The role of PDZ domain proteins in targeting, anchoring, and stabilizing membrane channels and receptors has been investigated using a variety of approaches such as peptide overlays, yeast two-hybrid assays, surface plasmon resonance, and cell transfection experiments. Many PDZ-ligand interactions have been well characterized in vitro, including AMPA receptors, which interact with the PDZ domain proteins GRIP (Dong et al., 1997
) and PICK-1 (Xia et al., 1999
); the cystic fibrosis transmembrane receptor, which binds the PDZ domain proteins CAP-70 (CFTR-associated protein, 70 kDA) (Wang et al., 2000
) and the Na+
exchanger regulatory factor (Raghuram et al., 2001
); and the PDZ domains of postsynaptic density protein-95 and other members of the MAGUK family of proteins, which bind to NMDA receptor (Kornau et al., 1995
) and to nNOS (Brenman et al., 1996
). However, recent findings from our lab suggest that at least in some cases PDZ-ligand interactions that occur in vitro may not occur in vivo. The PDZ domains of α, β1, and β2 syntrophin have been shown to bind nNOS with similar affinity in vitro (Gee et al., 1998b
), but in skeletal muscle of the α-syntrophin–null mouse β1 and β2 syntrophin do not associate with nNOS (Adams et al., 2000
). The demonstration of in vivo specificity of PDZ domains has prompted us to use in vivo approaches to investigate the physiological relevance of PDZ domain/ligand interactions.
Some PDZ-ligands have been characterized in vivo using invertebrate models. In Caenorhabditis elegans
, three PDZ proteins, Lin-2, Lin-7, and Lin-10, mediate the localization of the LET-23 receptor (Kaech et al., 1998
). This receptor fails to localize to the basolateral surface when the PDZ proteins are mutated. Mutational studies in Drosophila
show that the PDZ domain–containing protein, discs large, is required for the neuromuscular junction localization of Shaker potassium channels (Tejedor et al., 1997
). However, no direct test of PDZ-ligand interactions has been reported in whole animal mammalian systems. In this paper, we have described one approach that should be generally applicable to in vivo analyses in a broad range of systems.
We investigated the functional role of the α-syntrophin PDZ domain in vivo by using a dominant-negative approach. We developed a transgenic mouse line that expresses a PDZ-less form of α-syntrophin at high levels in skeletal muscle. The observation that ΔPDZ α-syntrophin is localized to the sarcolemma indicates that the PDZ domain is not required for plasma membrane association. This result is consistent with the demonstration that the PH2 and SU domains in tandem are required for binding to dystrophin and other family members (Ahn and Kunkel, 1995
; Kachinsky et al., 1999
). Furthermore, the ΔPDZ form of α-syntrophin competed with endogenous α-syntrophin for binding sites on the sarcolemma (dystrophin and dystrobrevin). This competition produced skeletal muscle that contained little or no sarcolemmal syntrophin PDZ domain as demonstrated by the reduction in labeling with mAb 1351, an antibody that binds to an epitope within the PDZ domain of α-syntrophin.
In addition to displacing endogenous α-syntrophin, the ΔPDZ α-syntrophin was able to displace β1-syntrophin from the sarcolemma as well. These data suggest that α- and β1-syntrophin compete for the same binding sites, at least under conditions of high α-syntrophin expression, to facilitate membrane association. Interestingly, the ΔPDZ form was more effective than the full-length Tg α-syntrophin in displacing β1-syntrophin. This suggests that the absence of the PDZ domain increases the affinity of syntrophin for dystrophin/dystrobrevin. Thus, the PDZ domain of syntrophin may act to regulate the interaction of syntrophin with the dystrophin complex. An interesting possibility is that binding of a ligand to the PDZ domain promotes association of the PH-SU domains with their binding sites on dystrophin. Such regulation is similar to that demonstrated for MAGUK family proteins (McGee and Bredt, 1999
) and could be important for localizing PDZ ligands to the sarcolemma via the dystrophin complex.
Two syntrophin PDZ ligands have been studied extensively in vitro. nNOS and voltage-gated sodium channels bind to α-syntrophin PDZ domains but via different mechanisms. nNOS contains a beta-finger immediately adjacent to its own PDZ domain that interacts directly with the PDZ domain of syntrophin (Hillier et al., 1999
). In contrast, sodium channels bind to syntrophin PDZ domains via the more conventional COOH terminal “SXV” motif (Gee et al., 1998a
). The function of this interaction is unknown, but syntrophin binding could potentially affect the ion conduction properties of the sodium channel or mediate its participation in a scaffold involving other channels and/or modifying enzymes. However, our data show that syntrophin binding is not required for normal sodium channel distribution on the sarcolemma. This conclusion is in agreement with the observation that sodium channel density is only slightly reduced in skeletal muscle of the mdx mouse which lacks dystrophin and, therefore, sarcolemmal syntrophin (Ribaux et al., 2001
In contrast, sarcolemmal localization of nNOS does depend on the presence of an α-syntrophin PDZ domain. In the absence of a sarcolemmal syntrophin PDZ domain, nNOS fails to localize to the sarcolemma. The localization of nNOS to the sarcolemma is important in regulating blood flow in skeletal muscle during muscle activity. Thomas et al. (1998)
have shown that in the absence of nNOS, attenuation of adrenergically stimulated vasoconstriction does not occur. Similar results were obtained from mdx mice which lack sarcolemmal nNOS (Thomas et al., 1998
). Therefore, it is likely that the α-syntrophin PDZ domain is not only necessary for proper localization of nNOS but is required for proper function of nNOS in skeletal muscle as well.
In addition to syntrophin, another membrane-associated protein, caveolin-3, has been shown to bind nNOS (Venema et al., 1997
). However, in muscle from the Tg ΔPDZ mice and the Tg ΔPDZ/ α-syntrophin–null cross, we could not detect nNOS on the membrane. Therefore, the interaction of nNOS with caveolin-3 must be insufficient for stable retention of nNOS at the sarcolemma. It has also been proposed that nNOS may localize to the sarcolemma by binding membrane-associated proteins directly via its PDZ domain (Abdelmoity et al., 2000
). Because in the absence of a syntrophin PDZ domain we do not detect nNOS on the sarcolemma, if such interactions occur they are insufficient for stable retention. Alternatively, such interactions could be induced by association of the nNOS beta finger with the syntrophin PDZ domain.
A third putative ligand for the α-syntrophin PDZ domain is aquaporin-4. Aquaporin-4 is normally present on the sarcolemma of fast twitch fibers of skeletal muscle but is absent from muscle that lacks the dystrophin protein complex (Frigeri et al., 1998
). The COOH-terminal amino acid sequence of aquaporin-4 is -VLSSV, a potential class I PDZ domain interaction sequence. Therefore, we used our transgenic mice to determine if the membrane localization of aquaporin-4 depended on a syntrophin PDZ domain. In the absence of an α-syntrophin PDZ domain, aquaporin-4 is also absent from the sarcolemma. Aquaporin-4 is an integral membrane protein that contains six transmembrane domains and therefore does not need to be recruited to the sarcolemma like the soluble protein, nNOS. Rather, the syntrophin PDZ domain is likely involved in targeting or stabilizing aquaporin-4.
The in vivo studies presented here also provide insight into how α-syntrophin interacts with the sarcolemma. Data from the Tg ΔPDZ show that α-syntrophin is localized to the sarcolemma and binds directly to dystrophin in the absence of a PDZ domain. However, the immunofluorescence studies of muscle expressing full-length and ΔPDZ syntrophin on an mdx background show that, even when expressed at high levels, syntrophin does not associate with the sarcolemma in the absence of dystrophin. Apparently, interactions with sodium channels or phosphatidyl inositols are insufficient to localize syntrophin to the sarcolemma. In fact, our data indicate that interaction with the dystrophin complex is necessary and sufficient for sarcolemmal α-syntrophin localization.