Discussion
We have previously shown that the X-linked form of Opitz GBBB syndrome (OS) results from loss of function mutations in MID1, a gene that encodes a member of a new class of microtubule-associated proteins [5]. A highly related factor, termed MID2, has also been identified which shares 77% amino acid identity (92% similarity) with MID1 and is expressed in many of the same embryological tissues, albeit at a lower level in most [15,19]. Both MID1 and MID2 are members of the RBCC family of proteins, a group of proteins with diverse intracellular localisations and presumably varied functions. Despite these differences, an ability to function as part of large multiprotein complexes is common to RBCC proteins [29]. Consistent with this is the observation that the 67 kDa MID1 is mostly found in complexes of between 250 kDa and 450 kDa [17]. However, no function has been definitively ascribed to any of the motifs in the MID proteins or indeed either protein as a whole. In this study, we have undertaken yeast two-hybrid screens to identify interacting partners that were likely to constitute part of these MID-complexes as well as performed precise deletion analysis of the MID proteins to begin to elucidate the role of individual motifs.
The only apparent similarity between all the MID1 mutations identified in patients with OS is that the resultant mutant protein is no longer able to completely decorate the microtubules. This observation implies that the function of one or more proteins within the MID1 complex is absolutely required along the length of the microtubules. Identification of such bona fide interactors might therefore be expected to shed light on the possible role of the OS protein and thus provide a better understanding of the molecular pathogenesis of the disorder. To this end, we performed yeast two-hybrid screens using MID1 and MID2 as bait and a 10.5 dpc mouse embryo cDNA library as prey. Significantly, we identified the phosphoprotein, Alpha 4 (also known as IGBP1) as an interactor of both MID proteins. This interaction between the MID proteins and Alpha 4 was confirmed by both vector swapping in the two-hybrid system as well as by co-localisation of differentially tagged proteins to the microtubule network. The specificity of this interaction is also supported by the fact that mouse Mid1 (but not mouse Mid2 presumably because of its lower abundance) was recently identified in the reverse two-hybrid screen of a 9 dpc murine embryonic cDNA library where the mouse Alpha 4 was used as the bait [30]. To characterise this interaction further, we chose to exploit the observation that C-terminal mutations of MID1 form cytoplasmic clumps both in vivo[18] and in transformed cell lines over-expressing GFP-mutant MID1 fusion proteins [5,15,17,18] by co-transfecting constructs expressing the ΔCTD proteins (and the other domain-specific deletions of MID1 and MID2) as GFP fusions with the myc-tagged Alpha 4 construct. These experiments were complemented by yeast two-hybrid analysis of the same combinations of constructs.
Through a series of in-frame deletions of both MID1 and MID2 employed in both immunofluorescence and yeast two-hybrid assays, we have demonstrated functions for three domains of these RBCC proteins. We have shown a direct role for an RBCC B-box region in protein-protein interactions, that being the binding of Alpha 4, a regulatory subunit of the PP2-type phosphatases including the principal cellular phosphatase, protein phosphatase 2A (PP2A; [21-24]). In addition to the identified role for the B-boxes, we have shown that the coiled-coil domain not only mediates the homodimerisation of these proteins but also their ability to heterodimerise. Finally, our data, together with previous observations by others [18], support a role for the C-terminal domain in microtubule binding and to a lesser extent in dimerisation. Collectively, the immunofluorescence and yeast two-hybrid data indicate that MID dimerisation is a prerequisite for association of each MID-Alpha 4 complex with microtubules.
The interaction of MID1 and MID2 with Alpha 4 raised the possibility that these RBCC proteins, like Xnf7, are phosphoproteins. Indeed, western blot analysis of the transiently expressed MID-GFP fusion proteins using anti-phosphoserine and anti-phosphothreonine antibodies has confirmed these suspicions. Analysis of the domain-specific deletion proteins using these same antibodies showed that most serine phosphorylation of MID1 seems to occur at, or immediately adjacent to, the B-boxes, whereas threonine phosphorylation was likely to occur at residues in more than one domain. As a preliminary step towards identifying those residues that are phosphorylated, computer-based searches identified numerous potential sites for phosphorylation by serine/threonine kinases. However, only sixteen of these (see Fig 5C) are completely conserved across all MID species isolated to date. Of note are the potential phosphorylation sites at serine 96 (S96; [30]) and serine 92 (S92) that are located immediately amino-terminal to the B-boxes of MID1 and MID2 and are deleted in the ΔBB constructs. S92 is a consensus site for GSK3 that would be dependent on prior phosphorylation of S96. S96 falls within a consensus phosphorylation sequence for the MAP-kinase (MAPK), ERK2 (P-N-S/T-P; [31]), as well as casein kinases I and II. Therefore, either of these kinases could conceivably play a priming role for GSK3 phosphorylation of S92. Similar phosphorylation mechanisms have been observed for other proteins including some microtubule-associated proteins, eg. tau and MAP2, where it has been implicated in the regulation of specific functions of those proteins [32,33]. Notably, in the case of both eIF2Bepsilon (at serine residues 535 & 539) and tau (at serine residues 208 & 212), DYRK, a MAPK-related kinase, plays such a priming role for GSK3 [34]. However, DYRK is unlikely to be involved in phosphorylation of MID1 S96 because of a single amino acid difference in its surrounding sequence required for recognition of this site for phosphorylation. In fact, Liu et al [30] have provided evidence to support the involvement of a MAPK, although the target of this activity was not shown. Irrespective of the identity of the kinase(s) responsible, it is conceivable that phosphorylation at these or other sites in MID1 and MID2 could play an important role in the overall function of the MID proteins, for example: regulation of the Alpha 4-MID interaction and hence regulation of PP2-type phosphatase activity. However, our preliminary analysis of a MID1 Ser92/Ser96 double mutant has suggested that MID dimerisation and the MID-Alpha 4 interaction are not dependent on phosphorylation at these residues (unpublished observations).
Regardless of the role of MID phosphorylation, the implication of both MID1 and MID2 in the Alpha 4-mediated regulation of phosphatase activity may provide valuable clues as to the pathophysiological consequences of MID1 mutations that underlie Opitz syndrome. It can be envisaged that, in tethering Alpha 4 to the microtubules, MID1 (and MID2) could be affecting the activity of the PP2-type phosphatases and thereby, in-turn, modulating the rapamycin-sensitive signaling pathway. This could conceivably occur by one of a number of mechanisms. Firstly, the MID proteins may control the availability of Alpha 4 to the phosphatases either by its 1) sequestration or 2) turnover, facilitated by the possible role of the MID RING finger motif in ubiquitination. Secondly, the Alpha 4-MID interaction may direct PP2A (and PP2-related) phosphatase activity to specific targets on the microtubules, which may include MID1 and MID2 themselves. The fact that both endogenous MID1 and transiently expressed GFP-MID1 and GFP-MID2 decorate microtubules throughout the cell cycle ([17]; unpublished observations) suggests that binding of MID1 and MID2 to the microtubule network may not itself be regulated by targeted Alpha 4-dependent PP2A activity. However, it remains possible that dynamic regulation of MID1 microtubule binding may escape detection by immunofluorescence as endogenous levels of MID1 and Alpha 4 in examined cell lines are both low (unpublished observations). The fact that we have not been able to demonstrate PP2A(C) co-localisation with the MID-Alpha 4 complexes on the microtubules, despite a known microtubule-associated pool of PP2A [35], would perhaps support the former of these mechanisms. However, we cannot exclude the possibility that recognition of PP2A(C) by this antibody is blocked by the interaction of Alpha 4 and MID1/2. If this is indeed the case, then it could be envisaged that PP2A(C) is also a target of MID RING-mediated degradation through its independent interaction with Alpha 4. An alternate hypothesis is that MID function is indeed controlled by Alpha 4-PP2-type phosphatases but through the regulated binding of additional factors that might be components of the MID1 macromolecular complexes. The characterisation of additional interacting partners will therefore be important to assess this possibility.
Numerous studies in mice have demonstrated that Alpha 4, like its yeast homologue Tap42, plays an essential regulatory role within the cell through its regulated binding to the catalytic (C) subunit of PP2-type serine/threonine protein phosphatases [21-24]. The interaction of Alpha 4 with PP2A(C) has been most extensively studied and shown to be dependent on phosphorylation of Alpha 4 by the mTOR (target of rapamycin) kinase. Although PP2A has a wide range of biological functions, Alpha 4 regulates a distinct subset of events in a rapamycin-sensitive manner, including progression through the cell cycle and the regulation of protein biosynthesis (for a review see [36]). The implication that disruption to some aspects of this rapamycin-sensitive pathway might be associated with the pathogenesis of the developmental disorder, Opitz syndrome, raises two intriguing possibilities. Firstly, it can be envisaged that mutations in other components of the pathway may give rise to similar clinical phenotypes. It is therefore perhaps worthy to note that the Alpha 4 gene maps to Xq13 [26] in the vicinity of the critical interval for FG syndrome, a malformation disorder with some clinical overlap with that of Opitz GBBB syndrome. We are currently investigating whether molecular defects in Alpha 4 indeed underlie FG syndrome or other developmental disorders mapping to the proximal long arm of the X chromosome. Secondly, it is also feasible that genetic polymorphisms that reflect a variation in the level of expression or activity of one or more components of the rapamycin-sensitive pathway might also contribute to the clinical variability of OS. In this regard, the indication from our immunofluorescence studies that the microtubular localisation of Alpha 4 is likely to be limited by the level of the MID1 and MID2 proteins within a particular cell type provides indirect support for our earlier hypothesis that MID2 may be able to compensate, at least partially, for the loss of MID1 in Opitz syndrome [5].