Carboxyl-Terminal Modulator Protein (CTMP), a Negative Regulator of PKB/Akt and v-Akt at the Plasma Membrane

See allHide authors and affiliations

Science  12 Oct 2001:
Vol. 294, Issue 5541, pp. 374-380
DOI: 10.1126/science.1062030


The PKB (protein kinase B, also called Akt) family of protein kinases plays a key role in insulin signaling, cellular survival, and transformation. PKB is activated by phosphorylation on residues threonine 308, by the protein kinase PDK1, and Serine 473, by a putative serine 473 kinase. Several protein binding partners for PKB have been identified. Here, we describe a protein partner for PKBα termed CTMP, or carboxyl-terminal modulator protein, that binds specifically to the carboxyl-terminal regulatory domain of PKBα at the plasma membrane. Binding of CTMP reduces the activity of PKBα by inhibiting phosphorylation on serine 473 and threonine 308. Moreover, CTMP expression reverts the phenotype of v-Akt–transformed cells examined under a number of criteria including cell morphology, growth rate, and in vivo tumorigenesis. These findings identify CTMP as a negative regulatory component of the pathway controlling PKB activity.

PKB is a major downstream target of receptor tyrosine kinases that signal via the phosphatidylinositol 3-kinase (PI 3-kinase). PKB mediates a wide variety of biological responses to insulin and insulin-like growth factor 1 (IGF-1) and other growth factors (1–2). Upon cell stimulation, the kinase is translocated to the plasma membrane, where it is phosphorylated on two amino acids, Thr308 in the catalytic domain and Ser473 in the COOH-terminal regulatory domain (3–8). To date, three proteins have been shown to physically interact with PKB (9–11).

To identify new proteins that interact with PKB, a HeLa cell cDNA library was screened by yeast two-hybrid analysis, with the kinase domain plus the COOH-terminal regulatory domain of PKBα as bait. From 1.5 × 106 primary transformants screened, seven identical clones were identified. This cDNA encoded a protein that specifically interacted with PKBα, as demonstrated by activation of the reporters for histidine auxotrophy and lacZ activity [Web fig. 1, A and B (12)], and in a mammalian cell two-hybrid assay (13). This interaction was observed only with constructs containing the COOH-terminal regulatory domain of PKBα, encompassing amino acids 411 to 480 [Web fig. 1, C and D (12)]. Sequence analysis revealed the presence of an open reading frame encoding a protein of 240 amino acids with a predicted molecular mass of 27 kD that we termed COOH-terminal modulator protein, or CTMP (Fig. 1A). The cDNA contains the unusual feature of an Alu cassette at its 3′ end, a sequence usually found in intronic DNA. Database screening of mouse expressed sequence tags revealed a protein of 230 amino acids with a similar sequence (79% identity) that may be the mouse homolog of CTMP (Fig. 1A). The mRNA for human CTMP was detected predominantly in skeletal muscle, testis, uterus, brain, and kidney, with lower levels observed in heart, liver, and lung [Web fig. 2A (12)]. The presence of multiple bands after reverse transcription–polymerase chain reaction (RT-PCR) strongly suggested that the gene for CTMP undergoes alternate splicing in some tissues, generating multiple RNA transcripts [Web fig. 3, A and B (12)]. Endogenous CTMP protein was detected with an antibody (14) specific for human CTMP in extracts from HeLa and human embryonic kidney (HEK) 293 cells, with weaker expression found in COS-1 cells [Web fig. 2B (12)]. Weaker CTMP expression was seen in the human SJRH30 rhabdomyosarcoma cell line, and no signal was detected in the rat H9C2 myocardium cell line. These data were confirmed by blotting with a second antibody against CTMP (13). CTMP migrated with apparent molecular masses of 22 to 26 kD in the different cell lines, possibly owing to posttranslational modification (see below).

Figure 1

Structure and localization of human CTMP. (A) Alignment of deduced amino acid sequences from human (accession number AJ313515) and mouse CTMP. (B) NIH 3T3 cells were transfected with 5 μg of expression vectors for GFP (pGFP, left panel) or GFP-CTMP (pGFP-CTMP, right panel) (29). GFP fluorescence was then analyzed in living cells using confocal microscopy (30). The arrows indicate areas of high fluorescence intensity at membrane ruffles. (C) CTMP expression in glioblastoma cell lines was assessed by immunoprecipitation followed by immunoblotting with the same antibody against CTMP (99390). The position of CTMP is indicated by an arrow. Clone H is a cell-line stably expressing Flag-CTMP (see Fig. 5). The asterisk represents the IgG light chain. (D) Detection of endogenous CTMP protein by Western blotting of equivalent amounts (50 μg) of cytosolic (S100), membrane (P100) fractions (31) or total cell lysate (Total) from LN229 cells. (E) To determine the cellular distribution of endogenous PKBα and CTMP proteins, LN229 cells were fixed with 4% paraformaldehyde and incubated with PKB mAb (32), followed by staining with protein A/G coupled to FITC (green, PKB panel), or with the polyclonal CTMP antibody 99390 followed by staining with rhodamine-conjugated rabbit antibody (red, CTMP panel). Slides were analyzed by confocal microscopy, and the pictures represent the central section of the x-y plane. Areas of colocalization of CTMP and PKB at the membrane ruffles are visualized in yellow (merge panel) and are indicated by arrows. (F) CTMP from pHook2-CTMP transfected COS-1 cells (29), clone H cell extracts, or P100 fraction from LN229 cells was detected with CTMP antibodies 99390 and 89570 (14).

When HeLa cell extracts were subjected to centrifugation into cytosolic (S100) and membrane (P100) fractions, endogenous CTMP was predominantly detected in the P100 fraction [Web fig. 2C (12)]. CTMP expression was detected in a range of different human cell lines (15), most notably in a glioblastoma cell line (LN229, Fig. 1C). Again, two molecular species corresponding to CTMP were observed in LN229 cells, and these forms appeared to be differentially localized to the membrane and cytosolic fractions of these cells (Fig. 1D). Immunofluorescence analysis demonstrated that endogenous PKB and CTMP colocalized at the plasma membrane (Fig. 1E). Endogenous complexes of CTMP and PKB were also detected in the P100 fraction of these cells by Western blotting (16). Immunolocalization of green fluorescent protein (GFP)–CTMP fusion protein indicated that CTMP associated with intracellular structures similar to membrane ruffles (Fig. 1B), whereas staining of the GFP control protein was detected in all cellular compartments. Time-lapse cine-microscopy of moving NIH 3T3 cells expressing GFP-CTMP revealed that the fusion protein was predominantly localized to the leading edge of the moving cells, decorating the rapidly moving membrane ruffles [Web fig. 4 (12)].

Western blotting of lysates from a control cell line stably expressing CTMP indicated that two forms of CTMP exist in these cells (Fig. 1, C and F; also see Fig. 5 for details). Transfection of CTMP without an epitope tag also produced two species of CTMP, suggesting that this protein undergoes posttranslational modifications such as phosphorylation in cells (Fig. 1F; also see Fig. 2E). It is interesting that the lower of the two CTMP forms in transfected cells comigrated with endogenous CTMP detected in the P100 fraction of LN229 cells (Fig. 1D). This further suggests that CTMP localization in the cell may be regulated by posttranslational modifications such as phosphorylation.

Figure 2

Interaction of CTMP and PKBα in quiescent cells, and phosphorylation following stimulation. (A) Schematic diagram of recombinant proteins used in these experiments (33). (B) COS-1 cells were transfected with 7.5 μg of the indicated PKB expression vectors, lysed, and fractionated (31). Soluble (S100) and particulate (P100) fractions were analyzed by Western blotting using the HA mAb 12CA5 (14). (C) COS-1 cells were transfected with 7.5 μg of the indicated PKB (33) and CTMP (29) expression vectors. After serum starvation (24 hours) and stimulation with 100 μM pervanadate (15 min at 37°C, +Per.), cells were lysed and fractionated. Immunoprecipitations were performed from S100 and P100 fractions using an antibody against Flag (14). PKBα expression was analyzed using an HA antibody (upper panel), CTMP expression was analyzed using Flag antiserum (middle panel), and CTMP-bound PKBα was detected using an HA antibody (bottom panel). The asterisk represents the IgG heavy chain. (D) COS-1 cells were transfected with 7.5 μg of pF-CTMP and with pGST or pGST-PKB-RD. Cells were lysed and fractionated as in Fig. 2B, and immunoprecipitations were performed from the S100 fractions with a Flag antibody. Association of GST-PKB-RD with CTMP was analyzed using a GST antibody (14). (E) Cells stably expressing Flag-CTMP were labeled in vivo with [32P] orthophosphate (34). After immunoprecipitation, phosphorylated Flag-CTMP was revealed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and Western-blotting, followed by autoradiography of the membrane (bottom panel). Expression level of Flag-CTMP was revealed by immunoblotting of the membrane with antibody against Flag (upper panel).

To further explore the biological relevance of the PKBα-CTMP complex, this interaction was analyzed in mammalian cells by immunoprecipitation. In cell extracts from transfected COS-1 cells lysed in buffer containing 1% NP-40 (v/v), CTMP formed a complex with a COOH-terminal regulatory domain mutant of PKBα (GST-PKB-RD,Fig. 2A), but not with full-length PKBα. One interpretation of this result is that binding of CTMP and PKBα may require intact plasma membrane structures, because both proteins have the ability to localize at the plasma membrane [(17) and Fig. 1B]. We therefore lysed transfected COS-1 cells in Hepes/sucrose buffer, facilitating the preparation of cytosolic (S100) and membrane (P100) fractions for immunoprecipitation. The efficiency of the fractionation was confirmed by transfection of a membrane-targeted PKB (m/p-PKB), a construct containing sites for myristoylation and palmitoylation of PKB that result in constitutive membrane anchoring [(17) and Fig. 2A]. The m/p-PKB protein was exclusively localized in the P100 fraction of lysed COS-1 cells (Fig. 2B). In contrast, wild-type PKBα was equally distributed between the membrane and cytosolic fractions. Association of PKBα and CTMP was examined by cotransfection of COS-1 cells with constructs encoding epitope-tagged versions of both proteins [hemagglutinin-tagged (HA)-PKBα and Flag-CTMP, Fig. 2C]. Both PKBα (upper panel) and CTMP (middle panel) were detected in cytosolic (S100) and membrane (P100) fractions (Fig. 2C). In addition, PKBα was observed in CTMP immunoprecipitates from membrane, but not cytosolic fractions (Fig. 2C, bottom panel). Treatment of COS-1 cells with pervanadate, an inhibitor of protein tyrosine phosphatases that potently activates PKB (18), disrupted the PKBα-CTMP interaction in the membrane fraction (Fig. 2C, bottom panel). These data confirm that PKBα interacts with CTMP in quiescent or unstimulated cells and that activation of PKBα with pervanadate perturbs this interaction. In contrast to full-length PKBα, a truncation mutant containing the COOH-terminal regulatory domain of PKBα (GST-PKB-RD, Fig. 2A) was detected in CTMP immunoprecipitates from cytosolic fractions of COS-1 cells (Fig. 2D). These observations suggest that a region of PKBα not present in the GST-PKB-RD construct inhibits the binding of CTMP to cytosolic PKBα. Alternately, other proteins may be involved in the formation of PKB-CTMP complexes. These data further support the hypothesis that binding of full-length PKBα and CTMP occurs only at the plasma membrane.

To explore a potential mechanism for the inhibitory effect of pervanadate on PKBα-CTMP complexes, we monitored phosphorylation of CTMP during PKB activation. In vivo labeling of cells stably expressing Flag-CTMP with [32P]orthophosphate demonstrated a fourfold increase in CTMP phosphorylation after pervanadate treatment (Fig. 2E), supporting the idea that CTMP is regulated at the posttranslational level by as-yet-unidentified protein kinases.

Activation of PKBα occurs via phosphorylation of Thr308 in the activation loop of the kinase domain and of Ser473 in the COOH-terminal regulatory domain (5). To test the influence of CTMP binding on activation of PKBα, we assayed kinase activity in immune complexes from transfected COS-1 cells treated with pervanadate. Pervanadate-stimulated PKBα activity was decreased in a manner dependent on the amount of transfected CTMP (Fig. 3A). Therefore, CTMP binding appears to have an inhibitory effect on PKB. Increased CTMP expression led to decreased phosphorylation on both Ser473 and Thr308 residues of PKBα, most notably on Ser473 (Fig. 3B). Especially at late time points, kinase activity of PKBα was stimulated by pervanadate, despite the presence of CTMP (Fig. 3C). This increase in activity over time was reflected in increases in phosphorylation of PKBα residues Ser473 and Thr308 in the presence or absence of CTMP (Fig. 3D). These results show that binding of CTMP to PKBα inhibits, but does not completely abolish, pervanadate-stimulated phosphorylation of the key amino acids necessary for kinase activity of PKBα.

Figure 3

CTMP inhibits PKBα activity by preventing its phosphorylation by upstream kinases. (A) COS-1 cells were cotransfected with an HA-PKB expression vector (2.5 μg) and with the indicated amount of pF-CTMP. Cells were serum-starved (24 hours), and stimulated with vehicle control (white bars) or 100 μM pervanadate (black bars) for 10 min at 37°C. Cells were then lysed and analyzed for PKB kinase activity as described (35). (B) The phosphorylation status of PKBα was investigated by using antibodies against pSer473(panel II) and Thr308 (panel III) (14). Expressed PKBα (panel I) and CTMP (panel IV) proteins were detected with the indicated antibodies (14). (C) COS-1 cells were transfected with an HA-PKBα expression vector (2.5 μg) together with 5 μg control vector (control, white bars) or 5 μg pF-CTMP (CTMP, black bars). Serum-starved cells (24 hours) were then stimulated with 100 μM pervanadate at 37°C for the indicated times, lysed, and processed for immune-kinase assay (35). (D) Immunoprecipitations from COS-1 cells transfected with the indicated constructs were analyzed for phosphorylation on PKBα residues Ser473 or Thr308 as described in Fig. 3B.

To determine the effect of CTMP on PKBα activity in response to physiological stimuli, we treated HEK293 cells transfected with constructs expressing PKBα and various amounts of CTMP with insulin or IGF-1. The kinase activity of PKBα was stimulated by both insulin and IGF-1, and this stimulation was progressively inhibited by increasing amounts of CTMP expression (Fig. 4A). Furthermore, analysis of the phosphorylation status of PKB revealed that CTMP expression led to a decrease in Ser473 and Thr308phosphorylation induced by insulin and IGF-1 (Fig. 4B). These data reinforce the results seen with pervanadate (Fig. 3) and support the hypothesis that CTMP is an inhibitor of PKB in vivo.

Figure 4

Effect of CTMP on PKBα signaling. (A) HEK293 cells transfected with expression vectors for HA-PKBα (1 μg) and pF-CTMP (amounts indicated) were serum-starved (24 hours) and stimulated with either insulin (100 nM, 15 min) or IGF-1 (50 nM, 10 min) at 37°C. Cells were then processed for immune-kinase assay (35). (B) Phosphorylation status of PKBα residues Ser473 and Thr308 was investigated as described in Fig. 3B. (C) HEK293 cells were transfected with 200 ng of the p-foslucreporter gene together with pHA-PKB (200 ng) and pF-CTMP (1 μg) where indicated. Luciferase assays were performed 48 hours after transfection as described by the manufacturer (Promega luciferase assay kit). (D) HEK293 cells were transfected with increasing amounts of pAS-CTMP (1, 2, 4, and 10 μg), an expression vector encoding an antisense CTMP cDNA (29). Expression and Ser473 phosphorylation of endogenous PKB were assayed by direct immunoblotting (14). The amounts of antisense cDNA were measured by RT-PCR (36).

To explore the effect of CTMP on downstream effectors of PKB, we analyzed the consequence of CTMP expression on PKBα-mediated transcriptional regulation. PKBα reduces both basal and serum-stimulated transcriptional activity of the c-fospromoter (16). Coexpression of CTMP completely abolished the inhibitory effect of PKBα on c-fos–mediated transcription (Fig. 4C), showing that CTMP inhibitory activity is correlated with a reversion of PKB downstream target function. Indeed, CTMP expression increased c-fos promoter activity above control levels, possibly through inhibition of endogenous PKB (Fig. 4C). In addition, expression of CTMP reduced phosphorylation of glycogen synthase kinase–3β (GSK-3β) on Ser9, a PKB-mediated phosphorylation event (16).

To determine the consequence of disrupting CTMP function in vivo, HEK293 cells were transfected with an antisense CTMP expression vector (Fig. 4D). Increasing amounts of antisense CTMP cDNA, confirmed by RT-PCR, increased Ser473phosphorylation of endogenous PKB without changing its expression (Fig. 4D). These data demonstrate that inhibition of endogenous CTMP function increases the activation status of endogenous PKBα and indicate that CTMP acts as a negative regulator of PKBα in vivo. Supporting these data, similar increases in endogenous GSK-3β phosphorylation on Ser9 were also observed in the presence of the antisense CTMP construct, but not a control antisense construct (16). This effect is not due to an up-regulation of PI-3 kinase activity, because no increase in the activity of this enzyme was observed when antisense CTMP or a control luciferase antisense cDNA was expressed (16). Therefore, ablation of CTMP function increases the ability of PKB to activate its downstream effectors in cells.

CCL64 mink lung cells stably expressing v-Akt(AKT8 cells), the viral homolog of PKB isolated from mouse, are transformed and tumorigenic in vitro (19–21). To expand the hypothesis that CTMP negatively regulates PKBα in vivo and to analyze whether CTMP expression could revert the phenotype of v-Akt–transformed cells, AKT8 cells were transfected with the cDNA encoding CTMP (CTMP) or with a cDNA corresponding to the original clone isolated from the two-hybrid analysis (CTMPL). This clone contains an extra 15 amino acids at the NH2-terminus and was identical to wild-type CTMP in terms of PKBα binding and inhibition (13). Clones stably expressing CTMP were selected and analyzed on the basis of three criteria: cellular morphology, growth rate, and in vivo tumorigenesis. Expression of CTMP varied somewhat between each cell line (Fig. 5A), whereas amounts of v-Akt remained largely unchanged in all cell lines tested. Significantly, the amounts of endogenous PKBα were increased in v-Akt–transformed cells, a phenomenon not due to cleavage of v-Akt after the Gag NH2-terminal domain (13). The presence of PKBα-CTMP complexes in vivo was examined by immunoprecipitation using antibodies to either PKBα or CTMP (Fig. 5B). PKBα-CTMP complexes were detected using antibodies to PKBα or Flag, in membrane fractions of clone H (Fig. 5B). Furthermore, v-Akt–CTMP complexes were also detected in this fraction, demonstrating that CTMP interacts with both forms of PKB expressed in these cells, suggesting a conservation of interaction between PKB and CTMP proteins from different species.

Figure 5

Phenotypic reversion of v-Akt–transformed cells by stable expression of CTMP. (A) AKT8 cells were stably transfected with expression vectors for CTMP (pF-CTMP and pF- CTMPL), or with control vector (pSG5-FlagNt-puro). Puromycin-resistant clones were selected and expanded in culture for further characterization (37). Selected clones from the empty vector plate (clone A), the pF-CTMP plate (clones E, F, G, and H), or the pF-CTMPL plate (clones B, C, and D) were analyzed for Akt/PKB (upper panel) and CTMP (bottom panel) expression. (B) P100 membrane fraction (500 μg) from clone H prepared in HES buffer (+ lanes) or HES buffer alone were incubated with 10 μg of the indicated antibodies (14). Immune complexes were then analyzed by immunoblotting with the indicated antibodies. PKB and CTMP protein migration was confirmed by direct blotting of the P100 fraction (direct). The asterisk shows the IgG heavy chain position. (C) CCL64, AKT8, and cells from clones A to H were seeded in a 35-mm dish and photographed with a phase-contrast microscope before reaching confluency. (D) Cells from parental cell lines CCL64 and AKT8 [5 × 104 (10% serum, left panel) or 2 × 105(0.1% serum, right panel)], as well as clones A, B, E, and F were seeded in 35-mm dishes. Cells were trypsinized, and viable cells were counted following staining by trypan blue exclusion at the indicated time for the relevant cell lines.

Expression of CTMP clearly altered the morphology of AKT8 cells (Fig. 5C); cells were larger in appearance and formed mosaics, similar to wild-type parental CCL64 cells. The change in morphology induced by CTMP expression in these cells may occur via inhibition of v-Akt. In this regard, decreased phosphorylation of v-Akt on Ser473 was observed in cell lines stably expressing CTMP (13). Clones stably transfected with either form of CTMP grew more slowly than did mock-transfected or v-Akt–transformed cells in high serum. Moreover, clones B, E, and F, grew more slowly than untransformed CCL64 cells, suggesting that CTMP inhibited proliferation induced by either v-Akt or by endogenous PKB (Fig. 5D, left panel). In low serum, clones B and F show no significant cell proliferation, which is identical to the profile observed for the parental CCL64 cells (Fig. 5D, right panel). In contrast, AKT8 cells not transfected with the cDNA encodingCTMP (clone A) were still able to proliferate, showing that CTMP expression can restore cell-cycle arrest under low serum conditions.

AKT8 cells form colonies in soft agar (19), suggesting that these cells have tumorigenic properties. We injected nude mice with the different cell lines described in Fig. 5. Mice injected with puromycin-resistant AKT8 cells formed tumors 11 days after injection (clone A, Fig. 6). Tumor growth in these mice was identical to that in mice injected with AKT8 cells not transfected with CTMP, demonstrating that the presence of the puromycin-resistance gene did not influence tumor growth. Wild-type CCL64 cells did not form tumors when injected into these mice. Analysis of nude mice injected with AKT8 cells stably expressing CTMP revealed that tumor growth was either abolished or delayed to that in animals injected with control AKT8 cells (Fig. 6).

Figure 6

CTMP inhibits tumor growth in nude mice. AKT8 cells transfected with the puromycin-resistant plasmid (clone A) or stably expressing CTMP (clones B to H) were trypsinized, washed, and resuspended in PBS at a concentration of 107cells/ml. Cells (100 ml, 106 cells) were injected subcutaneously into the backs of the nude mice. Three mice were injected for each corresponding cell line. Mice were examined for tumor growth every 3 days. The length and width of each tumor are described at the indicated time after injection.

Our results identify CTMP as a new component in the control of PKBα signaling and suggest that this negative regulation, which occurs via a direct interaction of CTMP with PKB at the plasma membrane, may be an important cellular mechanism in preventing inappropriate kinase activation, as well as subsequent excess cell growth and proliferation. The role of PKB in cell survival is well established (2). A key role for PKB in the progression of cancer was illustrated by the discovery of the protein-lipid phosphatase PTEN protein, the most highly mutated tumor-suppressor gene identified since p53 (22, 23). Cells lacking PTEN show increased PKB activity, suggesting that negative regulation of the PI 3-kinase and PKB signaling pathway by PTEN acts to prevent unregulated cell proliferation. Our data demonstrate that CTMP may represent an additional mechanism to negatively regulate PKB in cells. Whereas PTEN inhibits PKB activity indirectly by reducing the amounts of phosphatidylinositol-(3,4,5)-trisphosphate at the cell membrane, CTMP mediates its inhibition by binding directly to PKB and preventing its phosphorylation. In glioblastoma cell lines with compromised PTEN function (U343MG, U87MG), small amounts of CTMP were detected, whereas in glioblastoma cell lines with functional PTEN alleles (LN229), CTMP is readily detected (Fig. 1C). These data suggest that the elevated amounts of PKB activation seen in glioblastoma and other cell lines may be due not just to ablation of PTEN function, but also due to a decrease in the levels of CTMP protein expression in these cells.

  • * Present address: Center for Research in Occupational and Environmental Toxicology, Oregon Health Science University, Portland, OR 97201–3098, USA.

  • Present address: Department of Biochemistry, Royal Perth Hospital, GPO Box X2213, Western Australia 6001, Australia.

  • To whom correspondence should be addressed. E-mail: hemmings{at}


Stay Connected to Science

Navigate This Article