Distinct Phospholipase C--Dependent Signaling Pathways in the Drosophila Eye and Wing Are Revealed by a New small wing Allele

Corresponding author: Justin R. Thackeray, Clark University, 950 Main St., Worcester, MA 01610., jthackeray{at}clarku.edu (E-mail)

Communicating editor: T. SCHÜPBACH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The Drosophila genome contains a single phospholipase C- (PLC-) homolog, encoded by small wing (sl), that acts as an inhibitor of receptor tyrosine kinase (RTK) signaling during photoreceptor R7 development. Although the existing sl alleles behave genetically as nulls, they may still produce truncated Sl products that could in theory still provide limited PLC- function. Both to identify a true null allele and to probe structure-function relationships in Sl, we carried out an F₁ screen for new sl mutations and identified seven new alleles. Flies homozygous for any of these alleles are viable, with the same short-wing phenotype described previously; however, two of the alleles differ from any of those previously isolated in the severity of the eye phenotype: sl⁹ homozygotes have a slightly more extreme extra-R7 phenotype, whereas sl⁷ homozygotes have an almost wild-type eye. We determined the mutant defect in all seven alleles, revealing that sl⁹ is a molecular null due to a very early stop codon, while sl⁷ has a missense mutation in the highly conserved Y catalytic domain. Together with in vitro mutagenesis of the residue affected by the sl⁷ mutation, these results confirm the role of Sl in RTK signaling and provide evidence for two genetically separable PLC--dependent pathways affecting the development of the eye and the wing.

PHOSPHOLIPASE C- (PLC-) is involved in regulating a diverse array of cellular processes, including proliferation, differentiation, and motility (reviewed by REBECCHI and PENTYALA 2000 ). PLC- activation is triggered by the binding of a wide variety of growth factors, cytokines, and immunoglobulins to their membrane-bound receptor. The activated enzyme hydrolyzes the membrane phospholipid phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] into two intracellular second messengers: inositol 1,4,5-trisphosphate [I(1,4,5)P₃], which increases intracellular [Ca²⁺] by release from the endoplasmic reticulum, and diacylglycerol, which, in combination with Ca²⁺, activates the classical isoforms of protein kinase C. More recently it has become clear that PLC- must be viewed as more than simply a phospholipase. One study showed that a PLC- protein lacking phospholipase activity was able to stimulate a partial mitogenic response (SMITH et al. 1994 ), and another study showed that this ability depends on an intact SH3 domain (P. S. HUANG et al. 1995 ). The PLC- SH3 domain has now been shown to bind to an enhancer of phosphatidylinositol-3-OH kinase [PI(3)K] called PI(3)K enhancer (PIKE), acting as a guanine nucleotide exchange factor for PIKE, and thereby augmenting nuclear PI(3)K activity (YE et al. 2002 ). PLC- is therefore able to stimulate mitogenesis even when incapable of phospholipid hydrolysis.

Two distinct but very similar PLC- isoforms have been described in mammals, PLC-₁ and PLC-₂, which are encoded by different genes. Both are widely expressed throughout development, but the ₂ isoform is most abundant in cells derived by hematopoiesis (HOMMA et al. 1989 ). Mice homozygous for knockout mutations of PLC-₁ die during the early stages of embryogenesis (JI et al. 1997 ), whereas PLC-₂ knockouts are viable but result in impaired B-cell development and a general failure of B-cell antigen receptor signaling (HASHIMOTO et al. 2000 ; WANG et al. 2000 ). Vertebrates have three other PLC types, PLC-ß, PLC-, and PLC-, which differ from PLC- in overall structure, pattern of expression, and mode of activation, but are thought to have similar catalytic abilities (reviewed by RHEE 2001 ).

A single PLC- gene has been identified in Drosophila, encoded by small wing (sl) (THACKERAY et al. 1998 ). This sequence is the only PLC- homolog present in the genome and is equally similar to mammalian PLC-₁ and PLC-₂ (MANNING et al. 2003 ). Four mutant sl alleles have been recovered since its original discovery by Bridges in 1915. Three of these alleles are extant: sl¹, sl², and sl³ are all recessive and homozygous viable alleles, with a slightly shortened wing and a mildly rough eye (MORGAN et al. 1925 ; SIVERTZEV-DOBZHANSKY and DOBZHANSKY 1933 ; SCHALET 1986 ). One additional allele, sl³⁴—isolated by Gottschewski in 1934 but now lost—was reported to differ from the other alleles in that it had shortened wings but normal eye morphology (LINDSLEY and ZIMM 1992 ). More recently it was shown that sl^1,2,3 homozygotes have extra-R7 photoreceptor cells in 30–50% of the ommatidia (unit eyes) of the compound eye, and their wings occasionally contain ectopic patches of wing-vein-like material (THACKERAY et al. 1998 ). Assignment of cell fate in the developing eye and wing of Drosophila depends on the correct timing of a number of signals, especially those derived from the Notch, Sevenless, and Egfr signaling pathways (DIAZ-BENJUMEA and HAFEN 1994 ; FREEMAN 1997 ; BRENNAN and MOSES 2000 ). In the case of the receptor tyrosine kinases (RTKs) Sev and Egfr, each passes a signal to the nucleus via the highly conserved Ras/Raf/mitogen-activated protein kinase (MAPK) "cassette" (reviewed by TAN and KIM 1999 ).

The mild effect of the sl alleles is surprising for a variety of reasons. First, null mutations in almost all the other members of the Ras/Raf/MAPK-mediated signaling pathways are lethal (ZIPURSKY and RUBIN 1994 ). Second, loss of mammalian PLC-₁, which probably has the most similar functional role to Sl among the two mammalian PLC- homologs, results in early embryonic lethality in mouse knockouts. One possible explanation of this puzzle is that PLC- is involved in mammalian RTK signaling both as a positive (J. HUANG et al. 1995 ) and as a negative (OBERMEIER et al. 1996 ) regulator. However, the three extant sl alleles have molecular defects that would truncate Sl, but not necessarily eliminate all Sl functions (THACKERAY et al. 1998 ); despite this, the sl¹ and sl² alleles both behaved genetically as nulls when made heterozygous with deficiencies or duplications for the region containing sl (THACKERAY et al. 1998 ). Genetic epistasis experiments in the same study showed that Sl acts as a negative regulator upstream of MAPK and is most likely activated by Egfr. However, the shortened wing phenotype appears to be due to disturbance of a MAPK-independent pathway, because it is not suppressed by rl¹, a hypomorphic allele of the MAPK encoded by rolled. In contrast, rl¹ suppresses both the extra-R7 and ectopic wing-vein phenotypes completely (THACKERAY et al. 1998 ).

Although the mechanism by which Sl inhibits RTK signaling is not yet determined, a strong positive genetic interaction between sl alleles and mutations of GTPase activating protein 1 (Gap1; POWE et al. 1999 ) suggests that it may be via indirect downregulation of Ras1. Gap1 was first identified as a negative regulator of Ras-mediated signaling in the eye (GAUL et al. 1992 ), where it is presumed to promote conversion of active Ras-GTP to the inactive GDP-bound form. On the basis of the genetic interaction between sl and Gap1 and other evidence, a model was proposed in which Gap1 is activated by the PLC--induced rise in concentration of both inositol 1,3,4,5-tetrakisphosphate [I(1,3,4,5)P₄] and Ca²⁺, resulting in the downregulation of Ras by the activation of its GTPase activity (POWE et al. 1999 ).

The "standard model" for mammalian PLC- activation involves binding of a src-homology-2 (SH2) domain in PLC- to the activated receptor, followed by phosphorylation of one or more tyrosine residues as a result of RTK activity or from a receptor-associated tyrosine kinase such as Src (reviewed by RHEE 2001 ). However, it has also been shown that PLC- can be activated without tyrosine phosphorylation by binding phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P₃] to either the N-terminal pleckstrin homology (PH) domain (FALASCA et al. 1998 ) or the C-terminal SH2 domain (RAMEH et al. 1998 ). PLC- contains several additional motifs common to many signaling proteins: a second PH domain, an EF hand region, an N-terminal SH2 domain, one SH3 domain, and a C2 domain. Although some attempts have been made to dissect the roles of these domains in mammalian PLC- by in vitro mutagenesis (CHATTOPADHYAY et al. 1999 ), such targeted approaches require that prior assumptions be made about the likely functional importance of particular amino acids. By contrast, a random mutagenesis in vivo can identify important residues purely on the basis of the phenotypic consequences of their loss. In this study we describe such a screen for new sl alleles, using ethyl methanesulfonate (EMS) mutagenesis. Seven alleles were isolated and characterized, revealing that the originally identified sl alleles may still retain some function and the phenotype of one allele demonstrates the presence of independently mutable pathways for PLC- signaling in the eye and wing.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly stocks:
Flies were raised on a standard cornmeal, molasses, yeast, agar medium at 25°. The w sl¹ stock was originally derived by recombination from a ClB balancer chromosome as described previously (THACKERAY et al. 1998 ). The M9/ClB balancer stock was generated from the ClB stock described above, and an M9 stock was provided by the late Abe Schalet (Yale University); M9 is a version of In(1)sc⁸ +dl49 and carries the following markers: y^31d sc⁸ w^a mei^9L1 vb v^0f f (LINDSLEY and ZIMM 1992 ). The y cv v f and FM6/FM7c stocks were originally obtained from the Indiana Stock Center.

EMS mutagenesis:
y cv v f males were collected within 24 hr of eclosion and aged for 2–5 days on yeasted vials. After being allowed to feed on a solution of 0.2% EMS (Sigma, St. Louis) in 1% sucrose overnight, the males were placed on fresh standard food vials for 24 hr and then mated with virgin w sl¹ females. A line was established from each F₁ female carrying any combination of rough eyes, altered wing veination, or shortened wing by crossing to M9 males. The male forked progeny (f lies 3 cM proximal to sl) in each line were rescored for sl-like phenotypes; if no forked males were recovered among the F₂, indicating a lethal hit at or near sl, virgin v f F₂ females (which will carry the mutagenized X chromosome balanced over M9) were backcrossed to w sl¹ males to allow hits at sl to be scored. Lines carrying a new sl mutation were outcrossed to wild type and all other visible X-linked markers were removed by standard recombination crosses before being balanced over FM7c. Histological analysis of wings and eyes was performed as described previously (THACKERAY et al. 1998 ).

Determination of sl mutations:
Genomic DNA was isolated from homozygous adults of each strain and the entire sl gene examined for mutations by single-strand conformational polymorphism (SSCP) essentially as described (ORITA et al. 1989 ). The gene was amplified in six overlapping PCR reactions, using the following primer pairs:

• chunk 1: GN5/SG3 (5' GTTTGTATGCATTGCACTTA 3'/5' GGCAATGGCTGCGCTACTGA 3', corresponding to positions 162–181 and 1094–1113 of the published sl sequence; EMORI et al. 1994 );

• chunk 2: CP5/CP1 (5' GCAGTTCCTCGAGACGGAAC 3'/5' GTTGCGTAAGCAGCATATCC 3', 1050–1590, includes an additional 358 bp of intron 1);

• chunk 3: CP4/CP6 (5' CCTGACGGGCGATCAGTTCTCCA3'/5' GGAGTTTCCGTTGCGTGACTCG 3', 1302–1935, also includes intron 1);

• chunk 4: CP7/CP3N (5' ACTTCAAGGATCCGGTCGATAA3'/5' GGACATTCGTTATGATGGCATGC 3', 1828–2849, includes 65 bp of intron 2);

• chunk 5: CP2N/GYB3 (5' CTTCGAGTCTCTGGTCTCGCTGA 3'/5' AGTTCAGTGCGATCATCTGGGA 3', 2565–3460, includes 59 bp of intron 3);

• chunk 6: CP8/GH3N (5' CTCCTACCATCGCAATCAGATT3'/5' TCATACGTTTCAGATCCTAAGCT 3', 3345–4191).

These amplifications include the whole sl open reading frame (ORF) from 115 bp upstream of the initiating methionine to 204 bp downstream of the stop codon, as well as all three introns.

Each PCR reaction was carried out essentially as described (THACKERAY et al. 1998 ), except that the reaction volume was 20 µl and contained 2 µCi [³²-P]dATP. Parameters used were 94° for 45 sec, 56° for 45 sec, and 72° for 2 min for 36 cycles. An aliquot of each reaction was then digested in two different single-enzyme digests in a 15-µl volume as follows: chunk 1 with AluI and DpnII, chunk 2 with TaqI and HpaII, chunk 3 with TaqI and HhaI, chunk 4 with AluI and FokI, chunk 5 with TaqI and HpaII, and chunk 6 with AluI and HinfI. Stop solution (10 µl 95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol) was added, the sample was denatured at 100° for 2 min, and 3 µl was separated on a nondenaturing mutation detection enhancement acrylamide gel according to the manufacturer's instructions for standard SSCP (Cambrex, East Rutherford, NJ). After autoradiography, banding patterns between sl mutants and the y cv v f parental line were compared; a single region with altered mobility compared to the parental line was found for each mutant allele. The PCR reaction containing the band with altered mobility was repeated from a fresh genomic DNA sample and sequenced without subcloning using the fmol sequencing kit (Promega, Madison, WI) with internal primers predicted to be close to the site of mutation. A parallel reaction from the y cv v f parent was also sequenced with the same primers. A single mutation was found in each allele.

Genomic rescue constructs and germline transformation:
A Drosophila genomic library in EMBL3 (provided by Iain Dawson, Yale University) was screened with an sl cDNA probe. A positively hybridizing clone (G1) was purified by two rounds of rescreening, genomic DNA was prepared, and a restriction map of the G1 insert was generated (SAMBROOK et al. 1989 ). An 10-kb SalI/XbaI fragment (X10) was identified within G1, which includes the entire sl transcription unit and no others, by comparing the restriction map to a published map (JONES and RUBIN 1990 ) and to the genomic sequence of the region (ADAMS et al. 2000 ). The X10 fragment was subcloned into pBluescript KS, reisolated by a KpnI/NotI digest (enzymes that do not cut within X10, but which cleave on either side of the XhoI and XbaI sites in the pBluescript polylinker), and ligated into the KpnI and NotI sites of pCaSpeR-4. Modified versions of X10 were produced by site-directed mutagenesis using the Quikchange kit (Stratagene, La Jolla, CA), following the protocol supplied by the manufacturer. Mutagenic primers (5' GATTAGCCGAGTGTATGTGAAGGGTCAACGCCT 3' and 5' AGGCGTTGACCCTTCACATACACTCGGCTAATC 3' for the P1035V mutation; 5' GATTAGCCGAGTGTATGCGAAGGGTCAACGCCT 3' and 5' AGGCGTTGACCCTTCGCATACACTCGGCTAATC 3' for P1035A) were purchased from IDT (Coralville, IA) and full-length oligonucleotides were purified by PAGE as described (SAMBROOK et al. 1989 ). The mutations were generated in a 1.7-kb BamHI fragment subcloned from X10 and confirmed by sequence analysis of the entire 1.7-kb fragment. A modified X10 fragment was then reconstructed in pBluescript using the mutated 1.7-kb BamHI fragment and recloned into the KpnI and NotI sites of pCaSpeR-4. The plasmid DNA preparation for each modified construct to be injected was sequenced just prior to injection to confirm the presence of the expected mutation. Germline transformations were performed essentially as described (CRIPPS and BERNSTEIN 2000 ); the injection buffer had a final concentration of 1 µg/µl pCaSpeR-4::X10 and 0.1 µg/µl 2-3 helper plasmid in a 0.1 mM sodium phosphate buffer pH 7.8, 5 mM KCl. Injections were carried out in w¹¹¹⁸ embryos and the presence of the transposon was detected in the progeny of the injectees by the presence of the w⁺ gene marker of pCaSpeR-4. The transposon insertion was mapped using w;Bl/CyO or w;Bsb/TM6 stocks and balanced lines were established. The transgenes were crossed into a w sl¹ or w sl⁹ background for phenotypic analysis.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of seven new sl alleles:
We screened 20,000 EMS-mutagenized X chromosomes for new sl mutations in a standard F₁ screen, crossing mutagenized sl⁺ males to sl¹ females. Seven lines were recovered that showed a similar, X-linked recessive, sl-like phenotype both in trans-heterozygous combination with sl¹ and in a homozygous condition. The simplest interpretation is that each line represents a new sl allele; the seven mutations were therefore designated sl⁴ ... sl¹⁰.

sl⁷ and sl⁹ eye phenotypes differ from previously isolated alleles:
Adult eyes from females homozygous for all seven new alleles were sectioned and stained with toluidine blue to reveal photoreceptor cell fate. We found that five of the seven new alleles (sl^4,5,6,8,10) showed a similar percentage —ranging between 30 and 50%—of ommatidia with extra-R7 cells, as we previously described (THACKERAY et al. 1998 ) for the older alleles (Fig 1A–E, and Fig 2A). However, one allele, sl⁹, showed a slightly higher frequency, with a mean of 58.9% of mutant ommatidia. Indeed, of the six sl⁹ eyes in our sample, three had >80% of ommatidia with extra-R7 cells; by contrast, only one other eye of any other genotype even showed 65% of mutant ommatidia. Although the small size of the samples prevents a meaningful statistical comparison, these results and those described below suggest that the sl⁹ allele has a more extreme eye phenotype than that of any other previously characterized sl allele. The remaining allele, sl⁷, showed a very low level of mutant ommatidia compared to that of all the others, with a mean of only 7.1% showing one or more extra-R7 cells.

View larger version (0K): In this window In a new window Download PPT slide   Figure 1. Eye and wing phenotypes in the new sl alleles. Plastic sections of adult eyes stained with toluidine blue (left column) and wings (right column) are shown from females of the Canton-S strain used for outcrossing (A and F) and from females homozygous for sl5 (B and G), sl7 (C and H), sl8 (D and I), and sl9 (E and J). Open arrowheads indicate a mutant ommatidium containing one or more extra-R7 cells. Note in E the very high frequency of mutant ommatidia in the sl9 homozygote. The solid arrowhead in I indicates an example of ectopic wing vein that appears in some wings from homozygous animals of all seven of the new alleles.

View larger version (0K): In this window In a new window Download PPT slide   Figure 2. Histograms of eye and wing defects in the new sl alleles. (A) Percentage of ommatidia containing one or more extra-R7 cells in homozygous females of each genotype; n = 6 for each genotype, and error bars show the standard error of the mean. The solid bars indicate data described elsewhere (THACKERAY et al. 1998 ). The numbers on the x-axis refer to the allele designation. (B) Wing lengths of homozygous adults; n = 30, and error bars indicate the standard error of the mean.

The wing length of females homozygous for all seven new alleles was comparable to the length previously described for sl¹ and sl² (THACKERAY et al. 1998 ) at 80% of the normal length (Fig 1, F–J, and Fig 2B). Adults homozygous for all three of the preexisting sl alleles contain ectopic wing veins, principally beside vein LII, as well as less frequently in the third posterior cell or near the posterior crossvein (THACKERAY et al. 1998 ). Ectopic veins were observed in wings from females homozygous for each of the seven new alleles at comparable or slightly lower frequency (10–40% of wings) and in positions identical to the original alleles. The frequency and extent of ectopic wing-vein formation is rather variable even within a given sl genotype (R. MANKIDY and J. R. THACKERAY, unpublished data), so we did not attempt to quantify this phenotype more precisely in the new alleles.

The sl⁷ phenotype shows that the eye and wing pathways are separable:
Because the sl phenotype can be partially suppressed by uncharacterized modifiers (R. MANKIDY and J. R. THACKERAY, unpublished observations), we outcrossed a stock carrying the sl⁷ mutation again to determine whether the very mild eye phenotype of this allele is genuine; as a control, we outcrossed a stock carrying the sl⁶ mutation in a parallel set of crosses. After two further rounds of outcrossing and reisolation of the chromosomes carrying the sl⁶ and sl⁷ mutations, the eye and wing phenotypes were determined again. In each case the results were consistent with the original analysis: wing length was in the typical range seen for sl homozygotes (sl⁶ mean = 1.07 mm; sl⁷ mean = 1.24 mm, n = 30 for both genotypes), while 43.1% of sl⁶ (n = 4, SEM = 4.3) and 3.7% of sl⁷ (n = 6, SEM = 1.7) ommatidia from homozygous females had one or more extra-R7 cells. This confirms that the unique combination of a null-length wing but almost wild-type eye is the true sl⁷ homozygous phenotype and shows that the eye and wing phenotypes are genetically separable.

To characterize the nature of the sl⁷ allele in greater detail, we examined it in trans-heterozygous combination with two other alleles, sl⁸ and sl⁹. Trans-heterozygous adults in all three combinations—sl⁷/sl⁸, sl⁷/sl⁹, and sl⁸/sl⁹—had wing lengths indistinguishable from each other and typical of homozygotes for any single allele (Fig 3, E–H). In the eye, however, the sl⁸/sl⁹ trans-heterozygotes showed a phenotype equivalent to sl⁸ homozygotes: 45.9% of ommatidia had extra-R7 cells (SEM = 9.0, n = 6), whereas sl⁷/sl⁸ and sl⁷/sl⁹ trans-heterozygotes both showed a phenotype indistinguishable from sl⁷ homozygotes: 10.2% (SEM = 1.7, n = 6) and 7.8% (SEM = 2.6, n = 5) of ommatidia had extra-R7 cells, respectively (Fig 3, A–D and I). Because the severity of the eye phenotype of sl⁷/sl⁸ and sl⁷/sl⁹ females is not simply intermediate between the values seen in the homozygotes (Fig 2A), these results suggest that the sl^7,8,9 alleles do not differ simply by their quantitative level of Sl activity; instead, the sl⁷ allele appears to provide a PLC- function that is missing from both sl⁸ and sl⁹ during photoreceptor cell development. Furthermore, because sl⁷ is clearly equivalent to sl⁸ and sl⁹ with respect to the developmental pathway(s) affecting the length of the wing blade, these results imply that the role of Sl during development differs in some way in the ommatidia and the wing.

View larger version (0K): In this window In a new window Download PPT slide   Figure 3. Eye and wing phenotypes of sl trans-heterozygotes. Plastic sections of adult eyes stained with toluidine blue (left column) and wings (right column) are shown of females from the following genotypes: Canton-S (A and E), sl7/sl8 (B and F), sl7/sl9 (C and G), and sl8/sl9 (D and H). The arrowheads in B and C indicate relatively rare ommatidia containing an extra-R7 cell when sl7 is trans-heterozygous with one of the null alleles. (I) Histogram showing the frequency of photoreceptor R7 defects in sl trans-heterozygotes. Error bars indicate standard error of the mean; n = 6 for sl7/sl8, n = 5 for sl7/sl9, and n = 6 for sl8/sl9. Data for homozygotes are taken from Fig 2. Note the much greater frequency of extra-R7 cells in the sl8/sl9 trans-heterozygote, indicated in D by arrows.

The sl mutations are scattered throughout the ORF:
We scanned the sl transcription unit between the start and stop codons in each of the new alleles by SSCP, looking for changes relative to the parental X chromosome. A single change was found in each allele, as illustrated in Fig 4A. Four of the seven alleles, sl⁴, sl⁶, sl⁹, and sl¹⁰, contain a lesion expected to produce a truncated Sl product. Among this group sl⁹ is particularly revealing, because the stop codon in this allele occurs in codon number 54 out of 1236. Any mutant Sl protein produced in sl⁹ homozygotes would therefore lack all recognized PLC- domains, clearly indicating that sl⁹ is a null allele. The mutation in the sl⁵ allele is an in-frame deletion of 15 bp that removes a five-amino-acid segment from the C2 domain, indicating that this domain is indispensible for Sl function. The two remaining alleles both contain missense mutations in a catalytic domain: sl⁷ has a P1035L mutation within region Y and sl⁸ has a G385D mutation within region X.

View larger version (0K): In this window In a new window Download PPT slide   Figure 4. The location and nature of molecular lesions in the new sl alleles. (A) The position of each mutation is indicated by an arrowhead either above or below a schematic representation of the Sl protein. The position of all recognized domains within PLC- homologs is indicated by shaded boxes. PH, pleckstrin homology domain (note that the central PH domain is split, which is indicated by the boxes containing "P-" and "-H"); EF, EF hands; X/Y, phospholipase catalytic domains; SH2, src homology 2 domains; SH3, src homology 3 domain; C2, C2 domain. The amino acid replacements generated by the sl7 and sl8 mutations are indicated by P1035L and G358D, respectively. (B) Alignment of Pro1035 in Sl with the homologous part of region Y in 11 other PLC proteins. This was produced by augmenting a previously published alignment (ELLIS et al. 1998 ) with two recently identified PLC- homologs and Sl. Note that the proline homologous to Pro1035 of Sl is conserved across all species in all PLC homologs described to date.

Further mutagenesis of Pro¹⁰³⁵:
Because the sl⁷ mutation may identify a region that is differentially involved in eye and wing PLC- signaling, we investigated this mutation further. Position 1035 in Sl is homologous to a site within region Y that contains a proline in all PLC subtypes (ß, , , and ) described to date from plants, yeast, Drosophila, and mammals (Fig 4B; ELLIS et al. 1998 ). Analysis of the crystal structure of rat PLC-1, the only PLC for which a structure has been determined, places the homologous proline (Pro⁵⁵²) immediately adjacent to one of three loops that form a hydrophobic ridge surrounding the active site. The ridge is thought to insert into the membrane, allowing phospholipid substrates access to the active site of the enzyme (ESSEN et al. 1996 ). Because the X and Y catalytic domains are highly conserved in all PLC proteins and are therefore likely to be structurally and functionally similar in all four PLC subtypes, this proline probably serves the same structural role in Sl. The sl⁷ mutation might therefore affect the ridge in at least two ways: loss of the proline might twist part of the ridge into a less favorable orientation or the increased hydrophobicity due to the introduction of a leucine could alter its interaction with the membrane or substrate in some way.

To determine whether the sl⁷ phenotype is generated by the loss of the proline at 1035 or by the presence of the leucine at the same site, we followed an in vitro mutagenesis strategy to reveal the effect of two alternative changes to Pro¹⁰³⁵. First, we generated a germline rescue construct (X10) containing 10 kb of genomic DNA that includes the sl open reading frame and several kilobases of sequence upstream and downstream of the translation start and stop codons, respectively. A single copy of X10 rescued the extra-R7 phenotype of sl¹ and sl⁹ homozygotes completely; for example, among 10 sl⁹ homozygotes containing X10 (five individuals each from two independent insertions), only one ommatidium containing an extra-R7 cell was observed (Fig 5A and Fig B, and Fig 6A). The wing length of the X10-containing lines was also rescued (Fig 5E and Fig F, and Fig 6B), but because the size of adult tissues is heavily influenced by environmental conditions, it is intrinsically difficult to be certain whether or not the rescue is complete. However, we have previously found that the normal range for a wild-type wing is between 1.35 and 1.5 mm when measured along vein LIII from the anterior cross-vein to the wing margin (THACKERAY et al. 1998 ). The mean wing length of the two lines we examined containing a single copy of X10, in either an sl¹ or an sl⁹ background, was 1.49 and 1.34 mm, respectively (Fig 6B), suggesting complete rescue. These results suggest that the X10 fragment contains the complete sl transcription unit and all sequences necessary for its normal expression.

View larger version (0K): In this window In a new window Download PPT slide   Figure 5. Eye and wing phenotypes of sl9 homozygotes expressing various sl genomic constructs. Plastic sections of adult eyes stained with toluidine blue (left column) and wings (right column) are shown from females of the following genotypes: sl9 (A and E) and an sl9 sibling expressing a wild-type sl genomic construct, X10 (B and F); sl9 expressing a P1035A version of X10 (C and G); and sl9 expressing a P1035V version of X10 (D and H). The arrowhead in C shows an ommatidium containing an extra-R7 cell due to incomplete rescue by the P1035A mutation; note the complete rescue evident in the eye section from the line expressing the P1035V construct.

View larger version (0K): In this window In a new window Download PPT slide   Figure 6. Histograms of eye and wing defects in sl backgrounds expressing sl rescue constructs. (A) Frequency of photoreceptor R7 defects among siblings in backgrounds homozygous for either sl1 or sl9, lacking (-) or expressing (+) the X10 sl genomic construct, or the X10-P1035A or X10-P1035V modified constructs. Shaded bars also indicate sibs that contained the transgene. For each background the results from analysis of two independent insertion lines are shown. Error bars indicate standard error of the mean. (B) Wing lengths of the same animals shown in A; n = 30 for each genotype, and error bars indicate standard error of the mean.

Next we made two alterations to the X10 construct, replacing Pro¹⁰³⁵ with either valine (X10-P1035V) or alanine (X10-P1035A), and introduced each construct into sl mutant backgrounds. The X10-P1035V construct rescued the extra-R7 phenotype completely, with only four ommatidia among 20 heads containing an extra-R7 cell (Fig 5D and Fig 6A). By contrast, the X10-P1035A transgene showed a phenotype similar to sl⁷ in the eye with a mean of 9.3% of ommatidia with one or more extra-R7 cells (n = 10, SEM = 1.1) in two independent lines in an sl⁹ background and 2.9% (n = 10, SEM = 0.7) with the same two insertions in an sl¹ background (Fig 5C and Fig 6A). Wings from lines carrying both the X10-P1035V (Fig 5H and Fig 6B) and the X10-P1035A contructs (Fig 5G and Fig 6B) were at the lower end of the range typical of wild-type animals, suggesting complete rescue of the wing phenotype in each case. These data show that the sl⁷ phenotype cannot be attributed solely to the loss of proline or to the gain in hydrophobicity resulting from the P1035L mutation, because the P1035V construct in which both changes also occur is able to rescue sl null strains completely. The pivotal nature of Pro¹⁰³⁵ is underscored by these results, because the P1035L (i.e., sl⁷), P1035A, and P1035V mutations each generate a different combination of phenotypes: P1035L gives a weak phenotype in the eye and is null in the wing, P1035A is weakly mutant in the eye but wild type in the wing, and P1035V is wild type in both tissues.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

One of the major goals of this study was to determine whether the previously characterized sl alleles indeed represent the loss-of-function phenotype, because the three previously characterized alleles could produce a truncated protein with several domains intact. Five of the seven new alleles showed a homozygous phenotype indistinguishable from that previously described for sl¹, sl², or sl³ homozygotes. We found that the molecular defects in the new alleles vary widely in location within the protein, occurring at both N and C termini with different degrees of truncation and including an in-frame deletion and a single amino acid substitution among them. The most telling allele of all in this regard is sl⁹, which contains a nonsense codon at amino acid 54. Because mRNAs containing premature stop codons tend to be degraded by the nonsense-mediated decay pathway (HENTZE and KULOZIK 1999 ), it is unlikely that many sl⁹ transcripts would survive to be translated; however, even if some sl⁹ mRNAs escaped degradation, Sl products translated from them would have no recognized domains remaining. Therefore, the sl⁹ mutation clearly represents a true null allele. The sl⁹ homozygous phenotype is qualitatively the same as the previously characterized sl alleles (i.e., with extra-R7 cells and ectopic wing veins), which confirms our previous results showing that Sl is a negative regulator of the RTK pathways involved in cell-fate decisions during ommatidial and wing-vein development.

The phenotype of sl⁹ homozygotes is identical to that of sl^{1,2,3,4,5,6,8,10}, except that it is slightly more extreme in the eye. Why sl⁹ homozygotes should show a more extreme phenotype than that of the other alleles is uncertain; one possibility that we cannot rule out is that a closely linked enhancer of the extra-R7 phenotype is present in the sl⁹ background. However, this allele is the only one unlikely to produce any protein; every other allele could produce either a mutated full-length protein or a truncated protein containing intact copies of the N-terminal PH domain, EF hand region, region X, and the N-terminal SH2 domain. Because mammalian PLC- does not depend on its lipase function for its role in mitogenesis (SMITH et al. 1994 ; P. S. HUANG et al. 1995 ; YE et al. 2002 ), truncated Sl proteins lacking intact X and Y catalytic domains (such as those that could be generated by the alleles other than sl⁹) might still be able to participate at some level in certain aspects of PLC- function. For example, the intact N-terminal SH2 domain in each of the truncated Sl proteins would likely still be able to bind to an activated RTK at the membrane, where either its N-terminal PH domain or its EF hands might have an effect on signaling, perhaps by binding phospholipids or Ca²⁺.

Three of the new alleles, sl⁵, sl⁷, and sl⁸, contain either a missense mutation or an in-frame deletion, identifying functionally important residues. In the case of sl⁵, a five-amino-acid deletion at the N-terminal end of the C2 domain removes five conserved residues within ß-strand 2/1 of the eight ß-strand C2 structure (RIZO and SUDHOF 1998 ). C2 motifs typically bind calcium and phospholipids, but have proven to be capable of interacting with a wide range of other molecules, sometimes independently of calcium binding (RIZO and SUDHOF 1998 ). The C2 domains of PLC- proteins appear to lack several residues shown to be necessary for calcium binding in the PLC-1 C2 domain (ESSEN et al. 1996 ; REBECCHI and PENTYALA 2000 ), suggesting that the PLC- C2 domain has another (as yet unidentified) function. Combined with the fact that the C2 domain is well conserved in all PLC- proteins, including within the genus Drosophila (MANNING et al. 2003 ), the near-null phenotype observed in sl⁵ homozygotes indicates that the C2 domain does retain an indispensible function in Sl. The sl⁸ mutation replaces Gly³⁸⁵ with aspartate within the region X catalytic domain. There is a glycine at the homologous position in all described PLC proteins (MANNING et al. 2003 ), and the immediately adjacent histidine on the N-terminal side of this glycine is required for PIP₂ hydrolysis in both PLC- (CHENG et al. 1995 ) and PLC- (SMITH et al. 1994 ), playing a direct role in catalysis by acting as a proton donor (ESSEN et al. 1996 ). It is therefore not surprising that the introduction of an aspartate at this critical location would disable the phospholipase activity of the enzyme.

By far the most intriguing of the new alleles is sl⁷, because flies homozygous for this allele show the null wing-length phenotype, but have an almost wild-type eye. A trivial explanation of this phenotype might be the presence of a closely linked modifier that suppresses the extra-R7 phenotype, but not the wing-length phenotype. However, the fact that the X10-P1035A construct produced an almost identical phenotype in the eye, in an unmutagenized background, suggests that the sl⁷ phenotype is a genuine reflection of altered PLC- activity rather than an artifact produced by an interacting mutation. Furthermore, there is a precedent for an sl allele with an sl⁷-like phenotype: sl³⁴ was isolated by Gottschewski in 1934 and was described as having short wings, but normal eyes (LINDSLEY and ZIMM 1992 ); unfortunately, sl³⁴ has been lost. The fact that an independently isolated allele shows this same combination of eye and wing phenotypes tends to suggest that the sl⁷ mutation is not simply a rare hypomorph with threshold activity, but may in fact be a member of a class of sl mutations that are able to separate the extra-R7/ectopic wing-vein and wing-length phenotypes. This model is also consistent with evidence for two distinct Sl-mediated pathways in the eye and wing: a partial loss-of-function mutation of the rl-encoded MAPK, rl¹, is able to suppress both the extra-R7 and ectopic wing-vein phenotypes of sl homozygotes, but does not suppress the short-wing phenotype (THACKERAY et al. 1998 ).

The sl⁷ mutation is a replacement of Pro¹⁰³⁵ by leucine within the region Y catalytic domain. We found that an alanine at the same position recapitulated the phenotype of sl⁷ in the eye, but not in the wing, in which P1035A rescued the sl wing phenotype. By contrast, an sl construct containing valine at position 1035 was equivalent to one containing the wild-type proline, rescuing both the eye and wing defects of either sl¹ or sl⁹. First of all, the ability of the X10-P1035V construct to rescue flies lacking sl function is very surprising, because this proline is one of a small number of absolutely conserved sites across all PLC isoforms in region Y. In PLC-1 the homologous proline is predicted to form a turn leading into one loop of a hydrophobic ridge that lines the active site, a role that might be expected to depend on a proline at this position. A possible explanation is that because valine has a side chain slightly more hydrophobic and compact than that of leucine it is drawn into the ridge by hydrophobic clustering, thereby partially overcoming the loss of proline. The slightly lower hydrophobicity and longer side chain of leucine may be unable to achieve this effect, possibly because of steric clashes due to its length. In contrast, alanine is much less hydrophobic than either valine or leucine and so may be unable to overcome the loss of the turn resulting from replacement of the proline. In any event, the fact that three different amino acids at position 1035 can produce three different combinations of eye and wing phenotypes clearly indicates that Pro¹⁰³⁵ has a key modulatory role in Sl signaling.

Pro¹⁰³⁵ of Sl is homologous to Pro⁵⁵² of human PLC- in a region close to the active site where several mutations have been made and tested in vitro (WANG et al. 1996 ; ELLIS et al. 1998 ). One of the most interesting of these PLC- mutations was a replacement of Arg⁵⁴⁹ by alanine; this change dramatically reduces PIP₂ hydrolysis, but has little effect on hydrolysis of PI, demonstrating that changes in this part of region Y can alter the substrate specificity of PLC-. All PLC enzymes are thought to be able to hydrolyze PIP₂, PIP, and PI (RYU et al. 1987 ), although the relative physiological importance of the different substrates is unknown for any PLC. If the sl⁷ mutation has an effect similar to the Arg⁵⁴⁹ mutation of human PLC-1, hydrolysis of one substrate necessary for Sl-mediated signaling in the wing might be reduced, whereas the hydrolysis of a different substrate more important for signaling during photoreceptor development is less affected. An alternative explanation is that in sl⁷ homozygotes Sl is partially functional at a threshold of activity that is almost sufficient for wild-type function in the eye, but is not quite enough in the wing. However, the phenotype of sl⁷/sl⁸ and sl⁷/sl⁹ trans-heterozygotes suggests that there is a qualitative difference between these alleles, consistent with the sl⁷-encoded PLC- protein lacking a function needed in the wing, but retaining a function needed in the eye. Direct assays of phospholipid hydrolysis will be required to determine whether differential substrate use by Sl occurs in different physiological contexts.

	ACKNOWLEDGMENTS

We thank Tony Ip and Peter Clyne for their invaluable advice on germline transformation, Robert Reenan for suggestions on running SSCP and for assistance with the EMS screen, and David Thurlow for his comments on the manuscript. This work was supported by grant nos. R15 GM-55883-01 and R15 GM-55883-02 from the National Institutes of Health to J.R.T. and facilitated by a Multi-User Equipment grant DBI-0070241 from the National Science Foundation to J.R.T. and three other members of the Clark University Biology Department.

Manuscript received October 25, 2002; Accepted for publication February 25, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAMS, M. D., S. E. CELNIKER, R. A. HOLT, C. A. EVANS, and J. D. GOCAYNE et al., 2000  The genome sequence of Drosophila melanogaster.. Science 287:2185-2195.[Abstract/Free Full Text]

BRENNAN, C. A. and K. MOSES, 2000  Determination of Drosophila photoreceptors: timing is everything. Cell. Mol. Life Sci. 57:195-214.[Medline]

CHATTOPADHYAY, A., M. VECCHI, Q. JI, R. MERNAUGH, and G. CARPENTER, 1999  The role of individual SH2 domains in mediating association of phospholipase C-1 with the activated EGF receptor. J. Biol. Chem. 274:26091-26097.[Abstract/Free Full Text]

CHENG, H.-F., M.-J. JIANG, C.-L. CHEN, S.-M. LIU, and L.-P. WONG et al., 1995  Cloning and identification of amino acid residues of human phospholipase C1 essential for catalysis. J. Biol. Chem. 270:5495-5505.[Abstract/Free Full Text]

CRIPPS, R. M., and S. I. BERNSTEIN, 2000 Generation of transgenic Drosophila melanogaster by P element-mediated germline transformation, pp. 93–125 in Gene Transfer Methods: Introducing DNA Into Living Cells and Organisms, edited by P. A. NORTON and L. F. STEEL. BioTechniques Press, Philadelphia.

DIAZ-BENJUMEA, F. J. and E. HAFEN, 1994  The sevenless signalling cassette mediates Drosophila EGF receptor function during epidermal development. Development 120:569-578.[Abstract]

ELLIS, M. V., S. R. JAMES, O. PERISIC, C. P. DOWNES, and R. L. WILLIAMS et al., 1998  Catalytic domain of phosphoinositide-specific phospholipase C (PLC). Mutational analysis of residues within the active site and hydrophobic ridge of plc1. J. Biol. Chem. 273:11650-11659.[Abstract/Free Full Text]

EMORI, Y., R. SUGAYA, H. AKIMARU, S.-I. HIGASHIJIMA, and E. SHISHIDO et al., 1994  Drosophila phospholipase C- expressed predominantly in blastoderm cells at cellularization and in endodermal cells during later embryonic stages. J. Biol. Chem. 269:19474-19479.[Abstract/Free Full Text]

ESSEN, L. O., O. PERISIC, R. CHEUNG, M. KATAN, and R. L. WILLIAMS, 1996  Crystal structure of a mammalian phosphoinositide-specific phospholipase C. Nature 380:595-602.[Medline]

FALASCA, M., S. K. LOGAN, V. P. LEHTO, G. BACCANTE, and M. A. LEMMON et al., 1998  Activation of phospholipase C by PI 3-kinase-induced PH domain-mediated membrane targeting. EMBO J. 17:414-422.[Medline]

FREEMAN, M., 1997  Cell determination strategies in the Drosophila eye. Development 124:261-270.[Abstract]

GAUL, U., G. MARDON, and G. M. RUBIN, 1992  A putative Ras GTPase activating protein acts as a negative regulator of signaling by the Sevenless receptor tyrosine kinase. Cell 68:1007-1019.[Medline]

HASHIMOTO, A., K. TAKEDA, M. INABA, M. SEKIMATA, and T. KAISHO et al., 2000  Cutting edge: essential role of phospholipase C-2 in B cell development and function. J. Immunol. 165:1738-1742.[Abstract/Free Full Text]

HENTZE, M. W. and A. E. KULOZIK, 1999  A perfect message: RNA surveillance and nonsense-mediated decay. Cell 96:307-310.[Medline]

HOMMA, Y., T. TAKENAWA, Y. EMORI, H. SORIMACHI, and K. SUZUKI, 1989  Tissue- and cell type-specific expression of mRNAs for four types of inositol phospholipid-specific phospholipase C. Biochem. Biophys. Res. Commun. 164:406-412.[Medline]

HUANG, J., M. MOHAMMADI, G. A. RODRIGUES, and J. SCHLESSINGER, 1995  Reduced activation of RAF-1 and MAP kinase by a fibroblast growth factor receptor mutant deficient in stimulation of phosphatidylinositol hydrolysis. J. Biol. Chem. 270:5065-5072.[Abstract/Free Full Text]

HUANG, P. S., L. DAVIS, H. HUBER, P. J. GOODHART, and R. E. WEGRZYN et al., 1995  An SH3 domain is required for the mitogenic activity of microinjected phospholipase C- 1. FEBS Lett. 358:287-292.[Medline]

JI, Q. S., G. E. WINNIER, K. D. NISWENDER, D. HORSTMAN, and R. WISDOM et al., 1997  Essential role of the tyrosine kinase substrate phospholipase C-1 in mammalian growth and development. Proc. Natl. Acad. Sci. USA 94:2999-3003.[Abstract/Free Full Text]

JONES, K. R. and G. M. RUBIN, 1990  Molecular analysis of no-on-transient A, a gene required for normal vision in Drosophila. Neuron 4:711-723.[Medline]

LINDSLEY, D. L., and G. G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press, San Diego.

MANNING, C. M., W. R. MATHEWS, L. P. FICO, and J. R. THACKERAY, 2003  Phospholipase C- contains introns shared by src homology 2 domains in many unrelated proteins. Genetics 164:433-442.[Abstract/Free Full Text]

MORGAN, T. H., C. B. BRIDGES, and A. H. STURTEVANT, 1925  Genetics of Drosophila. Bibliogr. Genetica 2:1-262.

OBERMEIER, A., I. TINHOFER, H. H. GRUNICKE, and A. ULLRICH, 1996  Transforming potentials of epidermal growth factor and nerve growth factor receptors inversely correlate with their phospholipase C affinity and signal activation. EMBO J. 15:73-82.[Medline]

ORITA, M., Y. SUZUKI, T. SEKIYA, and K. HAYASHI, 1989  Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 5:874-879.[Medline]

POWE, A. C., JR., D. STRATHDEE, T. CUTFORTH, T. D'SOUZA-CORREIA, and P. GAINES et al., 1999  In vivo functional analysis of Drosophila Gap1: involvement of Ca2+ and IP4 regulation. Mech. Dev. 81:89-101.[Medline]

RAMEH, L. E., S. G. RHEE, S. SPOKES, A. KAZLAUSKAS, and L. C. CANTLEY et al., 1998  Phosphoinositide 3-kinase regulates phospholipase C- mediated calcium signaling. J. Biol. Chem. 273:23750-23757.[Abstract/Free Full Text]

REBECCHI, M. J. and S. N. PENTYALA, 2000  Structure, function, and control of phosphoinositide-specific phospholipase C. Physiol. Rev. 80:1291-1335.[Abstract/Free Full Text]

RHEE, S. G., 2001  Regulation of phosphoinositide-specific phospholipase C. Annu. Rev. Biochem. 70:281-312.[Medline]

RIZO, J. and T. C. SUDHOF, 1998  C2-domains, structure and function of a universal Ca2+-binding domain. J. Biol. Chem. 273:15879-15882.[Free Full Text]

RYU, S. H., P. G. SUH, K. S. CHO, K. Y. LEE, and S. G. RHEE, 1987  Bovine brain cytosol contains three immunologically distinct forms of inositol phospholipid-specific phospholipase C. Proc. Natl. Acad. Sci. USA 84:6649-6653.[Abstract/Free Full Text]

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SCHALET, A., 1986  The distribution of and complementation relationships between spontaneous X-linked recessive lethal mutations recovered from crossing long-term laboratory stocks of Drosophila melanogaster. Mutat. Res. 163:115-144.

SIVERTZEV-DOBZHANSKY, N. P. and T. DOBZHANSKY, 1933  Deficiency and duplications for the gene bobbed in Drosophila melanogaster. Genetics 18:173-192.[Free Full Text]

SMITH, M. R., Y. L. LIU, N. T. MATTHEWS, S. G. RHEE, and W. K. SUNG et al., 1994  Phospholipase C-1 can induce DNA synthesis by a mechanism independent of its lipase activity. Proc. Natl. Acad. Sci. USA 91:6554-6558.[Abstract/Free Full Text]

TAN, P. B. and S. K. KIM, 1999  Signaling specificity: the RTK/RAS/MAP kinase pathway in metazoans. Trends Genet. 15:145-149.[Medline]

THACKERAY, J. R., P. C. W. GAINES, P. EBERT, and J. R. CARLSON, 1998  small wing encodes a phospholipase C- that acts as a negative regulator of R7 development in Drosophila.. Development 125:5033-5042.[Abstract]

WANG, D., J. FENG, R. WEN, J. C. MARINE, and M. Y. SANGSTER et al., 2000  Phospholipase C2 is essential in the functions of B cell and several Fc receptors. Immunity 13:25-35.[Medline]

WANG, L. P., C. LIM, Y. KUAN, C. L. CHEN, and H. F. CHEN et al., 1996  Positive charge at position 549 is essential for phosphatidylinositol 4,5-bisphosphate-hydrolyzing but not phosphatidylinositol-hydrolyzing activities of human phospholipase C delta1. J. Biol. Chem. 271:24505-24516.[Abstract/Free Full Text]

YE, K., B. AGHDASI, H. R. LUO, J. L. MORIARITY, and F. Y. WU et al., 2002  Phospholipase C1 is a physiological guanine nucleotide exchange factor for the nuclear GTPase PIKE. Nature 415:541-544.[Medline]

ZIPURSKY, S. L. and G. M. RUBIN, 1994  Determination of neuronal cell fate: lessons from the R7 neuron of Drosophila. Annu. Rev. Neurosci. 17:373-397.[Medline]

This article has been cited by other articles:

				C. M. Manning, W. R. Mathews, L. P. Fico, and J. R. Thackeray Phospholipase C-{gamma} Contains Introns Shared by src Homology 2 Domains in Many Unrelated Proteins Genetics, June 1, 2003; 164(2): 433 - 442. [Abstract] [Full Text] [PDF]