In the 1940s, the possibility that a thin endothelial glycocalyx layer may exist on the luminal surface of the endothelial cell lining of the blood capillary microvasculature was suggested by Danielli[78
], and Chambers and Zweifach[142
]; however, at that time, the glycocalyx layer was difficult to visualize by conventional light and electron microscopy staining techniques. Therefore, in 1959, when the morphological classification scheme for vertebrate blood capillaries was developed by Bennett and colleagues[84
], the capillary wall was still considered to be a two-layered structure, consisting of the endothelial cell lining and basement membrane layers. Although evidence for the existence of an approximately 20 nm thick glycocalyx layer was provided by Luft in 1966[69
] based on electron microscopy of fixed skeletal muscle capillaries stained with ruthenium red, it is now well-known that the endothelial glycocalyx layer is a 150 to 400 nm-thick polysaccharide-rich anionic matrix of sialyated and sulphated proteoglycans and glycoproteins in the physiologic state in vivo
]. Since the individual fibers of the glycocalyx are circumferentially spaced a maximum of 20 nm apart[61
], this relatively narrow interspacing of the individual glycocalyx fibers would restrict the transvascular passage of larger macromolecules through the endothelial glycocalyx layer. This, indeed, appears to be case in the non-sinusoidal fenestrated blood capillaries of the kidney glomeruli that possess open 'non-diaphragmed' fenestrae covered by an intact glycocalyx layer, as the physiologic upper limit of pore size in the capillary wall of this blood capillary type is approximately 15 nm[60
], and may be closer to between 13 and 14 nm if the upper limit of pore size of kidney glomerulus capillaries is interrogated using non-flexible spherical macromolecules. This being the case, the barrier to the renal filtration of macromolecules such as hemoglobin (diameter ~6.4 nm) and albumin (diameter ~ 7 nm) are the slit-diaphragms of podocyte foot processes on the abluminal face of the basement membrane layer.
In 1951, Pappenheimer and colleagues formulated the classic small pore theory of microvascular permeability on the basis of experimental data on the restriction to the transvascular flow of various-sized unlabeled endogenous lipid-insoluble molecules across the walls of cat hind-limb microvasculature[6
]. By the measurement of the osmotic transients generated by the respective test molecules using the isogravometric osmotic transient technique, Pappenheimer et al. determined that cat hind-limb microvasculature was permeable to inulin (diameter ~3 nm), but not to hemoglobin (diameter ~6.4 nm). On comparison of the experimental values for the restriction to diffusion to theoretically predicted values for uniform water filled cylindrical pores 6 nm in diameter (less restrictive pores), or alternatively, for uniform water-filled rectangular slit-pores approximately 3.7 nm in width (more restrictive pores), it was observed that the experimental data values fit better with the theoretically predicted values for cylindrical pores 6 nm in diameter. However, in lieu of the polydisperse nature of inulin, which was the smaller of the two macromolecular test substances, it was noted that the range of the upper limit of pore size could only be established with additional experimental data demonstrating the restricted transvascular flow of less polydisperse macromolecular test substances similar in size to inulin. In subsequent, similar perfused cat hind-limb osmotic transient experiments with myoglobin (diameter ~4 nm), the restricted transvascular flow of myoglobin was demonstrated, and it was established that the physiologic upper limit of pore size in cat hind-limb blood capillary microvasculature is between 4 and 6 nm[7
]. The blood capillary microvasculature of the cat hind-limb is comprised of the capillary microvasculatures of the several different tissue types in the limb, which includes the capillary microvasculatures of skin, muscle, nerve, bone, myeloid bone marrow, adipose, and connective tissues; of these limb tissues, muscle, bone, and adipose tissue capillary microvasculature constitutes one type (non-sinusoidal non-fenestrated) (Table ; Figure , panel A; Additional file 1
); skin, nerve, and connective tissue another type (non-sinusoidal fenestrated) (Table ; Figure , panel B; Additional file 2
); and myeloid (red) bone marrow yet another (sinusoidal reticuloendothelial) (Table ; Figure , panel D; Additional file 4
). The physiologic upper limit of pore size for skin, nerve, and connective tissue blood capillaries, being non-sinusoidal fenestrated capillaries, is within the 6 to 12 nm range (Table ). Therefore, the physiologic upper limit of pore size of cat hind-limb blood capillary microvasculature, as measured by the isogravometric osmotic transient method, would be a slight over-estimation of the 'actual' physiologic upper limit of pore size in skeletal muscle blood capillary microvasculature. This is the likely reason why the experimental data values fit better with the theoretically predicted values for aqueous cylindrical pores 6 nm in diameter. This conclusion is supported by the ultrastructural evidence that the interendothelial cell clefts in the capillary walls of the non-sinusoidal non-fenestrated skeletal muscle blood capillaries are lined by macula occludens interendothelial cell junctions, which, in series, restrict the transcapillary passage of macromolecules larger than horseradish peroxidase (diameter ~4.6 nm) across the interendothelial cell cleft[79
](Table ; Figure , panel A; Additional file 1
The interplay of tissue blood capillary and interstitial space hydrostatic and oncotic pressures favors the net filtration of fluids and lipid-insoluble molecules across the blood capillary walls, and this ultrafiltrate, is the tissue lymph[47
]. Since the openings of the terminal endings of tissue interstitium lymphatic capillaries, or initial lymphatics, are permeable to macromolecules upwards of several hundred nanometers, the level of the restriction to the passage of systemically administered macromolecules is the tissue blood capillary wall[148
]. These differences in the ultrastructure of the tissue blood capillary and lymphatic capillary walls is the basis on which differences in macromolecule plasma concentration and regional lymph concentration have been used as indexes of regional differences in blood capillary permeability. Non-endogenous macromolecules that are not rapidly cleared from blood circulation accumulate in the myeloid bone marrow interstitium and the periosteal fibrous tissue lymphatics[47
]. This information is most pertinent to the interpretation of lower extremity and cervical region lymph flow data as the presence of non-endogenous macromolecules larger than approximately 5 nm in diameter within the lymphatic drainage of these regions would be attributable to the accumulation of these non-endogenous macromolecules in the myeloid bone marrow sinusoidal transcapillary filtrate.
Upon the intravenous administration of radiolabeled native macromolecules, radio-iodinated albumin and immunoglobulin, to dogs with cannulated thoracic lymph ducts, and the measurement of the changes in the respective test molecule concentrations in plasma and thoracic duct lymph, Wasserman and Mayerson, in 1952, noted that: (1) the lymph concentration of radio-iodinated albumin increases approximately 1.6 times faster than that of radio-iodinated immunoglobulin; and that (2) when administered at high-doses, steady-state lymph levels of both radio-iodinated species are achieved within 90 minutes after administration; whereas, when administered at low-doses, steady-state lymph levels are achieved between 7 and 13 hours after administration[11
]. The faster rate of accumulation of radio-iodinated albumin than immunoglobulin in thoracic duct lymph is consistent with the fact that the pore sizes in the diaphragmed fenestrae of non-sinusoidal fenestrated blood capillaries in many visceral organs and tissues are more restrictive to the transcapillary passage of immunoglobulin than albumin, as the physiologic upper limits of pore size in non-sinusoidal fenestrated blood capillaries with diaphragmed fenestrae vary between 6 and 12 nm (Table ). However, the fact that the test substances employed were radiolabeled native macromolecules is notable, as these radio-iodinated test substances[54
] constitute non-endogenous macromolecules that would be phago-endocytosed by reticuloendothelial cells of hepatic and myeloid bone marrow sinusoidal capillary walls. Based on the observed dose-related differences in the rates of radio-iodinated albumin and immunoglobulin accumulation in the thoracic duct lymphatic drainage, which is primarily that of the hepatic region lymph, the great proportion of these non-endogenous macromolecules the administered at low-doses are likely phago-endocytosed at the level of the hepatic sinusoidal blood capillary walls, and do not have the opportunity to actually flow across the open fenestrae of the hepatic sinusoids to enter the hepatic interstitium by the transvascular convective route. However, at high doses, the phago-endocytic activity threshold of the lining reticuloendothelial cells of the hepatic sinusoid blood capillaries is reached; and as such, the proportion of the dose of the circulating macromolecules above and beyond the phago-endocytic activity threshold flows across the open fenestrae and enters the hepatic interstitial space via the transvascular convective route. Therefore, at high doses, the radio-iodinated albumin and immunoglobulin accumulate in the hepatic region lymph and in thoracic lymphatic drainage at faster rates, than when administered at low-doses.
In 1956, Grotte et al. performed a series of additional dog lymph flow studies by employing various-sized dextran and plastic nanoparticles, the findings of which were the basis for the formulation of the dual pore hypothesis of microvascular permeability[12
]. The experimental findings underlying the formulation of the dual pore hypothesis of capillary permeability are discussed herein from the physiologic perspective. When the findings are viewed in this light, it becomes apparent that the findings are not per se
a confirmation for the existence of a large pore population, but rather, are evidence for the role of the reticuloendothelial cells of the hepatic and myeloid bone marrow sinusoids in the phago-endocytosis non-endogenous macromolecules. Grotte employed dextran nanoparticles ranging in size from approximately 2.5 to 24 nm in diameter, and fluorescent spherical plastic (methylmethacrylate) nanoparticles ranging in size from 60 to 140 nm in diameter, and measured the steady-state concentrations of the respective non-endogenous macromolecules in the locoregional lymphatics; which, in case of the lymph flow studies with the dextrans were the hepatic, lower extremity, and cervical region lymphatics; and in the case of the plastic nanoparticles were the hepatic, lower extremity, cardiac, and bronchial region lymphatics. The lymph concentrations of the dextrans smaller than 8 to 10 nm in diameter were measured at 7 hours after the intravenous infusion to animals with renal occlusion, and those of the larger dextrans were measured at 24 hours, which would have been sufficient periods of time for the smaller and larger dextrans to have reached steady-state lymph concentrations. It is notable that the lymph concentrations of the plastic nanoparticles were measured immediately following infusions at various time points over a period of between 3 and 4 hours, since these larger plastic nanoparticles were quickly cleared from systemic blood circulation, which is attributable to the rapid removal of these immunogenic particles by phago-endocytic uptake by hepatic Kupffer macrophages, and splenic red pulp macrophages, as well as, the reticuloendothelial cells of the hepatic sinusoidal capillaries.
On review of lymph flow study experimental data, it is evident that the point at which the particle plasma:lymph concentration ratio first deviates from approximately 1, or unity, represents the pore size cut-off of the capillary ultrafiltrate and the lymphatic drainage of the capillary population with the lowest upper limit of pore size, which constitutes that of non-sinusoidal non-fenestrated capillaries; and, several graded decreases in the particle plasma:lymph concentration ratio over a range of particle sizes represent the pore size cut-offs of the capillary ultrafiltrates of multiple capillary populations with different upper limits of pore size, which constitutes that of fenestrated capillaries. In case of the lower extremity and cervical region lymphatic drainage, there are graded decreases in the dextran particle plasma:lymph concentration ratios between dextran particle sizes from 4 to 8 nm in diameter, from approximately 1 to approximately 0.15 for lower extremity lymph, and from approximately 1 to approximately 0.25 for cervical lymph, which indicates the presence of populations of non-sinusoidal non-fenestrated capillaries with the upper limit of pore size closer to 4 nm, and indicates the presence of populations of non-sinusoidal fenestrated capillaries in the region with the upper limit of pore size closer to 8 nm. In the case of the hepatic region lymphatic drainage, the dextran particle plasma:lymph concentration ratio decreases from approximately 1 to approximately 0.85 at a dextran particle size of 8 nm, which indicates the presence of a population of non-sinusoidal fenestrated capillaries with a physiologic upper limit of pore size of approximately 8 nm, consistent with the cut-off of pore size of the capillary ultrafiltrate of intestinal mucosal non-sinusoidal fenestrated blood capillaries (Table ).
The observation that the plasma:lymph concentration ratio for the larger dextran particles between 8 and 24 nm in diameter remains unchanged for various body regions has been cited as physiologic evidence for the existence of another population of large pores, or 'capillary leaks', at least 24 nm in diameter in blood capillary microvasculature. Further, this observation, coupled with the finding that the plastic particle plasma:lymph concentration ratio for the lower extremity, cardiac, and bronchial region lymphs is approximately 0, or unmeasurable, has been cited as the physiologic evidence for the existence of the large pore population in capillary microvasculature. Based on these findings, that particles as large as 24 nm accumulate in the lower extremity and cervical lymph (dextran nanoparticles), and that particles 60 nm, and larger, do not accumulate in the lower extremity and cervical lymph (plastic nanoparticles), it has been concluded that the physiologic upper limit of pore size of the large pore population in capillary microvasculature ranges between 24 and 60 nm. Furthermore, the differences in the regional dextran particle plasma:lymph concentration ratios for the 8 to 24 nm diameter dextran particles (approximately 0.15 for lower extremity region lymph, approximately 0.25 for cervical region lymph, and approximately 0.85 for hepatic region lymph) have been additionally cited as evidence for differences in the number of small and large pores in the blood capillary populations of the lower extremity, cervical, and hepatic regions, with the small-to-large pore ratios for the respective regions being 1:34,000, 1:18,000, and 1:340[12
In case of the dextran nanoparticles between 8 and 24 nm in diameter, the findings taken altogether, support the likelihood that the administered dose of the dextran nanoparticles was high enough to saturate the phago-endocytic capacity of the reticuloendothelial cells of the hepatic and myeloid bone marrow sinusoidal capillaries, resulting in the accumulation of the dextrans nanoparticles in the respective interstitial tissue spaces and regional lymph. With regards to the accumulation in the hepatic interstitium and lymph, this would be via transvascular flow through open fenestrae and the phago-endocytic route; and with regards to the myeloid bone marrow, this would be via the phago-endocytic route. It is postulated here that the measured dextran nanoparticle concentrations in the lower extremity and cervical regional lymphatic drainages constitute the dextran nanoparticle concentrations of the myeloid bone marrow sinusoidal capillary filtrates of the respective regions. The relatively high dextran particle plasma:lymph concentration ratio of the hepatic region lymph (approximately 0.85), as compared to that of lower extremity and cervical region lymphs (approximately 0.2), is consistent with the hepatic sinusoidal capillary filtrate constituting the major proportion of the lymphatic drainage of the hepatic region. In case of the plastic nanoparticles between 60 and 140 nm, the particle plasma:lymph concentration ratio for the hepatic region lymph only reaches approximately 0.2 over 3 hours, and does not maintain steady-state levels over longer time periods; whereas particle plasma:lymph concentration ratio for the lower extremity, cardiac, and bronchial region lymph is approximately 0, or unmeasurable. The relatively rapid clearance of the plastic particles from blood circulation is attributable to the rapid sequestration of these particles in the splenic red pulp followed by phagocytosis of particles by the pulp macrophages. Therefore, the low level of plastic particle accumulation in the hepatic interstitium and associated lymphatic drainage, and virtually no plastic particle accumulation in the lower extremity, cardiac, and bronchial region lymphatic drainages, reflects the low level of phago-endocytic particle uptake at the level of the myeloid bone marrow blood capillaries, secondary to the rapid clearance of plastic particles from systemic blood circulation.