Diffusion Of Transmembrane Proteins Research Articles

Semantic segmentation of anomalous diffusion using deep convolutional networks

Heterogeneous dynamics commonly emerges in anomalous diffusion with intermittent transitions of diffusion states but proves challenging to identify using conventional statistical methods. To effectively capture these transient changes of diffusion states, we propose a deep learning model (U-AnDi) for the semantic segmentation of anomalous diffusion trajectories. This model is developed with the dilated causal convolution (DCC), gated activation unit (GAU), and U-Net architecture. The study addresses two key subtasks related to trajectory segmentation and changepoint detection, concentrating on variations in diffusion exponents and dynamic models. Additionally, extended analyses are conducted on the segmentation of single-model trajectories, multistate biological trajectories, and anomalous diffusion with added correlation functions. By rationally designing comparative models and evaluating the performance of U-AnDi against these models, we discover that U-AnDi consistently outperforms other models across all segmentation tasks, thereby affirming its superiority in the field. This performance edge also sheds light on the interpretability of U-AnDi's core components: DCC, GAU, and U-Net. The clarity with which these components contribute to U-AnDi's success underscores their congruence with the intrinsic physics underlying anomalous diffusion. Furthermore, our model is examined using real-world anomalous diffusion data: the diffusion of transmembrane proteins on cell membrane surfaces, and the segmentation results are highly consistent with experimental observations. Our findings could offer a heuristic deep learning solution for the detection of heterogeneous dynamics in single-molecule/particle tracking experiments, and have the potential to be generalized as a universal scheme for time-series segmentation. Published by the American Physical Society 2024

Curvature-enhanced membrane asymmetry slows down protein diffusion

Diffusion of transmembrane proteins plays a vital role in various cellular processes, such as endocytosis, raft formation and signal transduction. Current understanding of protein diffusion dynamics in lipid membranes is mostly based on hydrodynamic analyses, in which lipid membrane is simply treated as a thin layer of viscous liquid, same for highly curved ones. Therefore, how the mechanical state of highly curved membranes may affect the transmembrane protein diffusion remains unclear. In this study, we employed molecular dynamics (MD) simulations to analyse membrane curvature effect on the diffusion of cylindrically shaped Aquaporin-0 (AQP0) protein. Slowing down of protein diffusion with the increase in membrane curvature was observed. The possible contributions from membrane tension and asymmetric pressure profile were systematically investigated by simulating the protein diffusion dynamics in planar membranes with various tension levels or with various lipid number ratios between the two leaflets, respectively. We found no significant effect of membrane tension on AQP0 diffusion. Instead, the asymmetric pressure profile present in highly curved bilayer membranes was identified as a key factor that contributes to the slowing down of transmembrane protein diffusion. Our work thus contributes to a more complete picture of transmembrane protein diffusion on highly curved biological membranes.

Field model for multistate lateral diffusion of various transmembrane proteins observed in living Dictyostelium cells.

The lateral diffusion of transmembrane proteins on plasma membranes is a fundamental process for various cellular functions. Diffusion properties specific for individual protein species have been extensively studied, but the common features among protein species are poorly understood. Here, we systematically studied the lateral diffusion of various transmembrane proteins in the lower eukaryote Dictyostelium discoideum cells using a hidden Markov model for single-molecule trajectories obtained experimentally. As common features, all membrane proteins that had from one to ten transmembrane regions adopted three free diffusion states with similar diffusion coefficients regardless of their structural variability. All protein species reduced their mobility similarly upon the inhibition of microtubule or actin cytoskeleton dynamics, or myosin II. The relationship between protein size and the diffusion coefficient was consistent with the Saffman-Delbrück model, meaning that membrane viscosity is a major determinant of lateral diffusion, but protein size is not. These protein species-independent properties of multistate free diffusion were explained simply and quantitatively by free diffusion on the three membrane regions with different viscosities, which is in sharp contrast to the complex diffusion behavior of transmembrane proteins in higher eukaryotes.

Development of Cell Membrane Electrophoresis to Measure the Diffusivity of a Native Transmembrane Protein.

The lateral diffusion of transmembrane proteins in cell membranes is an important process that controls the dynamics and functions of the cell membrane. Several fluorescence-based techniques have been developed to study the diffusivities of transmembrane proteins. However, it is challenging to measure the diffusivity of a transmembrane protein with slow diffusion because of the photobleaching effect caused by long exposure times or multiple exposures to light. In this study, we developed a cell membrane electrophoresis platform to measure diffusivity. We deposited cell membrane vesicles derived from HeLa cells to form supported cell membrane patches. We demonstrated that the electrophoresis platform can be used to drive the movement of not only a lipid probe but also a native transmembrane protein, GLUT1. The movements were halted by the boundaries of the membrane patches and the concentration profiles reached steady states when the diffusion mass flux was balanced with the electrical mass flux. We used the Nernst-Planck equation as the mass balance equation to describe the steady concentration profiles and fitted these equations to our data to obtain the diffusivities. The obtained diffusivities were comparable to those obtained by fluorescence recovery after photobleaching, suggesting the validity of this new method of diffusivity measurement. Only a single snapshot is required for the diffusivity measurement, addressing the problems associated with photobleaching and allowing researchers to measure the diffusivity of transmembrane proteins with slow diffusion.

Transport phenomena in fluid films with curvature elasticity

Abstract

Receptor Mobility is Regulated by the Cytoskeleton Connected to an Exoskeleton via Transmembrane Pickets: Role in Phagocytosis

Receptor mobility is regulated by the cytoskeleton connected to an exoskeleton via transmembrane pickets: role in phagocytosis. Phagocytosis is initiated by clustering of receptors upon exposure to target particles bearing multiple ligands. Clustering depends on the ability of receptors to move laterally in the plane of the membrane. Intrinsic proteins were originally thought to diffuse freely along the fluid mosaic of the membrane, but more recent observations indicate that mobility can be severely restricted by the existence of a cytoskeletal fence anchored to the plasmalemma via transmembrane “pickets”. However, the molecular nature of these putative pickets, the manner whereby they associate with the cytoskeleton, and their role in phagocytosis have not been investigated. Based on its abundance and structural features, we surmised that CD44 may serve as a picket in macrophages. We used single-molecule tracking to study its behavior and how it affects the mobility of phagocytic receptors. In addition, because its extracellular domain can bind hyaluronic acid, we found that CD44 serves as a transmembrane connector between the pericellular (glycocalyx) coat, which is akin to an exoskeleton, and the cytoskeleton. The pericellular coat contributes to limit the lateral diffusion of transmembrane proteins and, in addition, restricts access of target particles to phagocytic receptors, which are comparatively short molecules. The implications of this transmembrane structural frame to the phagocytic response will be illustrated and discussed.

Mechanics of a lipid bilayer subjected to thickness distension and membrane budding

We study the distension-induced gradient capillarity in membrane bud formation. The budding process is assumed to be primarily driven by diffusion of transmembrane proteins and acting line tensions on the protein-concentrated interface. The proposed model, based on the Helfrich-type potential, is designed to accommodate inhomogeneous elastic responses of the membrane, non-uniform protein distributions over the membrane surface and, more importantly, the thickness distensions induced by bud formations in the membrane. The latter are employed via the augmented energy potential of bulk incompressibility in a weakened manner. By computing the variations of the proposed membrane energy potential, we obtained the corresponding equilibrium equation (membrane shape equation) describing the morphological transitions of the lipid membrane undergoing bud formation and the associated thickness distensions. The effects of lipid distension on the shape equation and the necessary adjustments to the accompanying boundary conditions are also derived in detail. The resulting shape equation is solved numerically for the parametric representation of the surface which has one-to-one-correspondence with the membrane surface under consideration. The proposed model successfully predicts the bud formation phenomenon on a flat lipid membrane and the associated thickness distensions of the membrane and demonstrates a smooth transition from one phase to the other (including necking domains). It is also found that the final deformed configuration is energetically favorable and therefore is stable. Finally, we show that the inhomogeneous thickness deformation on the membrane in response to transmembrane protein diffusion makes a significant contribution to the budding and necking processes of the membrane.

Control of Transmembrane Protein Diffusion within the Postsynaptic Density Assessed by Simultaneous Single-Molecule Tracking and Localization Microscopy.

Postsynaptic transmembrane proteins are critical elements of synapses, mediating trans-cellular contact, sensitivity to neurotransmitters and other signaling molecules, and flux of Ca and other ions. Positioning and mobility of each member of this large class of proteins is critical to their individual function at the synapse. One critical example is that the position of glutamate receptors within the postsynaptic density (PSD) strongly modulates their function by aligning or misaligning them with sites of presynaptic vesicle fusion. In addition, the regulated ability of receptors to move in or out of the synapse is critical for activity-dependent plasticity. However, factors that control receptor mobility within the boundaries of the synapse are not well understood. Notably, PSD scaffold molecules accumulate in domains much smaller than the synapse. Within these nanodomains, the density of proteins is considerably higher than that of the synapse as a whole, so high that steric hindrance is expected to reduce receptor mobility substantially. However, while numerical modeling has demonstrated several features of how the varying protein density across the face of a single PSD may modulate receptor motion, there is little experimental information about the extent of this influence. To address this critical aspect of synaptic organizational dynamics, we performed single-molecule tracking of transmembrane proteins using universal point accumulation-for-imaging-in-nanoscale-topography (uPAINT) over PSDs whose internal structure was simultaneously resolved using photoactivated localization microscopy (PALM). The results provide important experimental confirmation that PSD scaffold protein density strongly influences the mobility of transmembrane proteins. A protein with a cytosolic domain that does not bind PSD-95 was still slowed in regions of high PSD-95 density, suggesting that crowding by scaffold molecules and perhaps other proteins is sufficient to stabilize receptors even in the absence of binding. Because numerous proteins thought to be involved in establishing PSD structure are linked to disorders including autism and depression, this motivates further exploration of how PSD nanostructure is created. The combined application PALM and uPAINT should be invaluable for distinguishing the interactions of mobile proteins with their nano-environment both in synapses and other cellular compartments.

Bud formation of lipid membranes in response to the surface diffusion of transmembrane proteins and line tension

We study the formation of membrane budding in model lipid bilayers with the budding assumed to be driven by means of diffusion of trans-membrane proteins over a composite membrane surface. The theoretical model for the lipid membrane incorporates a modified Helfrich-type formulation as a special case. In addition, a spontaneous curvature is introduced into the model in order to accommodate the effect of the non-uniformly distributed proteins in the bending response of the membrane. Furthermore, we discuss the effects of line tension on the budding of the membrane, and the necessary adjustments to the boundary conditions. The resulting shape equation is solved numerically for the parametric representation of the surface, which has one to one correspondence to the membrane surface in consideration. Our numerical results successfully predict the vesicle formation phenomenon on a flat lipid membrane surface, since the present analysis is not restricted to the conventional Monge representation often adopted to the problems of this kind for the obvious computational simplicity, despite its limited capability to describe the deformed configuration of membranes. In addition, we show that line tension at the interface of the protein-concentrated domain makes a significant contribution to the bud formation of membranes.

Confined diffusion of transmembrane proteins and lipids induced by the same actin meshwork lining the plasma membrane.

The mechanisms by which the diffusion rate in the plasma membrane (PM) is regulated remain unresolved, despite their importance in spatially regulating the reaction rates in the PM. Proposed models include entrapment in nanoscale noncontiguous domains found in PtK2 cells, slow diffusion due to crowding, and actin-induced compartmentalization. Here, by applying single-particle tracking at high time resolutions, mainly to the PtK2-cell PM, we found confined diffusion plus hop movements (termed "hop diffusion") for both a nonraft phospholipid and a transmembrane protein, transferrin receptor, and equal compartment sizes for these two molecules in all five of the cell lines used here (actual sizes were cell dependent), even after treatment with actin-modulating drugs. The cross-section size and the cytoplasmic domain size both affected the hop frequency. Electron tomography identified the actin-based membrane skeleton (MSK) located within 8.8 nm from the PM cytoplasmic surface of PtK2 cells and demonstrated that the MSK mesh size was the same as the compartment size for PM molecular diffusion. The extracellular matrix and extracellular domains of membrane proteins were not involved in hop diffusion. These results support a model of anchored TM-protein pickets lining actin-based MSK as a major mechanism for regulating diffusion.

Single Particle Tracking in Double Cushioned, Blebbed Supported Lipid Bilayers Enables Studies of Transmembrane Protein Diffusion

Supported Lipid Bilayers (SLB's) are effective models for studying some biomembrane phenomena. A thin layer of water between the substrate and the bilayer engenders 2D fluidity and enables studies of lipid diffusion and peripheral membrane protein diffusion. However, the water layer is not thick enough to prevent friction between most transmembrane proteins and the substrate. Because of this, it is difficult to study the diffusive properties of proteins that protrude significantly from the membrane. Here, we describe a double cushioning SLB platform that is easy to construct and supports the mobility of most transmembrane proteins. The platform makes use of streptavidin to passivate the substrate and PEGylated lipids (with biotin attached) to act as a cushion. The bilayer can be formed by vesicle fusion, allowing us to incorporate membrane proteins from cell blebs without using detergents or other artifactual methods. Atomic force microscopy images of each layer in this SLB reveal heterogeneities in the lateral distribution and height of the cushioning components. Single particle tracking of fluorescently tagged DHHC20, a ∼40 kDa acyltransferase with 4 transmembrane domains, demonstrates fluidity of large multi-pass membrane proteins. Detailed analysis of trajectories shows the effects of the cushion heterogeneities on the diffusive properties of DHHC20. We will discuss the biological implications of these results and the applications of this platform to studying cytoskeleton-mediated confinement of plasma membrane components. We will briefly discuss a model for the mechanism of cytoskeletal confinement based on hydrodynamic interactions with immobilized membrane proteins.

Visualizing the Compartmentalization of the Surface of Mammalian Cells by Cortical Actin with Superresolution

The plasma membrane is a complex fluid where proteins move by diffusion, a process that is critical to biochemical reactions. Despite the membrane fluidity, distinct compartments transiently form on the cell surface, suggesting the plasma membrane maintains a dynamic ordered environment. Single-molecule methods show that membrane proteins exhibit transient confinement mediated by the actin cytoskeleton. However, direct visualization of the membrane compartmentalization by the underlying cortical actin structure is experimentally challenging because of the need for spatial resolution beyond the diffraction limit and high temporal resolution. In order to overcome these challenges, we employ dynamic photoactivated localization microscopy (PALM). We image the cortical actin with 40-nm resolution for continuous periods longer than one minute, while we simultaneously track individual membrane proteins that interact with the actin cytoskeleton. Kv2.1 and Kv1.4 ion channels are labeled with quantum dots to investigate the effect of cortical actin compartmentalization on protein diffusion using single-particle tracking. We find that individual ion channels are confined within compartments formed by the cortical actin for times up to several seconds. Nevertheless, molecules are observed to permeate through actin barriers at locations where the cytoskeleton is retracted from the plasma membrane. The analysis of compartment morphological properties reveals a broad distribution of compartment sizes. Further, the imaged actin cortex appears to form a fractal structure, which explains the observed subdiffusion of various membrane proteins over very broad time scales. Our data provide evidence showing that the restriction to lateral mobility by an actin random fractal is one of the fundamental physical mechanisms for the anomalous diffusion of transmembrane proteins.

Imaging and quantification of trans-membrane protein diffusion in living bacteria.

The cytoplasmic membrane forms the barrier between any cell's interior and the outside world. It contains many proteins that enable essential processes such as the transmission of signals, the uptake of nutrients, and cell division. In the case of prokaryotes, which do not contain intracellular membranes, the cytoplasmic membrane also contains proteins for respiration and protein folding. Mutual interactions and specific localization of these proteins depend on two-dimensional diffusion driven by thermal fluctuations. The experimental investigation of membrane-protein diffusion in bacteria is challenging due to their small size, only a few times larger than the resolution of an optical microscope. Here, we review fluorescence microscopy-based methods to study diffusion of membrane proteins in living bacteria. The main focus is on data-analysis tools to extract diffusion coefficients from single-particle tracking data obtained by single-molecule fluorescence microscopy. We introduce a novel approach, IPODD (inverse projection of displacement distributions), to obtain diffusion coefficients from the usually obtained 2-D projected diffusion trajectories of the highly 3-D curved bacterial membrane. This method provides, in contrast to traditional mean-squared-displacement methods, correct diffusion coefficients and allows unravelling of heterogeneously diffusing populations.

Biochips for Cell Biology by Combined Dip‐Pen Nanolithography and DNA‐Directed Protein Immobilization

A general methodology for patterning of multiple protein ligands with lateral dimensions below those of single cells is described. It employs dip pen nanolithography (DPN) patterning of DNA oligonucleotides which are then used as capture strands for DNA-directed immobilization (DDI) of oligonucleotide-tagged proteins. This study reports the development and optimization of PEG-based liquid ink, used as carrier for the immobilization of alkylamino-labeled DNA oligomers on chemically activated glass surfaces. The resulting DNA arrays have typical spot sizes of 4-5 μm with a pitch of 12 μm micrometer. It is demonstrated that the arrays can be further functionalized with covalent DNA-streptavidin (DNA-STV) conjugates bearing ligands recognized by cells. To this end, biotinylated epidermal growth factor (EGF) is coupled to the DNA-STV conjugates, the resulting constructs are hybridized with the DNA arrays and the resulting surfaces used for the culturing of MCF-7 (human breast adenocarcinoma) cells. Owing to the lateral diffusion of transmembrane proteins in the cell's plasma membrane, specific recruitment and concentration of EGF receptor can be induced specifically at the sites where the ligands are bound on the solid substrate. This is a clear demonstration that this method is suitable for precise functional manipulations of subcellular areas within living cells.

Diffusion and Interaction Dynamics of Individual Membrane Protein Complexes Confined in Micropatterned Polymer‐Supported Membranes

Micropatterned polymer-supported membranes (PSM) are established as a tool for confining the diffusion of transmembrane proteins for single molecule studies. To this end, a photochemical surface modification with hydrophobic tethers on a PEG polymer brush is implemented for capturing of lipid vesicles and subsequent fusion. Formation of contiguous membranes within micropatterns is confirmed by scanning force microscopy, fluorescence recovery after photobleaching (FRAP), and super-resolved single-molecule tracking and localization microscopy. Free diffusion of transmembrane proteins reconstituted into micropatterned PSM is demonstrated by FRAP and by single-molecule tracking. By exploiting the confinement of diffusion within micropatterned PSM, the diffusion and interaction dynamics of individual transmembrane receptors are quantitatively resolved.

Bim-FCS Analysis of Membrane Protein Diffusion Reveals Dynamics of Membrane Cytoskeleton and Lipid Domains in Intact Cells

Signaling transmembrane proteins are experiencing a heterogeneous structure due to membrane cytoskeleton meshwork and selective association with cholesterol stabilized lipid domains throughout the cell membrane. The limited optical resolution due to diffraction limits direct observation of both cytoskeleton structure and cholesterol stabilized micro-domains.Fluorescence correlation spectroscopy (FCS) allows to measure membrane protein diffusion, but does not resolve the heterogeneity. We analyze the diffusion of GFP-tagged membrane proteins in multiple areas of increasing size simultaneous using TIRF illumination and camera based FCS. The binned-imaging FCS (bim-FCS) allows distinguishing free Brownian diffusing proteins from proteins interacting with the membrane cytoskeleton and from proteins transiently entering sub-microscopic domains of reduced mobility. Using bimFCS it is possible to study how subtle changes modulate the structural nanodomains or the functional interaction with them, and how these changes propagate with time.We investigate effect of membrane meshwork on diffusion of transmembrane proteins and on GPI anchored proteins which associate with cholesterol stabilized micro-domains in intact cells. We show in real-time how the protein-membrane ultrastructure interactions are modulated by protein dimerization or lipid cross-linking. We perform Monte Carlo simulations of bimFCS experiments with square pixels to analyze the effect of cytoskeleton barriers and cholesterol domains on the diffusion. We are able to quantitatively model the changes induced by protein or lipid cross-linking.

The neuronal K-Cl cotransporter KCC2 influences postsynaptic AMPA receptor content and lateral diffusion in dendritic spines

The K-Cl cotransporter KCC2 plays an essential role in neuronal chloride homeostasis, and thereby influences the efficacy and polarity of GABA signaling. Although KCC2 is expressed throughout the somatodendritic membrane, it is remarkably enriched in dendritic spines, which host most glutamatergic synapses in cortical neurons. KCC2 has been shown to influence spine morphogenesis and functional maturation in developing neurons, but its function in mature dendritic spines remains unknown. Here, we report that suppressing KCC2 expression decreases the efficacy of excitatory synapses in mature hippocampal neurons. This effect correlates with a reduced postsynaptic aggregation of GluR1-containing AMPA receptors and is mimicked by a dominant negative mutant of KCC2 interaction with cytoskeleton but not by pharmacological suppression of KCC2 function. Single-particle tracking experiments reveal that suppressing KCC2 increases lateral diffusion of the mobile fraction of AMPA receptor subunit GluR1 in spines but not in adjacent dendritic shafts. Increased diffusion was also observed for transmembrane but not membrane-anchored recombinant neuronal cell adhesion molecules. We suggest that KCC2, likely through interactions with the actin cytoskeleton, hinders transmembrane protein diffusion, and thereby contributes to their confinement within dendritic spines.

A model for surface diffusion of trans-membrane proteins on lipid bilayers

The equilibrium theory of lipid membranes is modified to include the effects of a continuous distribution of trans-membrane proteins. These influence membrane shape and evolve in accordance with a diffusive balance law. The model is purely mechanical in the absence of the proteins. Conditions ensuring energy dissipation in the presence of diffusion are given and an example constitutive function is used to simulate the coupled inertia-less interplay between membrane shape and protein distribution. The work extends an earlier continuum theory of equilibrium configurations of composite lipid-protein membranes to accommodate surface diffusion.

Polymer-Supported Membranes for Probing Transmembrane Protein Diffusion and Interaction by Single-Molecule Techniques

Polymer-supported membranes (PSM) are valuable models for cell membranes, since they lack the complexity of a cellular system and confine all membrane components in a two-dimensional space that can be investigated by a multitude of surface-sensitive spectroscopic techniques. However, functional reconstitution of transmembrane proteins into a PSM that show lateral diffusion in the membrane is challenging. Here, we present an approach for efficiently incorporating proteins into tethered PSMs. For this purpose, we have employed a hydrophilic, inert polymer cushion (polyethylene glycol) as a spacer between a glass surface and the membrane. The polymer was functionalized with lipophilic anchors for capturing of proteoliposomes to the surface. Incubation with a polyethylene glycol solution induced fusion of the bound vesicles into an extended bilayer. Efficient incorporation of single-spanning transmembrane proteins into these PSMs was achieved by their reconstitution into very small lipid vesicles, which were readily fused on the surface. Mobility of proteins and lipids in the membrane was demonstrated by fluorescence recovery after photobleaching that showed a mobile fraction of ∼80 % for transmembrane proteins and full mobility of lipids. Single molecule tracking experiments confirmed free diffusion of lipids and reconstituted membrane proteins. For the transmembrane proteins, we obtained diffusion constants of ∼0.6-0.8 µm2/s that are 5-fold lower than those obtained for lipids. Application of the system for functional analysis was shown by measuring the interaction of the transmembrane receptor IFNAR2 reconstituted into PSMs with its soluble ligand interferon-α2 by ensemble binding experiments and single-molecule co-diffusion analysis. Thus, we have established a versatile experimental platform to study mobility and interactions of transmembrane proteins in a defined membrane environment by ensemble and single-molecule techniques.

Electron spin resonance and biochemical studies of the interaction of the polyamine, spermine, with the skeletal network of proteins in human erythrocyte membranes

Spermine ( N, N−bis(aminopropyl)-1,4-butanediamine) is a polyamine thought to be important in several cell regulatory processes. Previous studies had shown that spermine prevented the lateral diffusion of transmembrane proteins in human erythrocyte ghosts (Schindler et al. (1980) Proc. Natl. Acad. Sci. USA 77, 1457–1461). In this paper, we present results of studies on the effect of spermine on erythrocyte membranes by employing electron spin resonance spin-labeling techniques in conjunction with spin labels specific for skeletal proteins, bilayer lipids or cell-surface sialic acid of the membrane and by employing SDS-polyacrylamide gel electrophoresis analysis of extracted spectrin and Triton shells. The major findings are: (1) spermine significantly decreases the segmental motion of protein spin-label binding sites ( P < 0.0001), which are predominantly on cytoskeletal proteins; (2) addition of spermine leads to a significant increase in the rotational motion of spin-labeled terminal sialic acid residues ( P < 0.001), most of which are located on glycophorin A, a result which may be secondarily caused by spermine-induced aggregation of cytoskeletal proteins and the cytoplasmic pole of this transmembrane sialoglycoprotein; (3) spermine completely inhibits the low-ionic strength extraction of spectrin, the major protein of the skeletal network which is attached to the bilayer proteins by two or more connecting proteins; (4) pretreatment of ghosts with spermine followed by Triton extraction resulted in the retention of significantly increased amounts of Band 3 and other skeletal and bilayer proteins including Bands 4.2, 6 and 7 in Triton X-100 shells relative to that of control-treated ghosts. These results suggest that spermine acts both to increase protein-protein interactions in the cytoskeletal protein network and to bridge skeletal and bilayer proteins and are discussed with reference to possible molecular mechanisms by which spermine may influence cell functions.