Single amino acid change in the fifth transmembrane segment of the TRP Ca2+ channel causes massive degeneration of photoreceptors.

The trp gene encodes subunits of a highly Ca(2+)-permeable class of light-activated channels of Drosophila photoreceptors. The recently characterized mutation in this gene, Trp(P365), is semidominant and causes massive degeneration of photoreceptors by making the TRP channel constitutively active. We show that a single amino acid change, Phe-550 to Ile, near the beginning of the fifth transmembrane domain of TRP channel subunits is necessary to induce, and sufficient to closely mimic, the original mutant phenotypes of Trp(P365). Hypotheses are presented as to why the amino acid residues at position 550 and its immediate vicinity might be important in influencing the regulation of the TRP channel and why the substitution of Phe for Ile at this position, in particular, could result in constitutive activity of the channel.

Drosophila photoreceptors have two classes of light-activated channels, the highly calcium-permeable TRP channels and the non-specifically cation-selective TRPL channels (1). The TRP channels, the subunits of which are encoded by the trp gene (1)(2)(3)(4), are thought to carry most of the phototransduction current (5). The TRP protein is widely conserved throughout animal phylogeny (6 -8). However, the mechanisms of activation or regulation of TRP or TRP-related channels are still obscure. Yoon et al. (9) recently described a new mutation in the trp gene, Trp P365 , which confers on the mutant a set of phenotypes unlike any seen in previously isolated trp mutants. The Trp P365 mutants do not display photoreceptor responses that terminate prematurely during light stimulus, a phenotype that had been considered to be the defining hallmark of mutants in this gene (10,11). Trp P365 responses are small but are maintained throughout the duration of light stimulus. The most striking feature of Trp P365 mutants, however, is that their photoreceptors undergo rapid degeneration. Moreover, unlike any other known trp mutants, these Trp P365 phenotypes are semidominant, i.e. Trp P365 heterozygotes also display mutant phenotypes though they are not as severe as those of homozygotes. From whole cell patch-clamp recordings, Yoon et al. (9) concluded that these phenotypes arise as a result of excessive Ca 2ϩ entry caused by constitutive activity of the TRP channel. We now show that the above phenotypes are because of a single amino acid change, Phe-550 to Ile, in the fifth transmembrane segment of the TRP channel. We present, in addition, a possible mechanism by which the phenotypes might arise.

EXPERIMENTAL PROCEDURES
Minigene Construction and Germ Line Transformation-A trp minigene was constructed and subcloned into the pCaSpeR-3 vector for germ line transformation. The minigene consisted of three DNA fragments, the trp promoter and 5Ј-untranslated regions (Ϫ681 to Ϫ1 of genomic DNA, where ϩ1 is the trp translation start site), the coding region (ATG to the stop codon), and the 390-bp region immediately downstream of the stop codon. The minigene was subcloned into the XbaI and StuI sites of the CaSpeR-3 transformation vector. The vector containing the minigene was injected into early embryos in the presence of helper plasmid (12).
The following nine mutant trp minigene constructs were generated, and transgenic flies carrying each mutant minigene were generated by P-element-mediated germ line transformation in a wild-type background: P[trp (1) In addition to these transgenic lines in a wild-type background, three of the lines, that carrying mutation 3, that carrying mutations 1, 2, and 4, and that carrying all four mutations, were also placed in a trp-null (trp P343 ) background by appropriate genetic crosses. The trp P343 mutant was generated in this laboratory and was shown by others (14) to be functionally null.
Confocal Microscopy-Fly eyes were dissected in a fixative (4% formaldehyde in phosphate-buffered saline with 0.3% Triton X-100), and the dissected retinas (photoreceptor layers) were allowed to remain in the same fixative for 1 h. After incubation in phosphate-buffered saline that contained 4% normal goat serum, the fixed retinas were stained with phalloidin-tetramethylrhodamine B isothiocyanate (Sigma) to label filamentous actin in the rhabdomeres. Transverse optical sections of ϳ1 m thickness were taken ϳ6 m from the distal tips of the rhabdomeres.

RESULTS
Yoon et al. (9) showed that the Trp P365 mutant carries four protein sequence-altering mutations within its trp gene, P500T, H531N, F550I, and S867F. To simplify notations, these mutations will be referred to as mutations 1, 2, 3, and 4, respectively. To determine which of these four mutations is responsible for the Trp P365 phenotype, site-directed mutagenesis was carried out on a trp minigene to generate mutants carrying each of the four mutations singly, three of the four in all possible triplet combinations, or all four mutations together. The resulting mutant minigene constructs were introduced into a wild-type background by P-element-mediated germ line transformation (see "Experimental Procedures"). A single copy of the trp minigene, which has no introns, was found to fully complement the trp mutant phenotype when introduced into the null trp P343 background (not shown).
Transgenic Flies Carrying Each of the Four Mutations-Shown in Fig. 1 are electroretinograms (ERGs), 1 extracellularly recorded, light-evoked mass responses of the eye, obtained from transgenic flies each carrying a single copy of the recom-binant trp minigene harboring mutation 1, 2, 3, or 4, i.e. P[trp (1)]/ϩ, P[trp (2)]/ϩ, P[trp(3)]/ϩ, or P[trp (4)]/ϩ, all in a wild-type background. Thus, these transgenic flies had two copies of the endogenous wild-type trp gene and one copy of the mutagenized recombinant trp minigene. ERGs of P[trp(1)]/ϩ, P[trp(2)]/ϩ, or P[trp(4)]/ϩ were indistinguishable from those of wild-type ( Fig.  1, a, b, and d). Only those obtained from P[trp(3)]/ϩ were distinctly different from wild-type ERGs in that they showed markedly slower kinetics of decay at stimulus off-set (Fig. 1c). There was little variation in the ERG decay times among individuals of P[trp(3)]/ϩ so that the P[trp(3)]/ϩ ERG could always be distinguished from that of wild-type by inspection. Yoon et al. (9) have shown previously that one copy of Trp P365 introduced into a wild-type background causes the ERGs of the transgenic host to become somewhat smaller in amplitude and slower in the kinetics of decay than those of wild-type. Thus, the slow time courses of decay observed in this study were consistent with the previous findings, although little or no change in amplitudes was detected in the present study.
We next made the P[trp (3)] transgene homozygous so that the transgenic flies now carried two copies of P[trp(3)] and two copies of the endogenous wild-type trp gene. The main effect of making P[trp (3)] homozygous on the ERG was to slow down the kinetics of response decay even further (Fig. 2a). Thus, mutation 3 was the only one of the four that had any effect at all on the ERG, suggesting that mutation 3, i.e. F550I, might be the primary contributor to the Trp P365 mutant ERG phenotype.
Transgenic Flies Carrying Triplet Combinations of Mutations-To examine to what extent the other three mutations might play a role in producing the Trp P365 phenotype, we generated transgenic flies carrying various triplet combina- 1 The abbreviation used is: ERG, electroretinogram.  Fig. 2, c and a). For this reason, an ERG obtained from only one of these three lines, P[trp(123)], is shown in Fig. 2. On the other hand, ERGs of transgenic flies of the P[trp(124)] line, the only line that did not carry mutation 3, were completely wild-type (Fig. 2d). Thus, whenever the transgenic flies carried mutation 3, whether by itself or in combination with the other mutations, their ERGs were mutant, and whenever they did not carry mutation 3 their ERGs were wild-type. The results were clear-cut in suggesting that 1) mutation 3 alone was sufficient to generate the mutant phenotype, and 2) the other mutations contributed very little to the mutant phenotype.
Transgenic Flies in a trp-null Background-All transgenic lines tested to this point were generated in a wild-type background. Therefore, they were not genotypically comparable with Trp P365 /ϩ because of the presence of endogenous trp ϩ , even though the ratio of mutant to wild-type trp gene copy number was one, as in Trp P365 /ϩ (2:2 in the transgenic flies and 1:1 in Trp P365 /ϩ). In fact, ERGs of all transgenic flies carrying two copies of mutation 3 tested were not nearly as severely mutant as that of Trp P365 /ϩ (compare Fig. 2, a-c and Fig. 3b), suggesting that the presence of the extra wild-type copies of the trp gene might be influencing the ERG phenotype.
To assess properly whether or not mutation 3 corresponds to Trp P365 /ϩ, we needed to carry out a complementation test in the absence of the endogenous wild-type trp. Accordingly, we placed the P[trp(3)] and P[trp(1234)] transgenes in a trp-null (trp P343 ) background. As a positive control, the P[trp(124)] transgene was also placed in a trp-null background. If mutation 3, indeed, is the same as Trp P365 , one would expect that transgenic flies homozygous for a transgene harboring mutation 3 would display phenotypes closely resembling those of Trp P365 . If mutation 3 is absent from the transgene, on the other hand, one would expect the transgene to fully complement the trpnull background. Fig. 3 compares the ERGs obtained from Trp P365 heterozygotes and homozygotes (Fig. 3, b and e, respectively) with those obtained from transgenic flies homozygous for the transgenes harboring mutation 3 alone (P[trp(3)]) (Fig. 3c), all four mutations together (P[trp(1234)]) (Fig. 3d), or the only triplet combination that does not include mutation 3 (P[trp(124)]) (Fig.  3a), all placed in a trp P343 background. ERGs were tested at three different ages, 0, 3, and 7 days post-eclosion. Because ERGs showed very little age-dependent variations, particularly between 3 and 7 days post-eclosion, results from 7 days were not included in Fig. 3. They were very similar to those at 3 days post-eclosion.
At all three ages tested, the ERGs obtained from transgenic flies carrying (P[trp(124)]) were completely wild-type (Fig. 3a). That is, the transgene completely complemented the null trp background when it harbored all the other three mutations but not mutation 3. The ERGs of transgenic flies carrying transgenes P[trp(3)] and P[trp(1234)], on the other hand, were both much more severely mutant than those of Trp P365 /ϩ heterozygotes and approached those of Trp P365 homozygotes in severity. There was a tendency for ERGs of flies carrying mutation 3 alone to be somewhat less severe than those of flies carrying all four mutations (compare Fig. 3, c and d), but even the latter did not quite match ERGs of Trp P365 in severity (compare Fig. 3, d  and e). At least four different transgenic lines were tested for each genotype, and the above generalizations held true across different transgenic lines, making it unlikely that the minor differences in phenotype between the transgenic flies and the original mutant are due to differences in the sites of insertion of transgenes.
Thus, the results of the complementation test were consistent with the idea that mutation 3, and it alone, corresponds to the original Trp P365 mutation, although mutation 3 alone never quite matched the severity of Trp P365 in ERG phenotypes. The slight disparity between behaviors of the transgenes and Trp P365 will be discussed under "Discussion." Confocal Microscopy-To determine whether the ERG phenotypes of these mutants are accompanied by photoreceptor degeneration and whether mutation 3 alone is sufficient to cause degeneration, we carried out confocal microscopy. Microscopy was performed on the retinas of the same five classes of flies that were tested for their ERGs (Fig. 3), as well as wild-type and null trp mutant (trp P343 ) flies, at 0, 3, and 7 days post-eclosion. The retinas were stained with phalloidin to label filamentous actin in the rhabdomeres. As may be seen in Fig.  4A, the rhabdomeres appeared normal in both wild-type flies ( Fig. 4A-a) and trp P343 null mutants (Fig. 4A-b), as well as transgenic flies carrying mutations 1, 2, and 4, (Fig. 4A-c), at all three ages. In Trp P365 /ϩ heterozygotes, however, the rhabdomeres of some ommatidia began showing abnormalities by day 3. Some rhabdomeres looked elongated in cross section, and some seemed to be breaking up in two ( Fig. 4B-a, middle  panel, arrows). By day 7, most rhabdomeres were diffusely labeled, indicating that they were no longer structurally intact, and most ommatidia no longer showed a full complement of seven rhabdomeres (Fig. 4B-a, right panel). In Trp P365 homozygotes, only faint diffuse images of a few isolated rhabdomeres were detectable even at day 0 post-eclosion (Fig. 4B-d, left  panel). The degeneration phenotypes of the two classes of transgenic flies, P[trp(3)] and P[trp(1234)] (Fig. 4B-b and - and Trp P365 homozygotes (e), all at day 0 (first column) and day 3 (second column) post-eclosion. ERGs were also obtained from flies of all five genotypes at 7 days post-eclosion, but they were very similar to those obtained from 3 days post-eclosion and are not shown. The flies were made homozygous for the transgene so that each class of transgenic flies carried two copies of the mutant transgene but no wild-type trp gene. In all transgenic lines, the background trp-null stock was marked with w Ϫ . The control Trp P365 hetero-and homozygous stocks were also marked with w Ϫ . ERGs recorded from transgenic flies of the genotype, P[trp(124)];trp P343 , (a) are indistinguishable from those of wild-type. The stimuli were as described in Fig. 1. respectively), were intermediate between those of Trp P365 homozygotes and heterozygotes but more closely resembled that of homozygotes than that of heterozygotes. The resemblance to Trp P365 homozygotes was much stronger than was evident from ERG recordings. Thus, the degree of degeneration in transgenic flies carrying all four mutations was almost indistinguishable from that of homozygotes, particularly at days 3 and 7 post-eclosion (compare Fig. 4, B-c and -d). Degeneration in the transgenic flies carrying only mutation 3 was somewhat milder. At day 0 post-eclosion, most rhabdomeres appeared nearly normal. However, the label was very faint, and some rhabdomeres appeared to be splitting up ( Fig. 4B-b, left panel,  arrow). By day 3, most rhabdomeres were no longer detectable, and only a few isolated, faintly labeling rhabdomeres remained. This was in strong contrast to Trp P365 /ϩ, in which most rhabdomeres labeled distinctly even at day 7. Transgenic flies carrying mutations 1, 2, and 4 had completely normal rhabdomeres at all three ages tested (Fig. 4A-c). These results were completely consistent with ERG results (Fig. 3) except that transgenic flies carrying mutation 3 alone more closely resembled the original Trp P365 mutation than in ERG recordings. DISCUSSION By whole cell recordings from dissociated photoreceptor cells, Yoon et al. (9) showed earlier that in Trp P365 the TRP Ca 2ϩ channels are constitutively open, suggesting that the excessive influx of Ca 2ϩ through the TRP channels causes the massive photoreceptor degeneration observed in the mutant. These au-  (19,20); DTRPL, D. melanogaster TRPL protein (21); DTRP␥, Drosophila TRP␥ protein (22) (accession no. CAB96205); CTRP1, C. elegans TRP identified in genome sequencing project (23); CTRP2, another C. elegans TRP identified (24) (accession no. AAK21447); HTRPC1 Human TRP protein 1 (25) (accession nos. X89066 -8); HTRPC4, human TRP protein 4 (26) (accession no. NP 057263); Shaker, the Shaker K ϩ channel protein (27)(28)(29). Five other mammalian TRP-related proteins of the TRPC1 subfamily examined also had the same S5 sequence as HTRPC1, and seven other mammalian TRP-related proteins of the TRPC2, 4, and 5 subfamilies examined had the same S5 sequence as HTRPC4. thors proposed that the mutation(s) in the trp gene of Trp P365 renders the regulation of the TRP channel unstable, leading to an increased likelihood that the channels open and remain open spontaneously in an age-dependent and Trp P365 dosagedependent manner. Their results thus suggested that the amino acid residues that are altered in the TRP channel of Trp P365 might be important for the regulation of the TRP channel activity. We have shown in the present work that, of the four amino acid alterations detected in the TRP channel of Trp P365 (9), the F550I change in the fifth transmembrane segment is the critical mutation for the Trp P365 phenotype.
The ERG and degeneration phenotypes of transgenic flies homozygous for mutation 3 alone or all four mutations together nearly approximate the phenotypes of Trp P365 homozygotes when the transgenes are placed in a trp-null background (P[trp (3)];trp P343 and P[trp(1234)];trp P343 ) (Figs. 3 and 4). However, these phenotypes do not quite attain the severity of Trp P365 homozygotes (Figs. 3 and 4). Moreover, the phenotypes of transgenic flies carrying only mutation 3 tend to be slightly milder than those of transgenic flies carrying all four mutations (compare Fig. 3, c and d; Fig. 4, B-b and -c). Because these behaviors of transgenes are observed in several different lines of transgenic flies for each transgene, they are not likely to be due to differences in expression levels arising from different sites of transgene insertion. On the other hand, the transgenes had a trp promoter sequence of only ϳ680 bp and were devoid of introns (see "Experimental Procedures"). This use of abbreviated gene constructs in the transgenes could potentially account for the minor differences in behavior between the mutation 3-harboring transgenes and the original Trp P365 mutation. The abbreviated gene construct could, for example, affect the level of protein expression or the processing of gene products. The effects appear to be weak because the observed differences are only minor. On the other hand, the slight difference in ERG and degeneration phenotypes between transgenic flies carrying mutation 3 alone and those carrying all four mutations cannot be explained either by differences in sites of transgene integration or the abbreviated gene constructs used in transgenes. It may well be that mutations 1, 2, and 4 have minor synergistic effects on mutation 3, even though alone they do not produce any phenotype. The fact remains, however, that mutation 3 must be present for the mutant phenotypes to appear at all (Figs. 1, 2, 3a, and 4A-c). Moreover, mutation 3 alone is sufficient to produce phenotypes closely mimicking those of Trp P365 (Figs. [1][2][3][4]. That is, mutation 3 alone is necessary for the ERG and degeneration phenotypes of Trp P365 and sufficient for phenotypes closely matching those of Trp P365 . We conclude that, in all likelihood, mutation 3 corresponds to the original mutation Trp P365 . The amino acid alteration caused by mutation 3 is surprisingly subtle considering the drastic effect it has on the mutant phenotype. It involves the substitution of one non-polar residue for another. Alignment of the TRP S5 sequence with those of other TRP-related channel subunits shows that Phe-550 of TRP is conserved in S5 sequences of all fly TRPs but not all other invertebrate TRP-related proteins tested (Fig. 5). In the two Caenorhabditis elegans TRP-related proteins examined, this position is occupied by Phe in one, as in flies, and Cys in the other. On the other hand, in all mammalian TRP-related proteins tested (a total of 14 mammalian TRP sequences) a conserved Leu residue occupies this position (Fig. 5). In no case, however, is this position occupied by Ile. The following are some tentative conclusions from these and the previous observations. 1) This position is important for the regulation of TRP channel opening. 2) The particular amino acid residue preferred for this position appears to be different for different species, Phe for Drosophila and Leu for mammals, perhaps because of slightly different channel environments. 3) Ile appears to be particularly poorly tolerated at this position. In the following, we explore speculatively why this position might be important for the regulation of TRP channel opening and why Ile, in particular, may be a poor residue to occupy this position.
In a classic study, Doyle et al. (16) determined the structure of the KcsA K ϩ channel of Streptomyces lividens by x-ray crystallography. The KcsA channel is formed by four subunits, each of which contributes two transmembrane ␣-helices, one (inner helix) facing the pore and the other (outer helix) facing the hydrophobic membrane; these ␣-helices form the backbone of the channel. The four inner helices are arranged like the posts of an inverted teepee and cross in a bundle to produce a small hole near the intracellular entrance of the channel. Results of Cys accessibility studies (17) on voltage-gated Shaker K ϩ channels suggest that the hole formed by the bundle crossing serves as the gate for these channels. This picture of the channel structure probably applies to such other ion channels as various voltage-gated channels and TRP-related channels. Most of these channels, however, have six transmembrane segments, rather than two, in each subunit or homology domain. For these channels, the sixth transmembrane segment, S6, corresponds to the inner helix, and the fifth transmembrane segment, S5, corresponds to the outer helix of the KacsA channel. Thus, while the S5 segments do not define the channel gate themselves, they are immediately adjacent to S6 segments and are in a position to influence the movements of S6 segments. Hydropathy analysis places Phe-550 near the N-terminal end of S5, corresponding to the intracellular end of the segment. This is also the region of the channel at which S6 segments are thought to make bundle crossing to form the gate. Thus, the residues at positions 550 or in its immediate vicinity may be in a position to critically affect channel gating.
Why would Ile at this position cause constitutive activity of the channel but not the residues normally found at this position? It turns out that Ile has a side chain property that it does not share with Phe, Leu, or Cys. Ile is branched at the ␤-carbon position, whereas none of the other three has ramifications at the ␤-position. Branching at ␤ is expected to decrease the flexibility of the main chain through steric hindrance provided by the ␤-branched side chains (15). The presence of a ␤-branched residue close to the gate may restrict the movements of the helices forming the gate and could lock the gate in its resting state. Agam et al. (18) showed that the default state of the TRP and TRPL channels is the open state and that a continuous energy input from an ATP-dependent process is required to keep the channels closed in the absence of light stimulus. Thus, constitutive opening of the TRP channel may represent locking the channel in the lower-energy, default state, i.e. its resting state.