An Aberrant Sequence in a Connexin46 Mutant Underlies Congenital Cataracts*

An increasing number of diseases have been mapped to genes coding for ion channel proteins, including the gap junction proteins, connexins. Here, we report on the identification of an amino acid sequence underlying the behavior of a non-functional mutant connexin46 (CX46) associated with congenital cataracts. The mutant protein, CX46fs380, is 31 amino acids longer than CX46 and contains 87 aberrant amino acids in its C terminus. When expressed in mammalian cells, the mutant CX46 was not found at gap junctional plaques, but it showed extensive co-localization with markers for ERGIC and Golgi. The severe reductions in function and formation of gap junctional plaques were transferred to other connexins by creating chimeras containing the last third (or more) of the aberrant C terminus of the CX46 mutant. This sequence also impaired trafficking of a CD8 chimera. Site-directed mutagenesis of a diphenylalanine restored appositional membrane localization and function. These results suggest a novel mechanism in which a mutation causes disease by generating a motif that leads to retention within the synthetic/secretory pathway.

rabbit immunization, and the serum obtained was affinity-purified through a Sulfolink column after coupling the CX46IL peptide following the manufacturer's directions (Pierce). To obtain anti-CX46fs380 and anti-CX56CT antibodies, glutathione S-transferase fusion proteins containing amino acids 379 -466 of CX46fs380 or amino acids 255-511 of CX56 (15) were produced and used for rabbit immunization. These rabbit sera were stripped of anti-GST antibodies using a GST-immobi-lized agarose column (Pierce). For the anti-CX46fs380 antibodies, the stripped serum was affinity-purified using the CX46fs380 fusion protein coupled to a Sulfolink column (Pierce).

Primer number Sequence
5Ј-GTCATGGCCTTACCAGTGACCGCCTTG-3Ј 20 5Ј-CTGCTGTCCGGAGTTGCAGTAAAGGGTGATAAC-3Ј Hela cells were fixed in 4% paraformaldehyde. After fixation, cells were permeabilized with 1% Triton X-100 in phosphate-buffered saline pH 7.4 (PBS), blocked in 4% normal goat serum (NGS), 1% Triton X-100 in PBS, and incubated overnight in primary antibodies at 4°C. Cells were then rinsed six times with PBS and incubated in Cy2-and/or Cy3conjugated secondary antibodies. After incubation at room temperature for 45 min, cells were rinsed six times with PBS, and coverslips were mounted with 2% n-propylgallate in PBS/glycerol (1:1). When using the FITC-conjugated mouse monoclonal anti-human CD8 antibody, HeLa cells stably transfected with CD8 chimeras were fixed in 100% methanol at Ϫ20°C, blocked in 10% NGS in PBS, and incubated in antibody for 45 min. Then, cells were washed four times with PBS and coverslips mounted as above.
Specimens were studied with a Zeiss Plan Apochromat ϫ40 objective (n.a., 1.0) in an Axioplan 2 microscope (Carl Zeiss, München, Germany) equipped with a mercury lamp. Images were captured with an Axiocam digital camera (Carl Zeiss) using Zeiss AxioVision software. Confocal images were obtained using an IX70 Olympus Fluoview 200 laser-scanning confocal microscope (Melville, NY) equipped with laser lines (488 nm, 543 nm, 633 nm) with a ϫ60 Plan Apochromat objective (n.a., 1.4). Overlap images were generated using Image J software and the colocalization threshold plugin created by Tony Collins, Wayne Rasband, and Kevin Baler (University Health Network). Composite figures were assembled using Adobe Photoshop software (Adobe Systems, San Jose, CA).
Immunofluorescence was performed in transiently transfected cells at least three times for each construct, and at least in four different clones for stably transfected cells. Specimens were examined by three different observers, one of whom looked at all the visual fields in the well whereas the others looked at a minimum of 20 fields for each construct/treatment.

Cell Treatments
Brefeldin A-HeLa cells stably transfected with CX46fs380, Cx32-fs380CT, or CD8-fs380CT were treated with 40 g/ml cycloheximide (CHX) for 16 h followed by incubation in the presence of 4 g/ml BFA or ethanol (used as a solvent for the BFA stock solution) for 8 h or 20 h. These experiments were repeated three times for the CX46fs380 clones, and twice for the other chimeric constructs using at least two different concentrations of BFA (4 and 10 g/ml).
Incubation at Reduced Temperatures-Cells were grown at 37°C until they reached 40 -50% confluence. Then, they were either transferred to 25°C or 30°C or left at 37°C and further incubated until they reached 80 -90% confluence as previously described (12). These experiments were performed three times.
After the treatments, cells were rinsed in PBS, fixed in 4% paraformaldehyde, and processed for immunofluorescence as described above.

Electrophysiological Measurements
The DNA constructs were subcloned into the RNA expression vector, SP64TII. The plasmids were linearized with SalI, and capped RNAs were synthesized using the mMessage mMachine SP6 in vitro transcription kit (Ambion, Austin, TX) according to the manufacturer's instructions. The amount of RNA was quantitated by measuring the absorbance at 260 nm.
Xenopus oocytes were prepared and tested as described previously (16). To measure the junctional conductance, cell pairs were studied using the dual two microelectrode technique (17). For simple measurements of gap junctional coupling, both cells of the pair were initially held at Ϫ40 mV and 5-10-mV steps were applied to one cell while holding the second cell at Ϫ40 mV. The junctional conductance (g j ) equals I j /V j , where V j ϭ V cell2 Ϫ V cell1 . Data acquisition and analysis were conducted using a personal computer running pCLAMP version 6.0.5 (Axon Instruments, Union City, CA). Microelectrodes were filled with 3 M KCl (pH 7.4) and had resistances of 0.2-2 M⍀. The bath solution was modified Barth's solution. All of the experiments were performed on oocytes previously injected with antisense Cx38 oligonucleotides to ensure that the observed currents were not caused by the expression of endogenous Cx38 (18). Data expressed as log(conductance) were analyzed for their statistical significance using the unpaired Student's t test.

Flow Cytometry Analysis
Batch clones of HeLa cells transfected with CD8 chimeras plated on 60-mm culture dishes were grown until 80% confluence. Cells were trypsinized, centrifuged, and the pellets incubated in 10% NGS in PBS for 10 min followed by incubation in FITC-conjugated mouse monoclonal anti-human CD8 antibody. After 20 min, cells were pelleted, rinsed twice with 10% NGS in PBS, and resuspended in 10% NGS in PBS. Prior to flow cytometry analysis, cells were stained with propidium iodide. Flow cytometry was carried out on a BD FACSCan flow cytometer (BD Biosciences, Rockville, MD). FITC fluorescence was acquired using the FL1 channel, and the data analyzed using FlowJo software. Similar results were obtained when repeating these experiments using different concentrations of antibody; the experiment was performed twice at the concentration shown.

RESULTS
The Mutant CX46 Is Impaired in Its Ability to Reach the Plasma Membrane-To test whether CX46fs380 formed non-functional channels or its trafficking to the plasma membrane was impaired, we examined the distribution of wild-type CX46 and CX46fs380 by immunofluorescence in transfected HeLa cells that are competent for formation of gap junctions. Whereas cells transfected with CX46 showed gap junctional plaques at appositional membranes, no gap junctional plaques were observed in cells transfected with CX46fs380 ( Fig. 1A, left and right panels). Rather, CX46fs380 localized mostly in an intracellular location that corresponded to ERGIC and Golgi compartments based on its overlap with ERGIC-53 and the Golgi 58K protein (Fig. 1B). A minor proportion of CX46fs380 overlapped with the ER-resident protein PDI (not shown). The localization of CX46fs380 in the Golgi was further confirmed when it redistributed together with the Golgi 58K protein after treatment of transfected cells with BFA (Fig. 1C).
Several mutant membrane proteins have trafficking/folding defects that can be overcome by incubation at lower temperatures or treatment with chemical chaperones (19,20). To test whether the intracellular distribution of CX46fs380 was caused by its inability to acquire the correct conformation, we analyzed the cellular localization of CX46fs380 after incubation of the transfected cells at 25 or 30°C or after treatment with sodium 4-phenylbutyrate. No CX46fs380 immunoreactive gap junctional plaques were observed after any of these treatments (data not shown).
The Aberrant Behavior Results from Gain of a Deleterious Function-Because CX46fs380 lacks amino acids 380 -435 of wildtype CX46, it was possible that the defects of CX46fs380 resulted from the absence of sequence signals contained in the missing wildtype region that allow efficient trafficking and formation of gap junctional plaques or to the presence of a sequence motif in the aberrant region that prevents proper trafficking/localization. To differentiate between these possibilities, we generated a truncated form of CX46 containing only the first 379 amino acids of wild-type CX46 (CX46tr380) and analyzed its distribution by immunofluorescence. Cells transfected with this construct showed gap junctional plaques (Fig. 1A, center panel). Therefore, we reasoned that the amino acid sequence of the CX46fs380 C terminus following the frameshift was responsible for the observed effects.
The Amino Acid Sequence after the Frameshift (fs380CT) Confers Abnormal Behavior to Connexin Chimeras-To test whether the aberrant localization and loss of function were transferable traits controlled by the abnormal C terminus of CX46fs380, we generated several constructs on the backbone of CX56 (the chicken orthologue of human CX46) and studied their cellular distribution. Immunofluorescent studies showed that anti-CX56 immunoreactivity localized at appositional membranes in HeLa cells transfected with wild-type CX56 (Fig. 2, top left panel). Cells transfected with CX56 truncated after amino acid 445 (CX56tr446), a position analogous to amino acid 380 in CX46, also showed anti-CX56 immunostaining at appositional membranes (Fig. 2, top right panel). Cells transfected with a chimera containing amino acids 380 -435 of wild-type CX46 (CX46CT) appended to the first 445 amino acids of CX56 (CX56-CX46CT) also localized to appositional membranes (Fig. 2, bottom left panel). However, a chimera containing the aberrant sequence of the C terminus of CX46fs380 (CX56-fs380CT) showed an intracellular localization; its distribution was similar to that of CX46fs380 (compare Fig. 2, bottom right panel with Fig. 1A, right panel). The CX56 constructs were also analyzed for their ability to form functional gap junction channels in pairs of Xenopus oocytes. Wild-type CX56 induced large gap junctional conductances between oocyte pairs (Fig. 3A). CX56tr446 and the CX56-CX46CT chimera also induced large junctional conductances that were not significantly different from those induced by CX56 (Fig. 3, A and B). However, when CX56-fs380CT was expressed in Xenopus oocytes, gap junctional conductances were reduced to 7.6% of wild-type CX56 and 6.3% of CX56-CX46CT conductances (Fig. 3B), but they were still significantly higher than the junc-   DECEMBER 9, 2005 • VOLUME 280 • NUMBER 49 tional conductance determined in pairs of antisense Cx38-injected oocytes (Fig. 3C).

Retention Signal in a Connexin Mutant
We also generated Cx32 chimeras in which CX46CT or fs380CT were appended to full-length Cx32 and characterized their behavior by immunofluorescence after transfection into HeLa cells. Whereas anti-Cx32 immunoreactivity localized to appositional membranes in cells transfected with Cx32-CX46CT (data not shown), it showed an intracellular localization in cells transfected with Cx32-fs380CT (Fig. 4A). The pattern of Cx32-fs380CT immunoreactivity resembled that of the ER showing a reticular staining throughout the cell. Extensive overlap of the immunoreactivities to anti-CX46fs380 and anti-PDI antibodies was observed (Fig. 4A). There was a rather poor correspondence between anti-Cx32 immunoreactivity and that of markers for ERGIC or Golgi compartment, but a minor amount of anti-Cx32 immunoreactivity overlapped with LAMP-1 in a small subpopulation of cells (not shown). When CHX-treated cells stably transfected with Cx32-fs380CT were allowed to recover in control media for 8 h, a weak diffuse signal throughout the cytoplasm with a brighter staining localized to the perinuclear region was observed. However, when BFA was present during the recovery phase, the Cx32-fs380CT immunoreactivity redistributed throughout the cell (not shown) implying that Cx32-fs380CT initially reaches a BFA-sensitive Golgi compartment to be subsequently relocated to the ER where it predominantly resides at steady state. Functional characterization of these chimeras showed a drastic decrease in junctional conductance in oocyte pairs injected with Cx32-fs380CT cRNA as compared with that induced in oocyte pairs injected with Cx32-CX46CT cRNA (Fig. 4B). The junctional conductances induced by Cx32-fs380CT were not significantly different than those induced in antisense Cx38-injected oocytes.
The Last Third of fs380CT Confers Aberrant Behavior-To define sequence elements responsible for loss of appositional membrane localization and function, several CX56 chimeras containing different regions of the C terminus of CX46fs380 were generated and analyzed by immunofluorescence. Chimeras containing the first one-third (CX56-fs380(1-29)) or two-thirds (CX5-6fs380(1-58)) of fs380CT localized to appositional membranes like the chimera containing CX46CT (Fig. 5A). However, a CX56 chimera containing only the last 29 amino acids of the aberrant C terminus of CX46fs380 (CX56-fs380(59 -87)) localized within the cytoplasm (Fig. 5A).
These chimeras were also tested for their ability to induce junctional conductance in the Xenopus oocyte system. The chimera containing the first two-thirds of fs380CT (CX56-fs380(1-58)) induced robust junctional conductances that were not significantly different from those induced by wild-type CX56 (Fig. 5B). However, the chimera containing the last third of fs380CT (CX56-fs380(59 -87)) induced significantly reduced junctional conductances (Fig. 5C).
Identification of Amino Acid Residues Critical for the Abnormal Behavior-Examination of the last 29 amino acids of the C terminus of CX46fs380 revealed sequences that resembled previously described trafficking motifs (Fig. 6A). To test the possibility that these residues were responsible for the altered behavior of CX46fs380, the localization and function of chimeras in which these putative motifs were mutated was analyzed. Mutation of the amino acids TKTK (a dilysine-like motif) to TATA yielded a CX56 chimera (CX56-fs380TATA) that still local-  ized intracellularly when transfected into HeLa cells (Fig. 6B). This chimera was not functional in oocyte pairs (Fig. 6C). However, distinct immunoreactive staining at appositional membranes was observed in HeLa cells expressing a CX56 chimera in which the FF in the C terminus of CX46fs380 was mutated to AA (CX56-fs380AA) (Fig. 6B). Moreover, CX56-fs380AA induced robust gap junctional currents; the macroscopic gap junctional conductance of oocyte pairs expressing the CX56-fs380AA chimera was not significantly different from the value obtained from wild-type CX56 (Fig. 6C).

The Aberrant Sequence in CX46fs380 Causes Retention of a CD8
Chimera-A strategy used to study amino acid sequences that act as "retention" or "export" motifs has been to create chimeras with single transmembrane domain proteins such as CD8 (21). To test whether fs380CT affected trafficking of this unrelated protein, we generated a CD8 chimera containing fs380CT and stably transfected it into HeLa cells. Immunofluorescent studies in paraformaldehyde-fixed cells revealed that CD8-fs380CT showed an intracellular localization with a significant overlap with PDI and some overlap with ERGIC-53. Most cells also showed some overlap between the CD8-fs380CT and LAMP-1 immunoreactivities (not shown) suggesting that a proportion of the chimera was targeted for degradation in the lysosome before or after reaching the plasma membrane (22).
To test whether the diphenylalanine in fs380CT was also responsible for the localization of CD8-fs380CT, we stably transfected HeLa cells with a chimera in which FF was mutated to AA. In contrast to CD8-fs380CT, CD8-fs380AA immunoreactivity localized to the plasma membrane (Fig. 8A).
Surface expression of the CD8 chimeras was assessed by flow cytometry. The fluorescence intensity in cells transfected with the CD8-fs380AA chimera showed a 10-fold increase over that of cells transfected with CD8-fs380CT (Fig. 8B), indicating that CD8-fs380AA was about 10 times more efficient in reaching the plasma membrane than CD8-fs380CT.

DISCUSSION
The data presented have shown that the cataract-associated mutant, CX46fs380, did not traffic properly and was retained within the Golgi compartment. Several lines of evidence indicate that the aberrant amino acid sequence of the C terminus of CX46fs380 is responsible for the effects observed: CX46fs380 did not traffic properly whereas a truncated form containing the first 379 amino acids of CX46 (also present in CX46fs380) formed gap junctional plaques; connexin chimeras containing the aberrant sequence of CX46fs380 showed a predominant intracellular distribution; the last 29 amino acids of CX46fs380 were sufficient to cause impaired trafficking of chimeric molecules; and, surface expression of a CD8-fs380CT chimera was markedly decreased.
Acquisition of correct protein conformation is a key step for targeting proteins to the right cellular compartment. The conformation of nascent proteins, and their subsequent processing and transport can be sensitive to temperature. Several disease-associated mutant proteins exhibit temperature-sensitive trafficking (i.e. they reach the plasma membrane when incubated at reduced temperatures) (19). Trafficking of CX46fs380 to the plasma membrane was not temperature-sensitive, because gap junctional plaques were not detected after incubation of transfected cells at lower temperatures. Chemical chaperones can also rescue trafficking of some mutant membrane proteins (20). However, 4-phenylbutyrate did not redistribute CX46fs380 to appositional membranes. These results suggest that the trafficking defect in CX46fs380 is not because of a problem in the conformation of the protein.
The export of proteins through the secretory pathway proceeds through a series of directed translocations with formation of cargocontaining vesicles and their fusion from one compartment into the next. These events are directed by signaling elements present in the amino acid sequence of the cargo protein. The observation of the truncated form of CX46, CX46tr380, at gap junctional plaques suggests that the first 379 amino acids of wild-type CX46 are sufficient (while amino acids 380 -435 are not essential) for targeting of the protein to gap junctional plaques. Similarly, the first 445 amino acids of CX56 are sufficient for localization at gap junctional plaques.
In many systems the presence of gap junctional plaques correlates with gap junctional conductance. Consistently, in the present study, all connexin constructs that localized at gap junctional plaques formed functional channels. CX46fs380 and chimeras containing fs380CT did not localize at gap junctional plaques and induced significantly reduced junctional currents. The observation that the junctional conductance induced by CX56-fs380CT was significantly higher than that in AS-injected control pairs suggests that the small amount of CX56-fs380CT that reached the plasma membrane was still functional. These results suggest that the reduced gap junction function of CX46fs380 and connexin chimeras containing fs380CT was caused by improper trafficking of the protein rather than an inability to form functional channels.
Several trafficking motifs (e.g. KKXX, FF, RXR, F(X) 6 I/LI/L) have been identified within the amino acid sequence of membrane receptors, ERbased proteins and transporters (23)(24)(25)(26). ER-based proteins contain a luminal K/HDEL or a dilysine motif in their cytoplasmic C termini (23,27). ERGIC-53, a protein that cycles between the ER and Golgi compartments contains both dilysine and diphenylalanine motifs (28). The last 29 amino acids of fs380CT, which were required for impaired trafficking and function, contain a diphenylalanine and a dilysine-like motif. Our mutational analysis of putative trafficking signals indicated that replacement of the dilysine-like motif with alanines had no effects on protein localization or channel function.
In the present study, two phenylalanines were demonstrated to be critical for the anomalies observed in proteins containing the fs380CT. Their replacement with alanines restored gap junctional plaque localization (and function) to CX46fs380 and the connexin chimeras; it also increased surface expression of the CD8 chimera by about 10-fold. These results suggest that the diphenylalanine in fs380CT acts as a retention signal. This conclusion seems rather paradoxical, because a diphenylalanine motif was initially described as an anterograde transport signal in the p24 family of proteins; the transit of a CD8 chimera from the ER to the Golgi was decreased after replacement of the diphenylalanine with dialanine (24). All members of this family contain a diphenylalanine motif within the cytoplasmic C terminus and localize to the early secretory pathway at steady state (29,30). Similar to the p24 family of proteins, all protein molecules containing fs380CT analyzed in the present study localized to the early secretory pathway at steady state and spent part of their lifetime in the Golgi. However, the exact role of the diphenylalanine as a trafficking signal is unclear. Conflicting effects have been observed after replacement of diphenylalanine with dialanine in several members of this family: impaired anterograde trafficking of chop24 (24), impaired retrieval of p23 from Golgi (31), and increased forward transport of Emp24p from the ER in yeast (32). Moreover, generation of a diphenylalanine by substitution of FF for LL in the ER export motif of the ␣ 2B -adrenergic receptor (F(X) 6 LL) results in a significant decrease in ER to plasma membrane transport (26). Thus, the published data together with our results suggest that an FF signal targets the protein to the early secretory pathway or it acts as a retrieval/retention signal.
The conflicting inferences regarding the role of FF in members of the p24 family of proteins could result from the presence of other trafficking motifs in the protein sequence. A combination of motifs and their amino acid context determine the cellular fate of proteins (24,28,(31)(32)(33)(34)(35). In chop24, the effect of the FF on transport depends on the presence of a glutamine and a glutamic acid in the transmembrane domain (33). Similarly, Duvernay et al. (26) have found that substitution of the F(X) 6 LL motif with A(X) 6 FF was necessary to abolish completely ER to plasma membrane trafficking of the ␣ 2B -adrenergic receptor. Thus, the FF in CX46fs380 could be part of a more complex motif. However, this possibility seems unlikely, because this sequence originated from a frameshift, not by natural selection. Furthermore, both the immunofluorescence and functional data suggest that the FF is necessary and sufficient to cause impaired protein trafficking. Whereas trafficking signals that direct localization to gap junctional plaques may be present in the truncated form of wild-type CX46 and CX56, they must not be strong enough to overcome the effects of the diphenylalanine present in fs380CT.
In summary, the abnormal C terminus generated as a consequence of the frameshift in CX46 leads to mislocalization of the protein within the secretory pathway and loss of function. The critical residues for these effects are two adjacent phenylalanines. The results suggest that not only loss of targeting signals but also their creation in mutant proteins may lead to disease. This represents a novel mechanism by which mutant proteins can cause disease.