We thank John Flanagan, Svetlana Mojsov, Paul Tempst, Cathie. Daugherty, Tom ... inson, M. E., Hogan, B. L.M. & Rutledge, J. C. (1990) Proc. Natl. Acad. Sci.
Proc. Natl. Acad. Sci. USA Vol. 90, pp. 5554-5558, June 1993 Biochemistry
Formins: Phosphoprotein isoforms encoded by the mouse limb deformity locus (mutants/limb morphogenesis/immunoprecipitation/DNA binding)
THOMAS F. VOGT*, LAURIE JACKSON-GRUSBYt, JOHN RUSH, AND PHILIP LEDER Department of Genetics and Howard Hughes Medical Institute, Harvard Medical School, Boston, MA 02115
Contributed by Philip Leder, March 10, 1993
ABSTRACT Mutations at the mouse limb deformity (Id) locus result in defects of growth and patterning of the limb and kidney during embryonic development. The gene responsible for this phenotype is large and complex, with the capacity to generate a number of alternatively spliced messenger RNA transcripts encoding nuclear protein isoforms called "formins." We have made polyclonal antibodies to specific formin peptides and have confirmed the authenticity of the antibodies' reactivity, using cell lines derived from mice with molecularly defined mutations at the Id locus. In addition, we have used these antibodies to detect and characterize polypeptides encoded by both wild-ype and mutant Id alleles. In so doing, we show that a formin isoform (i) is modified by posttranslational phosphorylation at serine and threonine residues and (it) when present in a crude nuclear extract, is retained by DNA-cellulose.
fail to show any phenotypic abnormality supports the notion that the existing alleles of Id may be reduction-of-function alleles rather than null alleles (16-18). Since the predicted primary sequence of the formins offers few clues regarding function, we have initiated a biochemical characterization to gain insight into this important question.
MATERIALS AND METHODS In Vitro Transcription and Translation. Constructs Ia, Ib, and II were assembled from overlapping ld cDNAs (16, 17) and subcloned into pBluescript KS(+) (Stratagene) (Fig. 1). All in vitro transcription reactions were performed by using the phage promoters present on the vector and included the 7-methylguanosine-modified nucleotide for capping (Stratagene). In vitro translation was performed by using rabbit reticulocyte extracts and L-[35S]methionine (NEN, 1000 Ci/ mmol; 1 Ci = 37 GBq) as recommended by the supplier (Promega). Cell Lines and Transient Transfection. All cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with L-glutamine and 10% (vol/vol) bovine calf serum. NIH 3T3 and COS cell lines are available from the American Type Culture Collection. Primary cell lines were isolated from homozygous ldTgHd mice, a wild-type litter mate, and homozygous IdJ, idoR, 1dTTgBri137, and id"n2 mice by explanting a small piece of tail tissue into culture and growing fibroblast cells. The cell lines used were between passages 5 and 20. Total cellular RNA was prepared from cell lines by using guanidinium isothiocyanate and centrifugation through a CsCl cushion (20). RNase protection assays were performed with 20 ug of total RNA and 2 x 105 input cpm of [32P]UTP-labeled probe (21). The RNase protection probes Bi and B2 (see Fig. 2) are derived from ld cDNAs. Probe Bi, plasmid PRO.65, corresponds to ld nucleotide positions 20322398 and probe B-2, pHindIII/SpeO.472, corresponds to Id nucleotide positions 4191-4663 (15, 16, 18). Transient transfection of COS cells was performed by a DEAE-dextran protocol (22). The transfected constructs correspond to ld cDNA encoding either formin isoform lb, pcDNA-ldIb, or formin isoform IV, pcDNA-ldIV, subcloned into the pcDNAI expression vector (Invitrogen). Antibodies and Immunoprecipitations. Synthetic peptides for ld-4 (Phe-Lys-Thr-lle-Trp-Lys-Arg-Glu-Ser-Lys-Asn-Ile-
The morphogenesis and patterning of the vertebrate limb is a classical paradigm for studying induction and the acquisition of positional information during embryonic development (1-4). A large number of mutations (>40) whose phenotype includes an abnormality of limb development have been identified and genetically mapped in the mouse (5, 6). The existence of this repository of mutants presents an opportunity to determine the genetic and biochemical relationships and the molecular identity of the molecules controlling limb morphogenesis (7). There are five mutant alleles at the mouse limb deformity locus, Id, that alter the anterior-posterior development and patterning of fore and hind limbs by reducing and fusing the distal bones of the appendicular skeleton (8-14). These mutations are pleiotropic; in addition to the limb defects, there are defects in kidney development that result in renal agenesis (8, 14, 15). The cloning of the limb deformity gene, facilitated by the transgene insertional event that resulted in the creation of the ldTgHd allele, provided the initial molecular description of a phenotypic limb mutant (11, 13). Furthermore, examination of the transcripts of the mouse limb deformity locus showed that the gene is a complex transcription unit that encodes a number of protein isoforms termed the "formins" (16, 17). For three of the five limb deformity alleles, a structural alteration in the cloned gene has been identified (15, 16, 18). The unexpected findings 3hat these structural alterations are physically clustered in tt e gene and that two of the alleles, ldTgHd and ldTgBril37 are the result of transgene-associated alterations raise the possibility that the ldgene is particularly sensitive to localized disruption (11, 14, 15). The clustering of Id mutations to a small region, the observed expression of ld transcripts in all three germ layers during mouse development, and their continued expression in many adult tissues (e.g., neurons and salivary glands) that
Ser-Lys-Glu-Arg-Leu-Lys-Met-Ala-Gln-Ala-Ser-Val-Ser-
Lys-Leu-Thr-Ser-Glu) and ld-5 (Ile-Asn-Pro-Thr-Ala-SerLeu-Lys-Glu-Arg-Leu-Arg-Gln-Lys-Glu-Ala-Ser-Val-AlaThr-Asn) were synthesized with an Applied Biosystems model 430 peptide synthesizer and purified by reverse-phase HPLC. Prior to injection, female rabbits were bled for preimmune sera. Immunization was intradermal with an *Present address: Department of Molecular Biology, Princeton University, Princeton, NJ 08544-1014. tPresent address: Whitehead Institute for Biomedical Research, Cambridge, MA 02142.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
5554
Proc. Natl. Acad. Sci. USA 90 (1993)
Biochemistry: Vogt et al.
A
Predicted AUG
UAA
Ia
AUG
11
Mr
Mr
8.7
164
180
8.7
160
177
9.3
149
160
9.8
69
78
5.4
134
165
UAA
1_
AUG
AUG -
IV
UAA _
_
_
B
C la
XL
Apparent
lb
11
III
IV
D pre xx-I d
pre a-I d ac-I d
200-
U
97-
68IV Ia la IV IaIa FIG. 1. Translation and immunoprecipitattion of Id cDNAs encoding formin isoforms. (A) Schematic repressentation of the formin isoforms (I-IV) shown with their isoelectric p(oint (pI) and molecular weight (MA; shown x 10-3) as predicted from ttheir primary sequence (University of Wisconsin Genetics Computer Group peptide map lala
5555
with nonessensupplemented per ml) serum albumin bovine L-glutaacids, vitamins, amino essential acids, tial amino mine, 2.5 mCi Of 32p; (NEN, 1 Ci/mmol), and 10% fetal calf serum (dialyzed against Hanks' balanced salt solution). Labeling was carried out for 2-3 hr. For lysis and immunoprecipitation, all buffers contained 2 mM sodium orthovanadate. For phosphoamino acid analysis, immunoprecipitated 32p_ labeled proteins from transiently transfected COS cells were resolved by electrophoresis on a SDS/5% polyacrylamide gel (26), transferred to Immobilon P (Millipore) with CAPS buffer [3-(cyclohexylamino)-1-propanesulfonic acid] as described by LeGendre and Matsudaira (27). After transfer and autoradiography, segments of Immobilon membrane containing 32P-labeled formin were hydrolyzed with 5.7 M HCl (28). Hydrolyzates mixed with nonradioactive phosphoamino acid standards were analyzed by TLC on cellulose plates developed with isobutyric acid containing 0.5 M NH40H (29). Assignments were confirmed by TLC on cellulose plates at pH 3.5 (29). DNA-Cellulose Binding. pcDNA-ldIV was transiently transfected into COS cells by DEAE-dextran, and the cells were labeled with L-[35S]methionine (0.5 mCi/ml). After cell labeling, all manipulations were carried out at 4°C as described by Pognonec and colleagues (30). Cell lysates were added to 1.0 ml of native double-stranded calf thymus DNAcellulose (Pharmacia, 1.2 mg/ml) preequilibrated in loading buffer (10 mM Tris, pH 7.5/50 mM NaCl/0.5% Triton). The column was washed extensively with loading buffer (>15 bed salt volumes) followed by stepwise elution with increasing concentrations (0.2, 0.4, 0.6, 0.8, and 1.0 M NaCl). Equivalent amounts of each fraction were adjusted to RIPA buffer conditions. The fractions were then immunoprecipitated and resolved by electrophoresis on a SDS/6% polyacrylamine gel.
1sv_rs !_____!_____.! _t_
program; ref. 19). The apparent Mr refers to the size of the proteins as estimated from the in vitro translation of these cDNAs as shown in B. (B) In vitro transcription and translation of the Id cDNAs encoding formin isoforms as labeled in A. The [35S]methioninelabeled proteins were resolved by electrophoresis on a SDS/5% PAGE. The numbers indicate the migration of the protein size markers in kDa. (C) Immunoprecipitation of in vitro translated 35S-labeled isoform Ia by either preimmune (lane pre) serum or a mixture of a-ld-4 and a-ld-5 sera (lane a-ld). (D) Immunoprecipitation of 35S-1abeled formins Ia and IV after transient expression of the respective cDNAs in COS cells by either preimmune serum (lane pre) or a mixture of a-ld-4 and a-ld-5 sera (lane a-4d). (D) Immunoprecipitation of 35S-labeled formins Ia and IV after transient expression of the respective cDNAs in COS cells by either preimmune serum (lane pre) or a mixture of a-ld-4 and a-ld-5 sera (lanes a-ld). Immunoprecipitated proteins were resolved by electrophoresis on a SDS/5% PAGE.
emulsion of unconjugated peptide, methylated bovine serum albumin, and Freund's adjuvant (23). Serum was collected after the second and third booster and used without additional purification. Immunoprecipitation of [35S]methioninelabeled proteins from rabbit reticulocyte lysates was performed as described by Anderson and Blobel (24). Immunoprecipitation of [35S]methionine-labeled proteins from cell lines or transiently transfected COS cells was performed essentially as described in Harlow and Lane (25). Cells were labeled with L-[35S]methionine (NEN, 1000 Ci/mmol), and immunoprecipitations were performed with protein-ASepharose (Pharmacia) in radioimmunoprecipitation assay (RIPA) buffer (144 mM NaCl/50 mM Tris HCl, pH 8.0/0.1% SDS/0.5% deoxycholate/0.5% Nonidet P-40/2 mM EDTA) containing 1 mM phenylmethylsulfonyl fluoride and 1% aprotinin (Sigma). For immunoprecipitation of 32P-labeled proteins, cells were treated as above except that the cell labeling was performed in 1 x Salts (107 mM NaCl/5 mM KCl/3 mM CaCl2/l mM MgSO4/44 mM Na2CO3/10 mM glucose/2 mg of
RESULTS AND DISCUSSION Characterization of Anti-Id Antibodies and the Formin Products of Mutant Alleles. The amino acid sequences of five formin isoforms were predicted from cDNA sequences, and their apparent molecular masses were determined by in vitro transcription and translation followed by resolution by SDS/ PAGE (Fig. 1 A and B). Formin isoforms I-III share a common amino terminus encoded by a single 2-kilobase (kb) exon. This exon has an estimated pI of 9.8 because of its high density of basic amino acids (16). In contrast, the alternative amino terminus of isoform IV is encoded by a single 1.6-kb exon that is highly acidic with an estimated pl of 4.5 (17). To begin examining the encoded products of the Id gene and to verify the open reading frames predicted by cDNA sequencing, cDNAs encoding isoforms I-IV were transcribed in vitro, and the resulting synthetic RNA was translated in rabbit reticulocyte lysates (Fig. 1B). In all cases, the apparent molecular mass of the translated protein appears to be greater than that predicted by the primary sequence. Such differences may be due to (i) the highly charged nature of all of the isoforms; (ii) the presence of an extensive proline-rich region (66 proline residues over a region of 100 amino acid residues) in a single exon included in isoforms I, II, and IV (16, 17); or (iii) posttranslational modification. An additional smaller translation product of formin isoforms I and II is apparently due to an internal initiation at the penultimate amino-terminal methionine. Additional studies with a series of truncated synthetic RNA templates support the relative apparent size and authenticity of the in vitro translated proteins (data not shown). Based on the cDNA sequences, two peptides termed ld-4 and ld-5 (Fig. 2A) were synthesized and used as immunogens to raise rabbit polyclonal antisera against the formins. A compelling reason for selecting these particular peptide se-
5556
Proc. Natl. Acad. Sci. USA 90 (1993)
Biochemistry: Vogt et al.
A
WT I dHd
B-1
w
8-2
dHd G_
AL Iv Probes
B-i
B-2
Peptides
B
C
pre WT
a-Id WT
Id
4 5
a-Id
C
WT WT
Hd
200-
4
Idi
IdOR
::s
200....
-)I
.a,
d*o
' -0
97-
97-
FIG. 2. Immunoprecipitation of formin IV from wild-type and mutant Id alleles. (A) The schematic is a representation of formin IV showing the relative positions of the cDNA RNase protection probes, labeled B-1 and B-2, and the carboxyl-terminal synthetic peptides, labeled 4 and 5, to the site of the transgene disruption in the IdTgHd allele, ldHd. The results of an RNase protection analysis after hybridization of RNA prepared from either the wild type (lane WT) primary fibroblast cell line or the homozygous ldTgHd (lanes ldHd) primary fibroblast cell line to ld cDNA probes B-1 and B-2 are shown above the schematic. (B) Immunoprecipitation of [35S]methioninelabeled proteins from wild-type (lanes WT) and homozygous ldTgHd (lanes ldHd) primary cell lines (-1 x 106 cells) with preimmune (lanes pre) and a-ld-4 sera (lanes a-ld). The arrow indicates the position formin IV at a molecular mass of e165 kDa. (C) Immunoprecipitation of [35S]methionine-labeled proteins from wild type (lanes WT) and homozygous IdJ and ldOR primary cell lines with preimmune serum (lane pre) compared with a mixture of a-ld-4 and a-ld-5 sera (lanes a-ld). The arrow indicates the position of formin IV at a molecular mass of e165 kDa. The weaker signal in lane ldoR is a reflection oflower expression of ld transcripts in this primary cell line and the labeling of less cells, -1 x 105. Immunoprecipitated proteins were
resolved by electrophoresis on a SDS/5% polyacrylamide gel. migration of the protein size markers in
The numbers indicate the kDa.
quences is that the insertional event associated with the creation of the ldTgHd allele abolishes the stable production of the RNA encoding these peptide epitopes (Fig. 2A; ref. 18). Therefore, mice segregating the 1dTgHd allele provide a powerful genetic control by which to evaluate the authenticity of immunoreactive protein. Subsequent molecular characterization of the ldTgBril37 allele, a second independently created transgene-associated deletion at the Id gene, revealed that these epitopes would also be absent, therefore providing another genetic control (14, 15). Antisera raised against the peptides were screened by ELISA (data not shown) and by their ability to immunoprecipitate in vitro translated formin protein Ia (Fig. 1C). In addition, antisera were assayed for their ability to specifically immunoprecipitate formin isoforms after transient overexpression of ld cDNA encoding either formin isoform Ia or isoform IV in COS cells (Fig. ID). No proteins are specifically immunoprecipitated from mocktransfected COS cell extracts (data not shown) or from transfected cell extracts incubated with preimmune sera (Fig.
1D). Formin isoform IV is consistently detected at higher levels than formin isoform I in both the in vitro translation and transient expression assays. To date, we have failed to achieve stable overexpression of ld cDNA after transfection into NIH 3T3 and HeLa cells or in transgenic mice using heterologous promoters (data not shown). The underlying cause is undetermined but may suggest cells overexpressing formin isoforms are at a selective disadvantage. Characterization of the reactivity of the ld peptide antisera demonstrated that their greatest utility was in immunoprecipitation assays and that they were not reactive in immunoblot and immunohistochemistry assays. Therefore, we assayed expression of the endogenous formin isoforms by metabolic labeling of cultured cells. In a survey of expression of ld transcripts in established cell lines, mouse fibroblast cells (e.g., NIH 3T3 and Balb 3T3) were found to express relatively high levels of ld mRNA when compared with mouse tissues. Characterization of their ld transcripts by RNase protection assays designed to distinguish alternatively spliced transcripts demonstrated that NIH 3T3 cells almost exclusively produce ld transcripts encoding isoform IV (Fig. 2A; data not shown). Based on ld transcript analysis in developing mouse embryos, isoform IV is predicted to be expressed in the developing limb bud. Transcripts encoding isoforms I-III appear to be absent from the limb bud, although along with transcripts encoding formin IV they are present in other regions ofthe mouse embryo (17). Therefore, studies of isoform IV in cultured cell lines may prove to be useful for understanding the role of this isoform in embryonic limb bud mesoderm. As a genetic control for validating the utility of our peptide antisera to assay for the expression of formin IV, we established primary fibroblast cell lines from a homozygous ldTgHd/ ldTgHd animal and a homozygous wild-type (+/+) littermate. The wild-type primary fibroblast cell line expressed ld transcripts indistinguishable from those found in mouse tissue and in NIH 3T3 cells (Fig. 2A; data not shown). Our previous analysis of the transgene insertion present in the ldTgHd mutant allele demonstrated that it disrupts the stable production of ld transcripts encoding the epitopes of the peptide immunogens (18). RNA from the ldTgHd/ldTgHd primary cell line was assayed with an RNase protection probe that includes this region of the ld mRNA (Fig. 2A). As found for tissue RNA, while expressing ld transcripts 5' to the transgene insertion site (albeit at lower levels), the homozygous kdTgHd primary fibroblast cell line did not express stable ld RNA transcripts 3' to the transgene insertion site (Fig. 2A, compare Bi versus B2). Therefore, the ldTgHd/ldTgHd cell line is a genetic null for the formin epitopes to which our antisera are directed. To identify endogenous formin IV protein, 35S-labeled proteins from wild-type and homozygous IdTgHd cell lines were assayed by immunoprecipitation. A protein of 165 kDa is specifically immunoprecipitated from the wild-type cell line, but is absent in both the homozygous ldTgHd mutant cell line and in the preimmune samples (Fig. 2B). The immunoprecipitated endogenous formin IV is indistinguishable in size from both the immunoprecipitated in vitro translated and COS cell-expressed isoform. Identification of formin IV at 165 kDa is also supported by the absence of this immunoprecipitated polypeptide in a primary cell line derived from a ldTgBri137/ldTgBr1.37 mouse. In this mutation, an independent transgene-associated deletion in the Id gene results in a carboxyl-terminal truncation eliminating the formin amino acid epitopes (ref. 15; data not shown). Having established an assay for endogenous formin, we next proceeded to examine isoform IV expression from two Id mutant alleles, ldJ and idoR, in which the underlying molecular alteration is unknown. Both these alleles were recovered as spontaneous mutations more than 30 years ago (8, 9). Searching for the mutation associated with these alleles
~~~~Proc. Nati. Acad. Sci. USA 90 (1993)
Biochemistry: Vogt et al.Biochemistry: Vogt et al. is diffi'cult because of the large size of the id gene (>200 kb) and its complex organization. Thus far, our ongoing analysis of genomic sequences and RNA transcripts of these two id alleles has not revealed any alterations (data not shown). This finding is consistent with these alleles being point mutations or other subtle mutations. Our antisera to the carboxylterminal formin epitopes provide a complementary opportunity for surveying the multiple isoforms that include this region. Therefore, we established primary fibroblast cell lines from tail tissue of homozygous ldJ/ldJ and ldPOR/ldOR animals. In contrast to the jdTgHd mutant, both the IdJ' and idOR mutants express an immunoprecipitable isoform IV that is very similar in size to wild type (Fig. 2C). This result appears to rule out the introduction of a nonsense or frame-shift mutation into the ld gene sequences encoding isoform IV and focuses future inspection of these mutated alleles on other types of alterations (e.g., substitutions, posttranslational modifi'cation differences) and to formin peptide sequences not included in isoform IV (e.g., formin III). Formins Are Subject to Phosphorylation at Serine and Threonine Residues. Using mouse ld clones as probes, Trumpp et al. (31) have recently isolated a chicken cDNA with extensive homology to mouse isoform IV. Nuclear staining in primary chicken embryo fibroblasts and in specific regions of the developing chicken embryo was observed by using a polyclonal antibody raised to a fusion protein containing a carboxyl-terminal portion of the encoded chicken Id peptide sequence (31). Given the nuclear localization of the chicken formin (31), the presence of potential phosphorylation sites within predicted formin sequences, and the fact that phosphorylation/dephosphorylation is used in the regulatory control of many nuclear proteins (32), we wished to test whether formin isoform IV could be posttranslationally modifi'ed in this way. Thus, primary wild-type and ldgdlTH fibroblast cell lines were labeled with 32P, followed by immunoprecipitation with anti-formin peptide antisera. Immunoprecipitated isoform IV was phospholabeled (Fig. 3A). The phosphophorylated formin IV comigrated with [35S]methionine-labeled protein from wild-type fibroblasts. Significantly, the phospho-labeled formin at 165 kDa was absent from immunoprecipitates of 32p-labeled ldTrgHd/1dTgHd cells (Fig. 3A). We next sought to determine the nature of the phosphorylation of formin IV. Formin IV was transiently overexpressed in COS cells and immunoprecipitated to generate suffi'cient amounts of 32p-labeled isoform for phosphoamino acid analysis (Fig. 3B). Phosphoamino acids of isoform IV separated by TLC demonstrated phosphorylation of serine and possibly threonine but no detectable labeling of tyrosine (Fig. 3C). This result suggests that formin isoforms are substrates for cellular senine/threonine protein kinase(s). Evolutionarily conserved homologies to the proposed substrate recognition consensus sequences for cAMP/cGMPdependent protein kinase, casein kinase II, protein kinase C, and p34 cdc2 kinase phosphorylation are found in the formin IV primary amino acid sequence of mouse and chicken [PROSITE program; ref. 19]. It is notable that a number of the conserved kinase recognition sites are located in close proximity to candidate nuclear localization sequences (33) that are conserved between mouse and chicken. The conserved p34 cdc2 kinase site in formin IV (amino acid residues 468-472: Arg-Thr-Pro-Gly-Arg, in which Tyr is the site of phosphorylation) lies in close proximity to a candidate nuclear localization signal (amino acid residues 478-489: Pro-Pro-ProLys-Thr-Lys-Asp-Thr-Glu-Glu-Lys-Val); both a cAMP! cGMP and a casein kinase II site lie close to a candidate nuclear localization signal (amino acid residues 410-414: Arg-Pro-ro-LyTs-Lys) -Ana protinkias C-Andcsi kinase II site are within a candidate nuclear localization signal (amino acid residues 1188-1199: Pro-Thr-Ala-Ser-LeuLys-Glu-Arg-Leu-Arg-Gln-Lys). This type of juxtaposition
A
at-Id
pre
at- Id
WT
WT
Id Hd
Ss 32P3~
3$~35s
3 2P3
5557
200-
97-
B
pre a-Id
C
200-
97-
FIG. 3. Formin IV is a phosphoprotein. (A) Immunoprecipitation of proteins labeled with either [35S]methionine or 32P1 from wild-type (lanes WT) and homozygous IdTgHNd (lanes ldHd) primary cell lines (- 1 X 106 cells) with preimmune serum (lanes pre) compared with a mixture of a-ld-4 and a-id-5 sera (lanes a-id). The arrow indicates the position of formin IV at a molecular mass of -'165 kDa. (B) Immunoprecipitation and immunoblot of 32p1-labeled proteins of an Id cDNA encoding formin IV transiently expressed in COS cells. The arrow indicates the position of formin IV at -165 kDa. Immunoprecipitated proteins were resolved by electrophoresis on a SDS/5% polyacrylamide gel. The numbers indicate the migration of the protein size markers in kDa. (C) Phosphoamino acid analysis of the 32p-labeled formin IV shown in B. The schematic to the right indicates the relative migration of the amino acid standards for serine, threonine, and tyrosine.
has been associated with the regulated nuclear translocation of certain transcription factors (32). Formins Are Bound to DNA-CeUlulose. The nuclear subcellular localization of the chicken formin along with the observation that mouse formin IV exists as a phosphoprotein raises the possibility that formins may function by either direct or indirect association with DNA. To begin to test this notion, we have examined the ability of formin IV overexpressed in COS cell extracts to be retained on DNA cellulose. [35S]Methionine labeled COS cell lysates were incubated with double-stranded DNA cellulose and then eluted with increasing salt concentration. The eluted fractions were then immunoprecipitated with the anti-Id peptide sera and resolved by SDS/PAGE (Fig. 4). In diffTerent experiments, 10-30% of the formin IV was found in the flow-through, whereas the majority of formin IV was eluted from the column between 0.2 M and 0.6 M NaCl. Similar results were obtained with single-stranded DNA-cellulose; however, in contrast, no formin IV was retained on control cellulose columns (data not
shown). Examination of the formin IV sequence does not reveal any
relationship to known DNA/RNA binding motifs with the possible exceptions of a weak nonspecific DNA-binding activity associated with p34 cdc kinase phosphorylation sites and of a region of high proline density (34, 35). The proline density observed in mouse Id is conserved in the chicken ld gene product and its presence suggests a number of possible functional roles. Proline-rich regions have been suggested to
5558
Proc. Natl. Acad. Sci. USA 90 (1993)
Biochemistry: Vogt et al. pre
a-
Id
FT FT 0.2 0.4 0.6 0.8 1.0
200-
by a fellowship from the American Cancer Society and a research associateship from the Howard Hughes Medical Institute. 1. 2. 3. 4. 5.
!i ....
6. 1:.
7. 8. 9. 10.
97- j *6.8.
lb.
11.
12. 13. FIG. 4. DNA-cellulose chromatography of formin IV from COS cell transient expression. Immunoprecipitation of [35S]methioninelabeled proteins of the flow-through fraction (lanes FT) and the fractions eluted with increasing NaCl concentrations (0.2, 0.4, 0.6, 0.8, and 1.0 M) from a DNA-cellulose column. The fractions from the flow-through were immunoprecipitated with either preimmune serum (lane pre) or a mixture of a-ld-4 and a-ld-5 sera (lanes a-ld). The eluted fractions were immunoprecipitated with a mixture of the a-ld-4 and a-ld-5 sera. The immunoprecipitated proteins were resolved by electrophoresis on a SDS/6% polyacrylamide gel. The longer arrow indicates the position of formin IV at a molecular mass -165 kDa, and the two shorter arrows at -115 kDa and "81 kDa indicate presumed degradation products of formin IV that retain the carboxyl terminus of the protein. The numbers indicate the migration of the protein size markers in kDa. act as transcriptional activation domains in a number of transcription factors (35). In c-myc, a proline-rich region has been shown to have a serine residue that is a target of growth-factor-stimulated phosphorylation (36). Alternatively, the proline-rich region may be providing a flexible "dmolecular hinge," separating in the case of formin isoform IV domains of an amphipathic polypeptide (acidic amino terminus/basic carboxyl terminus). Such a role has been proposed for a proline-rich region in vinculin (37). Lastly, very recent studies by Ren et al. (38) using a filter binding assay indicate that the proline-rich region of formin contains a motif that binds to SH3 domains. This raises the possibility that the formins associate with proteins bearing such domains, and the identity of these associated proteins may provide a crucial link to understanding formin function. Because of the assessment of formin DNA-binding in the context of a COS cell extract, it remains an open issue whether formin IV possesses intrinsic DNA binding activity or is being retained on the DNA cellulose column indirectly through an association with another protein. Studies using purified formin and the demonstration of sequence-specific binding will be required to address this issue. Such studies, the determination of the phosphorylating kinase(s), and the identification of physiologically relevant protein interactions of formin with proteins with SH3 modules will ultimately be linked to mutant ld alleles and will aid in defining the role of formins in development and the phenotypic consequences of their altered function. We thank John Flanagan, Svetlana Mojsov, Paul Tempst, Cathie Daugherty, Tom Roberts, and Briggs Morrison for advice and reagents and David Chan, Radek Skoda, Tom Vasicek, and members of the Leder laboratory for their comments. T.F.V. was supported
14.
Brockes, J. P. (1989) Neuron 2, 1285-1294. Maini, P. K. & Solursh, M. (1991) Int. Rev. Cytol. 129, 91-133. Tabin, C. J. (1991) Cell 66, 199-217. Tickle, C. (1991) Development, Suppl. 1, 113-122. Gruneberg, H. (1%3) The Pathology of Development (Blackwell, Oxford). Lyon, M. F. & Searle, A. G. (1989) Genetic Variants and Strains of the Laboratory Mouse (Oxford Univ. Press, Oxford). Reith, A. D. & Bernstein, A. (1991) Genes Dev. 5, 1115-1123. Cupp, M. B. (1962) Mouse News Lett. 26, 51. Green, M. C. (1962) Mouse News Lett. 26, 34. Kleinebrecht, J. J., Selow, J. & Winkler, W. (1982) Anat. Anz. 152, 313-324. Woychik, R. P., Stewart, T. A., Davis, L. G., D'Eustachio, P. & Leder, P. (1985) Nature (London) 318, 36-40. Zeller, R., Jackson-Grusby, L. & Leder, P. (1989) Genes Dev. 3, 1481-1492. Woychik, R. P., Generoso, W. M., Russell, L. B., Cain, K. T., Cacheiro, N. L. A., Bultman, S. J., Selby, P. B., Dickinson, M. E., Hogan, B. L. M. & Rutledge, J. C. (1990) Proc. Natl. Acad. Sci. USA 87, 2588-2592. Messing, A., Behringer, R. R., Slapeak, J. R., Lemke, G., Palmiter, R. D. & Brinster, R. L. (1990) Mouse Genome 87,
107. 15. Vogt, T. F., Jackson-Grusby, L., Wynshaw-Boris, A. J., Chan, D. C. & Leder, P. (1992) Mamm. Genome 3, 431-437. 16. Woychik, R. P., Maas, R. L., Zeller, R., Vogt, T. F. & Leder, P. (1990) Nature (London) 346, 850-853. 17. Jackson-Grusby, L., Kuo, A. & Leder, P. (1992) Genes Dev. 6, 29-37. 18. Maas, R. L., Zeller, R., Woychik, R. P., Vogt, T. F. & Leder, P. (1990) Nature (London) 346, 853-855. 19. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic Acids Res. 12, 387-395. 20. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J. & Rutter, W. J. (1979) Biochemistry 18, 5294-5299. 21. Krieg, P. A. & Melton, D. A. (1987) Methods Enzymol. 155, 313-324. 22. Seed, B. & Aruffo, A. (1987) Proc. Natl. Acad. Sci. USA 84, 3365-3369. 23. Mojsov, S., Heinrich, G., Wilson, I. B., Ravazzola, M., Orci, L. & Habener, J. F. (1986) J. Biol. Chem. 261, 11,880-11,889. 24. Anderson, D. J. & Blobel, G. (1979) Methods Enzymol. 96, 111-115. 25. Harlow, E. & Lane, D. (1988) Antibodies: A Laboratory Manual (Cold Spring Harbor Lab., Plainview, NY), pp. 424468. 26. Laemmli, U. (1970) Nature (London) 277, 680-685. 27. LeGendre, N. & Matsudaira, P. (1989) in A Practical Guide to Protein and Peptide Purification for Microsequencing (Academic, New York), pp. 49-69. 28. Kamps, M. P. & Sefton, B. M. (1989) Anal. Biochem. 176, 22-27. 29. Cooper, J. A., Sefton, B. M. & Hunter, T. (1983) Methods Enzymol. 99, 387-402. 30. Pognonec, P., Boulukos, K. E. & Ghysdael, J. (1989) Oncogene 4, 691-697. 31. Trumpp, A., Blundell, P. A., de la Pompa, J. L. & Zeller, R. (1992) Genes Dev. 6, 14-28. 32. Hunter, T. & Karin, M. (1992) Cell 70, 375-387. 33. Silver, P. A. (1991) Cell 64, 489-497. 34. Churchill, M. E. A. & Suzuki, M. (1989) EMBO J. 8, 41894195. 35. Mermod, N., O'Neill, E. A., Kelly, T. J. & Tjian, R. (1989) Cell 58, 741-753. 36. Seth, A., Alvarez, E., Gupta, S. & Davis, R. (1991) J. Biol. Chem. 266, 23521-23524. 37. Couto, M. D. & Craig, S. W. (1988) Proc. Natl. Acad. Sci. USA 85, 8535-8539. 38. Ren, R., Mayer, B. J., Cicchetti, P. & Baltimore, D. (1993) Science 259, 1157-1161.
