J. Phys. Chem. B 2007, 111, 12133-12135
12133
Molecular Orientation of Membrane-Anchored Mucin Glycoprotein Mimics Raghuveer Parthasarathy,*,†,| David Rabuka,† Carolyn R. Bertozzi,†,‡,§ and Jay T. Groves*,†,‡ Department of Chemistry, UniVersity of California, Berkeley, California 94720, Physical Bioscience and Materials Science DiVisions, Lawrence Berkeley National Laboratory, Berkeley, California 94720, and Department of Molecular and Cell Biology, and Howard Hughes Medical Institute, UniVersity of California, Berkeley, California 94720 ReceiVed: March 16, 2007; In Final Form: July 20, 2007
Mucin glycoproteins contribute to a wide range of cell-surface phenomena. Their dense glycosylation is believed to confer structural rigidity as well as molecular extension beyond the glycocalyx, crucial to interaction with the cellular environment. However, controlled investigations of the relationships between glycosylation, rigidity, and extension of membrane-bound mucins or similar macromolecules are lacking, largely because of the absence of tractable experimental models. We have therefore made use of recently developed synthetic mucin mimetics, in which the core R-GalNAc monosaccharides of natural mucins are conjugated to a lipidated polymer backbone and anchored to fluid, solid-supported lipid membranes, and fluorescence interference contrast microscopy, an optical technique that provides nanometer-scale topographic information about objects near a reflective interface, to measure the orientation of the mucin mimics relative to the membrane plane. Data from two independent probes, fluorophores conjugated directly to the polymer backbone and fluorescent proteins bound to the sugar groups, unexpectedly show that the mucin mimic molecules lie flat along the membrane. Rigidity and core glycosylation are therefore insufficient to ensure molecular projection outward from a membrane surface.
Introduction
Experimental Methods
The physical characterization of membrane proteins, especially in environments similar to their natural contexts, remains challenging for a variety of reasons. Protein sizes are large enough to put them beyond the reach of standard chemical synthesis techniques (techniques that are also hindered by the molecules’ often high degree of glycosylation) and are too small for conventional optical imaging to provide structural information. We have initiated an experimental program that circumvents these difficulties, determining the spatial orientation of chemically synthesized, membrane-anchored polymeric mimics of cell-surface mucin glycoproteins using powerful interferometric techniques. Mucins are characterized by dense O-linked glycans that initiate with N-acetyl galactosamine (GalNAc) conjugated to serine or threonine residues on the amino acid backbone.1 As secreted molecules, mucins are well known as contributors to the rheology of mucous, tears, and other fluids. However, mucins also exist as membrane-anchored molecules, sometimes as domains of other proteins.2,3 It is thought that their dense glycosylation confers structural rigidity, allowing proteins to extend out from the glycocalyx where they mediate interactions with other molecules, viruses, or cell surfaces. Despite this prevalent picture of mucins towering over the peri-cellular landscape, little is known about the structure and orientation of mucins at membrane surfaces.
To explore the generic physical properties of mucins, we made use of recently developed, chemically synthesized, lipidanchored mucin mimics.4 The mimics (Figure 1) reproduce the O-R-linkage of the core GalNAc sugar characteristic of all mucins, but substitute the amino acid backbone with a synthetic polymer (methyl vinyl ketone, MVK), allowing the rapid production of large glycosylated macromolecules (mean molecular weight 35 kDa). One end of the molecule is lipidated, allowing insertion into lipid bilayer membranes. A small fraction (1%) of the monomer units are occupied by a Texas Red fluorophore, conjugated to the polymer backbone concurrently and by similar chemical mechanisms as the sugars to provide a random distribution of fluorescent probes along the macromolecular length; the fraction of Texas Red is verified by NMR.4 The synthesis yields rod-like mucin mimics with an average length of approximately 30 nm.4 The hydrophobic tail allows incorporation into solid-supported lipid bilayers, formed on SiO2 substrates in phosphate buffered saline by standard vesicle fusion techniques.5 The bilayers are composed of 95 mol % DOPC (1,2-dioleoylsn-glycero-3phosphocholine), a zwitterionic lipid, and 5% DOTAP (1,2dioleoyl-3-trimethylammonium-propane), a cationic lipid analogue (Avanti Polar Lipids, Alabaster, AL). The positive charge from the DOTAP cancels the negative surface potential of the silica substrate, presenting an electrostatically neutral environment.6 Following its assembly, the bilayer is incubated with mucin mimics (∼20 µg/mL, 15 h), unbound polymer is washed away, and the resulting membrane is imaged by epifluorescence microscopy. Analysis of the fluorescence intensity reveals a sample-to-sample variation of mucin mimic densities in the range 100-6000 molecules/µm2, typically around 500 molecules/ µm2; the orientation values given below are independent of
* Correspondingauthors.E-mail:
[email protected](R.P.),
[email protected] (J.T.G.). † Department of Chemistry, University of California. ‡ Lawrence Berkeley National Laboratory. § Department of Molecular and Cell Biology, and Howard Hughes Medical Institute, University of California. | Present address: Department of Physics and Materials Science Institute, University of Oregon, Eugene, OR 94703-1274.
10.1021/jp072136q CCC: $37.00 © 2007 American Chemical Society Published on Web 10/04/2007
12134 J. Phys. Chem. B, Vol. 111, No. 42, 2007
Parthasarathy et al.
Figure 1. Lipid-anchored mimics of mucin glycoproteins. R-linked GalNAc (n ≈ 300) and Texas Red fluorophores are conjugated to a methyl vinyl ketone (MVK) polymer backbone. The phospholipid tail allows incorporation into lipid membranes.
density. Control experiments with mucin polymers incubated with bare-glass (notoriously “sticky” to many macromolecules) with no supported lipid membrane yielded a much lower polymer density, around 20 molecules/µm2, indicating specific anchoring of the polymers to the membrane. The molecular mobility was measured to be 3 µm2/s, identical to that of membrane lipids, consistent with a lipid anchor.4 To probe the orientation of the membrane-anchored mucin mimics, that is, whether they project perpendicular to the membrane surface, lie flat along it, or adopt some conformation in between, we use fluorescence interference contrast microscopy (FLIC), an optical technique capable of extracting spatial information in the few to few hundred nanometer range.7-10 The lipid-plus-mucin-mimic membrane is assembled on a SiO2 layer on a reflective silicon substrate. Interference of the fluorescence excitation light, as well as the emitted light, with its reflection from the substrate leads to height-dependent fluorescence intensity (Figure 2). Accounting for the optical parameters of the setup, the excitation and emission spectra of the fluorophore (peaked at 595 and 615 nm, respectively, for Texas Red and 495/520 nm for FITC), the bandpasses of the filters (530-580 nm and 605-675 nm excitation and emission ranges for Texas Red, and 455-500/510-560 for FITC, using filters from Chroma Technology, Rockingham, VT), the reflectivity of the silicon substrate (effective index of refraction 10.7), the numerical aperture of the microscope objective (0.45), the indices of refraction of SiO2 and water (1.46 and 1.33, respectively), and the thickness of the SiO2 terraces (described below), one can extract topographic information from the fluorescence data with few nanometer precision.7-10 All of these parameters are either well-known, for example, indices of refraction, or directly measured using standard materials characterization equipment, for example, silicon reflectivity and oxide thickness values. (We neglect the slight reflectivity of the bilayer itself.) Despite its precision, FLIC has rarely been used to probe molecular orientation at surfaces.8,11 The interferometric nature of FLIC relates the observed fluorescence intensity (I) to the height of the fluorophore relative to the reflective plane (z); both the oxide spacer and the molecules of interest contribute to z (Figure 2a). Providing multiple oxide spacer levels of known thickness reveals the shape of I(z), the analysis of which reveals the height of the fluorophores above the oxide (see below).7,8 For this reason, we microfabricated using standard lithographic techniques SiO2/ Si substrates with arrays of 16 oxide terraces with intended thicknesses distributed between 0 and 225 nm. In brief, photoresist on a substrate with oxide thickness D was exposed to ultraviolet light under a patterned mask, developed, and then the substrate was etched with hydrofluoric acid. Four repetitions of this procedure with different mask orientations and etch amounts yielded 16 levels spaced by approximately D/16. Variability in the acid etch rate limits the precision with which
Figure 2. (a) FLIC imaging of mucin mimics anchored to solidsupported lipid bilayer membranes. Optical interference leads to heightdependent fluorescence intensity, the analysis of which reveals the mucin mimic orientation. Two independent probes, Texas Red fluorophores conjugated to the polymer backbone (left, orange asterisks) and lectin proteins that bind R-GalNAc (right, green), provide topographic information. (b) Fluorescence image of Texas-Red-labeled lipid-membrane-anchored mucin mimics on a terraced oxide substrate. (There are no fluorescent lipid probes present.) Superimposed on each terrace is its measured oxide thickness (nm). (c) Fluorescence intensity as a function of oxide thickness (circles), fitted (curve) to extract the mean fluorophore height. (d) The interferometric data reveal that, rather than projecting perpendicular to the membrane, the mucin mimic molecules lie flat along it.
the actual terrace heights match the desired values. It is therefore important to subsequently measure the oxide thicknesses, which was done with subnanometer accuracy via reflectometry using a NanoSpec Film Thickness Measurement System (Nanometrics, Milpetas, CA). Marks made during the lithographic process allowed exact alignment of the characterized chip regions with the membrane areas observed by fluorescence microscopy. FLIC data consisted of background-subtracted fluorescence intensities of the fluorophores of interest on oxide terraces of various (measured) thicknesses. Using custom software written using MATLAB (Matworks, Natick, MA), we calculated intensity versus z relations, where the fluorophore height z equals the oxide thickness plus the mean height of the fluorophores distributed randomly throughout the mucin macromolecule. Because the mucin length, 30 nm, is considerably smaller than the periodicity of the FLIC intensity I(z), ∼200 nm, the extended distribution of fluorophores is well-approximated by their mean position, as was verified by calculations of FLIC intensities of extended objects. The physics of the FLIC model are discussed in detail in ref 10. For each sample, we identified the mucin height as that which best fit the measured data using standard ξ2 minimization (i.e., the height that minimizes the mean square difference between the measured and calculated intensity values). The only free parameters in the fit were the mucin height and an overall (irrelevant) fluorescence intensity. The
Membrane-Anchored Mucin Glycoprotein Mimics uncertainties noted in the following section are the standard deviations over several samples, the number being stated in the text. Results and Discussion Fluorescence images of membrane-anchored R-GalNAcMVK show discrete levels of intensity (Figure 2b), corresponding to the oxide terraces raising the membrane by different heights. The mucin mimics project from the oxide surface a distance, or a range of distances, z, to be determined. Fitting the fluorescence intensity versus the oxide thickness (Figure 2c), we find a value zR ) 4.5 ( 1.2 nm (averaged over N ) 17 samples) for the R-linked GalNAc-MVK mucin mimics. (The only fit parameter besides z is an irrelevant overall fluorescence intensity.) Given that the bilayer itself has a thickness of 4-5 nm and is separated from the oxide by a 1-3 nm thick hydration layer,7,12 the value of zR implies that the mucin molecules are lying flat on the membrane surface. Separate FLIC imaging of Texas Red labeled lipids in supported bilayers gives the value zbilayer ) 4.4 ( 1.6 nm (N ) 7), the same height, within uncertainties, as the mucin mimics. Experiments on mucin mimics with β-linked GalNAc gave zβ ) 6.5 ( 1.5 nm (N ) 12), similar to the R-linked molecules. Combining the data from both R- and β-linked samples, zmucin ) 5.3 ( 1.0 nm. We further characterized the mucin mimic orientation using a second, independent fluorescent probe. Lectins are a class of proteins that bind specific carbohydrates.13 The soluble FITClabeled lectin Helix Pomatia Agglutinin (HPA) (EY Laboratories, San Mateo, CA), known to multivalently bind R-linked GalNAc,14 was filtered through a 0.2 µm syringe filter and incubated with membrane-incorporated R-GalNAc-MVK at 10 µg/mL for 1 h; unbound lectins were removed by washing. The lectins’ specificity with respect to the carbohydrate linkage was verified by comparison of supported lipid membranes with Rand β-linked GalNAc polymers, described in detail in ref 4. Under identical conditions, lectin density (as measured by FITC fluorescence) was 8 times lower for the β-linked GalNAc mucin mimics as compared to the R-linked molecules. In addition, there was no detectable lectin biding (less than 100 times the density, as compared to the R-linked molecules) to lipid membranes in the absence of incorporated mucin mimic polymers. HPA is a large hexamer, about 360 nm3 in volume,14 and so it intrinsically contributes to an elevated topographic height signal. (The FITC fluorophores are conjugated to residues distributed throughout the 79 kDa lectin.) FLIC data from FITCHPA bound to mucin mimics give a height value of zHPA ) 18.0 ( 1.7 nm (N ) 34); nonspecifically adsorbed to bare SiO2 it gives zHPA,oxide ) 7.7 ( 1.8 nm (N ) 15). The bilayer plus mucin mimics therefore elevate the lectins by 10.3 ( 2.5 nm, consistent with the height of the supported bilayer plus the few nanometer width of the mucin mimic molecules, and not their 30 nm length. The topographic information provided by the lectins therefore confirms that the mucin mimics are lying flat on the membrane. FLIC imaging of the Texas Red label on the mucin mimics after lectin binding gives zmucin ) 5.5 ( 1.7 nm (N ) 15), that is, unchanged from its value without the lectins present. If the mucin mimic molecules were projecting perpendicular to the membranes, FLIC would reveal a mean height of zmucin ) L/2 + z0 ≈ 21 nm, where L ≈ 30 nm is the length of the GalNAc-MVK and z0 ≈ 6 nm is provided by the supported bilayer plus its hydration layer, well above the measured value of zmucin ) 5.3 ( 1.0 nm. Our measurements also rule out the zmucin ) L/4 + z0 ≈ 13 nm height that would result if the
J. Phys. Chem. B, Vol. 111, No. 42, 2007 12135 molecules were freely fluctuating between flat and normal orientations. We can estimate the difference in entropy (∆S) between this case, in which each molecule explores a full hemisphere of orientations, and the observed configuration, in which a wedge at most a few nanometers in height is accessible, as being roughly equal to the logarithm of the ratios of the solid angles spanned by the rod-like molecules. This gives ∆S ≈ 2. Attractive interactions between the mucin mimic and the lipid bilayer (whether enthalpic or from solvent entropy) must therefore be greater than 2 kBT per molecule, where kB is Boltzmann’s constant and T is the absolute temperature. Per sugar, this amounts to less than 0.01 kBT; a weak interaction is sufficient to tie the molecules to the membrane surface. The data show, surprisingly, that the rigidity conferred by densely packed GalNAc monosaccharides on a polymer backbone is insufficient to enable spatial projection from a lipid bilayer membrane. We stress that the significant structural differences between these glycosylated polymers and natural cell surface glycoproteins prohibit simple extrapolation between our experiments and the native cellular environment. Our conclusions do indicate, however, that the adoption of particular confirmations cannot be assumed on the basis of chemical structure, but must be experimentally determined. The physical mechanisms that control the spatial conformation of cell surface proteins need to be subjected to quantitative biophysical scrutiny. The experimental platform described above provides a starting point for such investigations. Of particular interest for future studies are delineating the roles of the macromolecular surface density, branching, and elaboration with higher order glycans including charged monosaccharides. Acknowledgment. All substrates were fabricated at the UC Berkeley Microfabrication Laboratory. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract DEAC03-76SF00098, and by the National Institutes of Health (GM59907). References and Notes (1) Strous, G. J.; Dekker, J. Crit. ReV. Biochem. Mol. Biol. 1992, 27, 57-92. (2) Kuchroo, V. K.; Umetsu, D. T.; DeKruyff, R. H.; Freeman, G. J. Nat. ReV. Immunol. 2003, 3, 454-462. (3) Rosen, S. D. Am. J. Pathol. 1999, 155, 1013-1020. (4) Rabuka, D.; Parthasarathy, R.; Lee, G. S.; Chen, X.; Groves, J. T.; Bertozzi, C. R. J. Am. Chem. Soc. 2007, 129, 5462-5471. (5) Boxer, S. G. Curr. Opin. Chem. Biol. 2000, 4, 704-709. (6) Parthasarathy, R.; Cripe, P. A.; Groves, J. T. Phys. ReV. Lett. 2005, 95, 048101. (7) Lambacher, A.; Fromherz, P. J. Opt. Soc. Am. B 2002, 19, 14351453. (8) Kiessling, V.; Tamm, L. K. Biophys. J. 2003, 84, 408-418. (9) Parthasarathy, R.; Groves, J. T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12798-12803. (10) Parthasarathy, R.; Groves, J. T. Cell Biochem. Biophys. 2004, 41, 391-414. (11) Lambacher, A.; Fromherz, P. J. Phys. Chem. B 2001, 105, 343346. (12) Bayerl, T. M.; Bloom, M. Biophys. J. 1990, 58, 357-362. (13) Paulson, J. C.; Blixt, O.; Collins, B. E. Nat. Chem. Biol. 2006, 2, 238-248. (14) Sanchez, J. F.; Lescar, J.; Chazalet, V.; Audfray, A.; Gagnon, J.; Alvarez, R.; Breton, C.; Imberty, A.; Mitchell, E. P. J. Biol. Chem. 2006, 281, 20171-20180.
