Isolation and characterization of murine Ajuba cDNA.
To identify candidate proteins interacting with a cytoplasmic C-terminal domain of the murine EPO-R, we used a yeast two-hybrid screen. From an E9.5 total mouse partial cDNA library (39
), we identified four clones that specifically interacted with the EPO-R-derived protein in yeast. Two of these clones were overlapping partial cDNA clones that upon DNA sequencing were found to contain an open reading frame encoding for two LIM domains. To isolate the full-length cDNA encoding for Ajuba, a 554-bp partial cDNA piece was used to screen an E6.5 total mouse embryo phage cDNA library (42
). The 2,984-nucleotide sequence of clone 3B is shown in Fig. .
FIG. 1 Nucleotide and derived amino acid sequences of murine Ajuba cDNA. Amino acids are numbered on the left, and nucleotides are numbered on the right. The three LIM domains are underlined. Potential SH3 recognition motifs are in bold type. A putative NES (more ...)
The ATG at position 267 was presumed to be the start codon since it is flanked by the nucleotides that conform to the canonical eukaryotic translation initiation consensus sequence (22a
), and stop codons are present in all three reading frames upstream of this site. In addition, upstream of this site is a GC (75%)-rich region of 266 nucleotides. The start site was followed by a single open reading frame of 1,638 nucleotides and a 3′ untranslated region of 1,074 nucleotides. The cDNA is predicted to encode a 547-amino-acid protein of approximately 58 kDa, which we have designated Ajuba (“curiosity” in Urdu, an Indian dialect). Examination of its deduced amino acid sequence revealed a number of salient features. First, the C-terminal third was rich in cysteine and histidine residues which were organized into three tandemly arrayed copies of LIM domains with highest homology to group 3 LIM proteins (44% identity and 65% similarity score with chicken zyxin) (9
). Second, the N-terminal two-thirds was rich in glycine and proline (16 and 11%, respectively). The abundance of glycine residues in the N terminus was unique and distinguishes Ajuba from other group 3 members, such as zyxin, Enigma, and paxillin. The abundance of proline residues is typical of group 3 LIM proteins; however, most have a higher percentage (15 to 25%). In addition, there were two stretches of proline-rich SH3 recognition motifs (1
). Third, there was a nuclear export signal (NES) motif, very much like the functional NES present in zyxin both in its location within the deduced sequence and in its sequence homology (7
). Finally, there were no stretches of hydrophobic amino acids, typically present in transmembrane proteins or indicative of signal sequences.
Tissue expression of Ajuba.
Northern blot analysis for Ajuba gene expression was performed on embryonic and adult tissues and cell lines (Fig. ). A single 3-kb transcript was present in totipotent ES cells, embryonic yolk sac endoderm and mesoderm cell lines, placenta, undifferentiated F9 teratocarcinoma cells, and E12.5 fetal liver. Induction of ES cell differentiation into embryoid bodies resulted in a threefold increase of Ajuba transcript. RNA in situ hybridization studies of developing mouse embryos (data not shown) revealed that in early postimplantation embryos (E7.5 to 8.5) Ajuba was present in all embryonic germ layers, in the extraembryonic yolk sac blood islands, and in the fetal components of the developing placenta. As development progressed, expression was dramatically restricted such that in maturing embryos (post-E12.5), Ajuba expression was limited to the skin, nervous system, and genitourinary tract. Among adult tissues, Ajuba was present in the skin, brain, and genitourinary organs (e.g., testis, epididymis, ovary, uterus, and kidney). No transcript was detected in adult subcutaneous tissue, bone marrow cells, liver, spleen, thymus, stomach, intestine, or skeletal muscle.
FIG. 2 Northern blot analysis for Ajuba mRNA expression in cell lines and tissues. Total RNA (12 μg) from each tissue was subjected to formaldehyde-agarose gel electrophoresis, transferred to a Zetabind membrane, and hybridized with a 32P-labeled Ajuba (more ...) Ajuba is a 55-kDa protein.
Rabbit polyclonal antiserum was generated against a carboxy-terminal peptide of Ajuba. In vitro-translated full-length Ajuba clone 3B produced a protein product of approximately 55 kDa (Fig. , lane 1). From the products of this in vitro translation reaction, anti-Ajuba immune serum precipitated a 55-kDa protein (lane 3), whereas preimmune serum did not react with any products (lane 2). Immune serum also specifically detected a 55-kDa protein in detergent-soluble cell extracts from F9 (data not shown) and ES (see Fig. A) cells. Thus, the immune serum specifically recognized a 55-kDa protein in cells expressing Ajuba mRNA.
FIG. 3 Ajuba cDNA encodes a protein of 55 kDa specifically recognized by anti-Ajuba immune serum. The cDNA of Ajuba, clone 3B, was subcloned into pBS K/S−; 1 mg of linearized plasmid was transcribed and translated in the presence of [35S]methionine. (more ...)
FIG. 5 Ajuba associates with Grb2 in vitro and in vivo in a serum-dependent manner. (A) ES cell extracts (ca. 10 million cells per lane) were incubated with approximately 5 μg of the indicated fusion proteins. Bound products were isolated with glutathione-agarose (more ...)
When the full-length EPO-R and Ajuba proteins were coexpressed in cell lines, we did not detect an interaction between the two proteins (data not shown). In addition, GST fusion proteins of each protein did not interact with the reciprocal protein from cell extracts in in vitro pull-down experiments (data not shown). Therefore, despite the interaction between a partial cDNA of Ajuba and a domain of the EPO-R cytoplasmic tail in yeast, the two full-length proteins did not interact in vitro or in vivo. Nonetheless, many features of the Ajuba protein sequence (e.g., LIM domains and amino-terminal SH3 recognition motifs) and developmental pattern of expression prompted us to determine the cellular functions for Ajuba.
Ajuba is a cytosolic protein, not found at sites of focal adhesion or associated with the actin cytoskeleton.
To determine the subcellular distribution of Ajuba, immunofluorescence analysis was performed. NIH 3T3 fibroblast cell lines expressing Myc-tagged full-length Ajuba (3T3.Ajuba), Myc-tagged pre-LIM domain of Ajuba (the N terminus) (3T3.PreLIM), or Myc-tagged LIM domains of Ajuba (all three LIM domains in the C terminus) (3T3.LIM) were generated. NIH 3T3 cells do not express Ajuba mRNA or protein (data not shown). Results of immunofluorescence studies with anti-Myc antiserum and Ajuba-containing 3T3 cell lines are presented in Fig. . No Myc expression was detected in cells transfected with an empty vector (Fig. A). In cells containing full-length Myc-Ajuba, the protein was found predominantly in the cytosol (Fig. B). There was no nuclear staining; however, trace staining of the cell surface was evident. Subcellular fractionation studies using F9 cells, which express endogenous Ajuba, gave identical results (data not shown). 3T3 cells containing Myc–pre-LIM Ajuba expressed the protein only in the cytosol, in a pattern similar to cells containing full-length Ajuba (Fig. C). In cells containing only the three LIM domains found at the C terminus of Ajuba, significant amounts of protein were detected in the nucleus as well as throughout the cytosol (Fig. D).
FIG. 4 Immunofluorescence analysis of NIH 3T3 fibroblasts expressing Ajuba isoforms. Immunofluorescence was performed as described in Materials and Methods. (A to D) Single immunofluorescence with anti-Myc antiserum; (E) dual immunofluorescence with anti-Myc (more ...)
Since two closely related group 3 LIM proteins, zyxin and paxillin, are found at sites of adhesion between cells and the substratum, we performed dual immunofluorescence on Myc-Ajuba-containing 3T3 cells transiently transfected with paxillin (Fig. E). No colocalization of Ajuba and paxillin was observed at sites of focal adhesion. Zyxin has been shown to also associate with the actin cytoskeleton (9
). To determine if Ajuba colocalized with actin filaments in the cytosol, 3T3 cells containing Myc-Ajuba were costained with anti-Myc antiserum and phalloidin to detect actin filaments (Fig. F). We did not observe any colocalization of Ajuba with the actin filaments.
This analysis demonstrated that like other group 3 LIM proteins, Ajuba was predominantly a cytosolic protein, and a small amount may be associated with the cell surface membrane. However, in contrast to other group 3 LIM proteins, Ajuba did not localize to sites of cellular adhesion to substratum, nor did it associate with actin filaments. Interestingly, removal of the amino terminus, including a putative NES, resulted in the accumulation of the LIM domains in the nucleus.
Ajuba associates with Grb2 in vitro and in vivo.
The amino-terminal half of Ajuba contained two proline-rich regions that correspond to potential SH3 recognition motifs (Fig. ) (1
). To determine if cytosolic SH3-containing proteins might associate with Ajuba, we added various SH3-containing GST fusion proteins to ES cell extracts. Following incubation, the GST fusion proteins were isolated with glutathione-agarose, the products were separated by SDS-PAGE and transferred to nitrocellulose, and immunoblotting with anti-Ajuba antiserum was performed. This in vitro analysis demonstrated that Grb2 (Fig. A, lane 3) and related Grap (14
) (Fig. ) associated with Ajuba, whereas GST alone, GST-Nck, GST-Lck, and GST-Vav did not (Fig. A, lanes 2 and 4 to 6).
FIG. 6 The pre-LIM domain of Ajuba and the SH3 domains of Grb2 mediate the association of Ajuba and Grb2. (A) Extracts from approximately 10 million 3T3.PreLIM Ajuba cells (lanes 2 to 5) or 3T3.LIM Ajuba cells (lanes 7 to 10) were incubated with approximately (more ...)
To determine if Ajuba and Grb2 associated in vivo, 3T3 cells containing Myc-Ajuba were deprived of serum overnight and then stimulated with serum for 15 min. Detergent-soluble cell extracts were prepared and immunoprecipitated with anti-Grb2 antiserum followed by immunoblotting of products with anti-Myc antiserum. This in vivo analysis indicated that cellular Ajuba associated with Grb2 following serum stimulation (Fig. B, lane 4), whereas under these conditions no association was detected in cells starved of serum (Fig. B, lane 3). This analysis was also carried out in ES and F9 cells, which contain endogenous Ajuba. As observed with 3T3.Ajuba cells, expressing exogenous Ajuba, there was an association between Ajuba and Grb2 (data not shown). Thus, Ajuba associated with Grb2 in vitro, and more importantly, in vivo.
The association between Ajuba and Grb2 was mediated by the N-terminal pre-LIM domain of Ajuba and either SH3 domain of Grb2.
To more precisely define the regions of Ajuba required for the interaction with Grb2, we performed in vitro pull-down experiments. Purified GST-Grb2 and GST-Grap fusion proteins were added to extracts from Myc–pre-LIM Ajuba-containing 3T3 cells or Myc-LIM Ajuba-containing 3T3 cells. As negative controls, GST alone and GST-Vav fusion proteins were added to the same cell extracts. Results of these experiments are presented in Fig. A. GST-Grb2 and GST-Grap interacted with the pre-LIM domain of Ajuba (lanes 3 and 4). There was no interaction between GST-Grb2 or GST-Grap and the LIM domains of Ajuba (lanes 8 and 9). GST alone (lanes 2 and 7) or GST-Vav fusion protein (lanes 5 and 10) did not interact with either domain of Ajuba. Next, GST-Ajuba, GST–pre-LIM Ajuba, and GST-LIM Ajuba fusion proteins were added to F9 cell extracts and following incubation were isolated by incubation with glutathione-agarose beads. Bound products were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with anti-Grb2 antiserum. Results of these experiments are presented in Fig. B. GST-Ajuba and GST–pre-LIM Ajuba bound Grb2 (lanes 3 and 4), whereas GST alone and GST-LIM domains of Ajuba did not (lanes 2 and 5). These analyses demonstrated that it was the pre-LIM domain of Ajuba, not the LIM domains, that mediate the interaction with Grb2.
We next determined which domains of Grb2 or Grap interact with Ajuba. GST fusion proteins containing full-length Grb2, Grb2 with inactivating mutations in the C-terminal, N-terminal, or both SH3 domains (36
), full-length Grap, the N-terminal SH3 domain of Grap, or the C-terminal SH3 domain of Grap (14
) were added to cell extracts from 3T3 cells containing Myc-tagged pre-LIM Ajuba. Anti-Myc immunoblotting of bound products was performed. Results of this experiment are presented in Fig. C. For an interaction between Grb2 or Grap and pre-LIM Ajuba to occur, a functional SH3 domain, either N terminal (lanes 5 and 8) or C terminal (lanes 4 and 9) or both (lanes 3 and 7), was required. When both SH3 domains of Grb2 were nonfunctional, pre-LIM Ajuba did not bind (lane 6). This result demonstrated that either SH3 domain of Grb2 could mediate the interaction between Grb2 and Ajuba.
Ajuba expression enhances MAP kinase activity.
Grb2 is an adapter protein that couples signals from activated cell surface growth factor receptors or other activated cytosolic signaling intermediates to the activation of Ras and subsequently to MAP kinase activation (24
). To determine if the interaction between Ajuba and Grb2 is functionally significant, we tested whether expression of Ajuba could affect MAP kinase activity.
Fibroblast 3T3 cell clones expressing Ajuba, the pre-LIM domain of Ajuba, or the three LIM domains of Ajuba, and control 3T3.Neo cells (derived from transfection with an empty vector), were deprived of serum and then stimulated with 10% serum for 15 min. Cells were then lysed, and MAP kinase activity present in equal amounts of protein from each sample was determined by immunoprecipitation of ERK and by in vitro kinase reactions performed with the bound material and MBP as a substrate. In control cells, minimal MAP kinase was present following serum starvation (Fig. A, lane 1). After stimulation with 10% serum, MAP kinase activity was increased (lane 4), as expected. In cells containing Ajuba, serum starvation overnight did not completely suppress MAP kinase activity (lane 3). MAP kinase activity was 10-fold greater than in control cells, a level of MAP kinase activity observed in control cells stimulated with serum. Addition of 10% serum to these cells stimulated further MAP kinase activity (lane 6) to a level fivefold higher than that in control cells following serum stimulation. This pattern of MAP kinase activity mapped to the pre-LIM domain of Ajuba since cells containing pre-LIM Ajuba (Fig. A, lanes 2 and 5) gave a response similar to that of cells expressing full-length Ajuba, whereas cells expressing only the LIM domains of Ajuba (Fig. B, lanes 2 and 4) exhibited MAP kinase activity profiles similar to those of control mock-transfected 3T3 cells (Fig. B, lanes 1 and 3). This result correlates precisely to the domain of Ajuba required to mediate the interaction between Ajuba and Grb2. Thus, in cells expressing exogenous Ajuba or the pre-LIM domain, but not the LIM domains, of Ajuba, MAP kinase activity persisted despite serum starvation, and the response to serum was exaggerated.
FIG. 7 Ajuba expression in 3T3 cells results in an augmentation of MAP kinase activity. Cells were incubated in serum-free medium overnight and then either lysed (−) or stimulated for 15 min at 37°C with medium containing 10% serum and (more ...) Ajuba expression in Xenopus oocytes promotes meiotic maturation in a Grb2- and Ras-dependent manner.
To demonstrate that the enhanced MAP kinase activity observed in fibroblasts expressing Ajuba was functionally relevant in a physiological context, we tested whether expression of Ajuba in fully developed Xenopus
oocytes could affect their meiotic maturation. In response to progesterone and insulin, fully grown Xenopus
oocytes (which are arrested at the first meiotic prophase of the cell cycle) resume the meiotic process leading to the production of the unfertilized egg. MAP kinase activation is an essential component of this response pathway (22
). Fully grown stage VI oocytes were isolated from Xenopus
ovaries and microinjected with in vitro-transcribed Myc-Ajuba, Myc–pre-LIM Ajuba, or Myc-LIM Ajuba mRNA or, as a negative control, mRNA produced by in vitro transcription of the parental pCS2 vector. Following recovery, oocytes were treated with progesterone or insulin and at various time points scored for the presence of GVBD, as an indicator of meiotic progression. In the absence of inducer, some batches of oocytes injected with Ajuba underwent spontaneous maturation (data not shown). However, reproducibly, the presence of Ajuba in oocytes was found to increase the number that underwent GVBD in response to progesterone (Fig. A) or insulin (Fig. B). Fifty-five percent of oocytes injected with control mRNA underwent GVBD, whereas 80% of oocytes injected with Myc-Ajuba mRNA underwent GVBD (Fig. A). This augmented response mapped to the pre-LIM domain of Ajuba, not the LIM domains, since 65% of oocytes injected with Myc–pre-LIM Ajuba underwent GVBD whereas only 30% of oocytes injected with Myc-LIM Ajuba underwent GVBD (Fig. A).
FIG. 8 Ajuba expression in Xenopus oocytes promotes meiotic maturation in a Grb2- and Ras-dependent manner. (A) Mature Xenopus oocytes were injected with 50 ng of in vitro-transcribed mRNA as described in Materials and Methods. After recovery, healthy oocytes (more ...)
To determine if Grb2 contributed to this result, oocytes were coinjected with Ajuba and Grb2 mRNA or with Ajuba and an inactivated isoform of Grb2 in which both SH3 domains are nonfunctional (Grb2 SH3-N,C). Following coinjection of Ajuba and Grb2, ca. 90% of oocytes underwent GVBD in response to progesterone (a value significantly greater than for oocytes injected with Ajuba alone), whereas coinjection of Ajuba and Grb2 SH3-N,C resulted in only 60% GVBD (a value comparable to that for control oocytes) (Fig. A). Looked at another way, Grb2 SH3-N,C inhibited the ability of Ajuba to promote GVBD in response to progesterone, whereas wild-type Grb2 further enhanced GVBD in oocytes injected with Ajuba mRNA. Grb2 coinjection with pre-LIM Ajuba also appeared to enhance GVBD; however, this difference was not significantly different from that for oocytes injected with pre-LIM alone.
Thus, Ajuba significantly enhanced meiotic progression of oocytes in response to progesterone. This response mapped to the pre-LIM domain of Ajuba and was further augmented when Grb2, but not an inactivated isoform of Grb2, was coinjected with Ajuba, indicating that the ability of Ajuba to promote GVBD was Grb2 dependent.
Since a major pathway by which Grb2 couples signals to the activation of MAP kinase is dependent on the activation of Ras, we tested whether a dominant inhibitory form of Ras (RasN17) would block the enhanced GVBD observed in oocytes injected with Ajuba (Fig. B). Insulin-induced GVBD occurs in a Ras-dependent manner, whereas progesterone-induced GVBD is Ras independent (35
) (Fig. B). In the presence of Ajuba, 25% more oocytes than controls underwent GVBD in response to insulin or progesterone. When RasN17 mRNA was coinjected with Ajuba mRNA, this 25% gain in GVBD was completely blocked (Fig. B). The number of progesterone-treated oocytes undergoing GVBD was similar to the number of control oocytes. Likewise, there was no difference between control/RasN17 and Ajuba/RasN17-injected insulin-treated oocytes. This analysis indicates that the ability of Ajuba to promote GVBD was mediated in a Ras-dependent manner.
To examine the kinetics of progesterone-mediated meiotic maturation of oocytes injected with Ajuba compared to controls, and to correlate this to expression of Ajuba protein and MAP kinase activation, we performed a detailed time course of GVBD development with concurrent biochemical analysis for Ajuba expression and MAP kinase activity. Compared with control oocytes, oocytes injected with Myc-Ajuba mRNA initiated GVBD and achieved maximal levels of GVBD at earlier times (Fig. A). Ajuba-injected oocytes expressed Myc-Ajuba protein as early as 4 h following progesterone addition, with maximal levels at 8 h, as determined by Myc immunoblotting of extracts from individual oocytes (Fig. D). MAP kinase activation paralleled the expression of Ajuba, as determined by immunoblotting individual oocyte extracts with an antibody that recognizes activated MAP kinase (New England Biolabs) (Fig. B). In addition, the peak of MAP kinase activity in Ajuba-injected oocytes preceded that observed in control oocytes (Fig. C). Expression of Ajuba protein did not affect the level of ERK protein expression (Fig. E). Expression of Ajuba protein and activation of MAP kinase activity directly correlated with the earlier onset of GVBD observed in oocytes injected with Ajuba mRNA (Fig. A).
FIG. 9 Ajuba expression in oocytes results in earlier onset of meiotic maturation (GVBD) coinciding with earlier activation of MAP kinase activity. Batches of oocytes were injected with water (control) or Ajuba mRNA. After recovery, healthy oocytes were cultured (more ...)