Variability between divers (bubblers compared with non-bubblers) is similar to the variability between sheep, as depicted in Figure 11. Sheep were grouped according to total bubble formation for all the blood vessels examined in each animal, normalized for area and time. Divers are presented according to their bubble score (data compiled from Lambrechts et al., 2013; Cialoni et al., 2015). The distribution is similar, with a high frequency of non-bubblers and a low frequency of heavy bubblers.

It has been suggested that the AHS are composed of lung surfactants deposited on the luminal aspect of blood vessels (Hills, 1992; Arieli, 2015). We recently confirmed the validity of this hypothesis, by showing that DPPC leaks from the lungs into the blood and settles to form AHS on the vessel (Arieli et al., 2015).

When a sequence of bubbles becomes detached from an AHS, the AHS is activated and the rate of bubble production increases (Figure 12, left panel). After seven successive detachments, the time interval between further detachments stabilizes over a short period. This activation of the AHS coincides with the phenomenon whereby more bubbles are seen in the blood after a second dive on the same day made using the same profile (Figure 12, right panel, adapted from Dunford et al., 2002). There are more activated AHS in the second dive.

Because AHS are to be found in blood vessels within the arterial as well as the venous circulation, with a distribution going as far as the cerebral capillaries (Hills, 1992), bubbles might develop within the arteries. This mechanism can explain symptoms of neurologic decompression sickness that occur without arterialization of venous blood, either via a patent foramen ovale or intrapulmonary arteriovenous anastomoses (Madden et al., 2015; Balestra and Germonpre, 2016). After decompression, at any distal bifurcation of the arterial tree, flow decreases for each vessel and the vessel's diameter is reduced. This increases the surface area available for diffusion with respect to blood volume, and the reduced wall thickness reduces the diffusion barrier. Thus, the diffusion of inert gas from the tissue into the blood will rise along the arterial tree. This may cause the expansion of bubbles at AHS within the distal arteries. Local reduction of blood flow would enhance the diffusion of inert gas into the artery. If arterialization of venous blood were the source of arterial bubbles, because the brain receives eight times the amount of arterial blood directed to the spinal cord, more arterial bubbles shunted from the venous circulation should reach the brain (Hallenbeck et al., 1975). However, spinal DCI is over three times more frequent than cerebral DCI. The internal veins which drain the spinal cord do not have valves (Stringer et al., 2012). It was also suggested that after decompression, spinal blood flow is reduced due to obstructions in the epidural vertebral veins (Hallenbeck et al., 1975). It is therefore possible that decompression bubbles develop in distal arteries within the nervous system, and more so in the spinal cord. A detailed discussion of arterial bubbles was presented in the paper by Arieli and Marmur (2017).

Various methods have been established as pretreatment before diving to reduce decompression stress. It has previously been hypothesized that exposure to hyperbaric oxygen at the beginning of a dive would result in replacement of the inert gas in the micronuclei by oxygen, with subsequent consumption of the oxygen by the mitochondria. This would shrink gas micronuclei having the potential to grow into bubbles and thus reduce the risk of DCI. The preoxygenation hypothesis was supported by experimental studies in prawns, rats and humans (Arieli et al., 2009; Castagna et al., 2009, among others). Pre-dive exercise, sauna, and vibration all resulted in a reduced bubble score following the dive (Blatteau et al., 2007, 2008; Germonpré et al., 2009; Balestra et al., 2016; Germonpre and Balestra, 2017). One of the commonly suggested explanations for the relief obtained using these methods is elimination or dislodgment of some of the gas micronuclei from the surface of blood vessels. According to Fang et al. (2016), it takes somewhat more than 2 h for nanobubbles to reappear. Thus, pretreatment was carried out within the time scale required to preempt the renewal of nanobubbles, and consequently the development of gas micronuclei.

Previous models of decompression presented investigators with difficulty when it came to matching the time required for bubble expansion by diffusion (a short process) with the development of DCI (a prolonged process). To overcome this problem, some investigators suggested an artificially low diffusion constant (Hugon, 2014), while others suggested a bubble skin as a barrier to diffusion (Yount, 1979; Yount and Hoffman, 1986; Yount et al., 2000; Kuch et al., 2011) or an envelope of surfactants (Wienke, 1990, 2009). A bi-phasic mechanism of bubble expansion, initiation (Figure 10) followed by diffusion-driven growth, against the background of the AHS, makes bubble expansion compatible with the development of DCI.

It is well-known that an increase in the exposure pressure in diving causes an increase in bubble production after decompression. An example of this is presented in Figures 13C,D (adapted from Eckenhoff et al., 1990; Cameron et al., 2007). It may be related to (1) Faster expansion of bubbles when there is a greater difference in gas tensions, with earlier detachment and thus an increase in the rate of bubble formation at the same AHS. (2) Activation of more AHS when gas tensions are higher. When a decompressed ovine blood vessel was taken from the hyperbaric chamber and placed in calm conditions, with a calculated gas tension of 228 kPa in the proximity of the tissue, no bubbles appeared. After activation of pulsatile flow of the saline, mixing yielded a gas tension of 600 kPa in the proximity of the tissue, and large numbers of bubbles appeared (Figure 13A). This was seen in 40% of the blood vessels, and we interpreted it as activation of gas micronuclei at the high gas tension. This is supported by our finding in the hydrophobic silicon wafer, that the density of bubbles increased with the decompression steps (Figure 13B). There will therefore be increased activation of AHS with the higher gas tensions when diving at elevated pressure.

Contact between bubbles and signaling receptors at the endothelial membrane can promote unwanted events. Sobolewski et al. (2011, 2012) exposed endothelial cells to contact with air bubbles. They demonstrated an influx of calcium through a stretch-activated channel, such as a transient receptor potential vanilloid family member, triggering the release of calcium from intracellular stores. They also showed activation of a calcium-independent, PKCα-dependent signaling pathway, resulting in mitochondrial depolarization. Mitochondrial dysfunction is likely a key contributor to the pathophysiology of DCI. Bubble contact with endothelial cells released microparticles which decreased cell viability, increased apoptosis, and caused over-expression of pro-inflammatory cytokines (Yu et al., 2017). Protection of the endothelium with escin (the main active compound in horse chestnut seed extract) reduced the incidence of DCI in the rat (Zhang et al., 2017a). Bubbles which develop at the AHS on the surface of the endothelium are liable to come into contact with the cell membrane. In our pulsatile flow experiments, large bubbles which rocked back and forth with the flow most likely did come into contact with the endothelial membrane. Proteoglyans, glypicans, and glycoproteins (the glycocalyx) protrude some tens of nm from the luminal aspect of the membrane. Plasma proteins bind to the glycocalyx to produce a 0.5–1.0 μm layer of macromolecules. This endothelial surface layer is removed by reactive oxygen species (Pries et al., 2000). It is possible that reactive oxygen species in diving remove this layer, exposing more AHS to the flow of plasma and thus enhancing bubble production. Markers of endothelial injury, such as cell adhesion molecule-1 (ICAM-1) and endothelin-1 (ET-1), were well correlated with the timing of bubble detachment and their number in the decompressed rat (Zhang et al., 2017b), a finding which supports a cause-and-effect relationship in which detached bubbles bring about endothelial injury.

Microparticles are present in the blood of any healthy individual. These MPs are tiny vesicles (0.1–1.0 μm) derived from pieces of cell membrane from leukocytes, erythrocytes, platelets, and endothelial cells. Exchange of markers between different MPs may cause a single MP to carry markers from various sources. Their number increases with stress, such as oxidative stress and apoptosis. After diving, there is an increase in circulating microparticles in the blood. Extensive research on this topic has been conducted in mice and divers (Thom et al., 2011, 2012, 2013). Some of the enlarged MPs contain gas (Yang et al., 2012; Thom et al., 2013). Thom et al. (2012) reported that there was a significant inverse correlation between post-diving bubble scores and MPs in divers. These MPs cause platelet aggregation and inflammation, and neutrophil activation. An association has been shown between MPs and decompression sickness (Thom et al., 2015). Madden and Laden (2009) even suggested that endothelial malfunction and MPs, not gas bubbles, may be the underlying cause of DCI. Enlarged MPs contain gas (Yang et al., 2012), are rich in iNOS, and contain NO₂ (Thom et al., 2013). Eight percent of the micronuclei from normal mice contained gas, and Thom et al. (2013) suggested these may be the gas micronuclei from which bubbles develop after decompression. Pathological effects were related to these enlarged MPs. Bubbles caused the release of endothelial MPs from endothelial culture, and these specific MPs were associated with severe endothelial injury and pro-inflammatory cytokines (Yu et al., 2017). However, I find it difficult to accept that even with high iNOS activity, the short-lasting NO and the more water-soluble NO₂, which easily crosses membranes and decomposes in water (Clean Air Technology Center, 1999; Signorelli et al., 2011), would create a pure gas phase from a solution of NO₂. There was also no mention of the mechanism whereby the MPs are torn from the membrane during and after decompression. According to my understanding, it is more probable that NO and its oxidized derivatives would diffuse into an already existing gas phase. Therefore, I would suggest that pieces of membrane are carried along with detached bubbles, some of which completely lose their gas in the lung, whereas others retain some of the gas, enabling them to cross the lung capillaries. It is possible that DPPC settled preferably on iNOS-rich membrane to form the AHS, which made the enlarged MPs rich in iNOS. A possible explanation for the negative relationship between MPs and bubbles (Thom et al., 2012) may be that at some time after the dive, most of the bubbles lost their gas in the lung and continued to circulate as MPs, whereas the number of newly formed bubbles in the venous circulation had already been depleted. Thom et al. (2013) suggested that gas-containing MPs serve to initiate bubble growth, and that once formed, it is the bubbles that cause widespread physical damage. I would suggest the opposite, that nanobubbles at the AHS and detached bubbles create the decompression MPs. It may be that when a bubble formed at an AHS loses most of its gas in the lung and continues to circulate as an enlarged gas-containing MP, it will serve as a secondary gas micronucleus for further bubble growth in both arterial and venous blood.

Completely comprehensive physiological decompression models should therefore take into account bubble formation at the AHS, both in the venous and distal arterial circulation, as well as the added risk of inflammation, blood clotting and neutrophil activation, due to venous bubbles which are transformed into MPs while passing through the lung, and secondary bubble expansion from these gas-containing MPs.

A nanobubble gas phase in the path of the circulating blood may have far-reaching effects. Proteins with hydrophobic regions circulating in the blood will adhere to the gas phase-plasma interface. Deformation of their secondary and tertiary configuration will present them as foreign molecules or autoantigens. Normally hidden components of the intact protein (therefore unrecognized during thymus education), which are also present in a deformed protein, may be recognized as foreign as well, i.e., epitopes. This process has been proposed as a trigger for autoimmune diseases (Arieli, 2015). The presence of an autoimmune disease in homoeothermic, air-breathing vertebrates (DPPC can aggregate only in warm-blooded animals) will serve to increase autoimmunity, and the elevated risk of DCI with age may also be matched with the appearance of AHS. The same is true for the variable sensitivity to both diseases. If this hypothesis is proven correct, eliminating the AHS may provide protection against autoimmune diseases and DCI.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.