In some tissues, the injection of antibodies to the β1-integrins leads to a reduction in interstitial fluid pressure, indicating an active role for the extracellular matrix in tissue pressure regulation. If perturbations of the matrix occur in the periventricular area of the brain, a comparable lowering of interstitial pressures may induce transparenchymal pressure gradients favoring ventricular expansion. To examine this concept, we measured periventricular (parenchymal) and ventricular pressures with a servo-null micropipette system (2-μm tip) in adult Wistar rats before and after anti-integrin antibodies or IgG/IgM isotype controls were injected into a lateral ventricle. In a second group, the animals were kept for 2 wk after similar injections and after euthanization, the brains were removed and assessed for hydrocephalus. In experiments in which antibodies to β1-integrins (n = 10) but not isotype control IgG/IgM (n = 7) were injected, we observed a decline in periventricular pressures relative to the preinjection values. Under similar circumstances, ventricular pressures were elevated (n = 10) and were significantly greater than those in the periventricular interstitium. We estimated ventricular to periventricular pressure gradients of up to 4.3 cmH2O. In the chronic preparations, we observed enlarged ventricles in many of the animals that received injections of anti-integrin antibodies (21 of 29 animals; 72%) but not in any animal receiving the isotype controls. We conclude that modulation/disruption of β1-integrin-matrix interactions in the brain generates pressure gradients favoring ventricular expansion, suggesting a novel mechanism for hydrocephalus development.
- communicating hydrocephalus
- intracranial pressure
- interstitial fluid pressure
- parenchymal pressure
- ventricular pressure
- cerebrospinal fluid pressure
- transmantle pressure gradients
- intramantle pressure gradients
hydrocephalus is a chronic brain disorder characterized by expansion of the ventricles and, in some cases, significant neurological damage. The conventional view of this disorder is founded on the assumption that ventriculomegaly is essentially a “plumbing” problem with the major defect associated with an impediment to cerebrospinal fluid (CSF) absorption. Although this concept may have merit under some circumstances, several conceptual problems are evident. In the communicating form of the disorder, it is not clear how impaired clearance through CSF absorption sites could lead to a dilation of the ventricles, since the ventricular and subarachnoid compartments are connected and pressure would increase theoretically in both compartments equally. In addition, investigators have largely failed to measure suitable transmantle pressure gradients or have measured gradients that were very small in magnitude (12, 31, 33). In the latter regard, Levine (12) has argued that very small transmantle pressure gradients (1 mmHg or less) may be able to expand the ventricles. However, even if one accepts that small gradients in pressure contribute to ventriculomegaly, the mechanisms that cause the gradient in the first place have yet to be explained.
One group has suggested that increases in the impedance of pulsation in the subarachnoid space elevate the pulsatile blood flow to the choroid plexus with reflection of pressure to the ventricular CSF (5). Although some of the mathematical concepts inherent in this model have been challenged (37), supporters of this idea believe that the ventricular pulse pressure amplitude exceeds the amplitude of the pulsation in the subarachnoid compartment, causing a transmantle pulse pressure gradient and ventricular dilation. Along these lines, pulse pressure amplitudes within ventricular CSF have been observed to be higher than those measured in the brain parenchyma in hydrocephalus patients (6).
In any event, the failure to measure suitable pressure gradients or to identify mechanisms that can generate suitable pressure differences provides a significant impediment to progress in the hydrocephalus field. However, two lines of evidence from the literature provide a potential key to understanding this problem. First, a mathematical model developed by Peña et al. (20) suggests that a drop in the periventricular pressure relative to that in the ventricles could theoretically cause the ventricles to expand. Second, experiments by Wiig et al. (39) provide a possible mechanistic explanation to explain how a drop in interstitial fluid pressure might occur. These studies illustrated that dissociation of β1-integrin-matrix interactions in skin (following injections of anti-β1-integrin antibodies) resulted in a lowering of interstitial fluid pressure. These results indicate that in some tissues at least, cell-matrix interactions are in a dynamic state with the capacity to modulate interstitial fluid pressure. Our objective in the present study was 1) to test whether the intraventricular injection of anti-β1-integrin antibodies could lower periventricular interstitial fluid pressure relative to that measured in the ventricular system and 2) whether similar injections could induce ventricular enlargement.
MATERIALS AND METHODS
All experiments were approved by the animal ethics committee at Sunnybrook Health Science Centre. In addition, all studies conformed to the guidelines set by the Canadian Council on Animal Care and the Animals for Research Act of Ontario.
Surgical Procedures for Pressure Measurements
A total of 31 adult Wistar rats (208–274 g) was used for this investigation (Charles River Canada). Rats were anesthetized initially by placement in a custom-built rodent anesthesia chamber using isofluorane (5%) in oxygen. For the experimental procedure, they were maintained with 2–2.5% isofluorane in oxygen delivered by a nose cone. The animals were placed on a heated water pad, and the heart rate and oxygenation status were monitored by a pulse oximeter placed on the hind foot (Nonin 8600V; Benson Medical, Markham, ON, Canada). The animal was then fixed into position in a small animal stereotaxic device (KOPF model 900; David Kopf Instruments, Tunjunga, CA). All stereotaxic coordinates were taken from a rat brain atlas (19).
The skin over the cranium was removed and the junction of the sagittal and coronal sutures identified using a stereomicroscope (Carl Zeiss OPMI 1-FC). A small high-speed microdrill with a rounded tip (Fine Science Tools, Vancouver, BC, Canada) was used to grind away the bone to expose the dural membrane at two locations: 0.92 mm posterior/1.4 mm lateral right to the bregma and 1.7 mm anterior/1.6 mm lateral right to the bregma. The former hole was used for the injection of anti-β1-integrin antibodies into the lateral ventricle, and the latter hole was used for the insertion of the micropipette into the parenchymal interstitium for pressure measurement. For ventricular pressure measurements, two holes were made at 0.92 mm posterior and 1.4 mm left and right lateral to the bregma for the antibody injection and ventricular pressure tip insertion, respectively.
Measurement of Pressures Using the Servo-null System
Pressures were measured using a servo-nulling pressure measuring system (Vista Electronics, Ramona, CA). Borosilicate glass capillaries (World Precision Instruments, Sarasota, FL) 1-mm outer diameter and 0.75-mm inner diameter were pulled with a two-step pipette puller (Narshige PP-830 puller; East Meadow, NY) to yield a tapered tip that had a 2-μm diameter. A pipette grinder (Narshige EG-400 micropipette grinder) was used to grind the pipette tips to a 40° angle to facilitate tissue penetration. Each tip was examined under the microscope to ensure uniform tip diameters. The micropipettes were then filled with a 1 M NaCl solution using a nonmetallic syringe needle (Microfil MF34G; World Precision Instruments).
For the experiment, the micropipette was connected to a pipette holder, which was mounted on a micromanipulator (MM-33B precision micromanipulator; Fine Science Tools, North Vancouver, BC, Canada). The servo-null system was calibrated to zero before the micropipette puncture by immersing the tip in a pool of Ringer solution at the site of the hole in the cranium. For parenchymal pressures, the micropipette was then advanced 3.0 mm beyond the dura into the corpus callosum. For intraventricular pressures, the micropipette tip was inserted at a depth of 3.4 mm with respect to the dura. Figure 1 illustrates the location of the micropipettes relative to the lateral ventricles. The pipette tip was ∼500–600 μm from the anterior horn of the lateral ventricle.
After a stable reading was achieved, the antibodies were injected slowly into one of the lateral ventricles (∼5 μl/s). The parenchymal interstitial pressure was monitored continuously on a computer-based data acquisition system (Daq Software; A-Tech Instruments, Toronto, ON, Canada). All data were collected at a frequency of 10 Hz. Pressure recordings had to meet the following criteria: baseline stability for at least 10 min and an unchanged zero of the system upon withdrawal of the pipette from the tissue and reimmersion in the pool of Ringer lactate.
Antibody Injection for Pressure Measurements
Antibodies against β1- and α2β1-integrins or nonimmune isotype controls (IgG and IgM) were injected into a lateral ventricle (outlined in Table 1). For injections, a 250-μl Hamilton syringe (Fisher Scientific, Toronto, ON, Canada) with a 27-gauge needle was used. The syringe needle was loaded with 50 μl of the antibody solution, and the needle tip was lowered into one of the lateral ventricles 3.4 mm deep from the dura. The mass of protein injected was as follows: anti-β1-integrin IgG (10 μg), anti-β1-integrin IgM (50 μg), anti-α2β1-integrin IgG (25–50 μg), rabbit control IgG (10 μg), purified hamster IgM (25 μg), and rat control IgG (20 μg).
Rats (56 animals, 206–285 g) were anesthetized and fixed in position as described earlier. Lube (Refresh Lacri-Lube; Allergan, Markham, ON, Canada) was applied to eyes to prevent dryness, and alcohol and iodine was applied to their shaved heads. Also, 0.2 ml of Duplocillin (intramuscular), 0.1 ml of Temgesic (subcutaneous), and 4–5 ml of saline (subcutaneously at the paralumbar fossa) were injected. The skin over the cranium was reflected, and the junction of the sagittal and coronal sutures was identified using the stereomicroscope. A small high-speed microdrill, described earlier, was used to grind away the bone to expose the dural membrane at 0.92 mm posterior and 1.4 mm lateral right to the bregma. The hole was used for the injection of the antibodies or isotype controls into a lateral ventricle (Table 2).
A 250-μl Hamilton syringe with a 30-gauge needle was loaded with the injectate and the needle tip lowered into the lateral ventricle 3.4 mm deep from the dura. The injections (25, 50 or 100 μl) were performed at a rate of ∼5 μl/s, and the needle was retracted 1 min after the injection had finished. The hole was covered with bone wax. Bupivacaine hydrochloride (Sensorcaine; Astra Zeneca, Mississauga, ON, Canada) was applied in the area of the incision for pain control. The deeper skin layer over the cranium was sutured with polysorb 4-0 sutures and the more superficial layer closed with 4-0 silk sutures.
The rats were fed lab rat chow (LabDiet 5001) until euthanization with Euthanyl at 2 wk postsurgery. At this point, the brains were removed and fixed in 10% formalin. The weight of the rats immediately after euthanization ranged between 225 and 303 g. A coronal section of the brain was made 6 mm from the frontal tip of the cortex.
Assessment of Function-Blocking Status of the Anti-β1-Integrin IgG Antibody
To determine whether the anti-β1-integrin IgG antibody was a function-blocking antibody (antibody 1 in Table 1), we assessed its impact on the integrin-mediated adhesion of U937 cells to vascular cell adhesion molecule 1 (VCAM-1). U937 cells are a monocytic cell line that expresses constitutive levels of α4β1-integrins in a low-affinity state. To induce a high-affinity conformation in this system, we treated U937 cells with manganese (Mn2+). Cells were pretreated with the anti-integrin antibody (10 μg) or IgG isotype control (10 μg) before adhesion. With the use of a parallel-plate flow chamber system, cells were then introduced onto VCAM-1-coated tissue culture plates and allowed to adhere for a period of 2 min. An incremental wall shear stress (0, 2, 4, 10, and 20 dyn/cm2) was applied, and cell adhesion was quantified by video microscopy.
Penetration of Immunoglobins into Parenchymal Tissues
Animals received 100-μl intraventricular injections of the same mass (20 μg) of antibodies directed against β1-integrins (rabbit IgG, n = 3), control isotype rabbit IgG (n = 3), or hamster IgG (n = 1) using methods described earlier. Two hours after injection, the animals were perfused transcardially with 80–100 ml of saline, the brains were removed, and an ∼1.5-mm3 block from the cortical areas dorsal to the injection site in the ventricles was dissected. The location of the tissue harvested (superficial to the location of the pipette tip) is illustrated schematically in Fig. 1B, inset (circled). The tissues were placed on dry ice before their storage in an 80°C freezer. Brain samples were sonicated in a lysis buffer (10× wt/vol: 50 mM Tris·HCl, pH 7.4, 10 mM EDTA, 4 M urea, 3% Triton X-100, and 1× protease inhibitor; Calbiochem 539131) and 200 μM phosphatase inhibitor sodium vanadate.
Immunoblots were performed according to standard procedures. Briefly, the samples (40 μg of total protein) were boiled, centrifuged for 5 min, separated by SDS-PAGE (10%), and transferred to nitrocellulose membranes (Bio-Rad). Membranes were blocked for 1 h in Tris-buffered saline (TBS) containing 0.1% Tween and 5% skim milk powder. After a 1-h incubation with anti-rabbit IgG secondary antibodies conjugated to horseradish peroxidase (1:5,000; Jackson ImmunoResearch), the membranes were washed with TBS-Tween buffer and the bands revealed using Immobilon Chemiluminescence (Millipore WBKL50500). To ensure that equal amounts of protein were loaded in each lane, the blots were stripped and reprobed with an antibody against glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 1:2,000; Biodesign, Saco, ME) for 1 h, followed by a horseradish peroxidase-conjugated antibody and chemiluminescence detection of the GAPDH bands.
To normalize the pressure values, the data were averaged over 120-s intervals. Subsequently, a baseline value was derived for each trace by averaging a minimum of 10 min of a stable pressure recording immediately before the injection time. Each postinjection data point was expressed as a ratio by dividing each point by the average baseline value.
Cell adhesion data.
These data were normalized by expressing each point as a percentage with respect to total number of adherent cells at zero shear stress. All averaged data are means ± SE. The data were assessed using Student's t-test, ANOVA, or repeated-measures regression analysis adjusted for within correlations over time. We interpreted P < 0.05 as significant.
In all experiments, the intraventricular injection of IgG (n = 5) or IgM antibodies to the β1-integrins (n = 5) resulted in a decline of parenchymal pressures below baseline (examples shown in Fig. 2). In the majority of experiments (8 of 10 cases), the reduction in interstitial fluid pressure was preceded by a bolus injection effect (examples shown in Fig. 2, A and B), but in two cases, little bolus effect was observed (example shown in Fig. 2C). In most experiments (7 of 10 cases), the reduction in pressure persisted for the duration of the study (Fig. 2, A and C). In three cases, however, the pressure-lowering effect was of shorter duration with pressures returning to preinjection levels at some point (example shown in Fig. 2B). In addition to the aforementioned studies, we performed experiments using antibodies to α2β1-integrin (n = 3). In each case, pressures declined below preinjection levels (data not shown).
The intraventricular injection of the appropriate isotype controls induced a very different response. In all cases (n = 7), parenchymal pressures increased above preinjection values and remained elevated for the duration of the experiment (example shown in Fig. 3A).
After injections of the anti-integrin antibodies (IgG based, n = 5; IgM based, n = 5), pressures in the ventricles increased and remained elevated over the course of the experiment in the majority of cases (example shown in Fig. 3B). In a few cases, pressures eventually returned to baseline, but not below. A similar pattern was observed after injection of the isotype rabbit IgG (n = 2) or hamster IgM (n = 2) controls (example shown in Fig. 3C).
In every experiment in which IgG or IgM antibodies to the β1-integrins were injected, interstitial fluid pressures declined below baseline. Therefore, we pooled the results for the ventricular (n = 10) and parenchymal tissue fluid pressures (n = 10). We also averaged the data from all the isotype control injections for parenchymal (n = 7) and ventricular pressures (n = 4). The averaged data are shown in Fig. 4.
Parenchymal pressures following injection of anti-β1-integrin antibodies or their isotype controls.
Approximately 20 min after injection of the anti-β1-integrin antibodies, the parenchymal pressures decreased below preinjection (baseline) levels and remained below baseline for the duration of the experiment (Fig. 4A). By the end of the monitoring period, interstitial fluid pressure had declined to ∼70% of the baseline value (Fig. 4A, filled circles). In contrast, the parenchymal pressures associated with the injection of the IgG/IgM isotype controls were elevated above baseline and remained so for the duration of the study (Fig. 4A, open circles). Pressures were significantly different between the two groups (P < 0.0001; repeated-measures regression analysis).
Comparison of ventricular and parenchymal pressures after the injection of antibodies to β1-integrins.
The pressures measured in the lateral ventricles induced with injections of the specific anti-β1-integrin antibodies (Fig. 4B, open circles) remained above baseline during the experiments and were significantly greater than pressures monitored in the periventricular interstitium (P = 0.038; repeated-measures regression analysis). This indicated a pressure gradient between the ventricles and surrounding tissues.
Comparison of ventricular pressures after injections of anti-β1-integrin antibodies or their isotype controls.
Ventricular pressures associated with the isotype control injections were higher than those induced by the specific antibodies to the β1-integrins (Fig. 4C). These differences were significant (P = 0.0002; repeated-measures regression analysis).
Comparison of pressures in the ventricles and parenchyma after injection of the IgG/IgM isotype controls.
Pressures in the ventricles and parenchyma following the injection of the isotype controls were not significantly different (Fig. 4D; P = 0.067; repeated-measures regression analysis).
The data from Fig. 4 can be used to estimate potential pressure gradients between ventricular CSF and periventricular interstitial fluid under the conditions of our experiments. The average baseline (preinjection) ventricular and parenchymal pressures in these studies were 6.6 ± 0.5 and 5.6 ± 0.5 cmH2O, respectively (not significantly different; Student's t-test). Averaging all baseline pressures, we obtained 6.0 ± 0.3 cmH2O, which is only slightly higher than the normal servo-null-based values of CSF pressure noted in the studies of Wiig and Reed (3.4 mmHg or 4.7 cmH2O; Ref. 38). Taking the data from Fig. 4B, at 24, 48, and 72 min after injection of the specific anti-β1-integrin antibodies, the average ventricular pressures had increased 24, 27, and 42% above baseline, respectively. In contrast, the average parenchymal pressures at the same times had declined to 88, 80, and 70% of baseline levels, respectively. Therefore, we can estimate average ventricular pressures of 7.4, 7.6, and 8.5 cmH2O and parenchymal pressures of 5.3, 4.8, and 4.2 cmH2O at the three respective times. This would give theoretical ventricle to parenchymal pressure gradients of 2.1, 2.8, and 4.3 cmH2O at the various times after injection. Using a similar approach, we estimated the average parenchymal pressure differences caused by injection of specific anti-integrin antibodies relative to the isotype controls (data from Fig. 4A). At 24, 48 and 72 min, the pressures induced with the specific anti-integrin antibodies were 4.3, 3.6, and 4.8 cmH2O lower, respectively, than those induced with the appropriate IgG/IgM controls.
Over the course of the 2-wk period following the injection of specific antibodies or isotype controls, we did not notice any obvious behavioral or weight changes that might be indicative of hydrocephalus development. However, on posteuthanization analysis, we observed hydrocephalus in many of the animals that received antibodies to the β1-integrins. Table 2 summarizes the results of these experiments. If we consider the results from injections of the antibodies to β1-integrins (IgG and IgM) and to α2β1-integrin, 21 of 29 animals developed some degree of ventriculomegaly. The magnitude of hydrocephalus was considerably greater in the animals that received IgG antibodies to β1- or α2β1-integrins compared with those injected with the IgM antibodies (examples provided in Fig. 5, C–F). Indeed, the IgM anti β1-integrin antibodies induced only minimal expansion. In contrast, none of the animals injected with isotype IgG/IgM controls exhibited hydrocephalus (examples in Fig. 5, A and B).
Assessing Characteristics of the IgG Anti-β1-Integrin
The functional status of the IgG anti-β1-integrin antibody (antibody 1 in Table 1) was not apparent from the manufacturer's data sheet. Because this antibody induced marked hydrocephalus in many of the rats and lowered parenchymal interstitial fluid pressure, we suspected that it was a functional blocking antibody with respect to integrin function. To investigate this issue using an independent measure, we employed a cell adhesion assay. Adherent cells were counted in a flow system that permitted the application of variable shear stress forces on the cells (Fig. 6). We observed that the percentage of adherent U937 cells treated with the anti-β1-integrin antibody was significantly lower than that noted with the isotype-treated control (2-factor ANOVA, group × shear stress; P < 0.0001). This result provided additional evidence that the IgG anti-β1-integrin antibody was function blocking with respect to the β1-integrins.
Penetration of Immunoglobulins into the Periventricular Parenchyma Following Intraventricular Administration
Our servo-null pressure experiments implied that the antibodies to the β1-integrins penetrated at least 500–600 μm from the anterior horn of the lateral ventricle (position of the micropipette tip), since we observed a pressure effect after injecting the antibodies. In addition, in the chronic studies some of the ventricles seemed to expand beyond the pressure measurement point, suggesting that the antibodies may have penetrated even further into the parenchyma. This supposition was supported by preliminary experiments.
Western blotting was performed on brain samples from animals injected with rabbit anti-β1-integrin, isotype control rabbit IgG, and anti-hamster IgG (example shown in Fig. 7, A–C). A 50-kDa band was detected in animals injected with anti-β1-integrin IgG (Fig. 7A) and the isotype control IgG (Fig. 7B) but not following the administration of anti-hamster IgG antibodies (Fig. 7C). These data indicated that the immunoglobulins injected into the lateral ventricles penetrated a considerable distance into the brain parenchyma. In all experiments, the β1-integrin bands (example in Fig. 7A) were more intense than those exhibited by the isotype rabbit IgG controls (Fig. 7B). Whether the antibodies to the β1-integrins were retained in the brain more than the isotype controls (perhaps due to binding integrins in the tissues) needs further evaluation.
The brain extracellular matrix has been of interest to investigators in the hydrocephalus discipline for some time. For example, Wyss-Coray et al. (40) demonstrated that overexpression of transforming growth factor-β1 increased the production of brain matrix elements and that these changes were associated with a hydrocephalus phenotype. What has been missing in this field, however, is a mechanism that links matrix changes with the biomechanical forces that favor ventricular expansion. In this regard, studies in non-central nervous system (CNS) tissues have provided a clue as to how this might occur, and in doing so have transformed our view of the interstitium (39). Rather than simply being a passive participant, the extracellular matrix assumes an active role in regulating interstitial fluid pressure. Our data would suggest that matrix components provide a similar dynamic function within the brain parenchyma.
Pressure Differences and Hydrocephalus
It is commonly assumed that a CSF absorption deficit or some impediment to the CSF flow contributes to the development of communicating hydrocephalus. Levine (12) has argued that an elevated resistance to CSF flow causes a small increase in the gradient of pressure (<1 mmHg) between the ventricles and subarachnoid space and that this gradient is able to overcome the resistance and balance CSF formation and absorption. He has postulated that ventricular expansion occurs at the expense of parenchymal extracellular fluid, which is driven into the capillaries. This is an interesting concept that incorporates the classic view of a CSF flow obstruction or absorption deficit. In addition, however, changes in the parenchymal interstitium may provide another mechanism that contributes to hydrocephalus development.
On the basis of a mathematical model, Peña et al. (20) have postulated that a relative reduction in brain interstitial fluid pressure in conjunction with low tissue elasticity could produce ventricular expansion. Of fundamental importance is the concept of a pressure gradient between the ventricles and periventricular tissues (intramantle or transparenchymal gradients). Ignoring the issue of tissue elasticity, our data would suggest a mechanistic approach to the generation of such a pressure gradient. Through whatever initiating factor, the modulation of β1-integrin-matrix interactions appears to alter brain tissue pressure in such a way as to encourage ventricular expansion.
Under normal conditions, ventricular and parenchymal pressures are closely coupled, suggesting fluid continuity between the two compartments (38). In our studies, ventricular and periventricular (parenchymal) pressures were virtually identical following the injection of IgG/IgM isotype controls into the ventricles (Fig. 4D). Nonetheless, we observed that pressures could become uncoupled under certain circumstances. The most obvious example of this relates to the pressure differences observed between the ventricles and periventricular parenchyma after injection of the antibodies to the β1-integrins (Fig. 4B). It seems likely that these differences facilitate ventricular expansion. Unexpectedly, however, we also noted some other pressure differences.
It would seem that there was a tendency for the ventricular pressures to be greater after injection of isotype controls compared with injection of the specific anti-integrin antibodies (Fig. 4C). If the integrin-matrix interactions in the periventricular area are disrupted, it seems possible that the matrix will expand and water will enter this tissue from the ventricular system. This may have the effect of reducing the ventricular pressure transiently compared with injections of the isotype controls, which are expected to have little impact on the brain interstitium.
The original concept of matrix tissue pressure regulation arose from studies on inflammation (23, 39). The antibodies we used were foreign proteins and, as such, might induce some inflammation in the brain. However, we injected three different antibodies to the integrins, and only those specifically directed against β1- or α2β1-integrins had the pressure/hydrocephalus effect. If a nonspecific inflammation were responsible for the pressure effects observed, we should have seen the effects with all antibody injections, including the isotype controls, but this was not the case.
Role of Integrins in Interstitial Fluid Pressure Regulation and Hydrocephalus Development
Integrins are cell surface glycoproteins that mediate cell-matrix interactions by providing a physical transmembrane link between the extracellular matrix and the cell cytoskeleton (10). In the rat paw, a polyclonal anti-β1-integrin IgG that inhibited fibroblast-mediated collagen adhesion in vitro lowered the interstitial fluid pressure and also induced edema formation, similar to that observed in inflammatory reactions (24) or following burn injury (13, 15). Further studies in the literature have implicated the α2β1-integrins (which bind collagen and laminin) in interstitial pressure regulation (25). The observation that the F-actin-disrupting agent cytochalasin D also lowered interstitial fluid pressure supports a role for extracellular and intracellular cytoskeletal linkages in pressure regulation (2). In contrast, the pressure effect was not mediated by fibronectin as polyclonal anti-rat fibronectin IgG, and the fibronectin receptor binding protein Arg-Gly-Asp (RGD) had no impact on interstitial pressures.
Rubin et al. (27) have proposed that fibroblasts in the skin regulate interstitial fluid pressure by exerting a tensile force on the collagen matrix, which restrains the interstitial gel from swelling. The β1-integrins provide the link between the extracellular matrix and the cytoskeletal contractile apparatus. Integrin function can be modulated by cytokines and other factors that regulate the balance between grip and release, leading to compaction or tissue swelling, which in turn affects interstitial fluid pressure (27). It should be noted, however, that in relative terms there are lesser amounts of fibrous proteins such as collagens, fibronectins, and vitronectins in CNS tissues (7, 29, 32). Prominent brain matrix elements include lectican, hyaluronic acid, and the tenascin family (28).
Cell Types and Integrin-Matrix Interactions in the Brain That May Be Relevant to the Pressure Effect
β1-Integrins are believed to have many important roles in CNS function (16) and are expressed on choroidal and ependymal cells and throughout the neuropil on glial cells and vascular structures (4, 9, 18). It seems likely that glial cells contribute to the pressure effects noted in this report, because they are the most numerous cells in the CNS and are believed (among other functions) to synthesize extracellular matrix factors that are essential for neurite outgrowth (9). A variety of β1-integrin receptor types are expressed on astrocytes (35, 36), and those in the focal and point contact areas are believed to facilitate cell attachment and spreading along the extracellular matrix.
Cells within the brain microvasculature also express β1-integrin receptors. Astrocyte fibers surrounding select microvessels in the adult primate express β1-integrins, as do the endothelial cells themselves (4). Endothelial cells and astrocytes maintain the basal lamina and support the barrier properties associated with the blood-brain barrier. At this point, we do not know whether injections of the anti-β1-integrin antibodies affected barrier function. However, if this was the case, one might expect that the integrity of the barrier would be lost and the interstitial pressure would increase as fluid and solutes moved from the blood into the parenchymal interstitium. Of course, it is possible that this effect blunted the reduction in pressure we observed.
The ependymal cells lining the ventricles would have been exposed to the anti-β1-integrin antibodies. Although these cells appear to provide a metabolic barrier at the CSF-brain interface (3), it is unlikely they restrict significantly the movement of immunoglobins or water between the ventricles and periventricular tissues. Nonetheless, the ependymal cell cilia play a role in CSF homeostasis and deficiency of regulatory factors necessary for ciliogenesis is associated with abnormal differentiation of ependymal cells and hydrocephalus in mice (1). Therefore, we cannot discount some effect of the antibodies on the ependymal cell layer.
When β1-integrin expression was ablated specifically in precursors of neuronal and glial cells using a cre-lox genetic strategy, the mice exhibited a strikingly convoluted cortex and a reduced brain size (8). This suggested that the β1-integrins had an important role in determining the structural integrity of the brain. In addition, disorganization of the cerebral cortex was observed in humans and mice with mutations in the laminin α2-chain (17). Children with mutations of the LAMA2 gene have reduced or absent laminin α2-chain. MRIs of these individuals show structural cortex abnormalities and ventricular dilation (22, 34). Since laminin is an important ligand for the β1-integrins, it is possible that the disruption of the β1-integrin-laminin interactions may be a contributor to the pressure patterns and hydrocephalus development noted in this report.
In this regard, it is of interest to note that the dystroglycans share laminin as a primary matrix ligand with a number of β1-integrin receptors (30). The dystroglycans are heterodimeric transmembrane receptors that link the intracellular cytoskeleton of select cells with the matrix elements. Enlargement of the lateral ventricles was observed in 45% of (epiblast-specific) dystroglycan null mice (30). The authors speculated that this could be due to the stenosis of the cerebral aqueduct and/or blockage of the arachnoid granulations. However, one might speculate that the epiblast-specific loss of dystroglycan might affect cell matrix interactions in a way analogous to that observed with the injection of β1-integrin antibodies.
Other Unresolved Issues and Limitations of Our Study
At present, we do not know the specific molecular mechanisms associated with the tissue pressure drop after the injection of the antibodies to the β1-integrins. This might be due to disruption of cell integrin-matrix interactions, as noted earlier, or possibly due to some as yet uncharacterized consequence of antibody-β1-integrin binding. In addition, we do not know whether we targeted the most appropriate cell-matrix interactions. It is possible that the modulation of more brain-specific components may have had an even greater effect on hydrocephalus development.
There are also issues related to the antibodies we used in this study. In our experience, both the IgG and IgM anti-integrin antibodies reduced parenchymal pressures by approximately the same degree. However, the ventriculomegaly with the anti-integrin IgG antibodies was considerably greater than that observed with the IgM counterparts, which induced modest ventricular expansion at best. There are a number of possible explanations for these findings. First, IgM is larger than IgG, and it is possible that less of this antibody entered the parenchyma, and consequently, the volume of brain affected may have been smaller. In addition, relatively less penetration of the IgM antibodies may have made more available for CSF absorption via lymphatic or arachnoid CSF drainage mechanisms (11). Second, the full time course of the pressure effects of the two antibodies is unknown, because our servo-null pressure experiments were of relatively short duration. It is possible that the pressure effects of the IgM anti-β1-integrin antibodies were exerted over a shorter period compared with those of its IgG counterpart. If so, a short-lived pressure gradient may have been insufficient to permit the development of marked hydrocephalus. The degree of parenchymal penetration and the duration of the pressure effects induced with the IgG and IgM anti integrin antibodies need to be evaluated further.
Finally, it will be important to determine whether the integrin-matrix concept has relevance to recognized animal models of hydrocephalus. In a kaolin-based dog model of hydrocephalus, Penn et al. (21) were unable to measure any pressure gradients among the ventricle, brain, and subarachnoid space with pressure measurements taken from each location continuously before and after kaolin administration. They used an InSite pressure-monitoring system (a capacitive-based sensor at the end of a pacemaker lead) for the measurements. Whether these differences relate to the technology to measure pressure or to unknown issues related to the complexity of hydrocephalus development remain to be elucidated.
Current treatments for hydrocephalus work by diverting cerebrospinal fluid to another space by implanting shunts or through the third ventriculostomy procedure. The shunts in particular have proven to be problematic. Assuming, however, that the matrix-pressure concept espoused in this report has relevance to human hydrocephalus, it may be possible to develop new molecular strategies to modulate integrin-matrix interactions in such a way as to reverse the drop in parenchymal interstitial fluid pressures. Indeed, in skin the pressure lowering effects associated with disruption of the β1-integrins can be undone with selected drugs such as the anti-inflammatory agent α-trinositol (d-myo-inositol 1,2,6-triphosphate) (14, 26) and an isoform of platelet-derived growth factor (PDGF-BB) (25). These data highlight the possibility that certain forms of communicating hydrocephalus may ultimately be treatable with molecular/pharmacological agents thus reducing the dependence on problematic shunts.
This study was funded by the Canadian Institutes of Health Research Grant 7925 and the BrainChild Foundation.
We are very grateful to Kelly Markham-Coultes, Lilian Weng, and Sara Moore for assistance in performing the experiments and to Dr. Myron Cybulsky and Jacob Rullo for help in conducting the cell adhesion studies. In addition, we thank Marko Katic and Dr. Alex Kiss (Department of Research Design and Biostatistics, Sunnybrook Health Sciences Centre) for help in the computational analyses of the data.
- Copyright © 2009 American Physiological Society