B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /b ra i n res

Research Report

The impact of agrin on the formation of orthogonal arrays ofparticles in cultured astrocytes from wild-type andagrin-null mice

Petra Fallier-Beckera, Jan Sperveslagea, Hartwig Wolburga,⁎, Susan Noellb

aInstitute of Pathology, University of Tübingen, Medical School, Tübingen, GermanybClinics of Neurosurgery, University of Tübingen, Medical School, Tübingen, Germany

A R T I C L E I N F O

⁎ Corresponding author. Institute of PatholoGermany. Fax: +49 7071294834.

E-mail address: hartwig.wolburg@med.unAbbreviations: AQP4, Aquaporin-4; BBB, Bl

particles

0006-8993/$ – see front matter © 2010 Elsevidoi:10.1016/j.brainres.2010.09.092

A B S T R A C T

Article history:Accepted 26 September 2010Available online 1 October 2010

Astrocytic endfeet membranes are studded with aquaporin-4 (AQP4) containing orthogonalarrays of particles (OAP) which can be visualized exclusively by the freeze-fracturing method.They are predominantly expressed where the astroglial membrane is in contact with thesuperficial andperivascularbasal lamina. This polarity seems tobeessential for the integrityofthe blood-brain barrier (BBB). The basal lamina containing many extracellular matrix (ECM)components such as collagen, laminin and heparansulfate proteoglycans like agrin is thoughtto influence this OAP-related polarity of astrocytes. Recently, we have shown that agrin, inparticular the neuronal isoform A4B8, is capable of influencing the formation of OAPs inastrocyteswhen cultured in the presence of agrin-conditionedmedia. In this paperwewantedto investigate whether coating with exogenous agrin compared to coating with other ECMcomponents would induce OAP formation in astrocytes of the agrin-null mouse. For thispurpose, we cultured astrocytes from agrin-null and wild-type mice on agrin- or ECM-coatedsurfaces. Immunofluorescent cytochemical staining of AQP4 indicated a higher AQP4expression level in cultures with agrin- or ECM-coated than in cultures with uncoatedsurfaces, whereas western blot analyses and PCR showed no differences. α-Dystroglycan isthought to be a potential receptor of agrin and was immunostained in wild-type as well as inagrin-null astrocytes. In freeze-fracture replicas, we observed an increase in OAP density inastrocytes when growing on agrin- and ECM-coatings. These results concurred with otherexperiments in which changes in volume were measured following hypotonic stress, whichsupported the positive influence of exogenous agrin onAQP4 insertion into themembrane, onOAP formation and on water transport.

© 2010 Elsevier B.V. All rights reserved.

Keywords:Blood-brain barrierExtracellular matrixCell cultureFreeze-fracturingAgrin-null mouse

gy, University of Tübingen, Medical School, Liebermeisterstraße 8, D-72076 Tübingen,

i-tuebingen.de (H. Wolburg).ood-brain barrier; ECM, Extracellular matrix; OAP, Orthogonal arrays of intramembranous

er B.V. All rights reserved.

Fig. 1 – Scheme of the agrin-coating method. Agrin producing HEK cells (red arrows) are grown to confluence (top). After lysis ofHEK cells, surface is coated with agrin (middle). Astrocytes are seeded on agrin-coated surface (bottom).

3B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

1. Introduction

Orthogonal arrays of particles (OAP) of astrocytes werediscovered in the 1970s (Dermietzel, 1973) using freeze-fracture techniques. But it was not until 30 years later that itbecame clear that they contain the water channel proteinAQP4 (Rash et al., 1998). A specific property of this waterchannel is the occurrence of different isoforms (Moe et al.,2008). The best examined isoforms are AQP4-M1 and AQP4-M23 which exhibit different water transport capacities (Silber-stein et al., 2004). Furman et al. (2003) investigatedM1 andM23transfected cells and found that the AQP4-M23 isoform wascapable of forming huge lattices in the cell membrane,whereas AQP4-M1-transfectants showed only few and smallones. Only the transfection of a mixture of M1 and M23resulted in the formation of orthogonal arrays which resem-bled the phenotype of arrays found in astrocytes.

In the healthy brain, perivascular as well as superficialastrocytic endfeet membranes are studdedwith OAPs, whereasparenchymal astrocyte membranes only reveal few OAPs(Rohlmann et al., 1992). It has been proposed that the contactof membranes with the basal lamina is responsible for the OAPformation and therefore the OAP-related polarization of theastrocyte.

Agrin as a component of the ECM is an extracellularheparansulfate proteoglycan (Tsen et al., 1995) which wasoriginally identified as being capable of clustering acetylcholinereceptors at the neuromuscular junction (McMahan, 1990;Bezakova and Ruegg, 2003). Barber and Lieth (1997) describedthe importance of agrin in the CNS for the integrity of the BBBshowing that during chick and rat development, agrin accumu-

latesonbrainmicrovesselsby the time the vasculaturebecomesimpermeable.

The agrin molecule exists in different splice variants; theendothelial isoform A0B0 is released, whereas the neuronalisoform A4B8 is inserted in the cell membrane. Noell et al.(2009) showed that in the agrin-null mouse the OAP formationat the superficial endfoot membrane of astrocytes is clearlyreduced, pointing to the role of agrin for the formation of OAPsfrom subunits.

In the present freeze-fracture study we wanted to find outwhether cultured astrocytes from the agrin-null mouse braingrowing on agrin- and ECM-precoated surfaces (Fig. 1) are ableto form OAPs. Furthermore, we studied the influence ofexogenous agrin on AQP4 expression and on changes in cellvolumes of agrin-null astrocytes during and after hypotonicchallenge.

2. Results

2.1. Freeze-fracture experiments

In freeze-fracture replicas of cultured astrocytes, we generallyfound an increase of OAP densities using culture dishes coatedwith either ECM or exogenous agrin from HEK cells comparedwith astrocytes cultured without ECM or agrin.

Wild-type and agrin-null astrocytes were cultured ondifferent coatings and subsequently fixed for freeze-fractureanalysis. In wild-type as well as in agrin-null astrocytes, theinfluence of agrin- or ECM-coating on the number and densityof OAPs was substantial. In freeze-fracture replicas of control

4 B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

astrocytes cultured without agrin or ECM coating of tissueculture dishes (Fig. 2A), we found an average density of5 OAPs/μm2 (Fig. 3). In agrin-null astrocytes (Fig. 2B) thedensity was 3 OAPs/μm2 (Fig. 3). In both the wild-type(Fig. 2C) and the agrin-null astrocytes (Fig. 2D) coating withagrin-isoform A4B8 resulted in an increase up to 15 OAPs/μm2

(Fig. 3), whereas wild-type astrocytes on agrin A0B0-coating(Fig. 2E) only showed 6 OAPs/μm2 (Fig. 3). However, agrin-nullastrocytes (Fig. 2F) revealed 19 OAPs/μm2 on agrin A0B0-coating (Fig. 3). The influence of ECM gel was even morepronounced: wild-type astrocytes reached an OAP density of17 OAPs/μm2 (Figs. 2G, 3) and agrin-null cells a density of25 OAPs/μm2 (Figs. 2H, 3).

Taken together, coating experiments using agrin A4B8,agrin A0B0 and ECM-gel revealed a 5-, 6-, or 8-fold increase ofOAPs in agrin-null astrocytes in comparison with wild-typeastrocytes, respectively (Table 1).

2.2. Immunostaining

Wild-type and agrin-null astrocytes were cultured on agrin- orECM-coated coverslips and processed for immunostainingwith an antibody against AQP4 as previously described forwild-type astrocytes (Noell et al., 2007). Here, we examined theimmunoreactivity against AQP4 in agrin-null astrocytes incomparison with wild-type astrocytes. In wild-type (Fig. 4A) aswell as in agrin-null astrocytes (Fig. 4B) cultured withoutcoating, AQP4 staining was diffusely distributed within thecytoplasm. Culturing the cells on agrin showed AQP4 stainingconcentrated in membrane areas in all cultures (Figs. 4C–F).ECM coating revealed intensive AQP4 fluorescence in mem-branes of wild-type cells (Fig. 4G) and in agrin-null cells(Fig. 4H). Thus, as reported in the freeze-fracture paragraph,also immunostaining against AQP4 revealed a strongerimmunoreactivity in agrin-null astrocytes when cultured onECM-, agrin A4B8- and agrin A0B0-coating (Table 1).

In order to address thequestionofwhether thepresence ofα-dystroglycan as an agrin-receptor is an indispensable precondi-tion for the formationofOAPs,we stainedwild-type (Fig. 5A) andagrin-null (Fig. 5B) astrocytes immunocytochemically for α-dystroglycan. Both cultures were found to be immunopositivefor α-dystroglycan. This finding explains the observation thateven agrin-null astrocytes are able to form OAPs when culturedon exogenous agrin (Table 1).

2.3. Western blotting

Western blot analysis showed that the AQP4-isoform M23-band was more abundant than the AQP4-isoform M1 bandindependent of the coating. For western blot analysis, wild-type and agrin-null astrocytes were grown on ECM- and agrin-isoform A0B0 and A4B8 coatings. Western blotting wasperformed with an AQP4 antibody and showed two distinctbands, the lower band (32 kDa) representing the AQP4-M23isoform and the upper band (34 kDa) representing the AQP4-M1 isoform. The western blot showed a stronger immuno-

Fig. 2 –Freeze-fracture replicasof culturedwild-type (A,C, E,G)andaindicated by the labels. The number of OAPs (circles) increases dep

reactivity for AQP4-M23 than for AQP4-M1 regardless of thetreatment and origin of the cells (Figs. 6A, B; Table 1).

2.4. PCR

PCR was used to prove the expression of both AQP4-isoformsM23 and M1 on RNA level independent of the presence of acoating and independent of the origin of the astrocytes (Fig. 7,Table 1). Agrin does not seem to influence the expression of theAQP4, neither on the mRNA nor on protein level.

2.5. Cell volume measurements

In swelling experiments we found differences between theuncoated astrocyte cultures and the cultureswhichwere coatedwith the neuronal agrin-isoform A4B8. Coating with theendothelial agrin-isoform A0B0 did not induce any alterationof water flow (Table 1).

Cultured wild-type and agrin-null astrocytes were grown oncoated and uncoated coverslips fitting into the flow-throughchamber.

Wild-type astrocytes grown on uncoated surfaces showeda volume increase of 8%, but in the presence of the agrin-isoform A4B8 we found a significant volume increase of 21%(Fig. 8). In contrast, in the presence of agrin A0B0, nosignificant difference to the uncoated controls was observed.Comparable results were achieved with agrin-null astrocytes.Coating with agrin A4B8 resulted in a significant cell volumeincrease of 26% compared to the uncoated controls, whichshowed only 14% of volume increase. A0B0-coating, however,revealed no significant difference to uncoated control cells(9%, Fig. 8). Taken together, agrin-null astrocytes did not showa significant difference in water transport compared to wild-type astrocytes.

3. Discussion

Agrin, a heparansulfate proteoglycan, is known to influence thedifferentiationof themotor endplate by clusteringacetylcholinereceptors (McMahan, 1990). Noell et al. (2007) were able to showcomparable effects of agrin on the clustering of AQP4-contain-ing OAPs in cultured astrocytes. In addition, Noell et al. (2009)detected a paucity of OAPs in the endfeet membranes in theagrin-null mouse, although the expression of the AQP4 proteinwas not down-regulated. This finding suggested amajor role ofagrin for the formation of OAPs fromAQP4molecules as seen inthe freeze-fracture replica.

In the present study we asked whether exogenouslyapplied agrin isoforms (endothelial A0B0 and neuronal A4B8)and extracellular matrix gel from Engelbreth–Holm–Swarmmouse sarcoma are capable of influencing OAP formation ofcultured agrin-null astrocytes. Freeze-fracture analysisshowed an effect of ECM- and agrin-coating on the occurrenceof AQP4-containing OAPs in agrin-null and wild-type astro-cytes. The number of OAPs in the coated cultures was

grin-null astrocytes (B,D, F,H) growingondifferent coatings,asending on different coatings. Scale bars: 100 nm.

5B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

Fig. 3 – Statistical analysis (box-blots) of OAP densities inwild-type and agrin-null astrocytes growing on different coatings. Thedifference between the wild-type control cells and the wild-type cells growing on agrin-isoform A4B8 and on ECM gel issignificant (p<0.05). The difference in OAP density between the agrin-null control cells and the agrin-null cells growing onagrin-isoform A4B8 and A0B0 and on ECM gel is significant (p<0.05).

Table 1 – Summary of results.

Cell cultures Wild-type astrocytes Agrin-null astrocytes

Coating ECM Agrin A0B0 Agrin A4B8 Control ECM Agrin A0B0 Agrin A4B8 Control

Freeze fracturing (number of OAPs/μm2) 17 6 15 5 25 19 15 3Immunofluorescence of AQP4 ++ + ++ + +++ + +++ +Immunofluorescence of α-dystroglycan n.d. n.d. n.d. + n.d. n.d. n.d. +Western blot M23>M1 M23>M1 M23>M1 M23>M1 M23>M1 M23>M1 M23>M1 M23>M1Cell volumemeasurements (Intensity dF/dF0) n.d. 7% 21% 8% n.d. 9% 26% 14%PCRa M23, M1 M23, M1 M23, M1 M23, M1 M23, M1 M23, M1 M23, M1 M23, M1

n.d. not done.a By trend, M23 was higher expressed than M1 in all samples.

6 B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

significantly increased compared to the uncoated controls.Immunofluorescent cytochemical staining of AQP4 revealedhigher AQP4 expression in the coated cultures than in theuncoated cultures. The intensity of immunofluorescencestaining correlated with the findings that cultured murineastrocytes in media conditioned with the neuronal agrin-isoformA4B8 and the endothelial agrin-isoformA0B0 revealeda stronger immunoreactivity within the cell membrane thanuntreated controls (Noell et al., 2007).

The AQP4 molecule exists in different isoforms. The M1isoform is 22 amino acids longer at the N-terminus than theM23 isoform (Jung et al., 1994). Furman et al. (2003) andSilberstein et al. (2004) showed that in cells transfected withAQP4-isoform M1 the membranes reveal very small or noOAPs, whereas the AQP4-M23 transfectants form huge OAPlattices. Transfection of the cells with both isoforms (M1 andM23) resulted in the formation of OAPs reminiscent of those inastrocytes in vivo.

As we could only find small or no OAPs in the astrocytemembranes of the agrin-null mouse we expected that theAQP4-isoform M1 was most abundant in western blots.Surprisingly, this was not the case. All western blots revealeda higher abundance of the M23 bands compared with the M1

Fig. 4 – Immunocytochemical staining of AQP4 in wild-type (A, C,coatings, as indicated. Scale bars: 20 μm.

bands regardless of the coatings. In freeze-fracture replicas ofastrocytes, cultured on agrin or ECM coatings, we alwaysfound an increased number of OAPs compared to those incontrol astrocytes grown on uncoated surfaces. Since wefoundmore M23 thanM1 in the western blot and by trend alsoin PCR analyses of all cultures, we propose that the coatingof ECM or agrin is crucial for the formation of OAPs rather thanthe ratio of the AQP4-isoforms. Based on these results andthe experiments of Furman et al. (2003) who used cellstransfected with M23 and M1 AQP4-isoforms and Suzukiet al. (2008) who found that the size of AQP4-M23 up toAQP4-M17 is small enough to form OAPs, we assume that thepresence of AQP4-M23 could represent the preconditions forforming OAPs.

At the healthy BBB, the astrocytic endfoot membranereveals an extremely high density of OAPs at the contact siteof the membranes and the endothelial basal lamina consist-ing of collagen, laminin and heparansulfate proteoglycanslike agrin, whereas only few OAPs are found at theparenchymal membranes that do not contact the basallamina. Furthermore, it is known that under pathologicalconditions, the number of OAPs decreases at the endfootmembranes (Neuhaus, 1990) despite an upregulation of AQP4

E, G) and agrin-null astrocytes (B, D, F, H) growing on different

7B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

Fig. 5 – Immunocytochemical staining of α-dystroglycan is positive in both wild-type (A) and in agrin-null (B) astrocytes. Scalebars: 10 μm.

8 B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

expression (Saadoun et al., 2002; Warth et al., 2004). Rascheret al. (2002) showed a loss of agrin at multiple sites of a braintumor. This suggests that agrin and other components of theECM enhance the formation of OAPs and the polarity ofastrocytic endfeet.

Agrin cannot be the only factor participating in OAPformation, because agrin-null astrocyte cultures are able toform OAPs also in the absence of exogenous agrin (Fig. 2B,

Fig. 6 – (A) Western blot showing the distribution of AQP4 inagrin-null-astrocytes and wild-type astrocytes grown on PLL(poly-L-lysine, control), ECM, agrin-isoform A0B0 and A4B8.The lower bands (32 kDa) represent theM23 isoform of AQP4,the upper bands (34 kDa) represent the AQP4-M1 isoform.(B) Statistical analysis of (A) including SEM showing relativeamounts of AQP4-M23 and -M1.

control). Since agrin does not bind to AQP4 directly, butindirectly via the dystrophin–dystroglycan complex (Gee et al.,1994), α-dystroglycan may serve as a receptor for exogenousagrin in agrin-null astrocytes as well. Indeed, α-dystroglycancould be detected immunocytochemically in wild-type as wellas in agrin-null astrocytes (compare Figs. 5A and B). Further-more, immunocytochemical staining of AQP4 revealed afluorescent signal in the cell membranes, which was moreintensive in coated than in uncoated cultures. This observationwas independent of the origin of the cells. The staining of allcoated cultures showed the samepattern supporting the freeze-fracture results that exogenously applied ECM components areable to influence the occurrence of OAPs.

To test the function of water transport, cell volumemeasurements of agrin-null and wild-type astrocytes wereperformed and revealed an increased swelling capacity ofcells grown on agrin-A4B8 coating independent of theirorigin. Volume measurements of ECM-coated astrocytescould not be tested due to the experimental design. Never-theless, we can speculate that the swelling behavior ofastrocytes should be similar as shown by astrocytes grownon the endothelial agrin-isoform A0B0, because this agrinisoform is present in the ECM of the basal lamina in the brain.Taken together, we were not able to find a difference in watertransport between wild-type and agrin-null astrocytes grow-ing on A4B8 agrin.

4. Conclusions

Two preconditions must be fulfilled to form OAPs of a certainsize and number: AQP4-M23 isoform must be expressed andenough agrin or ECM should be present. No OAPs should beexpected if one of these components was absent or reduced.Noell et al. (2009) have shown that in the absence of agrin theastrocytic endfoot membranes in vivo contained a very smallnumber of OAPs. However, the amount of AQP4-M23 wasnormal. In addition, α-dystroglycan as a binding partner ofagrinmight be important for the localization of the OAPs at the

Fig. 7 – PCR analyses of astrocyte cultures. AQP4-M1 (upperbands) and AQP4-M23 (middle) and HPRT (lower bands)representing the load control.

9B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

endfoot membranes. Wolburg-Buchholz et al. (2009) were ableto show that at sites of inflammatory cuffs during experimen-tal autoimmune encephalomyelitis, dystroglycanwas lost andthe OAP-related polarity was reduced. This polarity reductionappears to involve the appearance of OAPs on membranedomains which normally did not contain them. Takentogether, the size and number of OAPs formed in astrocytesdepend on agrin containing ECM and the presence of AQP4-M23 and dystroglycan in the astroglial membrane.

5. Experimental procedures

5.1. Culture of murine brain wild-type- and agrin-nullastrocytes

All animal care and experimental protocols conformed tothe University of Tübingen Animal Ethics Committeeguidelines and the German legislation regulating the useof animals in research. The use of animals in this study

Fig. 8 – Statistical analysis (box-blots) of cell volume measuremedifferent coatings.

was minimized to the necessary numbers for quantitativeand qualitative analyses. Murine brain astrocytes wereseparately isolated and cultured from individual E19.5BALB/c mice embryos derived from the breeding ofheterozygous agrin mice (Noell et al., 2009). Briefly, afterremoval of meninges, the tissues were digested with 0.1%trypsin and 0.02% EDTA (Lonza, Verviers, Belgium) in PBSfor 15 min at room temperature. Then, the cell suspensionwas transferred into culture medium, centrifuged (400×g;5 min), resuspended in culture medium and filteredthrough a 70 μm nylon mesh. Cells were seeded in T25culture flasks (Corning, Heidelberg, Germany) and culturedin Dulbecco's modified Eagle's medium supplemented with20% fetal calf serum, penicillin (10,000 units/ml), andstreptomycin (10,000 μg/ml) under standard conditions upto passage two (all Lonza, Verviers, Belgium). All littermatesused for culture of astrocytes, cyto- and histochemistry andfreeze-fracturing were genotyped by PCR with lysates of tailbiopsies. For coating experiments wild-type and agrin-nullastrocytes were used in passage 3. After 7 days cell cultureswere stopped and analyzed using freeze-fracture, immu-nostaining, western blotting, PCR and cell volume detectiontechnique.

5.2. Coating with agrin

Transfected HEK 293 cells producing the neuronal splicingvariant A4B8 or the endothelial splicing variant A0B0 of agrin(kindly provided by Stephan Kröger, LMU Munich) werecultured in 0.01% poly-L-lysine (PLL)-precoated flasks and oncoverslips under standard conditions. PLL was purchasedfrom Sigma (Deisenhofen, Germany). After reaching con-fluency the cells were lysed with 0.5% Triton X-100 solution(15 min at 37 °C) and washed with PBS. Using an antibodyagainst agrin (kindly provided by Stephan Kröger, LMUMunich), the equal distribution of agrin coating was demon-strated. Subsequently, wild-type or agrin-null astrocytes wereseeded on precoated agrin (A0B0 and A4B8) surface andcultured for 3–7 days (Fig. 1).

nts of cultured wt- and agrin-null astrocytes growing on

10 B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

5.3. Coating with ECM

Precooled culture flasks and coverslips were coated with ECMgel originated from Engelbreth–Holm–Swarm mouse sarcoma(Sigma, Deisenhofen, Germany). Wild-type and agrin-nullastrocytes were seeded in the coated culture flasks andcultured for 3–7 days.

5.4. Freeze-fracture experiments

Monolayers of cultured cells were fixed with 2.5% glutar-aldehyde in 0.1 M cacodylate buffer (pH 7.4) for 2 h at roomtemperature. The specimens were then cryoprotected forfreeze-fracturing in 30% glycerol and snap-frozen innitrogen slush (−210 °C). Subsequently, they were fracturedin a Balzer's freeze-fracture device (BAF400D; Balzers,Liechtenstein) at 5×10−6 mbar and −150 °C. The fracturefaces were shadowed with platinum/carbon (3 nm, 45°) forcontrast and carbon (30 nm, 90°) for stabilization of thereplica. After removal of the cell material in 12% sodiumhypochlorite, the replicas were rinsed in double-distilledwater several times and mounted on Pioloform-coatedcopper grids. The replicas were observed by using a ZeissEM10 electron microscope (Zeiss, Oberkochen, Germany).OAP densities were determined as OAPs per squaremicrometer at a magnification of 100,000:1 in 30 replicasof three different assays in each experiment. Statisticalevaluation was performed by using the Wilcoxon/Kruskal–Wallis one-way analysis of variance (ANOVA) on ranksfollowed by Tukey–Kramer method for comparisons (JMP,Cary, NC, USA).

5.5. Immunocytochemistry

Coverslip cultures of wild-type and agrin-null astrocytes werefixed in 4% paraformaldehyde for 15min. Fixed astrocytes wereincubated at 4 °C overnight with a rabbit polyclonal AQP4antibody (Santa Cruz, Heidelberg, Germany) 1:100 diluted in0.25% Triton X-100 and 1% DMSO in PBS. AQP4 was detectedusing a secondary goat anti-rabbit Cy3-labeled antibody (Dia-nova, Hamburg, Germany, 1 h at room temperature). Immu-nostaining of α-dystroglycan was performed using a rabbitpolyclonal antibody (Santa Cruz, Heidelberg, Germany). Ascontrol for nonspecific staining or autofluorescence, the prima-ry antibody was omitted. Nonspecific binding was blocked byincubation for 30min in 4% normal goat serum and 1% BSA inPBS. Specimens were mounted in Mowiol (Calbiochem, Merck,Darmstadt, Germany). Fluorescence was visualized with a

Table 2 – Primer sequences.

Target mRNA (mouse) Primer sequence

HPRT s GCT GGT GAA AAG GACas CAC AGG ACT AGA ACA

AQP4 M23 s GGA AGG CTA GGT TGGas TGG TGA CTC CCA ATC

AQP4 M1 s CTC CCA GTG TAC TGGas TGG TGA CTC CCA ATC

LSM510 META confocal laser scanning microscope (Zeiss,Oberkochen, Germany) using a HeNe laser for excitation at543 nm with appropriate filter sets and a 40× oil immersionobjective (N.A. 1.3).

5.6. Electrophoresis and immunoblotting

Confluent astrocytes were lysed and prepared for westernblotting as described by Neely et al. (1999). Briefly, cellcultures were lysed with Laemmli-buffer, protein wasmeasured using the method of Bradford. 5 μg total proteinof each sample was used for electrophoresis with 12.5% SDS-PAGE. The samples were blotted on a nitrocellulose mem-brane and stained with an antibody against AQP4 (SantaCruz, Heidelberg, Germany) and a secondary antibodylabeled with horseradish peroxidase (Sigma, Deisenhofen,Germany). Western blots were densitometrically quantifiedusing ImageJ software (NIH, Bethesda, MA, USA; available athttp://rsb.info.nih.gov/ij). Absolute optical density (OD) wasnormalized to the ODs of the corresponding bands of β-tubulin loading control and expressed as relative abundancein arbitrary units. Each experiment was performed at leastnine times.

5.7. Reverse Transcriptase (RT)-PCR

RT-PCR analysis of AQP4-isoformM23, M1 and HPRT (hypoxan-thine guanine phosphoribosyl transferase) mRNAs was per-formed as described previously (Ivanova et al., 2002). The HPRTgene has been reported as a constitutively expressed house-keeping gene (Frericks and Esser, 2008) and has been identifiedas the single best reference gene (De Kok et al., 2005). Total RNAwas isolated using the peqGold RNApure extraction kit (Peqlab,Erlangen, Germany) according to the manufacturer's protocol.cDNA was synthesized from 1 μg of total RNA with 1 μl dNTPs(0.8 nM), 1 μl MMLV reverse transcriptase, 5 μl 5×-buffer, 1.5 μlhexanucleotide (10 pmol/μl, all reagents from Invitrogen, Karls-ruhe, Germany) for 1 h at 37 °C followed by enzyme inactivationfor 5 min at 95 °C. PCR was conducted with 1.5 μl of the RTreaction, 0.5 μl sense and antisense primers (10 pmol/μl), 1 μldNTPs (5 mM), 10×-PCR buffer (+15 mM MgCl2) and 0.3 μlAmpliTaq Gold DNA Polymerase (Applied Biosystems, FosterCity, California, USA). PCR conditions for M23-, M1-, and HPRT-primers were: 35 cycles of denaturation for 1 min at 95 °C,annealing for 1 min at 62 °C, extension for 1 min at 72 °C,followed by a final elongation step at 72 °C for 5 min (FlexCycler,Analytik Jena). Concurrent RT-PCR amplification of HPRT wascarried out asan internal control for variations in the efficiencies

(5′→3′) Product size (bp) Annealing (°C)

CTC T 250 62CCT GCTGA CTT C 460 62CTC CAA CAGC CCG 510 62CTC CAA C

11B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

of RNA isolation and RT. The primer sequences are shown inTable 2.

The PCR products were separated by electrophoresis on a1.5% agarose gel stained with ethidium bromide and wereanalyzed by a UV transilluminator.

5.8. Cell volume measurements

For volume measurements, cells were grown on round cover-slips fitting into an incubation chamber as described previously(Noell et al., 2007). Briefly, cells were incubated in 5 μM calcein(Invitrogen, Karlsruhe, Germany) in culture medium at 37 °C for30 min. The coverslips were then mounted in a closed flow-through incubation chamber (LaCon,Staig,Germany).Cellswereimaged with a Zeiss LSM 510 META confocal microscope (Zeiss,Oberkochen, Germany) via a 63× long-distance objective. The488-nmargon laser linewasused for excitationwith appropriatefilter sets, the laser intensity being kept low to avoid cell damageandbleachingduring the experiment. To test for cellular volumechanges, the following paradigm was used. To establish a baseline, the chamber was perfused with isotonic Hanks' balancedsalt solution (HBSS; 300mOsm, Invitrogen, Karlsruhe, Germany)at a flow rate of 300 μl/min. After 180 s, the solution wasswitched to hypotonic (200 mOsm) HBSS for an additional 180 sand finally back to isotonic buffer for at least another 200 s. Forcell volume changes, the increase or decrease in fluorescenceintensity was correlated with cell shrinkage or swelling,respectively. Subsequent calculations were performed withMicrosoft Excel (Microsoft, Seattle, USA). The fluorescencemeasurements were plotted as dF/F0, where F0 is the averagedinitial fluorescence intensity and dF is the difference of themeasured fluorescence to the initial fluorescence. The initialfluorescence values were compared with fluorescence minima(highest volume) and tested for statistical significance byapplying the nonparametric Wilcoxon rank test (n>30 cellswere measured in each experimental group) (Table 1).

Acknowledgments

Yeliz Donat-Krasnici and Ria Knittel are thanked for the skilfultechnical assistance in cell culturing, freeze-fracturing, andmicroscopy. We thank Drs. Markus Rüegg (Biozentrum,University of Basel, Switzerland) and Urban Deutsch (TKI,University of Bern, Switzerland) for providing us with theagrin-null mice. Dr. Stephan Kröger (LMUMunich, Germany) isthanked for providing agrin-transfected HEK cells. Withsupport by the Deutsche Krebshilfe-Mildred Scheel-Stiftungto HW and SN (grant numbers 107686 and 109219).

R E F E R E N C E S

Barber, A.J., Lieth, E., 1997. Agrin accumulates in the brainmicrovascular basal lamina during development of theblood-brain barrier. Dev. Dyn. 208, 62–74.

Bezakova, G., Ruegg, M.A., 2003. New insights into the roles ofagrin. Nat. Rev. Mol. Cell Biol. 4, 295–308.

De Kok, J.B., Roelofs, R.W., Giesendorf, B.A., Pennings, J.L., Waas,E.T., Feuth, T., Swinkels, D.W., Span, P.N., 2005. Normalization

of gene expression measurements in tumor tissues:comparison of 13 endogenous control genes. Lab. Invest. 85,154–159.

Dermietzel, R., 1973. Visualization by freeze-fracturing of regularstructures in glial cell membranes. Naturwissenschaften 60,208.

Frericks, M., Esser, C., 2008. A toolbox of novel murinehouse-keeping genes identified by meta-analysis of largescale gene expression profiles. Biochim. Biophys. Acta 1779,830–837.

Furman, C.S., Gorelick-Feldman, D.A., Davidson, K.G., Yasumura,T., Neely, J.D., Agre, P., Rash, J.E., 2003. Aquaporin-4 squarearray assembly: opposing actions of M1 and M23 isoforms.Proc. Natl Acad. Sci. USA 100, 13609–13614.

Gee, S.H., Montanaro, F., Lindenbaum, M.H., Carbonetto, S., 1994.Dystroglycan-alpha, a dystrophin-associated glycoprotein, is afunctional agrin receptor. Cell 77, 675–686.

Ivanova, T., Mendez, P., Garcia-Segura, L.M., Beyer, C., 2002. Rapidstimulation of the PI3-kinase/Akt signalling pathway indeveloping midbrain neurones by oestrogen. J. Neuroendocrinol.14, 73–79.

Jung, J.S., Bhat, R.V., Preston, G.M., Guggino, W.B., Baraban, J.M., Agre, P., 1994. Molecular characterization of an aqua-porin cDNA from brain: candidate osmoreceptor andregulator of water balance. Proc. Natl Acad. Sci. USA 91,13052–13056.

McMahan, U.J., 1990. The agrin hypothesis. Cold Spring Harb.Symp. Quant. Biol. 55, 407–418.

Moe, S.E., Sorbo, J.G., Sogaard, R., Zeuthen, T., Petter Ottersen, O.,Holen, T., 2008. New isoforms of rat aquaporin-4. Genomics 91,367–377.

Neely, J.D., Christensen, B.M., Nielsen, S., Agre, P., 1999.Heterotetrameric composition of aquaporin-4 water channels.Biochemistry 38, 11156–11163.

Neuhaus, J., 1990. Orthogonal arrays of particles in astroglial cells:quantitative analysis of their density, size, and correlationwith intramembranous particles. Glia 3, 241–251.

Noell, S., Fallier-Becker, P., Beyer, C., Kroger, S., Mack, A.F.,Wolburg, H., 2007. Effects of agrin on the expression anddistribution of the water channel protein aquaporin-4 andvolume regulation in cultured astrocytes. Eur. J. Neurosci. 26,2109–2118.

Noell, S., Fallier-Becker, P., Deutsch, U., Mack, A.F., Wolburg, H.,2009. Agrin defines polarized distribution of orthogonalarrays of particles in astrocytes. Cell Tissue Res. 337,185–195.

Rascher, G., Fischmann, A., Kroger, S., Duffner, F., Grote, E.H.,Wolburg, H., 2002. Extracellular matrix and the blood-brainbarrier in glioblastoma multiforme: spatial segregation oftenascin and agrin. Acta Neuropathol. 104, 85–91.

Rash, J.E., Yasumura, T., Hudson, C.S., Agre, P., Nielsen, S., 1998.Direct immunogold labeling of aquaporin-4 in square arraysof astrocyte and ependymocyte plasma membranes in ratbrain and spinal cord. Proc. Natl Acad. Sci. USA 95,11981–11986.

Rohlmann, A., Gocht, A., Wolburg, H., 1992. Reactive astrocytesin myelin-deficient rat optic nerve reveal an altereddistribution of orthogonal arrays of particles (OAP). Glia 5,259–268.

Saadoun, S., Papadopoulos, M.C., Davies, D.C., Krishna, S., Bell, B.A.,2002. Aquaporin-4 expression is increased in oedematoushuman brain tumours. J. Neurol. Neurosurg. Psychiatry 72,262–265.

Silberstein, C., Bouley, R., Huang, Y., Fang, P., Pastor-Soler, N.,Brown, D., Van Hoek, A.N., 2004. Membrane organization andfunction of M1 and M23 isoforms of aquaporin-4 in epithelialcells. Am. J. Physiol. Ren. Physiol. 287, F501–F511.

Suzuki, H., Nishikawa, K., Hiroaki, Y., Fujiyoshi, Y., 2008.Formation of aquaporin-4 arrays is inhibited by palmitoylation

12 B R A I N R E S E A R C H 1 3 6 7 ( 2 0 1 1 ) 2 – 1 2

of N-terminal cysteine residues. Biochim. Biophys. Acta 1778,1181–1189.

Tsen, G., Halfter, W., Kroger, S., Cole, G.J., 1995. Agrin is a heparansulfate proteoglycan. J. Biol. Chem. 270, 3392–3399.

Warth, A., Kroger, S., Wolburg, H., 2004. Redistribution ofaquaporin-4 in human glioblastoma correlates with loss of

agrin immunoreactivity from brain capillary basal laminae.Acta Neuropathol. 107, 311–318.

Wolburg-Buchholz, K., Mack, A.F., Steiner, E., Pfeiffer, F., Engelhardt,B., Wolburg, H., 2009. Loss of astrocyte polarity marks blood-brainbarrier impairment during experimental autoimmuneencephalomyelitis. Acta Neuropathol. 118, 219–233.