Atomic Force Microscopy in Cell Biology Episode 2 Part 5 pptx

The outlined BR monomer represents a section close to the cytoplasmic surface of the lipid membrane, and the positions of the transmembrane α-helices A to F were obtained after merging s

Fig 5 Imaging a polypeptide loop grafted onto bacteriorhodopsin (A) Secondary structural model of

bacteriorhodopsin from Halobacterium salinarum (B) The polypeptide loop connecting transmembrane

α-helices E and F (loop EF) of the bacteriorhodopsin molecule was replaced by loop EF from bovine rhodopsin

to produce the mutant IIIN (indicated by the filled circles, the numbering in the boxes give the residue in IIIN first and the residue in the rhodopsin loop second) V8 protease cleaves this loop after the glutamates indicated

by the arrows (see Fig 6) (C) Topograph of the mutant IIIN containing the rhodopsin EF loop (D) Threefold symmetrized correlation average of mutant IIIN trimer (E) Standard deviation map of the average (F)

Three-fold symmetrized correlation average of the bacteriorhodopsin trimer imaged elsewhere (M¨uller, Sass et al.,

1999) (G) Standard deviation map of the bacteriorhodopsin trimer The outlined BR monomer represents

a section close to the cytoplasmic surface of the lipid membrane, and the positions of the transmembrane

α-helices A to F were obtained after merging six atomic models of BR (Heymann et al., 1999) Vertical

brightness range of contact mode topographs corresponds to 1.8 nm Minima and maxima of the SD maps were 0.32 and 0.43 nm for (E) and 0.07 and 0.19 nm for (G) respectively.

13 Atomic Force Microscopy and Spectroscopy of Membrane Proteins 267

rhodopsin (Rho EF) also interacts with rhodopsin kinase, which phosphorylates light-activated rhodopsin, and with arrestin, which displaces transducin from light-light-activated phosphorylated rhodopsin To directly observe the rhodopsin loop, purple membrane containing the mutant bacteriorhodopsin (called IIIN) was imaged by AFM under phys-iological conditions to a resolution of 0.7 nm (Fig 5C) It was found that the modifi-cation of loop EF changed neither the crystallographic lattice nor the extracellular surface

(Heymann et al., 2000) This was not unexpected, because fragments of bacteriorhodopsin

separated in the EF loop can be reconstituted with the bacteriorhodopsin chromophore

(Kataoka et al., 1992; Liao et al., 1984; Sigrist et al., 1988) Thus, the bacteriorhodopsin

framework was not affected by the loop replacement which provided a stable founda-tion for studying the Rho EF loop The major difference in the topographs between the cytoplasmic surfaces of the mutant and bacteriorhodopsin purple membrane is the much larger EF loop projecting toward the C-terminus (Figs 5D and 5F)

C Identification by Removal of a Polypeptide Loop

From Fig 5 it is clear that structural changes of membrane proteins induced by the replacement of individual polypeptide loops can be directly observed by AFM under physiological conditions Alternatively, individual loops can be removed to identify structural features of membrane protein surfaces

Digestion of the rhodopsin loop EF from mutated IIIN bacteriorhodopsin (Fig 5B) with V8-protease did not affect the purple membrane crystallinity (Fig 6A) The AFM topograph showed a significant reduction in the major protrusion compared to the undi-gested surface (Fig 6B) and the ends of helices E and F became clearly visible This structural change was consistent with mass spectrometry indicating that a 10-residue

fragment of loop Rho EF had been removed (Heymann et al., 2000) Interestingly, AFM

topographs of purple membrane did not show any indication of the largest polypeptide residue located at the cytoplasmic surface, the C-terminus (24 aa), and the simplest interpretation is that this is too unstructured to allow imaging

D Identification by Removal of Polypeptide Ends

This alternative method to identify the surfaces of membrane proteins is illustrated by the selective cleavage of terminal polypeptide sequences of aquaporin Z (AqpZ) from

E coli (Scheuring, Ringler et al., 1999) and of major intrinsic protein (MIP) from sheep lenses (Fotiadis et al., 2000).

Figure 7A shows the AFM topograph of recombinant AqpZ tetramers reconstituted into a bilayer and assembled into a two-dimensional (2D) crystal The AqpZ had an N-terminal fragment of 26 amino acids located on the cytoplasmic surface After over-night treatment with trypsin, the N-terminal fragment had been removed from the protein

at the trypsin cleavage site Arg26 While the uncleaved sample allowed only one AqpZ surface to be imaged, the digested sample clearly showed substructures of the cytoplasmic (circle) and extracellular (square) AqpZ surface (Fig 7B)

Topology prediction and antibody labeling of MIP places the approximately 5-kDa C-terminal region on the cytoplasmic surface of the lens fiber cell membrane The native

Fig 6 Imaging bacteriorhodopsin after removal of the EF loop (A) Topograph of the bacteriorhodopsin surface after cleavage of the rhodopsin EF loop with V8 protease For cleavage sites of the V8 protease, see Fig 5B The topograph is displayed as a relief tilted by 5 ◦ (B) Threefold symmetrized average of the

bacteriorhodopsin trimer imaged in (A) (C) Standard deviation map of the average Vertical brightness range

of contact mode topographs corresponds to 1 nm Minima and maxima of the SD map was 0.1 and 0.17 nm.

cytoplasmic surface of MIP tetramers, reconstituted into a bilayer and assembled into a 2D crystal, exhibited maximum globular protrusions of 0.8 ± 0.1 nm (Fig 7C) After

removal of the C-terminal tail with carboxypeptidase Y the cytoplasmic surface changed its appearance (Fig 7D) The cytoplasmic surface appeared coarser, and the averaged structure revealed the partial loss of four prominent protrusions leaving a central cavity within the MIP tetramer This structural change is emphasized by the difference map (Fig 7F) calculated between the unit cell of the digested (Fig 7D) and of the native (Fig 7E) cytoplasmic MIP surface It is important to note that neither the extracellular surface of AqpZ nor that of MIP appeared to be structurally affected by the enzymatic digestion of the cytoplasmic surface

13 Atomic Force Microscopy and Spectroscopy of Membrane Proteins 269

Fig 7 Identification of protein structures by removal of polypeptide ends (A) Identifying the cytoplasmic surface of aquaporinZ (AqpZ) Topograph of AqpZ tetramers assembled into a 2D crystal In this crystal form, each tetramer is neighbored by four tetramers oriented in the opposite direction with respect to the membrane plane B, The same crystal imaged after trypsin digestion Since trypsin removes most of the 26-amino-acid long C-terminal region, the surface structure of the cytoplasmic surface changed drastically The extracellular surface of the AqpZ tetramers was unchanged by this treatment (circles) Vertical brightness range of contact mode topographs corresponds to 3 nm Topographs are displayed as reliefs tilted by 5 ◦ (images courtesy

of Simon Scheuring, University of Basel) (C) Imaging the removal of the C-terminal region of the major intrinsic protein (MIP) Averaged topograph of the cytoplasmic MIP surface imaged in buffer solution The MIP tetramers from sheep eye lenses were reconstituted into a lipid bilayer where they assembled into a 2D

crystal (Fotiadis et al., 2000) The same unit cell (white square) contained one MIP tetramer and had a side

length of 6.4 nm (D) Unit cell of the cytoplasmic MIP surface after removal of the C-terminal region (E) Unit cell of the native cytoplasmic MIP surface (F) Difference map calculated between topographs of digested (D) and of native MIP revealing the location of the C-terminal regions of the cytoplasmic surface; major protrusion Vertical brightness range of contact mode topographs corresponds to 1 nm (images courtesy of Dimitrios Fotiadis, University of Basel).

From the previous AFM measurements it follows that the conformation of the AqpZ N-terminal fragment (26 amino acids) was too flexible to be imaged with subnano-meter resolution (blurred protrusion) but was structurally sufficiently stable to distort the scanning stylus, thereby preventing the visualization of other substructures on the cytoplasmic surface The C-terminal fragment of MIP existed in a structurally more stable conformation and was imaged by the AFM stylus In contrast, the C-terminal

region of bacteriorhodopsin, consisting of 25 amino acids, is not observed by AFM and does not influence the visualization of surrounding substructures by AFM (compare

to Fig 6) The results illustrate that the polypeptide ends of distinct proteins exist in conformations of different stability; the conformation of the C-terminal region of MIP

is stable enough to be reproducibly imaged at subnanometer resolution; the N-terminal region of AqpZ is structurally less stable than the C-terminal end of MIP but more stable

than the disordered C-terminal domain of bacteriorhodopsin (Belrhali et al., 1999; Essen

et al., 1998; Grigorieff et al., 1996; Heymann et al., 1999; Luecke et al., 1999b; Mitsuoka

et al., 1999) which is not detected by AFM (M¨uller, Sass et al., 1999) Accordingly, a

structural change caused by the cleavage of a polypeptide can only be observed by the AFM if it could be reproducibly detected before of its removal

IV Observing the Oligomerization of Membrane Proteins

α-Hemolysin is a water-soluble protein that undergoes several conformational changes from the time it is released from Staphylococcus until it interacts with a plasma

mem-brane Initially hemolysin is a monomer, which undergoes oligomerization into a homo-oligomeric ring finally inserting into the lipid bilayer forming a pore Interestingly, this pore which facilitates water permeation across the membrane can be genetically

en-gineered to sense a range of different organic molecules (Gu et al., 1999) For some years

it has been discussed whether the hemolysin oligomer exists in a hexameric or in a hepta-meric stoichiometry since different techniques have shown different oligohepta-meric states

of the complex (Gouaux et al., 1994; Song et al., 1996) Avoiding problems which may

arise determining the oligomeric stoichiometry of proteins imaged by diffracting tech-niques, the high signal-to-noise ratio of the AFM allows subunits of protein complexes

to be imaged directly and their oligomeric stoichiometry to be determined

Staphylococcalα-hemolysin inserted into phospholipid bilayers is shown in Fig 8 The

AFM topograph shows unambiguously the hexameric state of the oligomeric complex

which assembled into a two-dimensional array (Czajkowsky et al., 1998) Interestingly,

additional data have been recently published onα-hemolysin mutants locked in their open state (Malghani et al., 1999) In contrast to the previously described data, the mutants appear to exist in a heptameric stoichiometry As already pointed out (Czajkowsky et al.,

1998), it may be possible that α-hemolysin may form stable oligomers which differ

in their stoichiometry However, it remains a challenge to determine those factors that influence the oligomeric state of biochemically indistinguishableα-hemolysins.

ATP synthases are large protein complexes that convert the energy of a transmembrane proton (or Na+) gradient into the biological energy source ATP Its integral membrane complex Fo(∼170 kDa) couples the transmembrane flow of protons to the rotation of a

molecular stalk (Kato-Yamada et al., 1998; Noji et al., 1997; Sabbert et al., 1996) The

rotational force expels the spontaneously formed ATP from the three catalytic sites of the water-soluble F1 complex (∼400 kDa) While the catalytic subcomplex α3β3γ as well

the isolated subunitsδ and ε of F have been solved to atomic resolution (Abrahams et al.,

13 Atomic Force Microscopy and Spectroscopy of Membrane Proteins 271

Fig 8 Contact mode AFM topograph ofα-hemolysin oligomers The α-hemolysin inserted into the

phos-pholipid bilayer in 10 mM sodium phosphate (pH 7.2) at room temperature and assembled into a 2D lattice.

Raw image displayed as relief tilted by 5 ◦ The scale bar represents 7.5 nm Topograph displayed as relief

tilted by 5 ◦(image courtesy of Daniel Czajkowsky, University of Virginia).

1994; Bianchet et al., 1998; Wilkens et al., 1995; Wilkens et al., 1997), the structure of the

Focomplex still awaits elucidation To gain insight into the mechanochemical coupling synthesizing ATP, the arrangement of the transmembrane Focomplex assembled from subunits I1, II1, IIIx, and IV1in chloroplast ATP synthase (CFoF1) or from subunits a1,

b2, cxin bacterial and mitochondrial ATP synthase (EFoF1) is a matter of investigation Several subunits, IIIx and (cx), form the “proton turbine” of the ATP synthase The mechanism determining the exact number of subunits, IIIx, however, is a topic of debate and remains to be answered

Atomic force microscopy of the IIIxoligomer of the most abundant ATP synthase from chloroplast revealed the surface at a sufficient resolution to allow the number of III sub-units to be counted Thus, topographs of the reconstituted cylindrical complex assembled from subunit III of the chloroplast FoF1-ATP synthase provided compelling evidence that

this proton-driven turbine comprises 14 subunits (Fig 9) (Seelert et al., 2000) This find-ing is in contrast to the stoichiometry of the E coli Focomplex which is postulated to be

a dodecamer of subunit c, mainly based on crosslinking experiments (Jones et al., 1998), genetic engineering (Jones and Fillingame, 1998), and model building (Dmitriev et al.,

1999; Groth and Walker, 1997; Rastogi and Girvin, 1999) Interestingly, X-ray analyses

of yeast FF -ATP synthase crystals yielded a decameric complex (Stock et al., 1999),

Fig 9 Proton-driven rotor of the chloroplast ATP synthase reconstituted into a membrane bilayer As shown

by this unprocessed topograph, this cylindrical oligomer comprises 14 subunits (Seelert et al., 2000) The dense

packing of oligomers required an alternating orientation vertical to the membrane plane Thus, the distinct

wide and narrow rings represent the two surfaces of the cylindrical complex Imaging buffer: 25 mM MgCl2 ,

10 mM Tris–HCl, pH 7.8 Vertical brightness range of contact mode topograph corresponds to 2 nm The raw

image is displayed as a relief tilted by 5 ◦.

indicating that polymorphic stoichiometries of Fo complexes may have a biological origin which is not yet understood

V Unraveling the Conformational Variability

of Membrane Proteins

A Force-Induced Conformational Changes

The cytoplasmic bacteriorhodopsin surface, imaged with a force of 100 pN applied to the AFM stylus, revealed trimeric structures arranged in a trigonal lattice of 6.2 ± 0.2 nm side length (Fig 10A, top; (M¨uller, Sass et al., 1999) Each subunit in the trimer features

a particularly pronounced protrusion extending 0.83± 0.19 nm above the lipid surface This protrusion is associated with the loop connectingα-helices E and F (Fig 10B; (M¨uller, Buldt et al., 1995)) Increasing the applied force to about 200 pN during imaging

changed the AFM topographs significantly The prominent EF loops were bent away and the shorter loops of the bacteriorhodopsin monomers were visualized (Figs 10A, bottom;

and 10D) This conformational change was fully reversible (M¨uller, Buldt et al., 1995),

Fig 10 Force-induced conformational change of the cytoplasmic purple membrane surface (A) At the top of the topograph the force applied to the AFM stylus was 100 pN While scanning the surface line by line, the force was increased until it reached 150 pN at the bottom of the image This

force-induced conformational change of bacteriorhodopsin was fully reversible (M¨uller, B¨uldt et al., 1995) Correlation averages of the cytoplasmic

surface recorded at 100 pN (B) and at 200 pN (D) The correlation averages are displayed in perspective view (top, shaded in yellow brown) and in top view (bottom, in blue) with a vertical brightness range of 1 nm and exhibited 9.2% (B) and 14.1% (D) RMS deviations from threefold symmetry Structural flexibilities were accessed by SD maps (C and E corresponding to B and D, respectively) which had a range from 0.08 (lipid) to 0.19 nm (EF loop region) Surface regions exhibiting a SD above 0.12 nm are superimposed in red-to-white shades in top of figure (B and D) The contact

mode topograph was recorded in buffer solution (100 mM KCl, 10 mM Tris–HCl, pH 7.8) The outlined bacteriorhodopsin trimer representing sections close to the cytoplasmic surface of the lipid membrane was obtained after merging six atomic models of bacteriorhodopsin (Heymann et al., 1999).

Topographs (A), (B), and (E) are displayed as relief tilted by 5 ◦.

suggesting that loop EF is a rather flexible element of the bacteriorhodopsin molecule At this force of 200 pN, the maximum height difference between the protein and the lipid membrane was 0.64 ± 0.12 nm Four distinct protrusions were recognized in almost every monomer, and a further distinct protrusion was present at the center of the trimers The calculated diffraction pattern of this topograph documents an isotropic resolution out to 0.45 nm (not shown)

While the standard deviation of the height measurements was around 0.1 nm for most morphological features of the topography, the EF loop exhibited an enhanced

SD of 0.19 nm (Fig 10C), consistent with the high-temperature factor observed by

electron microscopy (Grigorieff et al., 1996) and the structural variation among the atomic bacteriorhodopsin models (Heymann et al., 1999) When the major protrusion

representing loop EF had been pushed away by applying a force of 200 pN to the stylus, the cytoplasmic surface of the bacteriorhodopsin molecule appeared different and exhibited details of the shorter loops connecting helices AB and CD (Fig 10D) The protrusion between helices F and G together with the minor elevation between helices E and F likely represents what remained structured from loop EF and the protrud-ing parts of helices E and F that are compressed by the AFM stylus (Fig 10D) However,

it cannot be excluded that the protrusion between helices F and G included a small part

of the C-terminal domain This uncertainty arises because the AFM height signal in this area exhibited a significant standard deviation (Fig 10E; red shaded in Fig 10D) The other protrusions in the AFM topograph may be assigned by comparison with the atomic models derived from the bacteriorhodopsin trimer (see following section) In these mod-els, helix B protrudes out of the bilayer, and helix A ends below the bilayer surface Therefore, the protrusion close to helix B is likely to represent the short loop connecting helices A and B (Fig 10D) In addition, the discrete protrusion between helices C and D corresponds to their connecting loop A further protrusion of 0.2 nm height was present

at the threefold axis of the bacteriorhodopsin trimer and probably arises from structured

lipid molecules (Grigorieff et al., 1996).

To further analyze the conformations of the cytoplasmic surface, the unit cells of topographs recorded at applied forces of 100 and 200 pN were extracted, aligned with

respect to a reference, and classified by principal component analysis (Frank et al., 1987;

van Heel, 1984) The threefold symmetrized averages of the major classes shown in Figs 11A to 11E reveal the movement of the flexible structures The classes A, B, and

C, D were closely related to the force gradient Increasing the force to 120 pN resulted in

a slight deformation of the EF loop and enhanced the details of the surrounding protein structure (Fig 11A; compare to Fig 10B) Increasing the force to approximately 150 pN further pushed the EF loop away (Fig 11B), whereas at about 180 pN the conformational change of the loop was complete (Figs 11C to 11E) A central protrusion was apparent

in some bacteriorhodopsin trimers when imaged at 180 pN (Figs 11C and 11E) Most probably, this protrusion represented lipid headgroups which were absent or disordered

in some bacteriorhodopsin trimers Increasing the applied force to 300 pN resulted in deformation of the peripheral protrusions of the trimer The structural information of

these areas was lost (M¨uller et al., 1998), and when imaged at applied forces above

300 pN the bacteriorhodopsin trimers were irreversibly deformed (data not shown)

13 Atomic Force Microscopy and Spectroscopy of Membrane Proteins 275

Fig 11 Structural variability of the cytoplasmic bacteriorhodopsin surface The threefold symmetrized aver-ages were calculated from unit cells classified by multivariate statistical analysis using the algorithm kindly provided by J.-P Bretaudiere (Bretaudiere and Frank, 1986) (A) PM imaged at slightly enhanced forces of

120 pN (compare to Fig 11B) (B) Same membrane imaged at an applied force of approximately 150 pN In (C), (D), and (E) three conformations of the membrane are imaged at approximately 180 pN The last three averages differ in their central protrusion and in that of the EF loop (compare to Fig 10) The correlation averages are displayed in perspective view with a vertical brightness range of 1 nm Topographs are displayed

as relief tilted by 5 ◦.

VI Comparing AFM Topographs to Atomic Models

A Comparing Topographs of OmpF Porin to the Atomic Model

OmpF porin is present as stable trimeric structures in the outer membrane of E coli.

Each 340-amino acid (aa) OmpF monomer is folded into 16 antiparallelβ-strands to

form a large hollowβ-barrel structure which perforates the membrane The

transmem-brane pore facilitates the passages of hydrophilic solutes up to an exclusion size of

≈600 kDa (Nikaido and Saier, 1992) It is suggested that the charges of the porin chan-nel primarily modulate the pore selectivity (Klebba and Newton, 1998; Schirmer, 1998), and recent calculations have shown that the OmpF pore may establish an electrical po-tential which increases with decreasing electrolyte concentration (Schirmer and Phale, 1999) As observed in the AFM topograph of the periplasmic surface (Fig 12A), each trimer (outlined circle) is compromised of tripartite protrusions and three transmem-brane channels that are separated by 1.2-nm-thick walls The transmemtransmem-brane channel

has a characteristic elliptical cross section of a = 3.4 nm and b = 2.0 nm The arrows

point out individual polypeptide loops of a few aa size each connecting two antiparallel ß-strands lining the transmembrane pore Most features recorded in this AFM topograph

were correlated directly to the atomic model of the OmpF (Cowan et al., 1992)

sur-face rendered at 0.3 nm resolution (Fig 12B) Correlation averaging of the porin trimer enhanced common structural details among individual trimers (Fig 12C) but blurred vari-able areas of the subdomains (compare to porin trimers shown in raw data, Fig 12A) However, the characteristic shape of the transmembrane channel was more pronounced

showing an elliptical cross section of a = 3.4 nm and b = 2.0 nm Structural areas of

the periplasmic OmpF trimer surface exhibiting an enhanced variability are recovered

by the standard deviation map of the average (Fig 12D) Enhanced values of the SD map can directly correlate to surface structures which are expected to have enhanced flexibility (Fig 12B)