use of surface plasmon resonance for real-time analysis of the interaction of zo-1 and occludin
TRANSCRIPT
UA
AI
R
attogpocicbgcm3siptltfp
t
iascTam(AM
9
Biochemical and Biophysical Research Communications 288, 1194–1199 (2001)
doi:10.1006/bbrc.2001.5914, available online at http://www.idealibrary.com on
0CA
se of Surface Plasmon Resonance for Real-Timenalysis of the Interaction of ZO-1 and Occludin
nke Schmidt, Darkhan I. Utepbergenov, Gerd Krause, and Ingolf E. Blasig1
nstitute of Molecular Pharmacology, Robert-Rossle-Strasse 10, 13125 Berlin, Germany
eceived October 11, 2001
mologues) family which is known to assemble multi-pP3PpspZtatftu
ii2bHmfua
(btdcstdMZc
M
lo
Surface plasmon resonance (SPR) spectroscopy waspplied to study in real time, the interaction betweenhe tight junction proteins ZO-1 and occludin. To imi-ate the morphology of tight junctions, a cytosolic tailf mouse occludin was immobilised at the sensor anduanylate kinase-like domain (Guk) was allowed toass over the modified chip surface. The Guk domainf ZO-1 (residues 644–812) was found to bind to theytoplasmic, carboxy-terminal region of occludin (res-dues 378–521). This interaction was systematicallyharacterised with respect to the concentrations ofoth proteins and the binding conditions. Under theiven experimental conditions, association and disso-iation showed saturation kinetics, with affinity inicromolar range: ka 5 4.14 6 0.52 3 103 M21 s21, kd 5
.04 6 0.38 3 1023 s21, KD 5 639 6 51 nM. The resultsupport the hypothesis that the Guk domain of ZO-1 isnvolved in the recruitment of the transmembranerotein occludin at tight junctions by interacting withhe cytosolic carboxy-terminal sequence of occludin,ocated far from the cell membrane. We demonstratehe use of SPR spectroscopy as an effective approachor characterisation of the interactions of junctionroteins. © 2001 Academic Press
Key Words: surface plasmon resonance; tight junc-ions; protein interaction; occludin; ZO-1; Guk domain.
Tight junctions (TJ) play a critical role in maintain-ng the integrity of epithelial tissues. Their uniquenatomical feature, including a complex lipid-proteintructure, forms an impermeable seal between adja-ent cells. The transmembrane protein components ofJ are occludin, JAM (junctional adhesion molecule),nd the claudin family of proteins, while the cytoplas-ic counterpart includes zona occludens proteins
ZO-1, ZO-2, ZO-3), cingulin, symplekin, 7H6, Rab3B,F-6, and ASIP (1, 2). ZO-proteins are members of theAGUK (membrane-associated guanylate kinase ho-
1 To whom correspondence should be addressed. Fax: (149-30)4793-243. E-mail: [email protected].
1194006-291X/01 $35.00opyright © 2001 by Academic Pressll rights of reproduction in any form reserved.
rotein complexes. ZO proteins are comprised of threeDZ (PSD95/dlg/ZO-1) domains, an SH3 (src-homology) domain, and a guanylate kinase-like (Guk) domain.DZ, SH3, and Guk domains are involved in protein-rotein interactions many of which have been de-cribed at tight junctions (3, 4). For example a ZO-1olypeptide was reported to interact with ZO-2 andO-3 (5, 1), occludin (3), cingulin (6), claudins (7), junc-ion adhesion molecule (8), ZO-1 associated kinase (9),nd a transcriptional factor ZONAB (10). Therefore,he development of an effective in vitro binding assayor the identification and characterisation of interac-ions between junction proteins would facilitate ournderstanding of TJ function at the molecular level.Interactions of ZO-1 with occludin are assumed to be
nvolved in the assembly and regulation of TJ via bind-ng motifs on both proteins. It has been reported that a50 amino acid sequence of mammalian ZO-1 (4) mayind to a 150 amino acid region of chicken occludin (3).owever, it is unclear whether these coprecipitationeasurements on a heterologous system can be trans-
erred to a homologous mammalian system, and it isnknown exactly which sequences and mechanismsre involved in the interaction.In the present study, surface plasmon resonance
SPR) spectroscopy has been adopted to investigate theinding between TJ-proteins, as well as their associa-ion and dissociation (11). With this effective proce-ure, binding of the complete Guk domain of ZO-1 at aytoplasmic part of occludin has been demonstrated,tandardised, and characterised with respect to pro-ein concentration dependence, binding kinetics, repro-ucibility, specificity, and measurement conditions.oreover, the acidic region adjacent to Guk domain of
O-1 was shown to be unnecessary for binding to oc-ludin.
ATERIALS AND METHODS
Expression constructs of occludin and ZO-1. Total RNA was iso-ated from 50 mg mouse kidney using TRIzol reagent (Life Technol-gies, Eggenstein, Germany). cDNAs were synthesised from 1 mg
Rr3cdfGA(i5t
65CCZtTdwAlTmgp(r
c
wettcacNinwtwmmevgaqC(
(iM(taomm
(Staef
Vol. 288, No. 5, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
NA by RT-PCR (200 U MuMLV reverse transcriptase) with 10 mMandom hexamers (Promega, Madison, WI) using 22 cycles (94°C for0 s, 68°C for 1 min, 72°C for 45 s). For occludin, a cDNA fragmentorresponding to the complete cytoplasmic, C-terminal tail of occlu-in (residues 264–521, 777 bp) was amplified by PCR using theorward primer 59-TTG TCG ACT AAG TTT CCG TCT GTC ATATC-39 and the reverse primer 59-AAA GGA TCC AAA ACC CGAGA AAG ATG GAT CGG-39 containing SalI and BamHI sites
underlined, bold: stop codon). Subsequently, the fragment compris-ng the occludin residues 378–521 (429 bp) was amplified using9-AAA GGA TCC AAA AGG GCT CCC ACG AAG GGG AAA-39 ashe reverse primer.
For ZO-1, a cDNA fragment encoding to the amino acid sequence01–890 (867 bp) was amplified by PCR using the forward primer9-GGG GGT CGA CCT AAT GAT GCA TTC CAG AGG AAT CCTTC-39 and reverse primer 59-CCC GGA TCC TTC TGG AGG TTTGA GGT CTT CGC-39. The cDNA sequence of the GukAR domain ofO-1 (residues 644–890) was subsequently amplified (738 bp) usinghe reverse primer 59-AAA GGA TCC CTA AGA CCT GTA ACC ATCTT GGA-39. The cDNA sequence of the GuK domain of ZO-1 (resi-ues 644–812) was subsequently amplified (504 bp) using the for-ard primer 59-GGG GTC GAC CTA GTC ATC ACT TGT AGC ACCTC CGC-39 containing SalI and BamHI restriction sites (under-
ined, bold: stop codon). The amplified products were cloned into theOPO TA vector (Invitrogen, Carlsbad, CA). The BamHI/SalI frag-ents of these constructs were subcloned into pMAL-c2x (New En-
land Biolabs, Beverly, MA) in frame with MBP (maltose-bindingrotein). The final expression plasmids were verified by sequencingBig Dye terminator cycle sequencing kit, Applied Biosystems, War-ington, UK).
Overexpression and purification of occludin and ZO-1. Escherichiaoli cells (TOP10F9, Invitrogen, CA) containing appropriate plasmids
FIG. 1. Purification of a cytoplasmic sequence of mouse occludin,maltose-binding protein) after transfection and overexpression in E.DS–polyacrylamide gel electrophoresis of MBP-GukAR (residuesransfected with MBP-Guk(AR) (F3, F4), cell lysate of E. coli after infter overexpression plus amylose affinity chromatography (F1, Flectrophoresis gel (colloidal blue staining) of cell lysate of occludin (ollowing anion exchange chromatography; F9, MBP; F10, MBP-occl
1195
ere cultivated in rich medium containing 10 g/l tryptone, 5 g/l yeastxtract, 125 mM NaCl, 10 mM glucose with 100 mg/ml ampicillin untilhe absorbance at 600 nm reached 0.5–0.8. Expression was induced byhe addition of isopropyl-b-D-thiogalactopyranoside (IPTG) to a finaloncentration of 0.3 mM and the cells were further incubated for 1.5 ht 37°C. After harvesting by centrifugation (3000g) for 7 min at 4°C, theell pellet was resuspended in column buffer SP1 (20 mM Tris, 200 mMaCl, and 1 mM EDTA) and sonicated 3 3 15 s with pauses of 10 s on
ce. Lysed cells were centrifuged for 2 min at 4°C (10.000g). The super-atant was filtered through a 0.45 mm filter and MBP-fused proteinsere purified over an amylose resin column according to the manufac-
urer‘s instructions (New England BioLabs, Inc., USA). The MBP-ZO-1sere eluted with buffer SP2 (20 mM Tris, 200 mM NaCl, 10 mMaltose). MBP-occludin was eluted with buffer SP3 (20 mM Tris, 10M maltose) and applied on HiPrep 16/10 DEAE column that had been
quilibrated with buffer SP3. The column was washed with two-columnolumes of buffer SP3 and the proteins were eluted with 200 ml linearradient of 0–0.3 M NaCl in buffer SP3. The protein fractions werenalysed by 10% SDS–polyacrylamide gel electrophoresis and subse-uent staining with the colloidal blue staining kit (Novex, San Diego,A). Protein concentrations were determined using a Lowry kit
P56756; Sigma Diagnostic, St. Louis, MO).
ZO-1/occludin binding studies. A BIACORE 2000 instrumentBIACORE AB, Uppsala, Sweden) was used to control the immobil-sation steps and characterise the binding of MBP-Guk(AR) to the
BP-occludin modified sensor surface. The surface of sensor chipCM5, BIACORE AB, Uppsala, Sweden) was activated with a mix-ure 400 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)nd 100 mM N-hydroxysuccinimide (NHS) for 5 min. MBP-ccludin.378-521 was concentrated and desalted by ultrafiltration onicrocon YM filters (Millipore Corp.; Bedford, MA), redissolved in 10M sodium acetate pH 5.0 and allowed to pass over the activated
the guanylate kinase-like (Guk) domain of mouse ZO-1, and of MBPi. (A) Occludin and ZO-1 in ephithelial and endothelial cells. (B) 10%–890) and MBP-Guk (residues 644–812): supernatant of E. coli
ction of overexpression with IPTG (F2, F5), and cell lysate of E. coliM, low molecular weight marker. (C) 10% SDS–polyacrylamide
, of eluate after amylose affinity chromatography (F8), and of eluaten.378–521.
ofcol644du6).
F7)udi
cpmwMnroNet(wsS(W
R
tijw
MZ8cfFGwMattu(
oActbtg
Ts(tCe7aa
Vol. 288, No. 5, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
hip surface for 8 min at a flow rate of 5 ml/min. Ethanolamine (1 M,H 8.5) was used to saturate the remaining active ester groups (12l, 5 ml/min). The immobilized occludin was equilibrated for 20 minith running buffer (200 mM NaCl, 20 mM Tris–HCl pH 7.8, 50 mMgCl2, and 10 mM maltose) at a flow rate of 5 ml/min (Fig. 2). Where
ot stated otherwise, association of MBP-Guk and MBP-GukAR wasegistered by perfusing in the running buffer for 4 min at a flow ratef 8 ml/min. To regenerate the chip surface the solution of 50 mMa2CO3 and 300 mM NaCl was infused for 5–10 min. BIACORE
valuation software 3.1 was used for fitting of binding kinetics usinghe 1:1 Langmuir binding model (12). To estimate binding constantsKD), the association and dissociation phases of the sensorgrammesere used for k a and k d calculation, based on nonsaturation and
aturation concentrations of MBP-Guk. Data represent mean 6.E.M. Significance values were calculated by One-way ANOVA
comparison between different groups) and two-tailed Mann–hitney Rank Sum Test (comparison between two groups).
ESULTS
In this study, the interaction between two tight junc-ion proteins occludin and ZO-1 has been character-sed. Occludin is a transmembrane protein of tightunctions with no similarity to any other proteinshereas ZO-1 is a multidomain protein belonging to a
FIG. 2. Immobilization of a cytoplasmic sequence of mouse occludhe C-terminal part of occludin was used as fusion proteins of MBPensor surface was activated and deactivated by means of 1-ethyl-NHS) and ethanolamine, respectively, in a buffer containing 200 mMions was 5 mM. RU, resonance units. (B) Association and dissociatio-terminal tail of mouse MPB-occludin.378–521 immobilized on theration of the occludin tail. MBP-GukAR was infused at concentrati.8, 50 mM MgCl2, and 10 mM maltose. Regeneration was performssociation and dissociation experiment followed. (C) Addition of diffend phosphorylase B), 5 mM bovine serum albumin (BSA), 5 mM MB
1196
AGUK family (Fig. 1A). The complete Guk domain ofO-1 with the adjacent acidic region (GukAR.644–90), Guk domain of ZO-1 (Guk.644–812), andarboxy-terminus of occludin were overexpressed asusion proteins with maltose binding protein (MBP).igure 1B shows that essentially pure MBP-ukAR.644–890 (F1) and MBP-Guk.644–812 (F6)ere obtained after amylose affinity chromatography.BP-occludin was somewhat degraded, therefore
nion-exchange FPLC was applied to obtain pure pro-ein (Fig. 1C, F10). MBP-Guk showed some degrada-ion (approx. 15%, see Fig. 1B), but the degraded prod-ct could not be separated from intact protein by FPLCdata not shown).
Figure 2A shows the immobilisation steps of MBP-ccludin.378–521 at the dextran coated sensor surface.fter EDC/NHS (1-ethyl-3-(3-dimethylaminopropyl)arbodiimide and N-hydroxysuccinimide) activation ofhe chip surface MBP-occludin was covalently immo-ilised. The amount of immobilised MBP-occludin de-ermined after deactivation of the remaining esterroups with ethanolamine corresponded to approx.
on the sensor chip of a surface plasmon resonance spectrometer. (A)ltose-binding protein, MBP-occludin.378–521). The dextran-coated-dimethylaminopropyl) carbodiimide (EDC)/N-hydroxysuccinimideaCl and 20 mM Tris–HCl (pH 7.8); the applied occludin concentra-f the GukAR domain of mouse ZO-1 (MBP-GukAR.644–890) at thensor chip of a surface plasmon resonance spectrometer and regen-of 2.5 mM. Binding buffer was 200 mM NaCl, 20 mM Tris–HCl, pH
by infusion of 50 mM NaCO3 and 300 mM NaCl; then the nextt controls including a protein mix (egg albumin, carbonic anhydrase,galactosidase (MBP-Gal), and running buffer.
in(ma3-(3
Nn ose
oned
renP-
6aaGritc
tbtBciasspa
wswi
tMchibpfir
GaoBad3ababtL1
Miaao
Vol. 288, No. 5, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
400 resonance units (RU) when 5 mM of occludin waspplied. When MBP-GukAR was allowed to run overn MBP-occludin modified sensor chip binding of MBP-ukAR was clearly visible (Fig. 2B). After switching to
unning buffer the dissociation of MBP-GukAR frommmobilised MBP-occludin could be followed. Finally,he chip surface was completely regenerated by appli-ation of the regeneration solution (Fig. 2B).
To show the specificity of occludin-GukAR interac-ions, a set of proteins including MBP-galactosidase,ovine serum albumin and a mixture of standard pro-eins (egg albumin, carbonic anhydrase, phosphorylase) were allowed to pass over MBP-occludin modified
hip surface. In all cases a positive or negative bulkndex contribution was obvious (Fig. 2C). However, nossociation of tested proteins was seen and the SPRignal returned to the initial level immediately afterwitching to the running buffer. These data exclude theossibility that MBP binds to the modified sensor chipnd suggested a specific occludin-GukAR interaction.The sequence of ZO-1 necessary for the interactionith occludin has been located to the region corre-
ponding to the GukAR (4). We next determinedhether the acidic region, comprised of 78 amino acids,
s involved in the interaction. Figures 3A and 3B illus-
FIG. 3. (A) Concentration-dependent association and dissociationPB-occludin.378–521 using surface plasmon resonance spectroscop
n mM (data in brackets denote the amount of MBP-GukAR boundssociation and dissociation of MBP-Guk.644–812. (C) The curve shond the amount of MBP-Guk.644–812 bound at the end of the associf experiments.
1197
rates the binding properties of MBP-GukAR andBP-Guk to MBP-occludin. The recombinant protein
omprised of Guk domain alone binds with somewhatigher affinity compared to MBP-GukAR. These data
ndicate that acidic region of ZO-1 is unnecessary forinding to occludin, and subsequent experiments wereerformed with MBP-Guk.644–812. The decreased af-nity of GukAR could be due to the fact that acidicegion masks the Guk domain.The concentration-dependent association of MBP-uk.644-812 with immobilised MBP-occludin.378–521nd the subsequent dissociation of Guk from theccludin-modified chip surface is shown on Fig. 3B.oth the kinetics of Guk association and the totalmount of bound Guk showed a defined concentrationependence. A typical standard curve is shown on Fig.C. The lower detection limit was determined to bebout 0.3 mM of MBP-Guk, whereas the saturatedinding was observed at concentrations over 5 mM. k on
nd k off values were determined from sensorgrammesy analysing association and dissociation kinetics ofhe Guk-occludin interactions using the 1:1 model ofangmuir isotherm (11). The values of 4.14 6 0.52 303 M21 s21 and 3.04 6 0.38 3 1023 s21 were obtained
BP-GukAR.644–890 of mouse ZO-1 to the C-terminal tail of mouseBP-GukAR was infused as indicated. The concentrations are given
the end of the association interval). (B) Concentration-dependentthe relationship of the Guk concentration infused at the sensor chipn period; data represent mean 6 SEM, n $ 4 from separated series
of My. Mat
wsatio
fc
aiwcittst
D
ri
fmatmCsStatabts
cludin seems not to affect binding, indicating that thesitio5htcmt
ltp
oZro01n
ao5MfiM
Vol. 288, No. 5, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
or k on and k off, respectively. Thus, an apparent bindingonstant, KD, of 639 6 51 nM, was calculated.In Fig. 4 the effect of flow rate, Ca21 concentration,
nd pH at the sensor chip is expressed. The flow ratenfluenced association of MBP-Guk to MBP-occludinhen #6 or $14 ml/min medium were used. This indi-
ates that mass transport limitations were not a factorn the range between 8 and 12 ml/min. Ca21 concentra-ions between 0.1 and 100 mM did not influence eitherhe association or the dissociation. The pH was alsohown to have no effect on binding in the range pH 5o 9.
ISCUSSION
Here, we demonstrate the use of surface plasmonesonance spectroscopy for the first time, to character-se the interaction between tight junction proteins.
Both occludin and ZO-1 fragments were purified asusion proteins with MBP in relatively high yields (20g per litre of culture for MBP-GukAR and MBP-Guk
nd 10 mg for MBP-occludin). The cytoplasmic occludinail was immobilised at the surface of the SPR sensor,imicking the biological situation in which the-terminal tail of occludin is “immobilized” at the innerurface of the plasma membrane. The results of thePR investigations confirm the association of the pro-eins previously suggested by indirect morphologicalnd coprecipitation experiments on heterologous pro-eins (3, 4). Furthermore our results indicate that thecidic tail directly C-terminal to Guk is not required forinding and that the binding mechanism is specific forhe Guk domain. The rather low degree of amino acidequence conservation between chicken and mouse oc-
FIG. 4. The effect of (A) flow rate, (B) calcium ion concentration,nd (C) pH on the association and dissociation of MBP-Guk.644-812f mouse ZO-1 to the C-terminal tail of mouse MPB-occludin.378–21 analysed by means of surface plasmon resonance spectroscopy.BP-Guk.644–812 was infused in concentrations of 1 mM. The
gures show the exact perfusion conditions and associated amount ofBP-Guk.644–812 at the end of infusion.
1198
mall number of conserved residues contain the bind-ng information. A highly conserved region (90% be-ween mouse and human) appears between amino ac-ds 400 and 500 in occludin and is included in theccludin fragment used in this study (residues 378–21). Bioinformatic studies (Fig. 5) suggest potentialelical structures both in this occludin region and inhe Guk region of ZO-1, which could result in a coiled–oil interaction, a well known protein–protein bindingechanism (13, and G. Krause, personal communica-
ion).MBP-Guk has been found to bind to occludin in the
ow micromolar and submicromolar range. This andhe estimated binding constant of about 639 6 51 nMoints to a moderate, reversible interaction between
FIG. 5. Sequence alignment of the C-terminal part of MPB-ccludin.378–521 and the GukAR domain with the acidic region ofO-1 (MBP-GukAR.644–890). (A) Lane 0, prediction of coiled-coilegion; lanes 1–3, sequence alignment of mouse, human, and chickenccludin; lane N, number of residues in the mouse occludin. (B) Lane, prediction of coiled–coil region; lane G, Guk domain of ZO-1; lanesand 2, sequence alignment of mouse and human ZO-1; lane N,
umber of residues in the mouse ZO-1.
ZO-1 and occludin. Such an interaction would be ofrrfsesapi(j
jtttoArsacbmcccsstTu
ie
molecular interaction analysis is a valuable tool tos
A
csA
R
111
111
Vol. 288, No. 5, 2001 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
egulatory relevance, possibly allowing the appropriateeorientation of occludin and/or ZO-1 to initiate theormation of TJ. The ZO-1 association to occludineems to be highly specific: no association was observedven when proteins with coiled–coil helices (bovineerum or egg albumin) were tested. The associationnd dissociation are not influenced by pH or Ca21 inhysiological and pathological levels. This means open-ng of TJ as caused by changes in Ca21-concentrations14) or drop of pH (15) is not a direct effect of theunction protein interactions studied here.
A number of intermolecular interactions at tightunctions have been described (3, 4, 8). A majority ofhe interactions was identified by coimmunoprecipita-ion followed by immunoblotting analysis. The SPRechnology has a number of advantages compared tother methods to study macromolecular interactions.ssociation and dissociation can be analysed easily in
eal time and the information on the specificity andtrength of the interaction is obtained within minutesfter the starting of experiment. The immobilised oc-ludin can be regenerated for hundreds of differentinding measurements over consecutive weeks. Thiseans that a complete series of binding experiments
an be performed with the same chip under identicalonditions. In addition to yielding the kinetics of asso-iation and dissociation, the calculation of binding con-tants is possible. By using these parameters it is pos-ible to test similar sequences of junction proteins andhe influence of different effectors on the interaction.hese data provide valuable information on the molec-lar architecture and regulation of tight junctions.In conclusion SPR was adopted to characterise bind-
ng of occludin to ZO-1. The results indicate that thexpression of recombinant proteins combined with bio-
1199
tudy the interactions of proteins at tight junctions.
CKNOWLEDGMENTS
We thank Gislinde Hartmann for technical assistance. We appre-iate the assistance of Dr. Linda Ball in preparation of the manu-cript. This work was supported by DFG BL308/6-1, DFG SFB 5072, and BMBF BEO 10015B/1466C.
EFERENCES
1. Wittchen, E. S., Haskins, J., and Stevenson, B. R. (1999) J. Biol.Chem. 274, 35179–35185.
2. Gonzalez-Mariscal, L., Betanzos, A., and Avila-Flores, A. (2000)Cell and Developmental Biol. 11, 315–324.
3. Furuse, M., Itoh, M., Hirase, T. Nagafuchi, A., Yonemura, S.,Tsukita, S., and Tsukita, S. (1994) J. Cell Biol. 127, 1617–1626.
4. Fanning, A. S., Jamson, B. J., Jesaitis, L. A., and Anderson, J. M.(1998) J. Biol. Chem. 273, 29745–29753.
5. Haskins, J., Gu, L., Wittchen, E. S., Hibbard, J., and Stevenson,B. R. (1998) J. Cell Biol. 141, 199–208.
6. Cordenonsi, M., Datri, F., Hammar, E., Parry, D. A., Kendrick-Jones, J., Shore, D., and Citi, S. (1999) J. Cell Biol. 147, 1569–1582.
7. Itoh, M., Morita, K., and Tsukita, S. (1999) J. Biol. Chem. 274,5981–5986.
8. Bazzoni, G., Martinez-Estrada, O. M., Orsenigo, F., Cordenonsi,M., Citi, S., and Dejana, E. (2000) J. Biol. Chem. 275, 20520–20526.
9. Balda, M. S., Anderson, J. M., and Matter, K. (1996) FEBS Lett.399, 326–332.
0. Balda, M. S., and Matter, K. (2000) EMBO J. 19, 2024–2033.1. Malmqvist, M. (1993) Nature 361, 186–187.2. Morton, T. A., and Myszka, D. G. (1998) Methods in Enzymol.
295, 268–294.3. Lupas, A. (1997) Curr. Opin. Struct. Biol. 7, 388–393.4. Lacaz-Vieira, F. (2000) J. Membr. Biol. 178, 151–161.5. Morton, T. A., Bennett, D. B., Appelbaum, E. R., Cusimano,
D. M., Johanson, K. O., Matico, R. E., Young, P. R., Doyle, M.,and Chaiken, I. M. (1994) J. Mol. Recognit. 7, 47–55.