Development of the blood-brain barrier: A historical point of view


  • Domenico Ribatti,

    Corresponding author
    • Department of Human Anatomy and Histology, University of Bari Medical School, Piazza G. Cesare, 11, Policlinico, 70124 Bari, Italy
    Search for more papers by this author
    • Drs. Ribatti and Nico are in the Department of Human Anatomy and Histology, University of Bari Medical School, Bari, Italy.

    • Dr. Ribatti is Full Professor of Human Anatomy at the University of Bari Medical School, Italy. His primary interest is the study of angiogenesis in several physiological and pathological conditions,

    • Fax: 39-080-5478310

  • Beatrice Nico,

    Search for more papers by this author
    • Drs. Ribatti and Nico are in the Department of Human Anatomy and Histology, University of Bari Medical School, Bari, Italy.

    • Dr. Nico is Full Professor of Histology at the University of Bari Medical School, Italy. She has extensively studied the development of the blood brain barrier and actually she studies the blood-brain barrier alterations occurring in Duchenne Musculary Dystrophy.

  • Enrico Crivellato,

    Search for more papers by this author
    • Dr. Crivellato is in the Department of Medical and Morphological Researches, Anatomy Section, University of Udine Medical School, Udine Italy,

    • Dr. Crivellato is Associate Professor of Human Anatomy at the University of Udine Medical School, Italy. He taught gross anatomy for several years and currently his research interests are in the areas of cell degranulation and the history of anatomy.

  • Marco Artico

    Search for more papers by this author
    • Dr. Artico is in the Department of Pharmacology of Natural Molecules and General Physiology, Anatomy Section, University of Rome “La Sapienza,” Rome, Italy.

    • Dr. Artico is Associate Professor of Human Anatomy at the Faculty of Pharmacy of the University “La Spaienza” of Rome, Italy. Specialist in neurosurgery, his scientific activity is devoted to the study of several aspects of clinical anatomy.


Although there has been considerable controversy since the observation by Ehrlich more than 100 years ago that the brain did not take up dyes from the vascular system, the concept of an endothelial blood-brain barrier (BBB) was confirmed by the unequivocal demonstration that the passage of molecules from blood to brain and vice versa was prevented by endothelial tight junctions (TJs). There are three major functions implicated in the term “BBB”: protection of the brain from the blood milieu, selective transport, and metabolism or modification of blood- or brain-borne substances. The BBB phenotype develops under the influence of associated brain cells, especially astrocytic glia, and consists of complex TJs and a number of specific transport and enzyme systems that regulate molecular traffic across the endothelial cells. The development of the BBB is a complex process that leads to endothelial cells with unique permeability characteristics due to high electrical resistance and the expression of specific transporters and metabolic pathways. This review article summarizes the historical background underlying our current knowledge of the cellular and molecular mechanisms involved in the development and maintenance of the BBB. Anat Rec (Part B: New Anat) 289B:3–8, 2006. © 2006 Wiley-Liss, Inc.


The original concept of a barrier preventing the movement of certain materials between the blood and the adult brain stemmed from studies of dye injections made into circulation. In 1878, the German scientist Paul Ehrlich (Fig. 1) obtained his doctorate in medicine with a dissertation on the theory and practice of staining animal tissues. In the same year, he was appointed assistant to Professor Frerichs at the Berlin Medical Clinic, who gave him every facility to continue his work with these dyes and the staining of tissues with them.

Figure 1.

A portrait of the germal scientist Paul Ehrlich, Nobel Prize in 1908 with Metchnikoff.

In 1885, Ehrlich reported that after parental injection in adult animals of a variety of vital dyes, practically all animal organs were stained, except the brain and spinal cord. An early conclusion from these experiments was that the specific feature of the central nervous system (CNS) was a lack or low affinity for vital dyes. Although Ehrlich himself described the observation that after intravenous application of some aniline dyes, most of the animal tissues were stained with the exception of the CNS, he thought that this difference was due to different binding affinities.

The existence of a barrier at the level of cerebral vessels was first postulated by Bield and Kraus (1898) and Lewandowsky (1900) based on the observation that the intravenous injection of cholic acids or sodium ferrocyanide had no pharmacological effects on the CNS, whereas neurological symptoms occurred after intraventricular application of the same substances. Lewandowsky (1900) introduced the term “blood-brain barrier” (BBB) to describe this phenomenon, but the idea of the existence of a barrier was not cleared until much later, when it was shown that the principal reason why certain dyes do not penetrate from the blood into the brain is because they are bound to plasma proteins, mainly albumin (Tschirgi,1950).

Further experiments by Goldmann (1909,1913), an associate of Ehrlich, however, indicated that after injection of the acidic dye Trypan blue into the brain ventricular system of dogs and rabbits, the brain tissue became colored, while by intravenous injection of Trypan blue the whole animal turned blue with the exception of the brain and spinal cord, which remained unstained; the choroid plexuses were stained. These experiments demonstrated that the lack of staining was not attributable to a lack of affinity of the dye for brain tissue. Goldmann (1913) hypothesized that the vehicle for substance transport was the cerebrospinal fluid (CSF), which gained access to the brain tissue via the choroid plexuses. This theory of ubiquitous material transport in the CNS by means of the CSF is also referred to as the concept of the “way of the spinal fluid.”


One of the factors limiting greater understanding of the BBB was the limited resolution of the light microscope, which approximates the thickness of a cerebral endothelial cell. With the development of the electron microscope, it became possible to define the ultrastructural features of blood vessels throughout the body.

Early ultrastructural investigation questioned the selective role of endothelial cells in the barrier function and the basement membrane and the surrounding astrocytic endfeet were claimed to be an essential part of the barrier (Dempsey and Wislocki,1955; Van Bremen and Clemente,1955; Luse,1956; Gerschenfeld et al.,1959).

Van Harrefeld et al. (1965,1966,1967) showed that the lack of an extracellular space in the CNS in ultrastructural studies was due to the swelling of tissue components, especially of astrocytic endfeet before fixation. The studies were attempted to preserve water distribution in CNS by rapid freezing followed by substitution fixation at low temperature and electron micrographs of tissue frozen shortly after circulatory arrest revealed the presence of an appreciable extracellular space.

The site of mammalian BBB was not determined until the late 1960s, although some observations performed at the light microscopic level indicated that macromolecular tracer horseradish peroxidase (HRP) penetrates into the brain parenchyma (Krzyzowska-Gruca,1966) (Fig. 2). HRP is a glycoprotein with a molecular weight of 40,000, which can be demonstrated at both light and the electron microscope level by cytochemical means. HRP injected intravenously into mice passed freely out of the capillaries in cardiac and skeletal muscle (Karnovsky,1967).

Figure 2.

Extravascular localization of the HRP reaction product in the chick brain with an immature BBB. [Color figure can be viewed in the online issue, which is available at].

In 1967, Reese and Karnovsky showed for the first time at ultrastructural level that the endothelium of mouse cerebral capillaries constitutes a structural barrier to the HRP (Fig. 3). This barrier is composed of the plasma membrane and the cell body of endothelial cells and of tight junctions (TJs) between adjacent cells. The TJs completely obliterate the narrow cleft between adjacent cells forming a continuous belts or rows of zonulae occludentes. Reese and Karnovsky (1967) found that HRP was able to enter the interendothelial spaces only up to, but not beyond, the first luminal interendothelial TJs in cerebral capillaries. Unlike muscle capillaries, these TJs appeared to be continuous, and pinocytotic vesicles were uncommon and not involved in the transport. Moreover, a relatively scarce number of vesicles in the cerebral adult endothelia might be one of the morphological features of the functioning BBB, whereas the richness in vacuoles, vesicles, and luminal and abluminal expansions and invaginations displayed by endothelial cells of noncerebral microvessels could be the indication of a high endo- exocytotic activity allowing transendothelial transport.

Figure 3.

The first page of the seminal paper published in 1967 by Reese and Karnovsky.

Similar results were obtained with the intravenous injection of smaller protein tracers such as microperoxidase, with a molecular weight of 1,900 and diameter of 2 nm (Feder,1971), and the even smaller ion lanthanum (Brightman and Reese,1969), the smallest available electron-dense tracer, with an ionic radius of 0.115 nm. The TJs enable the endothelium to exclude, actively, specific solutes from the interstitial fluid and to facilitate the transfer of other solutes from plasma to CNS fluid.

Almost simultaneously, experiments performed by Brightman (1965,1968) demonstrated that either ferritin or HRP injected intraventricularly passed between ependymal cells into the brain interstitial fluid and through the astrocyte endfeet gap junctions into the area of the basement membrane, where the tracers were then stopped by the endothelial cells. The results of these experiments constitute a valuable supplement to the above-mentioned observations of Reese and Karnowsky (1967), showing that the anatomical site of the BBB was neither the astrocytic endfeet nor the basement membrane, but rather the endothelium itself.

These results were supplemented with observations that in the choroid plexus, HRP was restrained from passage from the blood to the CSF by TJs connecting the apical regions of the epithelial cells, whereas the tracer passed freely through the ependyma, into the interstitial space, and from there passed between the gap junctions joining the almost complete sheath of astrocytic processes surrounding cerebral vessels, through the basal lamina, and up to, but not beyond, the abluminal TJs (Brightman and Reese,1969).

By freeze fracture, it could be shown (Nagy et al.,1984; Shivers et al.,1984) that the TJs between endothelial cells of CNS capillaries and venules are arranged in 6–8 parallel strands with complex net-like anastomoses all along the upper circumference of the endothelial cell. The complexity of the TJs is comparable to that observed in tight epithelia and restricts the passage of low-molecular-weight substances down to a diameter of 10–15 Å. This is also reflected by a very high transendothelial electrical resistance comparable to the values obtained in the skin or the urothelium (Crone and Olsen,1982; Butt et al.,1990).


Anatomical examination of the brain microvasculature shows that the endfeet of astrocytic glia are closely apposed to the outer surface of the endothelium. This close anatomical apposition led to the suggestion that inductive influences from astrocytic glia could be responsible for the development of the specialized BBB phenotype of the brain endothelium (Davson and Olendorf,1967).

Early evidences in support of this hypothesis came from grafting experiments in which brain vessels growing into grafts of peripheral tissue became less tight to intravascular tracers, while the relatively leaky vessels of peripheral tissues became tighter on growing into grafts of brain tissue (Bauer and Bauer,2000). Other evidences derived from in vitro studies, such as those showing the formation of extensive TJs in endothelial cells cultured with astrocytes, depend on a competent, endothelium-derived extracellular matrix (Arthur et al.,1987; Shivers et al.,1988).


What is known of BBB development in vertebrates is derived from a large body of data on a spectrum of organisms ranging from hagfish to sheep. Data derived from human fetal tissue are rare. Comparisons among vertebrate classes indicate that the structural and functional aspects of the BBB are remarkably conserved (Cserr and Bundgaard,1984).

Wislocki (1891) was the first to inject a dye, Trypan blue, in a mouse fetus. The dye stained almost all of the tissues of the body within striking exception of the brain. Behnsen (1905) injected Trypan blue into neonatal and adult mice and showed dye accumulation in the brain. While the dye uptake was clearly visible in neonatal mouse brain up to about 4 weeks of age, less staining was observed in 5- to 8-week-old and in adult mouse brain. Behnsen (1905) used a 1% Trypan blue solution, 0.5 ml per 20 g was injected subcutaneously in the back region three times with a 1-day interval between injections. Thus, the neonatal mice weighing a few grams received about 0.6 mg Trypan blue. Those mice that survived were killed 5 days after the last injection. Behnsen suggested that the BBB was not fully developed in immature mice. Careful examination of Behnsen's findings shows a pattern of staining in neonatal mice similar to that obtained by Broadwell and Brightman (1976) after HRP injection in adult mice and by Sparrow (1981), who examined the distribution of albumin within adult rat CNS by immunohistochemistry.

Stern and Rapoport (1927) injected rabbits, cats, rats, and mice with Trypan blue on the day of birth or during the first few days of life and the animals were examined 1–2 hr after injection and the dye did not penetrate into the brain.

Penta (1932) used an entirely different experimental design. By means of a daily injection of Trypan blue over the first postnatal 10–20 days of rats and guinea pigs, he observed some staining of brain tissue and concluded that the barrier to Trypan blue was immature. This confusion probably arose because in the same year Stern and Peyrot (1927) published a report showing that sodium ferrocyanide did penetrate into the developing brain but not in the adult.

The experiments claiming to demonstrate barrier immaturity involved enormous concentrations of dye that were toxic (one-third of Behnsen's experimental animals died) and probably exceeded the dye-binding capacity of the fetal plasma proteins, thus allowing free dye to penetrate into the brain. Stern et al. (1929) appeared to have been well aware of this problem and stressed the importance of taking account of large differences in the concentrations of dyes used by different authors.


As regards the exact time of maturation of the BBB to macromolecules, the data from literature are controversial. Accordingly to Mollgard and Saunders (1986), the brain barrier systems to proteins are tight in many species from the earliest stages of development. The biological significance of the early development of a BBB to protein might be that it allows these specific cell populations to use some glycoprotein for particular developmental mechanisms.

Wakai and Hirokawa (1978,1981) as well as Roncali et al. (1986) reported that the cerebral endothelium becomes completely impermeable to circulating macromolecules just before birth. By analyzing the permeability of CNS capillaries for HRP during embryonic chick development, it was shown that capillaries gradually became impermeable starting at embryonic days 13–14 (Wakai and Hirokawa.,1978; Roncali et al.,1986). Similar data have been presented for mouse embryos (Risau et al.,1986). Following injection of 1–20 mg HRP into the allantoic vein of chick embryos between day 7 and day 21 of incubation, Wakai and Hirokawa (1981) observed reaction products of HRP in the interepithelial clefts at both luminal (ventricular) and abluminal sides. At day 9, penetration was blocked at most apical junctional complexes of the choroidal epithelia. At day 10 and every subsequent stage, HRP molecules were completely impeded at the apical TJs.

In contrast, Lossinsky et al. (1986) as well as Vorbrodt et al. (1986a) found that the maturation of the BBB occurs in mouse between postnatal days 12 and 24. This process coincides with the appearance of cytochemically detectable alkaline phosphatase activity in luminal plasma membrane of the endothelial cell. Moreover, by evaluating the junctional clefts, Stewart and Hayakawa (1987) clearly showed that the cleft index, or the proportion of the junction that has unfused outer leaflets, is higher in fetuses than in older animals and declines significantly with age.

Together with the structural maturation, a differentiation-dependent expression of certain enzymes was also observed (Joo et al.,1967; Risau et al.,1986; Vorbrodt et al.,1986b). From these results, it could be concluded that the expression of alkaline phosphatase, butyrylcholinesterase, γ-glutamyl transpeptidase, and transferrin receptor seems to be tightly coupled with the BBB development.

Localization of anionic sites and glucoconjugates recognized by various lectins in brain endothelial during development was studied by Vorbrodt et al. (1986a). It was shown that the functional maturation of the BBB, occurring in the mouse after birth between days 12 and 24 of life, was accompanied by a disappearance of vesicular transport from the capillaries and by the formation of a uniform, thin, negatively charged layer on the surface of endothelial cells.

Engelhardt and Risau (1995) proposed two phases of endothelial-neuroectodermal interactions leading to BBB differentiation. Their model is based on the observation that early during brain angiogenesis, induction of specific genes can be observed in brain endothelial cells. In a subsequent phase, secondary interaction of “committed” brain endothelial cells in the developing neuroectodermal elements induce further endothelial differentiation, which leads to the fully functional BBB.