C

3

4

5

6

7

8

A9

sSh(urIf

10

11

12

13

14

15

16

17

18

©19

K20

21

11

2

i3

s4

c5

e6

a7

r8

h9

10

s11

s12

f13

o14

i15

of

1 02 d

RO

OF

ARTICLE IN PRESS+ModelOLSUB 2970 1–6

Colloids and Surfaces B: Biointerfaces xxx (2007) xxx–xxx

Stability of S-layer proteins for electrochemical nanofabrication

Alvaro Presenda a, Daniel B. Allred b, Francois Baneyx a,Daniel T. Schwartz a,b, Mehmet Sarikaya a,b,∗

a Department of Chemical Engineering, University of Washington, Seattle, WA, United Statesb Department of Materials Science and Engineering, University of Washington, Seattle, WA, United States

Received 27 October 2006; received in revised form 13 January 2007; accepted 8 February 2007

bstract

Crystalline cell surface layer proteins (S-layers) can be used in electrochemical fabrication to create nanoscale arrays of metals and oxides onurfaces so long as the proteins maintain their long-range order during processing. We have explored the stability of the HPI layer protein (the-layer protein from the microorganism Deinococcus radiodurans) adsorbed onto platinum surface after immersion in sulfuric acid or sodiumydroxide electrolytes ranging in pH from 0 to 14 over time periods ranging from 1 to 1000 s. Topographic data obtained by atomic force microscopyAFM) was used to characterize the protein stability, judged by its retention of long-range order after immersion. The compiled data revealed that,nder these solution conditions and in this environment, the HPI layer protein has a dose-dependent structural stability “envelope” in the acidic

D Pange from 1 < pH < 4. The protein retains its long-range order up to 1000 s from pH 4 to 11, and has a sharp stability edge between pH 12 and 13.

nterestingly, the more stringent requirement of stability (i.e., retention of long-range order) defined in the context of electrochemical fabricationor this protein narrowed the window of stability in pH and time when compared to previous stability studies reported for this protein.

2007 Published by Elsevier B.V.

py; pH

e 16

c 17

v 18

f 19

f 20

o 21

s 22

i 23

1 24

e 25

m 26

r 27

p 28

RR

EC

TE

eywords: Deinococcus radiodurans; Protein stability; Atomic force microsco

. Introduction

Bacterial cell surface layer proteins are becoming increas-ngly popular systems for organizing and templating theynthesis of nanomaterials [1–9]. These proteins are genericallyategorized as the paracrystalline outer cell wall protein foundnveloping cells of many microbiological species [10]. Theyre well known for having a high degree of order even whenemoved from the cell surface, and having high tolerance toarsh chemicals and temperature variability [11].

Previously, we have demonstrated that S-layer proteins canerve as masks for electrochemical fabrication of metal andemiconductor nanoarrays [8]. Protein-based electrochemical

UN

CO

Please cite this article in press as: A. Presenda et al., Stability of S-layBiointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

abrication allows smaller and higher density packing of peri-dic elements than lithography-based methods widely used inntegrated circuit fabrication technology. Initial results relied on

∗ Corresponding author at: Department of Chemical Engineering, Universityf Washington, Seattle, WA, United States. Tel.: +1 206 543 0724;ax: +1 206 543 6381.

E-mail address: sarikaya@u.washington.edu (M. Sarikaya).

w 29

t 30

s 31

c 32

33

aanr

927-7765/$ – see front matter © 2007 Published by Elsevier B.V.oi:10.1016/j.colsurfb.2007.02.011

stability study; S-layer proteins; Electrodeposition; HPI layer

lectrolyte formulations from the literature or those availableommercially. The quality of the resulting patterned materialsaried from exceptional, in the case of cuprous oxide, to poor,or platinum, to non-existent, for copper. The initial hypothesisor these varied results was based primarily on poor optimizationf the electrodeposition process, which can be finely tailored toatisfy the high nucleation density required for the electrochem-cal fabrication of individual units at a density on the order of012 cm−2 on the work surface. However, an alternative hypoth-sis for poor pattern transfer could be that the various electrolytesay disrupt the long-range order of the S-layer protein of D.

adiodurans (also known as the HPI layer, or the hexagonallyacked intermediate layer, protein). This alternative hypothesisas initially rejected owing to the extensive characterization of

his protein [12,13] which indicated a broad stability over timecales (16 h or greater) far exceeding that needed for electro-hemical fabrication.

An early study on the HPI layer stability used precipitation

er proteins for electrochemical nanofabrication, Colloids Surf. B:

s an indicator that the proteins survived exposure to various 34

ggravating conditions [12]. However, precipitation alone does 35

ot differentiate between HPI layer proteins that have a long 36

ange ordered nanostructure (desirable for electrochemical 37

ED

IN+ModelC

2 faces

f38

e39

e40

[41

t42

H43

o44

t45

u46

47

t48

o49

f50

t51

T52

r53

o54

o55

f56

W57

p58

t59

260

261

62

o63

B64

S65

T66

a67

i68

s69

m70

w71

w72

n73

t74

[75

f76

p77

t78

m79

e80

m81

r82

i83

r84

285

86

d87

F88

(89

m 90

s 91

b 92

t 93

T 94

95

w 96

a 97

3 98

s 99

a 100

2 101

102

p 103

h 104

r 105

p 106

a 107

T 108

w 109

2 110

111

T 112

L 113

a 114

w 115

n 116

b 117

a 118

2 119

120

a 121

o 122

v 123

p 124

c 125

a 126

a 127

2 128

129

e 130

a 131

p 132

c 133

surface. The counter electrode (anode) was a piece of platinum 134

NC

OR

RE

CT

ARTICLEOLSUB 2970 1–6

A. Presenda et al. / Colloids and Sur

abrication) from disordered HPI layer protein aggregates largenough to precipitate. Later stability studies used transmissionlectron microscopy in conjunction with heavy metal staining13] to examine the fine structure of the protein, a more stringentest which may be more useful for electrochemical fabrication.owever, as the degradation of the HPI layer protein was testedn a time scale of several hours to overnight, and short contactimes are of much greater interest, this information may bennecessarily conservative for fabrication purposes.

Here, we explore the temporal stability of the HPI layer pro-ein adsorbed on surfaces as a function of electrolyte pH (onef the key variables in formulating baths for electrochemicalabrication). The figure of merit for this protein is its abilityo retain its long-range order upon exposure to the electrolyte.his refined definition narrows the window of stability of the D.adioradurans HPI layer protein when compared to the windowf stability against other modes of degradation such as prote-lysis or acid hydrolysis, but it permits a rational approach toormulating an electrolye to achieve successful pattern transfer.

e demonstrate the value of this approach by re-examining cop-er electrodeposition through the HPI layer proteins, a processhat previously failed when using conventional electrolytes.

. Experimental methods

.1. Protein preparation

Intact 2D HPI-layer protein sheets were prepared as previ-usly described [8], using methods adapted from the work ofaumeister et al. [14]. Briefly, A 1 L culture of D. radioduransARK (ATCC 35073) was grown to early stationary phase inGY medium (0.3% tryptone/0.1% glucose/0.5% yeast extract)nd washed in deionized water two times and finally resuspendedn 2% sodium dodecyl sulfate (SDS). The cell suspension washaken for 2 h in a rotary shaker at 60 ◦C. Cells were sedi-ented at 2200 × g for 15 min and discarded. The supernatantas centrifuged at 18,000 × g for 45 min and the protein pelletas resuspended in 5% SDS. This process was repeated untilo solid material was visible at the bottom of the tube after cen-rifugation at 2200 × g. SDS-polyacrylamide gel electrophoresis15] revealed a protein band at about 100 kDa as well as sub-ragments previously identified to be strain-specific cleavageatterns from an extracellular protease [16]. The final volume ofhe suspension was brought to 3 mL, corresponding to an esti-

ated S-layer protein concentration of 1 mg mL−1, which wasstimated by the Bradford assay [17] using bovine serum albu-in (BSA, Sigma–Aldrich) as a standard and a protein sample

esuspended in deionized water. The stock sample for all exper-ments was diluted into 5% SDS to about 0.25 mg mL−1 foroutine use.

.2. Protein adsorption

U

Please cite this article in press as: A. Presenda et al., Stability of S-layBiointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

Protein adsorption on surfaces was performed as previouslyescribed [8], and summarized here for the reader’s convenience.or atomic force microscopy (AFM), grade V-4 muskovite micaStructure Probe, Inc.) was mounted onto steel AFM pucks. The

fevS

PR

OO

F

PRESSB: Biointerfaces xxx (2007) xxx–xxx

ica surface was cleaved using Scotch® tape and immediatelyputter-coated in a Gatan PECS system with 10 nm of platinumased on in situ quartz crystal microbalance monitoring. Forransmission electron microscopy (TEM), platinum-coated goldEM grids (Structure Probe, Inc.) were used.

For AFM, 15–20 �L of stock protein suspension was used,hereas 2–3 �L was used for TEM imaging. The protein was

pplied on the surface by pipette and allowed to sit for about0–60 s. The surface was then gently rinsed by repeated immer-ion in deionized water and either dried under argon for AFMnalysis or wicked dry with a wipe for TEM experiments.

.3. Sample preparation

Test solutions of sulfuric acid or sodium hydroxide were pre-ared by adding concentrated (10 M) sulfuric acid or sodiumydroxide to deionized water. The pH was monitored using high-esolution pH indicator strips (colorpHast®, EM Science). Therotein-coated surface was exposed to a given test solution forpredetermined time and immediately blown dry under argon.he surface was then rinsed by repeated immersion in deionizedater and dried under argon.

.4. Atomic force microscopy (AFM)

AFM was performed on a Nanoscope® III AFM underapping Mode® conditions using amplitude feedback control.inear scans rates were typically 0.8–1.2 �m s−1. Probes wereluminum-coated Pointprobe® Plus (Nanosensors) SPM probesith a typical resonance frequency of 250–350 kHz and a nomi-al tip radius of less than 10 nm. Lateral size bars on images areased on calibrations with diffraction grating standards and areccurate to within 1%.

.5. Transmission electron microscopy (TEM)

TEM was performed on a Philips 420 TEM (Cs ∼ 1.3 mm)t 120 kV accelerating voltage using a tungsten filament. Nobjective aperture was used. Electron diffraction was used toerify the material’s crystal structure by simulating it with aowder diffraction analysis. The effective camera length wasalibrated by an aluminum foil standard. Size bars on imagesre based on calibrations with diffraction grating replicas andre estimated to be accurate to within 5%.

.6. Copper electrodeposition

Electrodeposition was performed in a quiescent, three-lectrode, single compartment cell operating at room temper-ture. The working electrode (cathode) was composed of therotein-adsorbed platinum-coated gold TEM grid held by self-losing anti-capillary tweezers at an acute angle to the electrolyte

er proteins for electrochemical nanofabrication, Colloids Surf. B:

oil angled nearly parallel to the working electrode. The refer- 135

nce electrode was a saturated calomel electrode (SCE) and all 136

oltages are reported as cathode potentials with respect to the 137

CE.

IN+ModelC

faces

138

a139

b140

a141

3142

143

a144

d145

o146

w147

“148

s149

o150

b151

w152

Ffrfw

s 153

m 154

“ 155

“ 156

157

( 158

w 159

o 160

t 161

m 162

i 163

t 164

fi 165

ARTICLEOLSUB 2970 1–6

A. Presenda et al. / Colloids and Sur

Electrolytes of two types were prepared: a 0.5 M sulfuriccid/0.5 M cupric sulfate electrolyte (pH ∼ 0) and a comparableath of 0.5 M sodium sulfate/0.5 M cupric sulfate, using sulfuriccid to adjust the pH to 3.

. Results and discussion

The S-layer protein of D. radiodurans (HPI) can be purifieds intact 2D protein “sheets”. The two faces of the sheet haveifferent characteristics. Upon adsorption to surfaces, either facef the sheet may be exposed; for labelling purposes, one faceill be referred to as face “A” (Fig. 1A) and the other as face

B” (Fig. 1B) in this article. Inspection of the images in Fig. 1

UN

CO

RR

EC

TED

Please cite this article in press as: A. Presenda et al., Stability of S-layBiointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

hows that, whereas the periodicity is the same for either “A”r “B” faces, the topography is greatly different. On a routineasis, the hexagonal arrangement of face “B” is rarely resolved,hile face “A” can in nearly all cases. Because the goal of this

ig. 1. Atomic force microscope image pair of S-layer protein cell wallragments from Deinococcus radiodurans adsorbed on platinum coated micaevealing the topography of both orientations. Protein sheets will have theiraces designated as face “A” (A) or face “B” (B). Proteins are about 5–6 nm highith respect to the platinum surface.

(tpi

FH(me

PRESSB: Biointerfaces xxx (2007) xxx–xxx 3

tudy is to evaluate long-range order, there was a real danger ofislabeling an intact protein sheet as disordered because face

B” was exposed to the tip. To avoid this error, a standardizedblind trial” protocol was developed (Fig. 2A and B).

The first step consists in scanning at low enough resolutionfewer scans along the slow scan axis) so that the pattern –hether it is present or not – cannot be observed, but the outlinef protein sheets are still visible with height steps correspondingo single, double, or additional protein layers (Fig. 2A). Having

ultiple height steps in the scan maximizes the likelihood thatn the field of view, both faces of the protein sheet are goingo be exposed to the tip. Once a protein fragment is identi-ed, it is selected as a candidate for a higher resolution scan

PR

OO

F

er proteins for electrochemical nanofabrication, Colloids Surf. B:

Fig. 2B). An important feature in the high-resolution scan is 166

he imaging of the platinum-coated mica background, which 167

rovides a reproducible 2–5 nm cobblestone structure. The plat- 168

num surface itself, therefore, serves as an internal resolution 169

ig. 2. Standardized protocol for a single imaging experiment in which adsorbedPI protein has been exposed to a sample electrolyte. A low-resolution scan

A) is used to search for a characteristic region containing protein sheets withultiple height information. A subsequent high-resolution scan (B) is used to

xamine protein pattern quality.

UN

CO

RR

EC

TED

PR

OO

F

Please cite this article in press as: A. Presenda et al., Stability of S-layer proteins for electrochemical nanofabrication, Colloids Surf. B:Biointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

ARTICLE IN PRESS+ModelCOLSUB 2970 1–6

4 A. Presenda et al. / Colloids and Surfaces B: Biointerfaces xxx (2007) xxx–xxx

Fig. 3. Sample AFM data illustrating how quality was assessed. Images of multiple protein sheets obtained (left) and region outlining a protein sheet (as indicatedin schematics on the right) were identified. Only sheets exposing face “A” can be used for evaluation except under conditions of totally disordered (0% quality). Ifall proteins in the same sheet are ordered the quality is ranked 100%. Intermediate values were ranked into quartiles by judging the size of the region containinghexagonal order relative to the size of the entire sheet exposing the same face. Conditions and the assigned quartile values (for the particular image shown) areindicated in the lower left of the image.

D

IN+ModelC

faces

s170

t171

I172

o173

p174

v175

t176

p177

178

a179

S180

o181

2182

a183

m184

e185

b186

t187

t188

r189

M190

v191

o192

a193

a194

r195

i196

197

p198

b199

t200

i201

o202

r203

p204

e205

l206

o207

r 208

p 209

h 210

u 211

a 212

213

c 214

s 215

p 216

F 217

s 218

p 219

F 220

m 221

a 222

t 223

p 224

o 225

t 226

c 227

a 228

229

e 230

s 231

f 232

t 233

w 234

c 235

p 236

0 237

d 238

a 239

T 240

t 241

Fi0n

TE

ARTICLEOLSUB 2970 1–6

A. Presenda et al. / Colloids and Sur

tandard. If the platinum background cannot be imaged clearly,he image is not used for analysis and the AFM tip is replaced.n summary, the low-resolution scan allows image collectionbjectively without bias towards obtaining high quality data ariori. The need to have multiple protein layers in the field ofiew is essential to provide proteins exposing both faces to theip, because one face could be easily mistaken for disorderedrotein.

Fig. 3 illustrates how the quality of a protein sheet wasssessed after exposure to a given experimental condition.hown are a series of sample AFM images taken after exposuref adsorbed HPI layer proteins to a sulfuric acid solution at pHfor 1, 100, and 1000 s. Proteins constituting a face are outlined

nd labelled as either face “A”, face “B”, or “?” if undeter-inable. Only face “A” can be reliably used to evaluate quality as

xplained above. The quality of the protein sheet was estimatedy ranking the images into quartiles from unflawed (100%) tootally disordered (0%). Intermediate values were estimated byhe amount of protein lattice structure which could be discernedelative to the size of the entire sheet exposing the same face.

ore data was taken in the important “boundary” zone whereariability was greatest. As protein sheets that are totally dis-rdered (quality = 0%) have no determinable face, 0% can bessigned only when there is a clear fold of a protein resulting indouble step height from a height trace across the sheet so that

egardless of which face is exposed, both faces must be presentn the same image.

Fig. 4 shows the “phase diagram” of the complete HPI layerrotein stability obtained from repeated experiments. The num-ers in the spaces represent the total number of samples used forhe particular test. Most data collection was focused on identify-ng the “boundary” of the stability envelope, where the transitionccurs from useful protein to useless protein. Once identified, theest of the diagram was filled in based on extrapolation or inter-

UN

CO

RR

EC

Please cite this article in press as: A. Presenda et al., Stability of S-layBiointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

olation around the well-established data. Additionally, controlxperiments with 1 M sodium sulfate showed that the loss ofong-range order primarily results from the acidity or basicityf the medium. Also, if ionic strength or a particular salt plays a

tiq

ig. 4. The D. radiodurans S-layer protein stability “envelope” for electrochemical fnsignificant or did not occur and were performed most repeatedly along the “bounda% nor 100%. Numbers in the squares indicate number of repetitive measurementseighboring values. Error should be considered to be within one quartile of the value

PR

OO

F

PRESSB: Biointerfaces xxx (2007) xxx–xxx 5

ole in the degree of order, this happens so in a manner that cou-les with the pH of the solution. A worthwhile point to mentionere is that, though this study did not extend out to many hours,nder all conditions tested, protein sheets can still be found inbundance on the surface, whether disordered or not.

Fig. 4 provided the information needed to reformulate theopper electrolyte to fall within the envelope of the HPI layertability. Of course, the pH value is not the only electrolyte com-onent that may impact the stability of a given protein; therefore,ig. 4 provides valuable guidance, though it does not ensureuccess. Only a specific study with all electrolyte componentsresent could test a “point” in electrolyte formulation space.ig. 5 shows TEM results before and after pH stability infor-ation was used to optimize the use of the HPI layer protein asmask for the electrochemical fabrication of copper nanopat-

erns. In the previous work [8], we have shown that the HPI layerrotein is an effective mask for the electrochemical fabricationf metal and metal oxide nanopatterns on surfaces. However, athat time it was not clear why the technique would not work foropper, an important material in the integrated circuit industry,nd perhaps the most well studied metal for electrodeposition.

In the original experiments (unpublished), the use of a simplelectrolyte consisting of 0.5 M sulfuric acid and 0.5 M cupriculfate electrolyte (a common copper electrolyte) consistentlyailed to produce nanostructured films of copper. However, underhese conditions, geometric patches of missing electron-densityere observed, suggesting that electrodeposition did not suc-

essfully proceed through the S-layer protein mask (Fig. 5A). AH test of the electrolyte revealed that the pH level was betweenand 1. Although the fabrication is complete within 3–5 s, the

ata from Fig. 4 indicates that the integrity of the S-layer proteint such a low pH is entirely lost in the first seconds of exposure.he original experiments with copper were performed without

he availability of Fig. 4 as a guide.

er proteins for electrochemical nanofabrication, Colloids Surf. B:

Fig. 4 provides a guideline for reformulating an electrolyte 242

o fall within the stability of the protein. This will be especially 243

mportant for simple metal electrolytes as they are generally 244

uite acidic. To verify that the loss of HPI order was responsible 245

abrication with respect to pH. Measurements were repeated until outliers wereries” of the stability envelope where quality of the protein crystal was neither

. Where numbers are missing, data was interpolated or extrapolated based ons shown, and pH should be considered accurate to within 1 unit.

EC

TED

ARTICLE IN+ModelCOLSUB 2970 1–6

6 A. Presenda et al. / Colloids and Surfaces

Fig. 5. Transmission electron micrographs of copper electrodeposited for 3 s at−300 mV vs. SCE through S-layer proteins of D. radiodurans adsorbed ontoplatinum-coated gold TEM grids using electrolytes (A) 0.5 M H2SO4/0.5 MCuSO4 at pH 0 and (B) 0.5 M Na2SO4/0.5 M CuSO4 adjusted to pH 3 withH SO , showing the lack of and the presence of hexagonally structured electrondef

f246

s247

w248

o249

c250

i251

t252

f253

w254

s255

b

i 256

r 257

o 258

e 259

c 260

4 261

262

l 263

a 264

o 265

i 266

o 267

o 268

T 269

v 270

h 271

r 272

r 273

A 274

275

S 276

F 277

( 278

U 279

R 280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

[ 297

298

[ 299

[ 300

[ 301

302

CO

RR2 4

ense material, respectively. Electron diffraction studies (inset of B) showed thelectron dense material to be copper (with a trace of cuprous oxide, as expectedor an oxidizable metal).

or the inability to fabricate copper nanopatterns, an electrolyteimilar to the original one was prepared, except that the protonsould be exchanged for sodium ions to keep the ionic strengthf the electrolyte constant, i.e., 0.5 M sodium sulfate and 0.5 Mupric sulfate adjusted to pH 3. This pH was chosen because its a good compromise between the performance of protein andhe electrochemistry where the protein should be able to survive

UN

Please cite this article in press as: A. Presenda et al., Stability of S-layBiointerfaces (2007), doi:10.1016/j.colsurfb.2007.02.011

or 100–1000 s of exposure, and copper is still highly solubleithout the need for complex coordination chemistry. Fig. 5B

hows that, after electrodeposition for the similar time frame asefore under the stated conditions, the hexagonal superstructure

[

[[[

PR

OO

F

PRESSB: Biointerfaces xxx (2007) xxx–xxx

s evident in the geometric patches. Electron diffraction studiesevealed the presence of crystalline copper (along with cuprousxide) in the electron dense material seen in the figure, which isvidence of an electrodeposition process as opposed to simpleopper staining or complexation with the protein sheets.

. Conclusions

The ability of the HPI layer of D. radiodurans to retainong-range order was investigated over a wide range of pHnd following 1–1000 s of exposure to the electrolyte. The aimf this study was to develop a rational approach to formulat-ng electrolytes for electrochemical fabrication of dense arraysf nanomaterials. We are well aware that the pH is not thenly electrolyte component that can impact protein stability.hus, while results such as those compiled in Fig. 4 can pro-ide valuable guidance, they do not ensure success. Our resultsighlight the fact that data from biochemical studies cannot beeadily extrapolated to protein-aided electrochemical nanofab-ication.

cknowledgments

This work was funded by Genetically Engineered Materialscience and Engineering Center (GEMSEC), a National Scienceoundation-Materials Research Science and Engineering CenterNSF-MRSEC) at the University of Washington, Seattle, WA,SA.

eferences

[1] K. Douglas, G. Devaud, N.A. Clark, Science 257 (1992) 642.[2] W. Pompe, M. Mertig, R. Kirsch, A.A. Gorbunov, A. Sewing, H. Engel-

hardt, A. Mensch, Proc. SPIE 2779 (1996) 72.[3] W. Shenton, D. Pum, U.B. Sleytr, S. Mann, Nature 389 (1997) 585.[4] M. Mertig, R. Kirsch, W. Pompe, H. Engelhardt, Eur. Phys. J. D 9 (1999)

45.[5] S.R. Hall, W. Shenton, H. Engelhardt, S. Mann, Chem. Phys. Chem. 2

(2001) 184.[6] R. Wahl, M. Mertig, J. Raff, S. Selenska-Pobell, W. Pompe, Adv. Mater.

13 (2001) 736.[7] L. Malkinski, R.E. Camley, Z. Celinski, T.A. Winningham, S.G. Whipple,

K. Douglas, J. Appl. Phys. 93 (2003) 7325.[8] D.B. Allred, M. Sarikaya, F. Baneyx, D.T. Schwartz, Nano Lett. 5 (2005)

609.[9] S.S. Mark, M. Bergkvist, X. Yang, L.M. Teixeira, P. Bhatnagar, E.R. Angert,

C.A. Batt, Langmuir 22 (2006) 3763.10] U.B. Sleytr, P. Messner, D. Pum, M. Sara, Crystalline Bacterial Cell Surface

Proteins, Academic Press, Austin, TX, 1996.11] H. Engelhardt, J. Peters, J. Struct. Biol. 124 (1998) 276.12] P. Lancy Jr., R.G.E. Murray, Can. J. Microbiol. 24 (1978) 162.13] W. Baumeister, F. Karrenberg, R. Rachel, A. Engel, B. Ten Heggeler, W.O.

Saxton, Eur. J. Biochem. 125 (1982) 535.

er proteins for electrochemical nanofabrication, Colloids Surf. B:

14] W. Baumeister, O. Kubler, H.P. Zingsheim, J. Ultrastruct. Res. 75 (1981) 303

60. 304

15] D.E. Garfin, Methods Enzymol. 182 (1990) 425. 305

16] B. Emde-Kamola, F. Karrenberg, FEMS Microbiol. Lett. 42 (1987) 69. 306

17] M.M. Bradford, Anal. Biochem. 72 (1976) 248. 307