learning a generative motion model from sequence based on a latent motion matrix

7.pdf

Home >Biology homework help >learning a generative motion model from sequence based on a latent motion matrix

Cell, Vol. 90, 235–245, July 25, 1997, Copyright 1997 by Cell Press

Costal2, a Novel Kinesin-Related Protein in the Hedgehog Signaling Pathway

John C. Sisson,* Karen S. Ho, transmembrane protein Patched (PTC) (Hooper and Scott, 1989; Nakano et al., 1989), protein kinase A (PKA) (JiangKaye Suyama, and Matthew P. Scott

Departments of Developmental Biology and Struhl, 1995; Lepage et al., 1995; Li et al., 1995; Pan and Rubin, 1995), and the product of costal2 (cos2)and Genetics

Howard Hughes Medical Institute (Whittle, 1976; Grau and Simpson, 1987; Simpson and Grau, 1987; Forbes et al., 1993; Capdevila et al., 1994).Stanford University School of Medicine

Stanford, California 94305-5427 cos2 is a maternal effect gene and a zygotic lethal. Removal of the maternal and zygotic contributions of cos2 causes embryos to die with a cuticle pattern similar to that of ptc homozygous embryos (Grau and Simpson,Summary 1987). As in ptc mutants, cos2 mutant embryos have expanded transcription domains of the HH target geneThe Hedgehog (HH) signaling proteins control cell

fates and patterning during animal development. In wg (Forbes et al., 1993) and mirror-image pattern dupli- cations in the central part of each segment (Grau andDrosophila, HH protein induces the transcription of

target genes encoding secondary signals such as DPP Simpson, 1987). In adults, cos2 mutations cause ectopic dpp expression (Capdevila et al., 1994) and duplicationsand WG proteins by opposing a repressor system. The

repressors include Costal2, protein kinase A, and the of wings, halteres, legs, and antennae (Whittle, 1976; Grau and Simpson, 1987; Simpson and Grau, 1987). TheHH receptor Patched. Like HH, the kinase Fused and

the transcription factor Cubitus interruptus (CI) act adult phenotypes are similar to those seen when ptc function is eliminated or when HH is overexpressedpositively upon targets. Here we show that costal2

encodes a kinesin-related protein that accumulates (Basler and Struhl, 1994; Ingham and Fietz, 1995). Strong allele-specific genetic interactions with fu suggest thatpreferentially in cells capable of responding to HH.

COS2 is cytoplasmic and binds microtubules. We find cos2 may function in the cytoplasm of A cells in parallel with or downstream of fu (Préat et al., 1993).that CI associates with COS2 in a large protein com-

plex, suggesting that COS2 directly controls the activ- When HH signal is received by A cells, an increase in FU activity and a decrease in COS2 activity are thoughtity of CI. to allow CI to directly activate the transcription of HH target genes. CI levels are posttranscriptionally elevated

Introduction along the A/P border in response to HH signal (Motzny and Holmgren, 1995; Slusarski et al., 1995), an apparent

Animal development employs localized sources of HH indicator of its activation as a transcription factor signals to organize pattern by controlling cell fates in the (Alexandre et al., 1996; Dominguez et al., 1996; Hepker embryo. Early in Drosophila development, embryonic et al., 1997). However, most CI detectable by antibody segments and nascent imaginal discs are divided into staining is observed in the cytoplasm (Motzny and anterior compartment (A) cells that transcribe cubitus Holmgren, 1995; Slusarski et al., 1995). Even with high interruptus (ci), a zinc-finger-containing transcription level HH signaling, little, if any, CI is visible in the nucleus, factor (Eaton and Kornberg, 1990; Orenic et al., 1990) suggesting that either a small amount of CI is suffi- and posterior (P) compartment cells that transcribe en- cient to activate targets or nuclear CI is masked from grailed (en), a homeobox containing transcription factor antibody detection. (Poole et al., 1985). HH signal, emanating from P cells, In this paper, we focus on COS2 and its relationship maintains the transcription of the signaling molecules with CI. We have found that COS2 is related to the wingless (wg) in embryonic segments and leg imaginal kinesin superfamily of proteins and binds strongly to discs and decapentaplegic (dpp) in leg and wing imagi- microtubules. COS2 is present at high levels in the cyto- nal discs along the A/P border (Basler and Struhl, 1994; plasm of A cells where CI is expressed. COS2 is associ- Capdevila et al., 1994; Tabata and Kornberg, 1994). The ated with CI in a large protein complex, suggesting that precisely defined sources of WG and DPP near the A/P COS2 constitutes part of a system for regulating the boundary pattern surrounding tissue. activity of CI. These studies suggest a novel role for

HH signal transduction employs a set of activators kinesin-related proteins in regulating signal transduction. and repressors of HH target gene transcription (Ham- merschmidt et al., 1997). In addition to HH, components required for activation include a seven transmembrane

Resultsprotein, Smoothened (SMO) (Alcedo et al., 1996; van den Heuvel and Ingham, 1996), the kinase Fused (FU)

Molecular Identification of cos2(Préat et al., 1990; Thérond et al., 1993, 1996), and CI. Com- cos2 is located on the right arm of the second chromo-ponents required for target gene repression include the some within polytene interval 43B2; 43C1.2 (Grau and Simpson, 1987; Heitzler et al., 1993). A chromosome walk was initiated from a chromosome position proxi- mal to cos2 at 43B1 (Figure 1A). Df(2R)spleD1 (43A1.2;*Present Address: Department of Biology, Sinsheimer Laboratories,

University of California, Santa Cruz, California 95064. 43B2) and Df(2R)NCX11 (43C1.2; 44C1.2) complement

Cell 236

cos2 mutations and bracket the cos2 locus. The chro- mosome walk spans the distance between their adja- cent deficiency endpoints at positions 170 and 1150 kb. Df(2R)EW60 complements cos2 mutations and removes DNA centered over position 190 (Figure 1A). Df(2R)DrlR121

fails to complement cos2 mutations and lies distal of Df(2R)EW60. Together Df(2R)EW60 and Df(2R)NCX11 limit the DNA interval containing cos2 to 60 kb (Figure 1A, horizontal bracket).

An analysis of the 60 kb region with genomic DNA blots reveals restriction fragment length polymorphisms (RFLPs) for several cos2 mutations and places part or all of cos2 within a 5 kb interval of DNA (Figure 1B, open bar). Df(2R)cos22 behaves as an amorphic allele of cos2(Grau and Simpson, 1987) and has a 6.5 kb DNA deletion between positions 1108 and 1115. Df(2R)DrlR121

(Heitzler et al., 1993) has a proximal endpoint that lies within the Df(2R)cos22 deletion, between 1110 and 1111. Therefore, part or all of cos2 must lie within the 5 kb region of their overlap, between 1110 and 1115. Two additional cos2 alleles map to this region. cos2V1 is a viable allele that displays adult pattern duplications in the presence of semidominant alleles of Cos1 (Grau and Simpson, 1987; Simpson and Grau, 1987). cos2V1 is associated with a 9 kb insertion at position 1112. In addition, cos214, a strong hypomorphic allele (Heitzler et al., 1993), is associated with RFLPs between 1115 and 1117.

cos2 is maternally active (Grau and Simpson, 1987), so cos2 mRNA is likely to be present in early embryos prior to the onset of zygotic transcription at 2.5 hr after fertilization. Radioactive cDNA synthesized from 0–2 hr, 4–8 hr, or 8–16 hr embryonic poly(A)1 RNA was hybrid- ized to blots containing the 60 kb cos2 region. Two contiguous SalI fragments (1.2 kb and 6.3 kb) that over- lap the 5 kb cos2 region hybridize to the 0–2 hr cDNA probe (Figure 1B). cDNA clones overlapping the large SalI fragment were recovered for two adjacent, diver- gently transcribed, maternally expressed transcription units (Figure 1B).

To determine which transcription unit is cos2, geno-Figure 1. Molecular Map of the cos2 Region and cos2 Transcript mic fragments containing either the proximal or distalExpression transcription unit were tested for their ability to rescue(A) Thick horizontal bars (top) indicate the positions of four deficien- cos2 embryonic lethality. Transgenic flies were con-cies relative to polytene chromosome positions and corresponding

chromosome walk positions (middle). Thin horizontal bars (bottom) structed carrying either a 6.1 kb genomic KpnI fragment indicate the positions of overlapping cosmid clones. cos2 lies within (K6.1) containing the proximal transcription unit or a 9.5 a 60 kb interval between Df(2R)EW60 and Df(2R)NCX11. The thick kb genomic HindIII fragment (H9.5) containing the distal line within this interval, overlapping 43B3 (1110), is enlarged in (B).

transcription unit (Figure 1B). In a cross of cos22/CyO (B) Four cos2 mutations are close to two maternally expressed tran-

to cos212/CyO stocks, no homozygotes survive com-scription units. Thick horizontal bars above the restriction map indi- pared to 431 heterozygous progeny. When the samecate positions of cos2 mutations. Hatching indicates uncertain defi- cross was done in the presence of one copy of K6.1,ciency endpoints. Df(2R)DrlR121 and Df(2R)cos22 define a 5 kb interval

(open bar) containing a portion of cos2. cos2V1 is an insertion, and the adult progeny were composed of 1050 heterozy- cos214 is associated with RFLPs within a 1.9 kb EcoRI fragment. gotes, zero homozygotes without K6.1, and 297 homo- Thick arrows below the restriction endonuclease map indicate the zygotes carrying K6.1. The expected number of cos2 positions and directions of transcription of cos2 and tuII. A 6.1

homozygotes was 525, and half of these should carry kb KpnI genomic fragment (K6.1, closed bar) fully rescues cos2

K6.1, so the results suggest that K6.1 contains all or aembryonic lethality to adulthood. A 9.5 kb HindIII genomic frag- substantial proportion of cos2. H9.5 does not rescuement (H9.5, hatched bar) fails to rescue cos2 embryonic lethality. cos2 embryonic lethality (data not shown).(B) 5 BamHI, (H) 5 HindIII, (K) 5 KpnI, (R) 5 EcoRI, (S) 5 SalI, and

(X) 5 XbaI. A 4.8 kb cos2 cDNA hybridizes to a 4.9 kb transcript (C) A blot containing total RNA from different embryonic stages present at high levels during the first four hours of em- and third instar larvae was hybridized to radioactive cos2 and rp49 bryogenesis (Figure 1C), moderate levels between four probes. The cos2 probe reveals a single 4.9 kb transcript. rp49

and twelve hours, and low levels for the duration of serves as a loading control (O’Connell and Rosbash, 1984).

embryogenesis. The transcript is also present during the third larval instar.

COS2, a Kinesin-Related Regulator of Development 237

cos2 Encodes a Kinesin Heavy Chain–Related Protein The complete sequence of a 4.8 kb cDNA clone for cos2 was determined, as was all of the genomic sequence flanking the cDNA in the rescuing transgene. The cDNA sequence reveals a single, large open reading frame (ORF). The putative translational start site matches the Drosophila consensus sequence well and contains co- dons common in other Drosophila genes (Cavener, 1987; Ashburner, 1989). Multiple stop codons in all three reading frames are present upstream of the putative start codon (data not shown). The surrounding genomic sequence contains three short ORFs that do not begin with methionine or match the usual pattern of Drosophila codon usage.

cos2 is predicted to encode a 1201 amino acid poly- peptide with a molecular weight of 133 kDa (Figure 2A). The N-terminal (residues 1–450) and C-terminal (resi- dues 1050–1201) regions are predicted to form globular structures consisting of alternating a helices and b sheets (Figure 2B, MacVector 4.1.1, Kodak). The central region (residues 643–990) contains 36 heptad repeats (Figure 2A, underlined) that are predicted to mediate the formation of a stable homodimer through a parallel coiled coil (Figure 2B) (Woolfson and Alber, 1995).

COS2 is similar to members of the kinesin protein family (Figure 2C). Over a span of 254 N-terminal amino acids (residues 136–389), COS2 is 25%–30% identical to the motor domains of different members of the kinesin gene family (Figure 2C, right column) (Higgins et al., 1992). Kinesins are molecular motor proteins that move along microtubules powered by ATP hydrolysis (re- viewed by Goldstein, 1993; Moore and Endow, 1996). Conventional kinesin consists of two kinesin heavy chains (KHC) and two kinesin light chains (KLC). KHC consists of an N-terminal motor domain, a central do- main made up of heptad repeats, and a C-terminal puta- Figure 2. COS2 Amino Acid Sequence, Analysis, and Alignments tive “cargo” domain thought to bind vesicles to move (A) cos2 is predicted to encode a 1201 amino acid protein. The N them. The motor domain of KHC is sufficient to mediate terminus contains three putative nucleotide-binding motifs (N1 [P

loop], shaded box; N2, thick underline; and N3, stippled underline)ATP-dependent movement along microtubules in vitro (Vale, 1996) and two putative microtubule-binding motifs (open(Yang et al., 1990). boxes) (Sablin et al., 1996). The central portion contains 36 heptadSeveral motor domain motifs implicated in nucleotide repeats arranged in eight clusters (thin underlining).

(N) or microtubule binding are highly conserved within (B) The predicted structure of COS2. Sequence analyses predict

the kinesin family (Goldstein, 1993; Sablin et al., 1996; that the N and C termini adopt globular conformations and that the Vale, 1996) and are generally conserved in COS2. For 36 heptad repeats mediate the formation of a homodimer by forming example, the nucleotide-binding motif 1 (N1 or P loop) a parallel coiled coil.

(C) Alignment of the putative cos2 P-loop motif (N1) with those ofin COS2 is 50% identical to the kinesin gene family representative members of the kinesin gene family. A consensusconsensus sequence (Figures 2A [shaded box] and 2C). sequence for the family is shown at the bottom. Closed and shadedFour residues strictly conserved in the family (Goldstein, rectangles indicate identity and similarity, respectively. The four

1993) are present in COS2 (Figure 2C, thin underlining), underlined residues are invariant within the kinesin gene family

but COS2 residues R177 and Q179 are significantly dif- (Goldstein, 1993). The minus sign indicates an acidic residue, and (X) ferent. The N2 motif, SSRSH, in COS2 is replaced by indicates the absence of a consensus residue. The percent identity SLPAH (Figure 2A, thick underline), while N3, DLAGS/ between cos2 and the indicated kinesin family members is shown

at the right.TE, is conserved in COS2 (Figure 2A, stippled underline). N4 is not present in COS2. At least two motifs have been tentatively implicated in microtubule binding (Sablin et

COS2 Expression Prior to Germ Band Extensional., 1996; Woehlke et al., 1997 [this issue of Cell]): the Polyclonal rat antisera were raised against N- andstrictly conserved DLL motif and the L12 motif (Figure C-terminal portions of COS2. Both antisera were affinity2A, open boxes) (Sablin et al., 1996). The L12 consensus purified and used to probe blots of embryo protein ex-sequence is FI/VPY/FRN/D (F 5 hydrophobic residues), tracts. Both antisera reveal a single band of 175 kDaand both the P and R residues are strictly conserved (Figure 3A). Preimmune antisera do not detect any pro-(Goldstein, 1993). In COS2, the DLL motif is present tein on these blots (data not shown). COS2 migrateswhile L12 is partially conserved, with the expected R

being absent. much more slowly than its predicted size of 133 kDa,

Cell 238

Figure 3. The Spatial Distribution of COS2 in the Early Embryo

(A) Affinity-purified anti-N-terminal (N) and anti-C-terminal (C) COS2 antisera recognize a single major protein band of 175 kDa on protein blots. (B–E) Methanol-fixed embryos were incu- bated with affinity-purified anti-COS2 (red) and anti-lamin (green) antibodies and de- tected by indirect immunofluorescence using a scanning confocal microscope. Anterior is to the left and dorsal is up. (pc) 5 pole cells, (cf) 5 cephalic furrow, and (vf) 5 ventral fur- row. COS2 is present along the periphery of stage 4 (B) and stage 5 (C) embryos sur- rounding somatic nuclei (bracket). COS2 can be seen in the apical cytoplasm of newly forming cells ([C], arrowheads) and at en- hanced levels along the basal face of elon- gated nuclei in stage 5 embryos (arrows). COS2 is present in all cells of stage 6 embryos (D and E; arrowheads). In (E), an optical sec- tion through the apices of presumptive epi- dermal cells reveals uniform levels of COS2 (arrow). (F and G) Optical sections at apical (F) and basal (G) focal planes, parallel with the sur- face of an early stage 4 embryo, show COS2 uniformly distributed within the cytoplasm. (H and I) COS2 accumulates between and apical of nuclei in late stage 4 embryos. In (H), a lateral view shows COS2 as vertical rays (arrow). In (I), a surface view shows COS2 in a honeycomb pattern. Punctate staining is seen along the honeycomb pattern (ar- rowhead). (J) During early cellularization, COS2 staining is elevated in furrow canals (arrows). The ar- rowhead indicates the position of newly formed membrane. (K) During late cellularization, COS2 staining is elevated in expanded furrow canals along the basal face of nuclei (fc; arrows); in slightly older embryos, COS2 staining forms a line that circumscribes the inner yolk ([C], arrows). Embryos were staged according to Campos- Ortega and Hartenstein (1985).

perhaps due to posttranslational modification. Both an- COS2 is associated with furrow canals throughout cellularization (Figures 3J and 3K). Furrow canals (fc) aretisera also recognize endogenous and overexpressed

COS2 in the cytoplasm of Drosophila S2-cultured cells located at the leading edge of newly forming membrane between adjacent somatic nuclei (Foe et al., 1993). Dur-(data not shown).

Both affinity-purified antisera were used to assess the ing cellularization, furrow canals move toward the basal end of the nuclei where they broaden, forming expandedexpression of COS2 in early embryos, and both give the

same results. In syncytial stage embryos (stage 4), prior furrow canals (efc), and then fuse with one another in a process that will seal off the new cells from the embryo’sto cellularization, COS2 is distributed uniformly within

the cortical cytoplasm (Figure 3B, bracket), at apical interior. COS2 is present at relatively high levels within each early furrow canal (Figure 3J, arrows). At this time,(Figure 3F) and basal (Figure 3G) focal planes. Anti-lamin

antibody (green) outlines the nuclei. COS2 is detected COS2 is also distributed uniformly at lower levels throughout the cortical cytoplasm and along new mem-neither within nuclei (Figures 3G–3I) nor in association

with microtubule spindles (data not shown). In late syn- brane trailing each furrow canal (Figure 3J, arrowhead). COS2 is associated with expanded furrow canals priorcytial blastoderm embryos just prior to cellularization,

COS2 accumulates between, and apical to, nuclei (Fig- to (Figure 3K, arrows) and after their fusion during late cellularization (Figure 3C, arrows). In cellular blastodermures 3H and 3I). A lateral view shows COS2 accumulation

forming rays perpendicular to the surface of the embryo embryos (Figure 3C, arrowheads) and after the onset of gastrulation, COS2 is in the cytoplasm (Figures 3D and(Figure 3H, arrow). Surface views along the apices of

nuclei show COS2 accumulation forming a honeycomb 3E, arrowheads) and at the periphery of all cells (Figure 5C). cos2 transcripts are uniformly distributed in thepattern (Figure 3I). COS2 is punctate rather than uniform

within the honeycomb lattice (arrowheads). early embryo (data not shown).

COS2, a Kinesin-Related Regulator of Development 239

Figure 4. COS2 Is Expressed at Elevated Levels in Anterior Compartment Cells

(A) In situ hybridization of a cos2 antisense riboprobe to a stage 10 embryo showing uni- form levels of cos2 transcript. (B and C) COS2 staining in stage 10 embryos. Lateral (B) and dorsal (C) views show COS2 expressed in stripes. The arrowheads indi- cate morphologically visible parasegmental grooves. (D and E) In situ hybridization of cos2 sense (D) and antisense (E) riboprobes to wild-type wing imaginal discs. (F) COS2 staining in wild-type leg (L) and wing (w) imaginal discs. An arrow indicates the po- sition of the anterior-posterior compartment border. (G and H) The anterior-posterior compart- ment border of a ptc-lacZ wing imaginal disc stained with both b-gal and COS2 antibodies, respectively. The anterior compartment is at the left in both panels. In (G), b-gal staining is in a narrow band of cells within the anterior compartment along the anterior-posterior compartment border (arrow). In (H), elevated

COS2 staining is restricted to the anterior compartment (arrow). Arrows in (G) and (H) are in corresponding positions. (I) Protein blot of anterior (Ant.) and posterior (Post.) wing disc fragment extracts probed with antibodies to COS2 (C2), CI, a-tubulin (T), and Engrailed (E). COS2, CI, and Engrailed are normalized to tubulin.

COS2 Levels Are Elevated in the Anterior posterior disc extract, it is less abundant than in the anterior disc extract, in keeping with the histochemicalCompartments of Embryonic Segments

and Imaginal Discs staining result. The COS2 detected in the posterior disc extract has a slower mobility than the anterior protein,In contrast to the uniform distribution of cos2 mRNA

in the germband-extended embryo (Figure 4A), COS2 suggesting it is a distinct posttranslational form of COS2. As expected, CI and EN were detected only inprotein is present in a striped pattern. Faint stripes along

the germband are first observed in late stage 9 embryos anterior and posterior disc extracts, respectively (Korn- berg et al., 1985; Eaton and Kornberg, 1990; Motznyand become prominent by stage 10 (Figure 4B). Each

stripe is continuous along the dorsal–ventral axis (Figure and Holmgren, 1995; Slusarski et al., 1995). 4C) in both the ectoderm and the underlying mesoderm (Figure 4B). The stripes persist throughout stage 11 and COS2 and CI Associate with Microtubules

in Embryo Extractsdecay during germband retraction (stage 12). The stripes appear to form just anterior of parasegmental grooves A hallmark of kinesins is the ability to bind taxol-stabi-

lized microtubules (Saxton, 1994). We tested whetherin anterior compartment cells, but precise determination of boundaries is difficult due to weak signal (Figures 4B COS2 from fly embryos also binds microtubules. Embryo

extracts were supplemented with taxol and centrifugedand 4C, arrowheads). The accumulation of COS2 in imaginal discs is remi- to bring down microtubules and associated proteins. In

the absence of taxol (Figure 5A, lane 1), COS2, kinesinniscent of its expression in the germband-extended em- bryo. In situ hybridizations with single-stranded sense heavy chain (KHC), and a-tubulin are in the supernatant.

In the presence of taxol, a-tubulin is in the pellet, show-(Figure 4D) and antisense (Figure 4E) cos2 probes show that cos2 mRNA is uniform within wing discs. In contrast, ing that microtubules have formed efficiently (Figure 5A,

compare lanes 4 and 5). While COS2 pellets, KHC re-COS2 levels are elevated in the anterior compartment (Figure 4F). A ptc-lacZ enhancer trap stock (AT90), pro- mains in the supernatant because kinesin does not bind

microtubules in the presence of the ATP contributed byducing nuclear-localized b-galactosidase (b-gal) in a ptc-specific pattern (Phillips et al., 1990), was used to the embryo extract (Lasek and Brady, 1985; Vale et al.,

1985). In the presence of the nonhydrolyzable ATP ana-show that the position of the A/P border (Figure 4G, arrow) corresponds to the line of transition from high to log AMP-PNP and apyrase, which breaks down ATP,

both KHC and COS2 are in the microtubule pellet (Figurelow COS2 levels (Figure 4H, arrow). The apparent elevation of COS2 in the anterior could 5A, compare lanes 6 and 7). Therefore, COS2 binds

microtubules in a taxol-dependent, ATP-insensitivebe due to higher protein levels, to differential fixation of COS2, or to the accessibility of COS2 to antibodies. We manner, while KHC binds microtubules in a taxol-depen-

dent, ATP-sensitive manner. A bacterially expressedconfirmed that the amount of protein is regulated by dissecting anterior and posterior portions of wing discs COS2-GST fusion protein, containing the putative motor

domain, also binds to purified microtubules (data notand measuring protein levels on blots (Figure 4I, [Ant.] and [Post.]). The amount of COS2 (C2), CI, and Engrailed shown).

We also tested whether CI associates with microtu-(E) protein was normalized to the amount of a-tubulin (T) in the two fractions. Although COS2 is present in the bules, since so much CI is cytoplasmic. CI associates

Cell 240

COS2 and microtubules, but not a strict colocalization (Figure 5D). Presumably, not all of the COS2 is microtu- bule associated in vivo, consistent with the in vitro mi- crotubule-binding results (Figure 5A, compare lanes 4 and 5, and 6 and 7).

COS2 and CI Physically Associate The similar microtubule association of COS2 and CI sug- gested the two proteins might be in a protein complex. We tested whether COS2 and CI coelute from a gel filtration column. A S100 embryo extract was separated on a Sepharose 4B column, and fractions were assayed for COS2, CI, and a-tubulin by immunoblotting. The elu- tion profiles for COS2 and CI are virtually identical (Fig- ure 6A). Their common peak fraction is approximately 500–600 kDa. A homodimer of COS2 is expected to elute with an approximate peak of 350 kDa. a-tubulin elutes with an apparent molecular weight of 110 kDa, consis- tent with the expected size of a/b-tubulin heterodimers (Figure 6A). Because microtubules are efficiently depoly- merized under the conditions used, the coelution of

Figure 5. COS2 Binds to Microtubules In Vitro and Overlaps with COS2 and CI is not dependent on microtubule-mediated a-Tubulin in Embryos cross-linking of the two proteins. (A) Microtubule-binding assays. S100 extracts were either not sup- We tested whether COS2 and CI are associated in a plemented ([2], lanes 2 and 3) or supplemented with 40 mM taxol, protein complex using immunoprecipitation. Anti-COS2 or 40 mM taxol, 0.5 mM AMP-PNP (A), and 80 U/ml apyrase (apy)

and anti-CI antibodies nearly completely precipitate (Binding, lanes 4–7). Supernatant (S) and pellet (P) fractions were

COS2 and CI, respectively (Figure 6B, compare pelletsassayed for COS2, CI, KHC, a-tubulin and SGG/ZW3. The 0.5 M and supernatants). A significant fraction of CI is copre-KCl causes most microtubles to dissolve and enter the supernatant cipitated by anti-COS2 antibodies and vice versa (Figure(compare lanes 10 and 11).

(B–D) A single scanning confocal micrograph showing indirect im- 6B). COS2 preimmune antisera alone do not precipitate munofluorescence for (B) a-tubulin, (C) COS2, and (D) images (B) and COS2 or CI (Figure 6B), nor do protein G–Sepharose (C) merged. In (B), the three cells shown at the top are undergoing beads alone (not shown). postblastoderm mitoses. An arrowhead indicates a microtubule aster; (n) indicates nucleus. In (C), COS2 is distributed throughout the cytoplasm, showing heightened accumulation along the periph-

cos2 Somatic Clones Have Increased Cytoplasmicery of some cells. COS2 is not observed on mitotic asters. In (D), yellow indicates overlap between a-tubulin and COS2 within the CI Staining and Cause Pattern Duplications cytoplasm, COS2 visible at the periphery of cells does not colocalize We employed the FLP recombinase-FRT technique (Xu with a-tubulin. Stage 6 embryo magnified 11003. and Harrison, 1994) to generate homozygous clones of

cos2 in wing discs and examined the subcellular location with microtubules just as COS2 does, in a taxol-depen- and expression of CI. Approximately 50% of flies geneti- dent, ATP-insensitive manner (Figure 5A). The slight cally competent to form cos2 somatic clones display amount of CI sedimenting in the absence of taxol is not extra wing veins and/or dramatic mirror-image duplica- consistently observed. tions characteristic of cos2 mutants (Figure 7A) (Grau

The effect of exogenous ATP on COS2 binding to and Simpson, 1987; Simpson and Grau, 1987). cos2 microtubules was tested. Microtubule pellets containing clones, marked by the loss of the MYC epitope carried COS2, CI, and KHC were washed and resuspended in on the homologous chromosome (green in Figure 7), are the presence of taxol and 5 mM ATP or taxol, 5 mM frequently observed in both the A and P compartments ATP, and 0.5 M KCl and were then recentrifuged. KHC of wing discs. Elevated cytoplasmic CI staining is seen is partially extracted into the supernatant with just ATP, in cos2 clones in the A compartment (Figures 7B–7D). as expected (Figure 5, compare lanes 8 and 9). However, The level of CI staining is independent of the clone’s both COS2 and CI remain microtubule associated in the distance from the A/P border (data not shown) or size presence of ATP. COS2, CI, and KHC are completely (Figure 7C). Nuclear CI is not evident in the clones (Figure extracted from microtubules in the presence of 5 mM 7C). cos2 clones in the P compartment do not express ATP and 0.5 M KCl (Figure 5, compare lanes 10 and 11). ci (data not shown). Most microtubules dissolve in the high salt, but some remain intact. Shaggy/Zeste-white3 (SGG/ZW3) protein (Siegfried et al., 1990), a kinase not expected to bind Discussion microtubules, serves as a control. A slight amount of SGG/ZW3 cosediments with microtubules (Figure 5, COS2 Is a Divergent Member of the Kinesin

Gene Familylane 7). Embryos stained with antibodies to a-tubulin (Figure The COS2 sequence resembles kinesin, but COS2 does

not appear to belong to an existing kinesin subfamily5B) and COS2 (Figure 5C) reveal an overlap between

COS2, a Kinesin-Related Regulator of Development 241

Figure 6. COS2 and CI Are Physically Associated in Exbryonic Ex- tracts

(A) Coelution profile of COS2 and CI from a Sepharose 4B column. The proteins elute in a common peak between 500 and 600 kDa. a-tubulin elutes as a peak at 110 kDa, the expected size for a/b-tubulin heterodimers. Protein standard sizes and elution peak positions are indicated above fraction numbers. The position of the void volumn is indicated at the left; only odd numbered fractions were analyzed. (B) Immunoblots show pellets (P) and supernatants (S) from coimmu- noprecipitations with COS2 antisera (anti-COS2) and CI antisera (anti-CI). COS2 preimmune serum (Pre.) does not precipitate either COS2 or CI. Both COS2 and CI are found in the pellets of immunopre- cipitations done using either anti-COS2 or anti-CI antisera. S43 re- fers to embryo lysate. Blots were probed with either COS2 or CI antibodies as indicated.

and may have novel properties. Phylogenetic subfami- lies have been established based on structural and func- tional similarities between motor domains (reviewed by Goldstein, 1993; Moore and Endow, 1996). Some sub- families are implicated in microtubule-based vesicle or organelle movement, while others participate in assem- Figure 7. CI Protein Levels Are Elevated in Cells Lacking cos2 bly or force generation for mitotic or meiotic microtubule Function spindles. The motor domain motifs implicated in nucleo- (A) A mirror-image duplication of anterior wing pattern from a tide binding in other kinesins are different in COS2, so P[hsp70-FLP]; P[FRT] cos2W1 heat-treated fly. Arrows indicate

planes of mirror-image symmetry.COS2 may lack motor activity. Most kinesin motor pro- (B–D) These scanning confocal micrographs show a portion of theteins release microtubules when provided with ATP, an anterior compartment of a wing disc by indirect immunofluores-intrinsic property of the motor domain (Lasek and Brady, cence of (B) nuclear localized MYC, (C) cytoplasmic CI, and (D) both1985; Vale et al., 1985; Cole et al., 1994). In contrast, (B) and (C) merged. In (B), the absence of MYC expression marks

COS2 remains attached to microtubules when exoge- a large cos2W1/cos2W1 clone (arrows) along the anterior edge (aster-

nous ATP is provided. This suggests that unlike kinesin isk) of the disc. A small cos2W1/cos2W1 clone is also indicated by and many kinesin-related proteins, COS2 may not regu- arrows. In (C), elevated levels of cytoplasmic CI are seen within late its binding to microtubules by ATP hydrolysis. The both cos2W1/cos2W1 clones. MYC does not localize to the nucleolus

(arrowhead). Magnified 6303.nucleotide-binding motifs of COS2 may be unable to coordinate ATP.

The unconventional nature of COS2 is also manifested COS2 Levels Are Posttranscriptionally Elevatedin its localization in early embryos. Prior to somatic cell in the Anterior Compartmentformation, COS2 accumulates in a honeycomb pattern Because cos2 mRNA levels are uniform, the elevatedat the cortex of the embryo. A similar lattice pattern level of COS2 in the A cells must be due to differencesis characteristic of actin and actin-associated proteins between A and P cells in either the production or the(Miller et al., 1989). Slightly later, during cellularization, stability of COS2. The uniform level of COS2 throughoutCOS2 is associated with the actin-rich furrow canals, the anterior compartment of imaginal discs is inconsis-and with the periphery of cells after cellarization. SMY1, tent with HH signal regulating its accumulation. HH regu-a divergent kinesin-related protein, also localizes to ac- lates CI posttranscriptionally in the anterior compart-tin-rich regions of the cell (Lillie and Brown, 1994) and ment, but the limited range of HH (Basler and Struhl,has been implicated in two actin-based processes: 1994; Tabata and Kornberg, 1994) results in a gradedpolarized growth and secretion in yeast (Lillie and

Brown, 1992). distribution of CI across the anterior compartment quite

Cell 242

within cos2 clones (Capdevila et al., 1994; Sanchez- Herrero et al., 1996). We propose that COS2 and CI act in a large protein complex in the cytoplasm of A cells to mediate the regulation of HH target genes.

The control of HH target gene expression may depend on the level and/or posttranslational form of CI. When increased CI is produced in wing discs far from the A/P border, beyond the influence of HH, dpp and ptc transcription are activated in A cells (Alexandre et al., 1996; Dominguez et al., 1996; Hepker et al., 1997). Be- cause dpp and ptc are also activated in P cells, the CI- mediated activation of these targets does not depend on an A compartment–specific factor. CI may normally require a HH-dependent modification to activate HH tar- gets, but elevated CI seems sufficient to activate HH targets. Along the A/P border CI levels are posttranscrip-

Figure 8. A Model for the COS2/CI Complex in HH Signaling tionally elevated in response to HH signaling (Johnson Diagrams of two adjacent anterior compartment cells are shown; et al., 1995; Slusarski et al., 1995; Hepker et al., 1997). the A/P border is at the right. Both cells express PTC, SMO, FU,

This elevated level of CI is thought to allow it to enter COS2, and CI. The known or suspected subcellular localization of

the nucleus and directly activate HH targets (Alexandreeach protein is indicated. PTC and SMO are associated in the plasma et al., 1996; Dominguez et al., 1996; Hepker et al., 1997).membrane (PM). COS2 and CI are associated in a cytoplasmic com-

plex. In the absence of HH signal (left cell), PTC inhibits SMO func- Although endogenous CI is hard to see in the nucleus, tion. COS2 is active, repressing CI levels posttranscriptionally and when the C-terminal portion of an epitope-tagged CI retaining CI in the cytoplasm (transcription off). In the presence of is removed, leaving the zinc fingers intact, CI protein HH signal (right cell), PTC is directly inhibited by HH, allowing SMO

appears in the nucleus and the cytoplasm (Hepker etsignaling to inhibit COS2 activity either through FU or other cyto- al., 1997). CI therefore appears competent to enter theplasmic factors (question mark). This triggers CI activation of HH nucleus but is normally restricted to the cytoplasm bytargets (transcription on). The complex may be microtubule-associ-

ated through COS2 (small question mark). PKA and SU(FU) function the C-terminal tail. The absence of detectable CI in the in the pathway, but their relationships to the complex are unknown. nucleus may be the result of inadequate CI antibodies.

The protein complex we have identified could control the level of CI and its subcellular distribution. The COS2/ CI complex may control the level of CI either by increas-unlike the COS2 distribution (Motzny and Holmgren,

1995; Slusarski et al., 1995). A uniform anterior- or poste- ing CI production or decreasing its degradation. The rior-specific activity could establish the high uniform complex could protect CI from proteases only when HH level of COS2 in the anterior compartment. One possibil- signal is received, or the complex could associate with ity is that the moderate level of CI in all A cells is sufficient polysomes to facilitate translation of ci mRNA. Because to stabilize COS2 in a complex. In P cells, COS2 would a substantial fraction of COS2 and CI are associated, turn over more rapidly because it is not protected by COS2 may sequester CI in the cytoplasm, possibly by complex formation. Another possibility is that CI height- tethering it to the cytoskeleton (Figure 8). Because CI ens translation of cos2 mRNA, a possible role for the lacks an obvious nuclear localization signal (Orenic et CI zinc-finger protein in the cytoplasm. Alternatively, a al., 1990; Motzny and Holmgren, 1995) its movement to factor controlled by en could destroy COS2 in P cells the nucleus may be regulated by its ability to couple to or stabilize it in A cells. cos2 is not required for patterning a protein that carries it there. The transcription factor the posterior compartment (Whittle, 1976; Grau and dCBP may serve this function (Akimaru et al., 1997). Simpson, 1987; Simpson and Grau, 1987), so the low COS2 may render CI unavailable to such a protein ex- level of COS2 detected in the posterior disc extract may cept along the A/P border, where COS2 is inhibited. The be nonfunctional. absence of detectable nuclear CI in cells lacking cos2

function may indicate the need for a second activating event in addition to a release from COS2. CI may haveCOS2 May Directly Inhibit CI from Activating to be modified to be activated, or it may be tetheredHH Target Genes to a COS2-independent complex, or transport into thePrevious genetic evidence indicates that cos2 functions nucleus may require an activated cofactor. Alternatively,in A cells to regulate HH target gene expression (Forbes nuclear CI may be masked from the antibodies used.et al., 1993; Préat et al., 1993; Capdevila et al., 1994;

The identification of the COS2/CI complex helps toSanchez-Herrero et al., 1996). Our findings are consis- fill in missing steps in HH signaling by showing directtent with these genetic data. First, COS2 accumulates interactions among two of the five known signal trans-to high levels in A cells. Second, COS2 physically associ- duction components and by providing a cytoskeletalates with CI, which is expressed in A cells. Third, cos2 link. The importance of the complex is further under-activity reduces CI staining in A cells. cos2 somatic scored by the presence of a third component, fusedclones in the anterior compartment of wing discs ex- (Robbins et al., 1997 [this issue of Cell]). The subcellularpress high levels of CI and cause mirror-image duplica- distribution of the complex may be important for control-tions of the wing. These pattern duplications are pre-

dicted to result from CI-mediated activation of dpp ling HH targets and, consequently, cell differentiation.

COS2, a Kinesin-Related Regulator of Development 243

Materials and Methods mounted in Vectashield H-1000 (Vector Laboratories Inc.) and exam- ined by confocal microscopy. Antibodies used are as follows: COS2, rat polyclonal antisera (1:5); a-tubulin, mouse monoclonal (1:25, giftMolecular Cloning and Hybridizations

Molecular biology techniques were carried out according to Sam- of Drs. R. Sakowizc and L. Goldstein); lamin, mouse monoclonal (1:40, gift of Drs. B. Harmon and J. Sedat); b-gal, rabbit polyclonalbrook et al. (1989). The cos2 chromosome walk was initiated with

a genomic clone (lB47, kindly provided by Dr. Ed Stephenson) and (1:100, Cappel); and all fluorescent secondary antibodies (1:200, Jackson ImmunoResearch Labs).using a cosmid library (Tamkun et al., 1992) made from an isogenic

fly stock (iso-1). The progress of the walk and positions of deficienc- ies were determined by in situ hybridization of biotin-labeled DNA Microtubule-Binding Assays fragments to polytene chromosomes (Pardue, 1994). Overlapping This assay was carried out according to Kellogg et al. (1989), with iso-1 genomic l phage clones lying between Df(2R)spleD1 and some modifications. Briefly, 16 g of 2–10 hr Canton S embryos were Df(2R)NCX11 were isolated, and positions of cos2 mutations were homogenized in 32 ml of C buffer (50 mM HEPES [pH 7.6], 1 mM determined using blots of cos2 mutant genomic DNA. From 38 cos2 MgCl2, 1 mM EGTA), 0.5 mM DTT, and protease inhibitors (1.74 mg/ cDNA clones recovered from a lgt10, 0–3 hr embryonic cDNA library ml PMSF, 1 mM benzamidine, 2 mg/ml aprotinin, 1 mg/ml leupeptin, (Poole et al., 1985) and a plasmid-based imaginal disc cDNA library 1 mg/ml pepstatin, all from Sigma) on ice. A supernatant (S100) was (Brown and Kafatos, 1988), two approximately full-length clones, prepared, and five 5 ml aliquots were made. One aliquot received D12 and D13, were found and sequenced. In situ hybridization of 40 mM taxol (Sigma) and 1 mM GTP (binding, lanes 4 and 5); three riboprobes to embryos and imaginal discs was carried out as de- aliquots received 40 mM taxol, 1 mM GTP, 80 U/ml apyrase (Sigma), scribed (Mathies et al., 1994). and 0.5 mM AMP-PNP (Boehringer-Mannheim) (binding, lanes 6

and 7, and extractions); and one was not supplemented (2taxol). Aliquots were incubated at 258C for 20 min and then on ice for 10Germline Transformations min. Each sample (4 ml) was layered over a 10% sucrose cushionThe 6.1 kb KpnI (K6.1) and 9.5 kb HindIII (H9.5) genomic fragments and centrifuged at 48,000 3 g for 30 min at 48C. For the 2taxol andwere subcloned into pCaSpeR4. Transgenic flies were made ac- both binding samples, supernatants were saved, and pellets werecording to Spradling and Rubin (1982), using w1118 embryos as recipi- washed and resuspended in 4.5 ml of CX buffer (C buffer supple-ents. Seven independent K6.1 inserts and eight independent H9.5 mented with 10% glycerol, 25 mM KCl, and protease inhibitors). Forinserts were recovered. Rescue crosses were done at 258C. the extraction samples, pellets were resuspended in 1 ml of CX buffer (supplemented with 40 mM taxol, 1 mM GTP, and either 5Antibody Preparation and Immunoblotting mM Mg-ATP or 5 mM Mg-ATP and 0.5 M KCl) and incubated on iceAffinity-purified rat polyclonal antisera were prepared to two parts of for 10 hr before centrifugation, as before. The resulting supernatantsCOS2. A 1.5 kb SacI–EcoRI (SR1.5) fragment, including the putative were saved, and pellets were resuspended in 1 ml of CX buffer. Eachmotor domain, and a 0.8 kb EcoRI (R0.8) fragment, including the sample (15 ml in 1X sample buffer) was separated by SDS–PAGE andN-terminal 19 heptad repeats, were each subcloned into two differ- immunoblotted.ent plasmid expression vectors, pATH10 (Rimm and Pollard, 1989)

and pGEX-2T (128/129) (kindly provided by Dr. M. Blanar). The ChromatographypATH10 clones create E. coli TRPE-COS2 fusion proteins, which A Sepharose 4B (Pharmacia) column (48.5 cm 3 1.77 cm2 equalingwere used as immunogens. Each TRPE-COS2 fusion protein was a bed volumn [Vt] of 86 ml and a void volumn [Vo] of 28.5 ml) waspurified from the BL21 pLysS cell lysates as inclusion bodies, cut calibrated with protein standards (Pharmacia) and operated at afrom SDS gels, and injected into rats (Josman Labs). The pGEX- pressure head of 64 cm with a flow rate of 17.5 ml/hr. Embryos were2T (128/129) clones create glutathione S-transferase (GST)-COS2 homogenized in TNE buffer (40 mM Tris [pH 7.2], 250 mM NaCl, 0.5fusion proteins, which were used to affinity purify the rat antisera. mM EDTA, 10% glycerol, 0.05% NP-40, and 1 mg/ml nocodazole) 1Soluble GST-COS2 fusion proteins were purified from BL21 pLysS proteinase inhibitors (previously listed), and an S100 protein extractcells using glutathione-agarose beads and coupled to AminoLink was prepared as above and dialyzed against column running bufferPlus chromatography columns (Pierce). Antibodies were eluted from overnight at 48C. The S100 was recentrifuged at 100,000 3 g for 30columns with 4.5 M MgCl2 and dialyzed against 50 mM HEPES (pH min at 48C. The total protein concentration of the resulting S100 (347.5), 150 mM NaCl, 1 mM EDTA, and 0.01% NaN3. mg/ml) was determined, and 250 ml (8.5 mg) was loaded onto theImmunoblots were carried out as described by Harlow and Lane column. Column runs were monitored by a UV spectophotometer(1988). After 7.5% SDS–PAGE, proteins were transferred to Protran at ODA280, and 1.5 ml fractions were collected. Proteins were precipi-membrane (Schleicher and Schuell), and membranes were blocked tated with acetone and analyzed by immunoblot.with 5% nonfat dry milk for 2–6 hr. Antibodies used are as follows:

COS2, rat polyclonal antisera (1:50); CI, rat monoclonal (1:15, gift of Dr. B. Holmgren); a-tubulin, mouse monoclonal (1:100, gift of Coimmunoprecipitation Drs. R. Sakowizc and L. Goldstein); EN, mouse monoclonal (1:500, An embryonic extract (S43) prepared in TNE buffer 1 proteinase gift of Dr. T. Kornberg); DmKhc, DK410-7.1 mouse monoclonal (1:50, inhibitors was preincubated with protein G–Sepharose beads (Phar- gift of Drs. R. Sakowicz and L. Goldstein); SGG/ZW3, rabbit poly- macia) for 30 min at 48C with rocking. Beads were pelleted in a clonal (1:500, gift of Dr. K. Willert); and all Horseradish Peroxidase microfuge (30 sec), and the pellet was saved for immunoblotting. (HRP)–conjugated secondary antibodies (1:20,000, Jackson Immu- Aliquots (100 ml) of the supernatant were transferred to fresh tubes noResearch Labs). HRP was detected with Chemiluminescence re- and supplemented with 1 ml of rat polyclonal COS2 antisera, 1 ml agent (NEN). of rabbit polyclonal CI antisera (gift of Dr. T. Kornberg), or 1 ml

Anterior or posterior fragments of wing discs were dissected from of preimmune sera, and then rocked at 48C for 30 min. Protein third instar larvae and transferred to 40 mM Tris (pH 7.2), 250 mM G–Sepharose beads were added and samples rocked for 2 hr at NaCl, 5 mM EDTA, and 0.05% NP-40 on ice, 0.5 fragments/ml. Frag- 48C. Beads pelleted as before were washed three times with TNE ments were stored at 2808C, thawed, and homogenized. Approxi- buffer. Washed beads were centrifuged, and pellets and superna- mately 70 anterior and posterior disc fragment equivalents were tants were examined by immunoblotting. analyzed by immunoblotting.

Somatic clones cos2 mutant clones were made with cos2W1 (Whittle, 1976). BothProtein Detection in Embryos and Discs

Washed and dechorionated embryos were fixed, as described, with P[w1; FRT]G13 cos2W1/CyO flies and P[w1; FRT]G13 P[hsp70-MYC] (G13-pM) flies were crossed separately with yw P[ry1; FLP]12; CyO/either heat and methanol (Miller et al., 1989), methanol (Kellogg et

al., 1989), or formaldehyde (Theurkauf, 1992). After fixation, embryos Sco flies. yw P[ry1; FLP]12; P[w1; FRT]G13 cos2W1/CyO and yw P[ry1; FLP]12; G13-pM/CyO siblings were crossed, and after two days,were stored either at 2208C in methanol or taken through a rehydra-

tion series to prepare embryos for indirect immunofluorescence. adults were transferred to fresh vials. Larvae were heat-shocked on days 2, 3, and 4 for one hour at 378C. Imaginal discs were dissectedThird instar larval imaginal discs were prepared for indirect immuno-

fluorescence as described by Johnson et al. (1995). Samples were from third instar larvae 30 min after a fourth one-hour heat shock.

Cell 244

Discs were incubated with monoclonal antibodies 9E10 anti-MYC Goldstein, L.S. (1993). With apologies to Scheherazade: tails of 1001 kinesin motors. Annu. Rev. Genet. 27, 319–351.(Sigma, 1:500) and 2A1 anti-CI (gift of Dr. B. Holmgren, 1:5) and

prepared for indirect immunofluorescence. Grau, Y., and Simpson, P. (1987). The segment polarity gene costal- 2 in Drosophila. I. The organization of both primary and secondary

Acknowledgments embryonic fields may be affected. Dev. Biol. 122, 186–200.

Hammerschmidt, M., Brook, A., and McMahon, A.P. (1997). The Correspondence regarding this paper should be addressed to world according to hedgehog. Trends Genet. 13, 14–21. M. P. S. (e-mail: [email protected]). We sincerely thank P.

Harlow, E., and Lane, D. (1988). Antibodies: A Laboratory Manual Simpson, M. Bourouis, and P. Heitzler for providing the cos2 fly

(Cold Spring Harbor, NY: Cold Spring Harbor Laboratory). stocks, which were essential in cloning cos2. We thank R. Johnson,

Heitzler, P., Coulson, D., Saenz-Robles, M.T., Ashburner, M., Roote,R. Nusse, and O. Papoulas for suggestions on the manuscript and J., Simpson, P., and Gubb, D. (1993). Genetic and cytogenetic analy-M. Fuller, L. Goldstein, R. Sakowicz, and B. Holmgren for helpful sis of the 43A-E region containing the segment polarity gene costadiscussions. We are also grateful to D. Gubb, J. Tamkun, C. Good- and the cellular polarity genes prickle and spiny-legs in Drosophilaman, Ed Stephenson, M. Blanar, H. Goodson, B. Holmgren, T. Korn- melanogaster. Genetics 135, 105–115.berg, K. Willert, B. Dalby, R. Sakowicz, L. Goldstein, B. Harmon, Y. Hepker, J., Wang, Q.T., Motzny, C.K., Holmgren, R., and Orenic, T.V.Bellaiche, T. Chou, N. Perrimon, J. Sedat, R. Mayo, W. Smith, and (1997). Drosophila cubitus interruptus forms a negative feedbackF. Harris for materials and help and to D. Lee, J. J. Plecs and T. loop with patched and regulates expression of Hedgehog targetAlber for advice. The Stanford PAN facility helped with DNA se- genes. Development 124, 549–558.quencing. This work was initiated with the support of a grant from

the American Cancer Society. K. H. is supported by a National Higgins, D.G., Bleasby, A.J., and Fuchs, R. (1992). CLUSTAL V: Science Foundation predoctoral grant. M. P. S. is an investigator, improved software for multiple sequence alignment. CABIOS 8, and J. C. S. an associate, of the Howard Hughes Medical Institute. 189–191.

Hooper, J.E., and Scott, M.P. (1989). The Drosophila patched gene Received February 7, 1997; revised June 6, 1997. encodes a putative membrane protein required for segmental pat-

terning. Cell 59, 751–765. References Ingham, P.W., and Fietz, M.J. (1995). Quantitative effects of hedge-

hog and decapentaplegic activity on the patterning of the Drosophila Akimaru, H., Chen, Y., Dai, P., Hou, D.-X., Nonaka, M., Smolik, S.M., wing. Curr. Biol. 5, 432–440. Armstrong, S., Goodman, R.H., and Ishii, S. (1997). Drosophila CBP

Jiang, J., and Struhl, G. (1995). Protein kinase A and Hedgehog is a co-activator of cubitus interruptus in hedgehog signalling. Na-

signaling in Drosophila limb development. Cell 80, 563–572. ture 386, 735–738.

Johnson, R.L., Grenier, J.K., and Scott, M.P. (1995). patched overex- Alcedo, J., Ayzenzon, M., Von Ohlen, T., Noll, M., and Hooper, J.E. pression alters wing disc size and pattern: transcriptional and post- (1996). The Drosophila smoothened gene encodes a seven-pass transcriptional effects on hedgehog targets. Development 121, membrane protein, a putative receptor for the hedgehog signal. Cell 4161–4170. 86, 221–232.

Kellogg, D.R., Field, C.M., and Alberts, B.M. (1989). Identification of Alexandre, C., Jacinto, A., and Ingham, P.W. (1996). Transcriptional microtubule-associated proteins in the centrosome, spindle, and activation of hedgehog target genes in Drosophila is mediated di- kinetochore of the early Drosophila embryo. J. Cell Biol. 109, 2977– rectly by the cubitus interruptus protein, a member of the GLI family 2991. of zinc finger DNA-binding proteins. Genes Dev. 10, 2003–2013.

Kornberg, T., Siden, I., O’Farrell, P., and Simon, M. (1985). The en- Ashburner, M. (1989) Drosophila: A Laboratory Handbook (Cold grailed locus of Drosophila: in situ localization of transcripts reveals Spring Harbor, NY: Cold Spring Harbor Laboratory Press). compartment-specific expression. Cell 40, 45–53. Basler, K., and Struhl, G. (1994) Compartment boundaries and the Lasek, R.J., and Brady, S.T. (1985). Attachment of transported vesi- control of Drosophila limb pattern by hedgehog protein. Nature 368, cles to microtubules in axoplasm is facilitated by AMP-PNP. Nature 208–214. 316, 645–647. Brown, N.H., and Kafatos, F.C. (1988). Functional cDNA libraries Lepage, T., Cohen, S.M., Diaz-Benjumea, F.J., and Parkhurst, S.M. from Drosophila embryos. J. Mol. Biol. 203, 425–437. (1995). Signal transduction by cAMP-dependent protein kinase A in Campos-Ortega, J.A., and Hartenstein, V. (1985). The Embryonic Drosophila limb patterning. Nature 373, 711–715. Development of Drosophila melanogaster (Springer-Verlag). Li, W., Ohlmeyer, J.T., Lane, M.E., and Kalderon, D. (1995). Function Capdevila, J., Estrada, M.P., Sánchez-Herrero, E., and Guerrero, I. of protein kinase A in Hedgehog signal transduction and Drosophila

imaginal disc development. Cell 80, 553–562.(1994). The Drosophila segment polarity gene patched interacts with decapentaplegic in wing development. EMBO J. 13, 71–82. Lillie, S.H., and Brown, S.S. (1992). Suppression of a myosin defect

by a kinesin-related gene. Nature 356, 358–361.Cavener, D.R. (1987). Comparison of the consensus sequence flank- ing translational start sites in Drosophila and vertebrates. Nucleic Lillie, S.H., and Brown, S.S. (1994). Immunofluorescence localization Acids Res. 15, 1353–1361. of the unconventional myosin, Myo2p, and the putative kinesin-

related protein, Smy1p, to the same regions of polarized growth inCole, D.G., Saxton, W.M., Sheehan, K.B., and Scholey, J.M. (1994). Saccharomyces cerevisiae. J. Cell Biol. 125, 825–842.A “slow” homotetrameric kinesin-related motor protein purified from

Drosophila embryos. J. Biol. Chem. 269, 22913–22916. Mathies, L.D., Kerridge, S., and Scott, M.P. (1994). Role of the teash- irt gene in Drosophila midgut morphogenesis: secreted proteinsDominguez, M., Brunner, M., Hafen, E., and Basler, K. (1996). Send- mediate the combinatorial action of homeotic genes. Developmenting and receiving the hedgehog signal: control by the Drosophila 120, 2799–2809.Gli protein Cubitus interruptus. Science 272, 1621–1625. Miller, K.G., Field, C.M., and Alberts, B.M. (1989). Actin-binding pro-Eaton, S., and Kornberg, T.B. (1990). Repression of ci-D in posterior teins from Drosophila embryos: a complex network of interactingcompartments of Drosophila by engrailed. Genes Dev. 4, 1068–1077. proteins detected by F-actin affinity chromatography. J. Cell Biol.Foe, V.E., Odell, G.M., and Edgar, B.A. (1993). Mitosis and Morpho- 109, 2963–2975.genesis in the Drosophila Embryo: Point and Counterpoint. In The Moore, J.D., and Endow, S.A. (1996). Kinesin proteins: a phylum ofDevelopment of Drosophila melanogaster, M. Bate and A. Martinez motors for microtubule-based motility. BioEssays 18, 207–219.Arias, eds. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory

Press), pp. 149–300. Motzny, C.K., and R. Holmgren (1995). The Drosophila cubitus inter- ruptus protein and its role in the wingless and hedgehog signalForbes, A.J., Nakano, Y., Taylor, A.M., and Ingham, P.W. (1993). transduction pathways. Mech. Dev. 52, 137–150.Genetic analysis of hedgehog signalling in the Drosophila embryo.

Development 119 (Suppl.), 115–124. Nakano, Y., Guerrero, I., Hidalgo, A., Taylor, A., Whittle, J.R.S., and

COS2, a Kinesin-Related Regulator of Development 245

Ingham, P.W. (1989). A protein with several possible membrane- of Drosophila homeotic genes structurally related to the yeast tran- scriptional activator SNF2/SWI2. Cell 68, 561–572.spanning domains encoded by the Drosophila segment polarity

gene patched. Nature 341, 508–513. Thérond, P., Busson, D., Guillemet, E., Limbourg-Bouchon, B., Préat, T., Terracol, R., Tricoire, H., and Lamour-Isnard, C. (1993). MolecularO’Connell, P., and Rosbash, M. (1984). Sequence, structure and

codon preference of the Drosophila ribosomal protein 49 gene. Nu- organisation and expression pattern of the segment polarity gene fused of Drosophila melanogaster. Mech. Dev. 44, 65–80.cleic Acids Res. 12, 5495–5513.

Orenic, T.V., Slusarski, D.C., Kroll, K.L., and Holmgren, R.A. (1990). Thérond, P.P., Knight, J.D., Kornberg, T.B., and Bishop, J.M. (1996). Phosphorylation of the fused protein kinase in response to signalingCloning and characterization of the segment polarity gene cubitus

interruptus Dominant of Drosophila. Genes Dev. 4, 1053–1067. from hedgehog. Proc. Natl. Acad. Sci. USA 93, 4224–4228.

Theurkauf, W.E. (1992). Behavior of structurally divergent alpha-Pan, D., and Rubin, G.M. (1995). cAMP-dependent protein kinase and hedgehog act antagonistically in regulating decapentaplegic tubulin isotypes during Drosophila embryogenesis: evidence for

post-translational regulation of isotype abundance. Dev. Biol. 154,transcription in Drosophila imaginal discs. Cell 80, 543–552. 205–217.Pardue, M.-L. (1994). Looking at polytene chromosomes. In Dro-

sophila melanogaster: Practical Uses in Cell and Molecular Biology, Vale, R.D. (1996). Switches, latches, and amplifiers: common themes of G proteins and molecular motors. J. Cell Biol. 135, 291–302.Volume 44, L.S.B Goldstein and E.A. Fyrberg, eds. (San Diego, CA:

Academic Press), pp. 333–351. Vale, R.D., Reese, T.S., and Sheetz, M.P. (1985). Identification of a novel force-generating protein, kinesin, involved in microtubule-Phillips, R.G., Roberts, I.J.H., Ingham, P.W., and Whittle, J.R.S.

(1990). The Drosophila segment polarity gene patched is involved based motility. Cell 42, 39–50. in a position-signalling mechanism in imaginal discs. Development van den Heuvel, M., and Ingham, P.W. (1996). smoothened encodes 110, 105–114. a receptor-like serpentine protein required for hedgehog signalling.

Nature 382, 547–551.Poole, S.J., Kauvar, L.M., Drees, B., and Kornberg, T. (1985). The engrailed locus of Drosophila: structural analysis of an embryonic Whittle, J.R.S. (1976). Clonal analysis of a genetically caused dupli- transcript. Cell 40, 37–43. cation of the anterior wing in Drosophila melanogaster. Dev. Biol.

51, 257–268.Préat, T., Thérond, P., Lamour-Isnard, C., Limbourg-Bouchon, B., Tricoire, H., Erk, I., Mariol, M.C., and Busson, D. (1990). A putative Woehlke, G., Rudy, A.K., Hart, C.L., Ly, B., Hom-Booher, N., and serine/threonine protein kinase encoded by the segment-polarity Vale, R.D. (1997). Microtubule interaction site of the kinesin motor. fused gene of Drosophila. Nature 347, 87–89. Cell, this issue, 90, 207–216. Préat, T., Thérond, P., Limbourg-Bouchon, B., Pham, A., Tricoire, Woolfson, D.N., and Alber, T. (1995). Predicting oligomerization H., Busson, D., and Lamour-Isnard, C. (1993). Segmental polarity in states of coiled coils. Protein Sci. 4, 1596–1607. Drosophila melanogaster: genetic dissection of fused in a Suppres-

Xu, T., and Harrison, S.D. (1994). Mosaic analysis using FLP recom- sor of fused background reveals interaction with costal-2. Genetics

binase. In Drosophila melanogaster: Practical Uses in Cell and Mo- 135, 1047–1062.

lecular Biology, Volume 44, L.S.B. Goldstein and E.A. Fyrberg, eds. Rimm, D.L., and Pollard, T.D. (1989). New plasmid vectors for high (New York: Academic Press), pp. 655–681. level synthesis of eukaryotic fusion proteins in E. coli. Gene 75,

Yang, J.T., Saxton, W.M., Stewart, R.J., Raff, E.C., and Goldstein, 323–327.

L.S. (1990). Evidence that the head of kinesin is sufficient for force Robbins, D.J., Nybakken, K.E., Kobayashi, R., Sisson, J.C., Bishop, generation and motility in vitro. Science 249, 42–47. J.M., and Thérond, P.P. (1997). Hedgehog elicits signal transduction by means of a large complex containing the kinesin-related protein Costal2. Cell, this issue, 90, 225–234.

Sablin, E.P., Kull, F.J., Cooke, R., Vale, R.D., and Fletterick, R.J. (1996). Crystal structure of the motor domain of the kinesin-related motor ncd. Nature 380, 555–559.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989). Molecular Clon- ing: A Laboratory Manual (Cold Spring Harbor, NY: Cold Spring Harbor Press).

Sanchez-Herrero, E., Couso, J.P., Capdevila, J., and Guerrero, I. (1996). The fu gene discriminates between pathways to control dpp expression in Drosophila imaginal discs. Mech. Dev. 55, 159–170.

Saxton, W.M. (1994). Isolation and analysis of microtubule motor proteins. In Drosophila melanogaster: Practical Uses in Cell and Molecular Biology, Volume 44, L.S.B. Goldstein and E.A. Fyrberg, eds. (San Diego, CA: Academic Press), pp. 279–288.

Siegfried, E., Perkins, L.A., Capaci, T.M., and Perrimon, N. (1990). Putative protein kinase product of the Drosophila segment-polarity gene zeste-white3. Nature 345, 825–829.

Simpson, P., and Grau, Y. (1987). The segment polarity gene costal-2 in Drosophila. II. The origin of imaginal pattern duplications. Dev. Biol. 122, 201–209.

Slusarski, D.C., Motzny, C.K., and Holmgren, R. (1995). Mutations that alter the timing and pattern of cubitus interruptus expression in Drosophila melanogaster. Genetics 139, 229–240.

Spradling, A.C., and Rubin, G.M. (1982). Transposition of cloned P elements into Drosophila germ line chromosomes. Science 218, 341–347.

Tabata, T., and Kornberg, T. (1994). Hedgehog is a signaling protein with a key role in patterning Drosophila imaginal discs. Cell 76, 89–102.

Tamkun, J.W., Deuring, R., Scott, M.P., Kissinger, M., Pattatucci, A.M., Kaufman, T.C., and Kennison, J.A. (1992). brahma: a regulator