Evidences for a leaky scanning mechanism for the synthesis of the shorter M23 protein isoform of aquaporin-4: implication in orthogonal array formation and neuromyelitis optica antibody interaction

Abstract

Aquaporin-4 (AQP4) exists as two major isoforms that differ in the length of the N terminus, the shorter AQP4-M23 and the longer AQP4-M1. Both isoforms form tetramers, which can further aggregate in the plasma membrane to form typical orthogonal arrays of particles (OAPs) whose dimension depends on the ratio of the M1 and M23. In this study, we tested the hypothesis that the M23 isoform can be produced directly by the M1 mRNA. In cells transiently transfected with AQP4-M1 coding sequence we observed besides AQP4-M1 the additional presence of the AQP4-M23 isoform associated with the formation of typical OAPs observable by two-dimensional blue native/SDS-PAGE and total internal reflection microscopy. The mutation of the second in-frame methionine M23 in AQP4-M1 (AQP4-M1(M23I)) prevented the expression of the M23 isoform and the formation of OAPs. We propose "leaky scanning" as a translational mechanism for the expression of AQP4-M23 protein isoform and that the formation of OAPs may occur even in the absence of AQP4-M23 mRNA. This mechanism can have important pathophysiological implications for the cell regulation of the M1/M23 ratio and thus OAP size. In this study we also provide evidence that AQP4-M1 is mobile in the plasma membrane, that it is inserted and not excluded into immobile OAPs, and that it is an important determinant of OAP structure and size.

FIGURE 1.

Possible mechanism responsible for AQP4-M23 synthesis from AQP4-M1 mRNA. A, Western blot analysis of AQP4-M1 and AQP4-M23 in different tissues. A polyvinylidene difluoride membrane containing rat tissues and transfected cells, immunoblotted with anti-AQP4 antibodies after glycine/SDS-PAGE is shown. The AQP4-M1/M23 ratio is variable among the different tissues; in particular, the AQP4-M23 isoform is abundant in the cerebrum, cerebellum, and spinal cord. Note that in HeLa cells transiently transfected with the M1 isoform, two bands of 30 and 32 kDa are present. B, analysis of AQP4 mRNA at the level of the TISs surrounding M1 (TIS-1) and M23 (TIS-2). The 5′-proximal AUG corresponding to M1 is located in a suboptimal context, whereas the AUG corresponding to M23 is in an optimal context (42–45). C, mutated form of AQP4-M1 mRNA (M23I) was generated by mutagenesis in which the substitution methionine in isoleucine at position 23 will destroy the TIS-2, avoiding the leaky scanning mechanism.

FIGURE 2.

Expression of AQP4-M1, AQP4-M1^M23I, and AQP4-M23 in transiently transfected HeLa cells. A, AQP4 immunoblot of membrane proteins prepared from HeLa cells transfected as indicated. Note that the 30 kDa band is present in cells transfected with the wild type AQP4-M1 and is absent in those transfected with AQP4-M1^M23I. B, epifluorescence micrographs of HeLa cells transfected with AQP4-M1, AQP4-M1^M23I, and AQP4-M23. Scale bar, 10 μm. Note the dot-like plasma membrane staining when M23 is expressed.

FIGURE 3.

Visualization of AQP4 OAPs by TIRF microscopy and two-dimensional BN/SDS-PAGE. A, TIRF micrographs of Alexa Fluor 488-labeled AQP4-M1^M23I, AQP4-M1, AQP4-M1^M23I + AQP4-M23, AQP4-M1 + AQP4-M23, and AQP4-M23 in HeLa cells. Scale bar, 10 μm. B, immunodetection of AQP4 pools after a 3–9% gradient BN PAGE (first dimension) and 12% SDS/PAGE (second dimension). Note that in BN/SDS-PAGE experiments the expression of AQP4-M1 isoform induced the immunodetection of five different pools due to the concomitant expression of AQP4-M23 and the consequent formation of small OAPs. In contrast, the parallel analysis of AQP4-M1^M23I revealed the presence of the single smaller AQP4 pool corresponding to AQP4 tetramers.

FIGURE 4.

OAPs visualization using NMO-IgG serum. Double immunofluorescence experiments performed with rabbit polyclonal AQP4 antibodies (AQP4) and with NMO serum (NMO) are shown. Several cells transfected with AQP4-M1 were recognized by the NMO-IgG serum, whereas no positive cells were found when the mutated AQP4-M1^M23I isoform was used. Scale bar, 15 μm.

FIGURE 5.

FRAP experiments performed on AQP4-M1- and AQP4-M23-expressing HeLa cells. A, four serial images from a data set. The images displayed were recorded before and 1 and 6 min after photobleaching pulse. B, representative AQP4-M1-GFP and AQP4-M1-GFP + AQP4-M23 individual FRAP curves. Note that AQP4 organization in OAPs strongly slowed its plasma membrane lateral diffusion. C, data shown as reciprocal half-times (t_1/2) for fluorescence recovery (mean ± S.E. of separate sets of measurements, n = 3–7).