar SAPL, demonstrated in Fig. 12 as a solid lining supported by a solid subphase, would provide an ideal “first line of defence” (70). Although condensed DPPC at a liquid-air interface might also provide a solid barrier, this property would be severely compromised on inspiration when cracks appear in that floating layer (see Fig.6 A).

When bacteria invade a tissue, they do so by diapedesis through cellular junctions. Hence, by spanning these junctions (Fig. 12), the SAPL barrier is ideally positioned to prevent this from occurring. There is, once again, a corresponding industrial application of cationic surfactants known as “anti-rooting through agents” (70), in which an adsorbed layer prevents diapedesis in the form of root penetration between fibrous pots in contact in nurseries.

This airway mucosal barrier could extend over the entire upper respiratory tract where sinonasal cavities, trachea, and bronchi have a common respiratory mucosa (123). In normal experimental animals, it is difficult for viruses to breach this barrier unless it is pretreated with polyoxyethylene 9 lauryl ether, otherwise known as polidocanol (PODC) (114). Hence, it could be most pertinent that PODC is an excellent solvent for SAPL (unpublished observations), just as solvents such as ethanol are effective barrier breakers of the gastric mucosal barrier believed (80) to have a similar molecular structure.

Geodesic-Dome Model

An interesting alternative to the bubble model that retains the continuous AH is based on the contention (10) that the liquid-air interface is masked by a solid phase of DPPC. This “crust” cracks open on expansion/inspiration, as simulated in Fig.6 A, to expose a true liquid surface to which more surfactant can be recruited from the hypophase. On subsequent compression, this recruited surfactant can be condensed to a solid and compacted with the solid sides to form both uniform oligolamellar rafts (Fig. 6 A) and “craggy oligolamellar rafts” as demonstrated by Nag et al. (110), the process resembling the “rafting” of compressed ice caps. Bangham (10) proposes that, at end expiration, these floating solid plates come together to form a rigid structure resembling a geodesic dome, thus resisting further collapse and establishing FRC and alveolar stability. On expansion, the solid plates need an appreciable pressure difference before they are pried apart or crack open (as per Fig.6 A) to explain the inspiratory delay in volume increase, otherwise attributed to “alveolar recruitment/reopening” following closure at end expiration (56).

As the geodesic dome seals during expiration, it will tend to force the flaccid epithelial wall and any fluid into the same contour as the outer surface of the shell (Fig.13 B), consistent with the smooth contours visualized by EM after aldehyde fixation has removed the solid surfactant lining (Fig.13 C). It therefore becomes a moot point whether the surfactant structures shown in Fig. 12 are truly bound to the surface by adsorption or are solid rafts adjacent to tissue surface where intervening fluid has been largely squeezed out (Fig. 13 B). However, the bubble model remains the only one based on the air-liquid interface to be evaluated by its proponents in terms of physiological parameters, the shortcomings discussed earlier causing this writer (65) to investigate a role for the surfactant adsorbed to the tissue/solid surface.

Fig. 13.

“Geodesic-dome” model of the alveolus (9, 10). Rafts or plaques of solid surfactant at liquid-air interface are tensioned by fluid exposed by cracking of ice cap on inspiration (A).B: diagram showing how fluid could be “squeezed out” or pumped out to produce intermittent contact of the shell with tissue as the solid plates (see Fig.6 A) come together at end expiration to prevent alveolar collapse and establish functional residual capacity. C: diagram showing how the type of electron micrograph commonly displayed can arise as an artifact caused by elutriation of surfactant by the very common use of aldehydes to fix proteins, whereas the use of lipid fixatives (B) reflects the true situation. Note how tissue-fluid interface in Afollows that of the totally fluid-filled lung (49).

Wettability of the Alveolar Wall

In most accessible sites in the body where adsorbed oligolamellar layers of SAPL have been identified, the tissue surfaces appear hydrophobic after rinsing away of the adjacent fluid, as manifest by a contact angle (Fig. 1 C) of 100–103° on articular cartilage (22, 71), 60° on mucus-free bronchial epithelium (81), 63° on pleural mesothelium (73), and 75° on amnionic epithelium exposed to fetal surfactant in utero (76). These values compare with 115° for close-packed methyl groups or 104° for CH2 groups exposed to water-saturated air (15). Hence, such layers might encourage “dewatering” of the apparently fluid-free areas on the alveolar surface. It has been demonstrated how an oligolamellar layer of SAPL adsorbed to a surface as hydrophilic as glass can spontaneously rupture a supernatant aqueous layer of up to 900 μm in thickness (68), just as other adsorbed surfactants are known to demonstrate such a “disjoining pressure” in vitro (33). Unfortunately, the alveolar surface is inaccessible for direct measurement.

A very common example of an adsorbed layer of SAPL acting as a surface dewatering agent occurs in the eye, where it is now known (85) to cause breakup of the tear film, which occurs on the ocular surface if we do not blink every 20 s or so. Hence, there is a substantial body of evidence supporting the concept (65, 70) of fluid-free areas on the normal alveolar surface.

The Morphological Model

If the oligolamellar SAPL lining, shown in Fig. 12, is directly adsorbed to alveolar epithelium, then it can act as a mild dewatering agent by pushing water aside from the gas-exchange surface into septal pools, as depicted in Figs. 8 B and11 A.

However, the first test of any model is to interpret the pressure-volume curve of an excised lung. Whereas pressure-volume data for the bubble model give a γ-surface area loop, which does not even overlap that produced on the Langmuir-Adam trough (86) except at maximal inflation, the loops are compatible if the contact angle (Fig. 1 B) of fluid on the tissue also varies as predicted (67). It is important to remember that all EMs are taken at zero-retraction pressure when θ is zero or almost so. Thus the force with which the fluid pools grip and, hence, tension alveolar epithelium is a function of both γ and θ (see Fig.11).

If one dispenses with the concept of a continuous aqueous layer, then the problem of instability introduced by the bubble model is no longer relevant.

One consequence of a discontinuous fluid lining based on the morphological model is that, upon inflation, the additional surface area will be provided by expansion of the pools rather than dry surface. Thus area change of the liquid-air interface confined within the edges of the pools will be greatly amplified (see Fig.11 D), making it far easier to justify a near-zero surface contractile force than it is possible when the conventional bubble model is used. Thus pools amounting to 10–30% of the alveolar surface would still exert a tensiolytic role eliminated by liquid filling. The question now remaining is how this morphological model can account for fluid balance in the lung.

Alveolar Fluid Balance

There is an instinctive assumption that, because an inflated alveolus tends to recoil inward, the same forces must be tending to suck fluid inward. However, in the morphological model, it can be seen in Fig.11 B how pools rendered convex by accumulated edema fluid can still tension the walls, and yet the natural tendency for the interface to collapse toward the concave side now generates a positive pressure helping to resolve excess fluid (Fig.11 B). As excess fluid is resolved (Fig. 11 C), the convexity becomes less as the radius of curvature (r inEq. 1 ) increases, so ΔP decreases and eventually becomes negative as the pool becomes concave (Fig.11 A), until it balances the pressure generated by the physiological water pumps. Hence, in the morphological model (Fig. 8 B), the fluid system is self-regulating (65, 70).

If other factors enlarge the pools to the point at which they join up, then they form the bubble model but as a pathological rather than the normal physiological state (b→a in Fig. 8). Having now formed one-sided bubbles, the instability considerations discussed above would then apply, explaining atelectasis. It also explains the stage of alveolar edema at which return to normal capillary pressure cannot reverse flooding (41) (ΔP↑↑ asr↓↓ inEq. 4 ).

All of the alveolar models compared thus far have been essentially static, but the alveolar wall could be highly dynamic as red blood cells shown in Fig. 10 bulge their way through underlying capillaries. The resulting “oilcanning” of the alveolar surface could induce a net fluid pressure in the vicinity of 8 mmHg because surface tension in the expanded (convex) mode exceeds that in the compressed (concave) mode to generate a net driving force for resolving fluid (75). However, in the living state, it is a moot point whether the space between red blood cells is occupied by plasma to the extent that capillaries become cylindrical. If not, then many points on the surface will be oscillating at over 60 Hz. Even if plasma does “iron out” the gaps, it still leaves a microconvex profile imposed on the alveolar wall.

Neonatal Lung

The remarkable ability of the normal neonatal lung to expel fluid at birth has raised the issue of whether the normal physiological mechanism(s) that have been pumping fluid into the potential air spaces in the fetus can reverse sufficiently rapidly to explain this phenomenon. Physical assistance in the form of compression of the thorax during labor or dewatering of the alveolar surface by preadsorbed surfactant (83) have been proposed. On the other hand, it is claimed (139) that β-adrenergic stimulation of ion channels due to adrenaline release during labor can stimulate those water pumps to cope with the extreme water load at birth.

The question that this raises is why such an efficient pumping system should be so severely compromised when there is a deficiency of surfactant, as occurs in the RDS (4). Despite the administration of exogenous surfactant alleviating hypoxia, the so-called “pinking” of neonates with RDS, it is events occurring some 24–48 h later that often determine the clinical outcome (104, 112). The infant cannot be weaned from the ventilator until sufficient fluid is cleared (104), and this has been known to take 8 days or more.

This raises the issue of whether surfactant could be a vital component of the pumping system per se (84). It is particularly interesting when the proponents of ion-channel pumps (153) state that “for a secretory organ to be capable of generating a chemical gradient a barrier must be present to restrict molecular diffusion,” whereas, in the fetal lung at least, “this barrier resides within alveolar epithelium.” The barrier of oligolamellar SAPL shown in Fig. 12would seem ideal for this function (84), since not only does it span epithelial cell junctions to render the membrane tight but even a monolayer can reduce, by an order of magnitude, the permeability of solid surfaces to cations as small as hydrogen ions (80). Essentially, the same oligolamellar structures in the form of liposomes (6) are also excellent semipermeable membranes, as needed to maintain an ion gradient and shift water.

Such an airway mucosal barrier can also act as the semipermeable membrane by which large (153) protein gradients can pump water when, by spanning intercellular junctions (Fig. 12), they should also prevent protein leakage that generally occurs through intercellular gaps. Protein leakage is a major problem in RDS (44) and especially the acute form (ARDS) (99) where any overdistension of the alveolar wall by the ventilator is likely to exacerbate any deficiency in the SAPL bridges that lack the mechanical support of epithelial cells. The tight junctions between epithelial cells determine protein leakage (119). The ability of serum proteins to carry away highly insoluble lipids (128) in the type of reversible combination discussed above would further erode lipid barriers (Fig. 12), possibly accounting for their clinically inhibitory actions (136).

“Surfactant Rescue”

Surfactant rescue has enjoyed outstanding success in reducing infant mortality by 40–50%, as described in many clinical reviews (135,138), any difference in efficacy between formulations being relatively minor. This may have occurred because all formulations of exogenous surfactant are designed with the sole objective of spreading the highly insoluble DPPC rapidly over an expanding liquid-air interface. Spreading is facilitated by chemical dispersing agents, viz. hexadecanol and Tyloxapol in the case of Exosurf (30), proteolipids coextracted from bovine or porcine lungs as Survanta (96) and Curosurf (138), respectively, while compounding with a second phospholipid, viz. phosphatidylglycerol (PG) as a dry powder, enables artificial lung-expanding compound to spread rapidly. The much greater surface activity retained by dry over wet surfactant (Fig. 7) has been much extolled by Morley et al. (106-109) and Bangham and co-workers (12). It is only when surfactant is applied in the less surface-active wet form that such formulations benefit from the inclusion of surfactant proteins (see Fig. 7 B).

If the morphological model (Fig. 8 B) is appropriate to the normal air-filled lung then, according to the Langmuir isotherm described above, the same lack of surfactant causing initial hypoxia would also result in a deficient lining adsorbed to the epithelial surface, thus compromising both the protein-oncotic and ion-channel water pumps. Hence, the future formulation of exogenous surfactant might be directed more toward establishing the SAPL barrier adsorbed to epithelium (Fig. 12); i.e., repairing the water pumps, in addition to spreading across the surface of water to alleviate hypoxia.

In conclusion, more attention might be paid to the vast literature outside of respiratory physiology and medicine where surfactants are studied at many surfaces, in addition to the liquid-air interface, for the many highly desirable properties (70) of potential physiological benefits that they can impart.


  • Address for reprint requests and other correspondence: B. A. Hills, Paediatric Respiratory Research Centre, Mater Children's Hospital, South Brisbane, Qld 4101, Australia.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract