Selective Diffusion Barriers Separate Membrane Compartments

Abstract

We have investigated exchange of molecules between different membrane domains on a highly compartmentalized cell, the spermatozoon. Using Alexa Fluor 555-cholera toxin B-subunit we have observed clustering of preexisting GM1 gangliosides which diffused across the anterior acrosome-equatorial segment interface but did not access the postacrosome. By contrast, single lipid and protein molecules readily exchanged between all three domains, although they diffused more slowly on nearing and crossing to the postacrosome. Thus, two types of diffusion interfaces are present on sperm heads, an “open” interface and a “mass filter” interface. The latter seems to be due to a protein-cytoskeleton network.

The plasma membrane of polarized cells is characteristically compartmentalized into domains of various sizes that reflect specialized function. At the nanometer level are lipid “rafts” and “corrals” which are thought to be in dynamic equilibrium with the surrounding bilayer and confine molecules for only seconds or less (1). At the other end of the scale are micrometer-sized macrodomains, which are stable throughout the lifetime of the cell, and may or may not be delineated by physical structures within the membrane (2). Understanding how different domains remain distinct from one another and whether or not there is exchange of molecules between them, are major problems in membrane biology.

Cholera toxin β-subunit (CTXB) is a high affinity probe for GM1 gangliosides, which are commonly held to partition preferentially into lipid raft-like structures (1). Being pentameric, CTXB can cross-link GM1 molecules to form clusters. It has been argued (4) that if GM1 gangliosides existed in the membrane as single molecules then CTXB-induced clusters would not be sufficiently large to be visible. If, on the other hand, GM1 gangliosides were present in preexisting complexes, such as lipid rafts, then cross-linking them with CTXB would induce formation of even larger and more easily detectable structures.

Previously, we have used Alexa Fluor 555-CTXB to visualize GM1 gangliosides on live boar sperm ((5) and also see Supporting Material). Here we also observed formation of bright clusters (∼0.3–1.2 μm diameter) that diffused throughout the anterior head (known as the anterior acrosome (AAc) and equatorial segment (EqS) domains; see Fig. 1 A , and Movie S1 in the Supporting Material). Thus, under steady-state conditions, raft-like structures are present in the plasma membrane of live boar sperm and are susceptible to cross-linking by multivalent probes such as CTXB. We performed mean-square displacement (MSD) analysis of the cluster trajectories. This showed that diffusion was random with a mean of D = 0.10 ± 0.01 μm²/s (Fig. 1, B–D), with some deviation from linearity at longer times (see Supporting Material for details). Significantly, these diffusing clusters exchanged readily between the AAc and EqS domains but did not diffuse across the boundary between the EqS and postacrosome (PAc), suggestive of a diffusion barrier between the latter domains but not the former. This is contrary to the evidence from surface-imaging techniques, which show clearly discernable structural boundaries among all three domains (6). The simplest explanation for these findings is that the EqS-PAc domains are separated by a high-energy diffusion barrier whereas no such barrier exists between the AAc and EqS.

Figure 1.

Analysis of the diffusion of Alexa Fluor 555-CTXB clusters between membrane domains on the head of boar spermatozoa. Sperm heads are shown in outline with domains as marked. (A) Representative trajectories of Alexa Fluor 555-CTXB clusters diffusing over AAc and EqS. Clusters exchange readily between the AAc and EqS. Trajectories begin with dark blue to light blue, then transition to dark green to light green, and finally orange to red. Color changes every 10 s. (B) Overlay of all recorded trajectories (70) to illustrate that clusters do not cross from the EqS to the PAc; color indicates density of tracks in a region. (C and D) MSD analysis and frequency distribution of D values.

The number of molecules in the CTXB clusters is likely to be ∼10³, assuming close packing and an effective radius of 3.35 nm for CTXB gives 5000 molecules in a 500-nm diameter cluster (5). To investigate whether individual molecules were similarly restricted in their distribution, the diffusion of single protein and lipid molecules was then measured and analyzed using Atto 647-WGA (which binds to N-acetylglucosamine and sialic acid sugars on membrane glycoproteins) and Atto 647-DOPE (which inserts into the outer leaflet of the bilayer).

Single molecules of Atto 647-WGA were applied to the center of the EqS using a nanopipette and single trajectories video recorded and analyzed (3). Results showed that labeled glycoproteins diffused with equal probability in an anterior or posterior direction and were unrestricted in exchanging between the EqS and PAc or between the EqS and AAc (Fig. 2 and Movie S2). Consistent with our earlier observations (3), those that crossed onto the PAc diffused more slowly (D = 0.26 ± 0.02 μm²/s) than those on the EqS (D = 0.48 ± 0.02 μm²/s) or AAc (D = 0.43 ± 0.02 μm²/s). For Atto 647-DOPE, trajectories were selected that originated within the EqS and were followed until the fluorophore bleached. These trajectories were separated into a mobile (D > 0.05 μm²/s, 61% of trajectories) and an immobile fraction (D < 0.05 μm²/s), as observed previously (10). Only the mobile fraction was analyzed in the different domains. They were observed to diffuse across the boundary onto the PAc domain (Fig. 2 and Movie S3) where they had a mean D value of 0.5 ± 0.1 μm²/s, comparable to that on the AAc (0.48 ± 0.04 μm²/s) and EqS (0.36 ± 0.04 μm²/s). Collectively, these experiments indicate that single lipid and glycoprotein molecules exchange freely between the AAc, EqS, and PAc domains.

Figure 2.

Data for single molecules of Atto 647-WGA lectin and Atto 647-DOPE on the sperm head. (A) The lectin was applied by nanopipette to the center of the EqS domain. Trajectories readily cross the both the AAc-EqS and EqS-PAc junctions. Color scheme is same as in Fig. 1A. (B) Overlay of all trajectories (1138); color indicates density of tracks in a region. (C) Frequency distribution of D values calculated from MSD curves and (D) averaged MSD curve. (E) Representative trajectories for Atto 647-DOPE. (F) Overlay of all trajectories (411) illustrating that the probe exchanges across both the AAc-EqS and EqS-PAc junctions. (G) Frequency distribution of D values calculated from MSD curves and (H) mean MSD curves for mobile (red) and immobile (blue) DOPE molecules.

To understand more about the nature of the putative barrier between the EqS and PAc domains, trajectories of the above-mentioned probes were reanalyzed when they entered a border region within 1 μm of the boundary on the EqS side. Results showed that the D value of Alexa Fluor 555-CTXB-labeled clusters decreased from a mean of 0.095 ± 0.001 μm²/s in the anterior portion of the EqS to 0.015 ± 0.005 μm²/s in the border region (Fig. S1). Single molecule diffusion was also slower close to the boundary. For Atto 647-WGA, diffusion coefficients were D = 0.51 ± 0.02 μm²/s in the EqS and 0.24 ± 0.03 μm²/s in the border area, and for Atto 647-DOPE, 0.35 ± 0.04 μm²/s in the EqS and 0.23 ± 0.04 μm²/s in the border area (Fig. S1). These results suggest that the boundary between the EqS and PAc is actually a zone with a finite width instead of a line interface between, for example, two lipid phases. Furthermore, the comparable diffusion coefficients for WGA and DOPE in these different regions show that the organization of the sperm membrane, not molecular size, controls the rate of diffusion.

Various dissociating reagents, including the cytoskeleton-disrupting agent Latrunculin A, were used to perturb the EqS-PAc barrier but all were unsuccessful (Table S1), indicating unusually high stability and suggesting that diverse submembranous elements may be present. In boar sperm, the membrane at the EqS-PAc junction is characterized by a raised zone of intramembrane particles (7), a close association with the underlying perinuclear theca (8), and a band of F-actin that persists even after Latrunculin A treatment (Fig. 3). Staining for F-actin is especially strong throughout the PAc domain whereas it is undetectable in the EqS (Fig. 3). The concept of protein-cytoskeleton networks creating stable domains within a membrane is supported by an increasing number of single molecule tracking studies (9–11). Extending this concept to spermatozoa, a model more in keeping with the present data is shown in Fig. 3. In this scenario, a barrier is effectively created at the EqS-PAc junction by a high density of transmembrane glycoproteins (blue hexagons) in the PAc that are cross-linked via the actin skeleton to form a protein-protein network. Single lipids and proteins (red trajectories) percolate slowly through the network while large complexes (green trajectory) are sterically hindered and excluded from entering the PAc. A border zone of decaying glycoprotein density on the EqS side would account for the slower diffusion of single lipids and proteins as they approach the interface.

Figure 3.

Distribution of antibody-labeled F-actin in control and Latrunculin A-treated sperm (see Supporting Material). (A) In control sperm actin staining is strongest in the PAc region with a weaker signal on the apical ridge. No staining is apparent in the EqS. (B) Latrunculin A pretreatment causes loss of actin staining throughout the sperm head except for a thin layer at the junction between the EqS and PAc. Scale bar = 4 μm. (C) Proposed model for barrier function between EqS and PAc (see text).

In summary, by comparing diffusion of cross-linked lipid-raft-like clusters with that of single lipid and protein molecules, we have discovered significant differences between interfaces separating three membrane compartments on the same cell. One interface appears open, allowing free exchange of molecules of all sizes (and presumably different charge), whereas the other is selective and behaves as a mass filter slowing down but allowing the crossing of lipids and single proteins. Barriers using the same biophysical principles are also likely to be present on other specialized cells. We hypothesize that the former is akin to a lipid phase separation (see Supporting Material for justification) whereas the latter is created by a dense cytoskeletal network that effectively immobilizes transmembrane glycoproteins. Functionally selective diffusion barriers between membrane compartments on spermatozoa would enable them to assemble and retain molecular complexes in the appropriate place at the appropriate time for recognition and fusion with the egg.

Supporting Material

Methods and materials, one table, one figure, and three movies are available at http://www.biophysj.org/biophysj/supplemental/S0006-3495(10)00472-8.