Mirk Protein Kinase Is a Mitogen-activated Protein Kinase Substrate That Mediates Survival of Colon Cancer Cells

INTRODUCTION

Eukaryotic cells are constantly subjected to growth, differentiation, and stress signals and use cascades of protein kinases called MAPKs, 3 and their activators to respond to these varied signals. The protein kinase cascades translate signals from cell surface receptors through the cytoplasm to the nucleus, to cytoskeletal elements, and to other parts of the cell. There are three major MAPK cascades in mammalian cells, consisting of three sequentially activated protein kinases, named for the final kinase in the cascade: the mitogen-activated erk1/2 cascade, the stress-activated protein kinase/c-Jun NH₂-terminal kinase cascade, and the high osmolarity-activated p38 kinase cascade (1, 2, 3) . In earlier studies, we had identified a protein kinase that was selectively activated in colon tumor cells compared with normal colon epithelial cells (4 , 5) . We have purified and cloned this kinase and now report that it is a MAPK substrate, is broken down to a smaller form in cells with active MAPKs, and can mediate cell growth and survival only under conditions in which the erk1/2 class of MAPKs is not highly activated.

MATERIALS AND METHODS

Results and Discussion

We have cloned a novel gene using sequence data from a protein purified from colon carcinoma cells (see “Materials and Methods”). The deduced protein structure is shown in Fig. 1A ⇓ . BLAST search confirmed that this gene was a homologue of a Drosophila gene called minibrain (Mnb; Ref. 8 ) or dyrk (9 , 10) ; therefore, it was called mirk. A initial report of this work has been presented (11) . Mutant Mnb flies are characterized by a marked reduction in size of the optic lobes and central brain hemispheres (8) . Dyrk1/minibrain is a strong candidate for the gene duplicated in Down Syndrome (12) , but no function has been established for this gene aside from its kinase activity. More recently, several other members of the Mnb/dyrk family have been cloned, but no function for any of these genes has been elucidated (13) . One Mnb/dyrk gene, dyrk1B, has three splice variants, encoding putative proteins of 629, 601, and 589 amino acids (14) , the longest being identical to mirk. A mirk splice variant of 601 amino acids had been isolated in our initial screen but was not studied further because it did not encode the peptide found in mirk protein purified from colon carcinoma cells (see “Methods and Materials”). This splice variant was generated by alternative splicing within exon 9 of mirk (Fig. 1B) ⇓ , which encodes amino acids found in the conserved kinase domain between subdomains X and XI.

mirk encodes a 629-amino acid protein with a bipartite nuclear localization signal in the N′ terminal nonconserved region, the 11 canonical kinase subdomains, a PEST sequence common to rapidly degraded proteins following the kinase domain, and a consensus MAPK phosphorylation sequence in the nonconserved C′ terminus (Fig. 1A) ⇓ . The amino acid sequence of mirk is 56% identical within the conserved kinase domain to the human forms of dyrk1 (minibrain), dyrk2, and dyrk3, with much less identity in the N′ terminus and C′ terminus. All dyrk family members are serine/threonine kinases that also can autophosphorylate on tyrosine; therefore, they are dual function kinases (8 , 9 , 10 , 13) . Their activation sequence is YQY or YTY, with both tyrosines phosphorylated. This YxY sequence aligns with the TxY activation sequence in MAPKs. The cDNA sequence of mirk was compared by BLAST search to sequences deposited in GenBank from the Human Genome Project. The genome sequence of the mirk gene was found in chromosome 19q13.1, and intron/exon junctures were determined by comparison of the two sequences. mirk has 11 exons and 10 introns encompassing 8.8 kb of genomic DNA (Fig. 1B) ⇓ . mirk was not mutated in the SW480 colon carcinoma cells from which it was cloned, because its sequence did not differ from the genomic DNA sequence.

Recombinant mirk expressed in sf21 insect cells was purified and then either assayed for protein kinase activity on immobilized MBP in an in gel kinase assay or Western blotted with anti-phosphotyrosine antibody. Recombinant mirk had MBP kinase activity at M_r 70,000, the molecular weight of mirk deduced from its amino acid sequence, and was phosphorylated on tyrosine (Fig. 1C) ⇓ . Phosphotyrosine was also detected in mirk expressed in E. coli, indicating that mirk has tyrosine autophosphorylation capacity (data not shown). Mirk activity is dependent on phosphorylation of tyrosines 271 and 273 (Fig. 1C) ⇓ located between subdomains VII and VIII in the sequence YQY, which aligns with the erk1 activation sequence TEY (15) . A mirk mutant at the ATP-binding site (K140R) and a double mutant at the putative activation sequence YQY, to FQF, were generated. Both mutations greatly decreased the kinase activity of mirk on MBP, as well as the capacity of mirk for autophosphorylation (Fig. 1C) ⇓ and the autophosphorylation of mirk on tyrosine (data not shown). Western blotting confirmed that equal levels of the wild-type and mutant proteins were present in each reaction (Fig. 1C) ⇓ . Analysis of the phosphorylated amino acids in MBP by TLC and of labeled MBP for phosphatase sensitivity (Fig. 1C) ⇓ demonstrated that mirk phosphorylates only serine and threonine residues in vitro. Mirk-phosphorylated MBP was a substrate for the serine-threonine phosphatase PP1 but not for the phosphotyrosine-specific phosphatase, PTP-β (Fig. 1C) ⇓ . In addition, mirk had no detectable in vitro activity on two synthetic phosphotyrosine substrates while phosphorylating certain substrates of the Ser/Thr kinase MAPKAP kinase-2 (Ref. 16 ; data not shown). Therefore, although mirk kinase is activated by autophosphorylation on tyrosine at the Y271/Y273 site, mirk itself is a serine-threonine kinase on exogenous substrates.

mirk exhibits restricted mRNA expression in normal tissue (Fig. 2, A and B) ⇓ , with highest expression in skeletal muscle, testes, heart, and brain, and with little expression in normal colon (Fig. 2A ⇓ , Lane c) and normal lung (Fig. 2B ⇓ , Lane lu). mirk is highly expressed in SW480 colon carcinoma cells and A549 lung carcinoma cells (Fig. 2C) ⇓ and other tumor cell lines including HeLa ovarian carcinoma cells, K562 chronic myelogenous leukemia cells, molt4 lymphoblastic leukemia cells, and G361 melanoma cells. mirk is not detectably expressed in HL60 promyelocytic leukemia cells or Raji lymphoma cells (Fig. 2C) ⇓ . Thus, mirk is expressed in few normal tissues but in many types of cancer.

Two polyclonal antibodies to nonconserved regions of mirk were raised and affinity purified, one to amino acids 1–19 of mirk and the second, to amino acids 595–609 in the C′ terminus of mirk. Both sequences were unique by BLAST search. A monoclonal antibody to recombinant mirk was also raised. In Western blots with C′-terminal antibody (Fig. 3C ⇓ , upper panel), with N′-terminal antibody (Fig. 3C ⇓ , lower panel, upper arrow), or with monoclonal antibody 11 (Fig. 3B) ⇓ , mirk was detected at the expected translational size of M_r 70,000. A M_r 67,000 mirk form was detected in several lines by both N′-terminal and C′-terminal antibodies. Therefore, the M_r 67,000 mirk must be caused by an internal deletion. A splice variant caused by internal deletion within exon 9 was cloned in our initial screen (see“ Materials and Methods”). This splice variant encoded a M_r 67,000 mirk protein after in vitro coupled transcription/translation (data not shown) and therefore is very likely to be the source of the smaller mirk form. In addition to the splice variants, a mirk breakdown form of M_r 57,000 was detected by N′-terminal antibody and monoclonal antibody but was not detected by C′-terminal-directed antibody, indicating that it was generated by C′-terminal deletion.

Monoclonal serum-free media for 24 h (Lanes 5–8) increases the abundance of both endogenous and transfected mirk protein. E, the stable mirk transfectant T3 expresses about 4-fold the amount of mirk as vector control V1 cells after cells are grown in serum-free medium, and an increase in both endogenous mirk in V1 vector control transfectants and transfected mirk in T3 cells caused by serum-free growth is blocked by concurrent treatment with IGF-I. F, analysis of mirk mRNA levels by RT-PCR and by transient transfections of mirk promoter constructs (not shown) confirmed that much of the modulation of mirk protein levels seen in A–E was caused by posttranscriptional mechanisms. mirk mRNA levels were analyzed by RT-PCR in 9V2 vector control and T7 mirk transfectant cells cultured in control (CT) serum-containing medium or in serum-free media (SF) for 24 h, alone or with 10 μm PD98059 (MEK inhibitor) or 10 ng/ml IGF-I. Primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were included in every reaction mixture as an internal control. Added PD98059 had no effect and IGF-I had a minor effect on endogenous mirk mRNA levels in 9V2 vector control transfectants, whereas IGF-I had no discernible effect on the combination of exogenous and endogenous mirk mRNA levels in the T7 transfectant. G, upper panel, mirk expression in V1 and V2 vector control HD3 transfectants and mirk stable overexpressors, T19 and T24, was analyzed by Western blotting using affinity-purified antibody directed to the N′ terminus of mirk. Lower panel, mirk expression in V1 and V3 vector control U9 transfectants and mirk stable overexpressors, T7 and T22, was analyzed by Western blotting using affinity-purified antibody directed to the N′ terminus of mirk. This antibody also detects M_r 57,000 mirk cleaved at the C′ terminus (Fig. 3C ⇓ ; also data not shown). Cells were cultured in serum-containing growth media. Fractionation of cells into nuclear (N) and cytoplasmic (C) fractions demonstrated that a M_r 57,000 mirk breakdown product was found solely in the nucleus of each cell type, whereas higher levels of full-length, M_r 70,000 mirk are found in each mirk transfectant. The M_r 57,000 mirk breakdown product could not be detected with C′-terminal-directed antibody (not shown), indicating that the C′ terminus of mirk was deleted in the nuclear species. \. antibody raised to recombinant mirk was used to detect mirk in the stable mirk transfectant T24 and a vector control transfectant V1. Mirk was more abundant in the T24 overexpressing line than the control cells, and a M_r 57,000 degradation product of mirk was also detected in the overexpressing line (Fig. 3B) ⇓ . Monoclonal antibody activity was removed by preabsorption with recombinant mirk protein, and preimmune mouse serum did not detect mirk in parallel blots (Fig. 3B) ⇓ . The blots were stripped and reblotted for β-actin, demonstrating equal loading and transfer (Fig. 3B ⇓ , lower panels). When eight cases of colon cancer tissue with paired normal colon were examined by Western blotting with this monoclonal antibody, mirk was found to be much more abundant in the cancer tissue in three of eight cases (nos. 1, 3, and 6) when normalized to actin levels (Fig. 3A ⇓ , upper panel). Lack of degradation of protein within the clinical samples was demonstrated by reblotting for β-actin protein (Fig. 3A ⇓ , lower panel). Mirk was localized predominately within the cytoplasm of the colon carcinoma cells within sectioned colon cancer tissue from three additional cases, not within the stroma (data not shown), as shown by immunocytochemistry with monoclonal antibody to mirk. Therefore, the greater abundance of mirk protein by Western blotting in a subset of colon cancer tissues was attributable to increased mirk expression within the colon cancer cells themselves, not another cell type.

Recombinant mirk is phosphorylated by all three classes of MAPKs, the erks, the SAP/c-Jun NH₂-terminal kinases and p38 (Fig. 4A) ⇓ . Western blotting then showed that mirk protein levels are down-regulated by activated erks in vivo. Inhibition of erk activity by removal of serum mitogens and by addition of a MEK inhibitor increased endogenous mirk protein levels 18–20-fold. When U9 colon carcinoma cells were placed in serum-free medium for 1, 2, and 3 days, levels of endogenous mirk were increased 5-, 6-, and 7-fold, respectively (Fig. 4B) ⇓ . These data suggested that serum mitogens activated some signaling path that either down-regulated mirk mRNA, mirk protein levels, or both. U9 cells exhibit autocrine regulation by epidermal growth factor, fibroblast growth factor, and transforming growth factor β family members (17, 18, 19) ; therefore, U9 cells exhibit some activation of erks in the absence of serum mitogens (Fig. 4C) ⇓ . Treatment of U9 cells for 24 h in serum-free medium with the MEK inhibitor PD98059 induced a 4-fold, dose-dependent increase in mirk levels, above that caused by serum-free culture alone (Fig. 4B ⇓ , Lanes 1–4), probably because the MEK inhibitor blocked the activation of erks by autocrine growth factors. This total 20-fold increase in mirk above background levels was reversed by concurrent treatment of cells with the serum mitogen IGF-I (Fig. 4B ⇓ , Lanes 5–8), which activated erk1 and erk2 for at least 3 h (data not shown). Therefore, blocking erk activation by removal of serum mitogens and blocking any residual erk activation from autocrine growth factors by the MEK inhibitor elevated mirk levels 20-fold, whereas activation of erks with IGF-I reduced mirk levels to those seen in cells cultured in serum-containing medium. Similar data were seen with SW480 and HD3 colon carcinoma cell lines (data not shown).

A time course then demonstrated that mirk levels increase as the fraction of activated erk1 and erk2 decreases (Fig. 4C) ⇓ . The mirk double band is better resolved than in A or B because the gel fractionation was extended for a longer period. No variation in total erk1/erk2 abundance was seen (lowest panel). U9 colon carcinoma cells were placed in serum-free media with or without 10 ng/ml IGF-I. Parallel immunoblots were probed for mirk protein (upper blot), for phosphorylated, activated erk1 and erk2 using monoclonal antibody directed to the erk1 and 2 phosphorylated activation sequence TEY (middle blot), and for total erk1/erk2 using an antibody which did not distinguish between activated and inactivated erk forms (lower blot). Mirk levels began to increase by 3 h of serum-free culture and continued to rise for 24 h, coincident with the decrease in activated erk forms. The slow decrease in activated erks may be attributable to autocrine growth factors. However, when IGF-I was added to maintain activation of erk1 and erk2 (Fig. 4C ⇓ , Lanes 7–12), no increase in mirk levels was observed.

Serum-free culture increased the level of transfected mirk (Fig. 4D) ⇓ in the T3 and T7 stable, clonal mirk transfectants of U9 colon carcinoma cells, as well as endogenous mirk in the vector controls, 9V1 and 9V2. Addition of IGF-I to maintain activated erks caused a decrease in both transfected and endogenous mirk levels (Fig. 4E) ⇓ . The stable mirk transfectant T3 expressed about 4-fold the amount of mirk as vector control V1 cells after growth in serum-free medium (Fig. 4E) ⇓ .

mirk mRNA levels were analyzed by RT-PCR in 9V2 vector control and T7 mirk transfectant cells cultured in control serum-containing medium or in serum-free media for 24 h with or without 10 μm PD98059 (MEK inhibitor) or 10 ng/ml IGF-I. Primers for glyceraldehyde-3-phosphate dehydrogenase were included in every reaction mixture as an internal control. Added PD98059 had little effect on endogenous mirk mRNA levels in 9V2 vector control transfectants (Fig. 4F) ⇓ , while dramatically up-regulating mirk protein levels (Fig. 4B) ⇓ . IGF-I had no discernible effect on the combination of exogenous and endogenous mirk mRNA levels in the T7 transfectant (Fig. 4F) ⇓ but slightly decreased mRNA levels of endogenous mirk (seen in three experiments). Thus, mirk protein levels are primarily modulated at the posttranscriptional level by the MAPKs erk1 and erk2.

MAPKs dimerize after their activation, and dimers translocate to the nucleus, where they modify transcription factors (20) . Phosphorylation of proteins by MAPKs has been shown to be a trigger for their degradation (21 , 22) . Western blotting with a mirk N′-terminal antibody demonstrated that a mirk M_r 57,000 breakdown product was found solely in nuclear fractions of both mirk HD3 transfectants (T19, T24) and vector control HD3 transfectants (V1, V2; Fig. 4G ⇓ , upper panel), placing the mirk breakdown product in the nucleus with activated MAPKs. Similar cell fractionation data were observed with U9 cells (Fig. 4G ⇓ , lower panel). We hypothesize that after activation by serum factors and translocation to the nucleus, erks signal the rapid turnover of mirks by phosphorylation of mirk at the canonical MAPK S557 site in its nonconserved C′ terminus and possibly at a second site within the PEST region. 4

Mirk protein levels were generally higher in resected colon cancers than in established cell lines by Western blotting (data not shown), which is not surprising because carcinoma cell lines are established in serum-containing medium that keeps mirk levels low. Cell lines may up-regulate expression of autocrine growth factors to adapt to passage, and in this way also contribute to the down-regulation of mirk. We hypothesized that because mirk abundance is rather low in cell lines, we would have to overexpress mirk in established cell lines to study its function. Clonal stable transfectants expressing elevated endogenous levels of mirk were isolated using two human colon carcinoma cell lines, HD3 and U9 (23) . Mirk levels in these transfectants remained elevated 3–4-fold over vector controls in serum-free conditions (Fig. 4, E and G) ⇓ . Each of the three mirk HD3 cell transfectants and both of the mirk U9 cell transfectants proliferated when switched to serum-free media, whereas none of the five vector control transfectants was able to sustain growth under these conditions (Fig. 5) ⇓ . Mutation of mirk at the putative ATP binding site (KR) and at the YQY activation site (FQF) completely inhibited the kinase activity of mirk on MBP (Fig. 1) ⇓ . Stable transfectants of these mirk mutants lacked the ability to proliferate when switched to serum-free conditions (Fig. 5) ⇓ . Similar data were obtained by direct cell counting and by MTT assay. Overexpression of mirk by stable transfection increased cell numbers 3–5-fold in HD3 cells and 5–9-fold in U9 cells after 5 days of serum-free growth, whereas mutations that blocked the kinase activity of mirk also blocked this biological activity (Fig. 5, A, B, E, and F ⇓ ). After 7 days in serum-free medium, mirk transfectants remained viable and readily resumed rapid growth when serum was added to the cultures (Fig. 5H) ⇓ , whereas vector control transfectant cultures did not re-establish rapid growth, indicating that some cells had lost viability.

Mirk may mediate tumor cell growth and survival under conditions in which MAPKs are not highly activated by extracellular signals, such as early in colon cancer evolution when tumor vascularization is suboptimal, or at the tubular adenoma stage when few autocrine growth factors have been up-regulated. Mirk is expressed in a wide range of tumor types (Fig. 2) ⇓ and in a subset of colon carcinomas (Fig. 3) ⇓ , indicating that selection for mirk-expressing cells is maintained after cancers evolve from premalignant stages. Mirk may mediate tumor cell survival, especially in solid tumors, during periods when carcinoma cells temporarily outgrow their nutrient support.