The full-length human MIXL1 ORF, ΔCAD mutants with truncation of carboxy-terminal acidic domain and ΔN25 mutant with truncation of both amino-terminal 25 bp region and carboxy-terminal acidic domain were amplified by PCR and cloned into expression vectors CMV5 (a kind gift from David W. Russell, UTSW at Dallas, TX) and CMV2-flag (Sigma, St. Louis, MO) to generate CMV5-MIXL1, CMV2-Flag-MIXL1, CMV5-ΔCAD, CMV2-flag-ΔCAD and CMV2-flag-ΔN25. The amino terminal of CMV2-flag-ΔCAD was replaced with the 1–93 bp amino-terminal portion of MIXL1 to generate the construct CMV2-flag-ΔPC lacking both PRD and CAD domains. Y110F mutation was introduced by PCR with a primer containing an A-to-T (Tyr-to-Phe) point mutation. The MIXL1 carboxy-terminal portion (301–699) of the construct CMV2-flag – ΔCAD or CMV2-flag-ΔN25 was replaced with the amplified MIXL1 fragments carrying the A-to-T point mutation to generate the construct CMV2-flag-Y110F or CMV2-flag-Δ2Y. Accuracy of the generated constructs was confirmed by double stranded sequencing.

HEK 293T cells were grown in modified Eagle's medium (MEM, Invitrogen) supplemented with 10% FBS, 0.1 mM non-essential amino acids and 1 mM sodium pyruvate (Invitrogen) at 37°C in 10% CO2. For transfection, cells were plated at a density of 2 × 10⁵ cells per well in a 6-well plate 2 days prior to transfection. The transfections were performed using LipofectAmine (Invitrogen) according to the manufacturer's protocols. Total amount of transfected DNA was adjusted to 1 μg per well by using appropriate parental vectors. Cells were harvested for nuclear extraction 48 hours after transfection. For immunoprecipitation experiments, the transfections were scaled up to 100 mm plates.

Nuclear extracts were prepared from transfected 293T cells 8 and diluted to approximately 1.0 μg protein/μL lysate. Briefly, 500 μL of the fresh nuclear extracts were pre-cleared with 50 μL of protein-A-agarose bead slurry (50% v/v, Roche, Indianapolis, IN) at 4°C for 30 minutes in an orbital shaker. After centrifugation at 14,000 × g at 4°C for 10 minutes, supernatant was mixed with 5 μg of the murine monoclonal antibody 4G10 (Upstate) or the isotypic control (murine monoclonal antibody against the V5 epitope-Invitrogen) at 4°C overnight on the orbital shaker. The immune-complexes were captured by adding 50 μL of protein A agarose bead slurry (50% v/v) and gently rocking on the orbital shaker at 4°C for 2 hours. After pulse centrifugation at 14,000 rpm for 8 seconds, the immunecomplex-protein-A-agarose-bead pellets were collected and washed 5 times with 800 μL of ice-cold modified radioimmunoprecipitation (RIPA) buffer (50 mM Tris-HCl [pH 7.4]; 150 mM NaCl; 1% NP-40; 0.25% sodium deoxycholate; 1 mM EDTA; 1 mM PMSF; 2 μg/mL leupeptin; 2 μg/mL pepstatin A; 2 μg/mL aprotinin; 500 μg/mL benzamidine; 1 mM Na3VO4; 1 mM NaF) and once with 800 μL ice-cold 1× PBS. The pellets containing immune complexes were resuspended in 60 μL of 2× sample buffer (100 mM Tris-HCl [pH 6.8], 200 mM DTT, 4% SDS, 0.2% Bromophenol Blue, 20% Glycerol).

Immunoprecipitates or nuclear proteins were resolved (50 μg of protein per lane) on pre-cast 10% NuPAGE gels (Invitrogen). After electrophoresis, the proteins were transferred to Hybond P nylon membrane (Amersham Biosciences, Piscataway, NJ) at 30 V overnight. The protocol for immunobloting was essentially as detailed elsewhere 21. Rabbit polyclonal antibody anti-MIXL1-N 8 was used at a dilution of 1:100 initially or 1:500 for immunoblotting studies. Mouse monoclonal antibody anti-flag M2 (Sigma) 1:300; mouse monoclonal antibody 4G10 (Upstate) 1:3000 or 1:4000.

Nuclear extracts of HEK 293T cells transfected with CMV5-MIXL1 was prepared without the addition of phosphatase inhibitors sodium fluoride and sodium orthovanadate. Immediately after extraction, 15 μg of nuclear protein was incubated with 40 units of calf intestine alkaline phosphatase (CIAP, Roche) in a 20 μL reaction for 30 minutes at 30°C. As a control, the same reaction was performed in the presence of 100 mM sodium orthovanadate, which inhibits CIAP activity. In addition, a mock reaction was also performed with no CIAP. The reactions were terminated with NuPAGE sample buffer (Invitrogen). The changes in protein mobility were determined by probing the immunoblots with the anti-MIXL1-N antibody.

When the full-length MIXL1 expression driven under CMV promoter in the construct pCMV5-MIXL1 was tested by transient transfection into the HEK cell line 293T, we detected at least three species for MIXL1 proteins in nuclear extracts by immunobloting with the anti-MIXL1-N antibody. Similar experiments with other expression constructs containing full-length or truncated MIXL1 also detected at least two species for MIXL1 proteins. To test the possibility that the multiple bands were the MIXL1 proteins modified by phosphorylation, the nuclear extracts were treated with alkaline phosphatase (CIAP) to remove phosphate groups from phosphorylated proteins. MIXL1 proteins were detected by immunobloting with the anti-MIXL1-N antibody. As shown in Fig. 1, the antibody specifically recognized four species for MIXL1 in untreated nuclear extracts. However, the CIAP treatment resulted in the disappearance of two species, α and β at the top and greatly intensified the signals of the lower two bands (γ and δ). Interestingly, the addition of the phosphatase inhibitor orthovanadate restored the four species of MIXL1 proteins on the blot, demonstrating that the two slower isoforms of MIXL1 are phosphatase sensitive. The slowest migrating species of MIXL1 proteins (α) was also lost in the mock reaction without addition of CIAP. The loss could be due to endogenous phosphatase activity from nuclear extracts, as phosphatase inhibitors were not used in both the nuclear extraction and the mock reaction. Together, the phosphatase-sensitive properties indicated that MIXL1 protein is phosphorylated at multiple sites.

A phosphorylation prediction program called Netphos 2.0 in the public domain 22 was used to search for potential phosphorylation sites in the MIXL1 protein sequence. By comparing the MIXL1 protein sequence with phosphorylation consensuses for known kinases, the program predicted 14 potential phosphorylation sites including 10 serine residues, 3 threonine residues and 1 tyrosine residue. Since only one tyrosine residue out of two showed potential for phosphorylation, we chose to examine tyrosine phosphorylation first. The phosphotyrosine antibody 4G10, which can specifically detect the tyrosine-phosphorylated proteins, was used to see if MIXL1 is tyrosine phosphorylated. Nuclear extracts prepared from the 293T cells transfected with the constructs pCMV5-MIXL1 (Full-length MIXL1) and pCMV5-ΔCAD (CAD-deleted mutant, ΔCAD) was loaded in duplicate for immunobloting. One half of the blot was hybridized with the antibody 4G10 and the other half with the MIXL1-specific antibody to show protein integrity and the relative position of MIXL1 on the blot. As shown in Fig. 2, the phosphotyrosine antibody 4G10 detected a weak band in the nuclear extracts from the cells with the full-length MIXL1, which was not present in the nuclear extracts with the control vector. The weak species was at a position similar to that of the MIXL1 proteins detected with the anti-MIXL1-N antibody, suggesting that the MIXL1 might be tyrosine phosphorylated in HEK 293T cells. Consistent with this possibility, the phosphotyrosine antibody detected a unique band of much stronger intensity at a position corresponding to mutant ΔCAD, demonstrating that the deletion of the CAD domain might result in increased tyrosine phosphorylation on MIXL1 proteins, possibly due to enhanced stability or easier access to the phosphorylation site(s).

The detection of tyrosine phosphorylation by immunobloting with the phosphotyrosine antibody could not rule out the possibility that the antibody 4G10 was bound non-specifically to the abundant MIXL1 proteins overexpressed in 293T cells. To address this possibility, we examined if MIXL1 proteins could be immunoprecipitated with the antibody 4G10, which specifically binds to phospho-tyrosine residues. The anti-MIXL1-N antibody detected a weak but specific signal for full-length MIXL1 proteins in the immunoprecipitates with 4G10 (Lane 1, 2 and 3 in Fig. 3). Similarly, the mutant ΔCAD was specifically immunoprecipitated with the phosphotyrosine antibody 4G10 (Lane 4, 5 and 6 in Fig. 3). The results demonstrated that the phospho-tyrosine antibody specifically recognized some MIXL1 proteins in nuclear extracts. The immunoprecipitation analysis further confirmed that MIXL1 was tyrosine phosphorylated in 293T cells.

MIXL1 contains only two tyrosine residues. Of these, one is in the amino-terminal domain (Tyr20) and the other is in homeodomain (Tyr110). Therefore, it is easy to determine the tyrosine phosphorylation site(s) in MIXL1 by mutational analysis. Since the mutant ΔCAD showed a much stronger signal for tyrosine phosphorylation than the full-length MIXL1 in western blotting with the antibody 4G10, it was employed in mutation analysis. Based on the ΔCAD mutant sequence, we generated 3 MIXL1 mutants with either a point mutation on the residue Tyr110 or a small amino-terminal deletion including Tyr20 or both (Fig. 4A). Since the antibody anti-MIXL1-N could not detect the amino-terminally deleted mutant proteins, a small epitope Flag was added in frame to the amino-terminal ends of all the three mutants as well as the mutant ΔCAD (Fig. 4A). Nuclear extracts from the 293T cells transfected with the four constructs were prepared for immunobloting with the antibody 4G10. The same blot was stripped and rehybridized with the mouse monoclonal anti-Flag antibody, which detects the Flag epitope. Surprisingly, the antibody 4G10 detected tyrosine phosphorylation on the mutant Flag-Y110F containing the single tyrosine residue Tyr20, similar to that on the mutant Flag-ΔCAD (renamed as Flag-2Y here) with both tyrosine residues (Fig. 4B). In contrast, the antibody 4G10 failed to detect tyrosine phosphorylation on the amino-terminal deletion mutant Flag-ΔN25 containing the single tyrosine residue Tyr110 as well as the mutant Flag-Δ2Y lacking both tyrosine residues, although the protein levels for the four constructs were similar in the nuclear extracts (Fig. 4B). Moreover, the same pattern was detected in the COS-1 cells transfected with those four constructs (data not shown). Thus, these observations suggested tyr20 to be the target of phosphorylation in HEK 293T cells, although this is inconsistent with the prediction by the program Netphos2.0. Whether tyr20 is phosphorylated on endogenous MIXL1 proteins remains to be confirmed.

Recent studies reveal that the substrate proteins for phosphorylation often contain a docking domain to recruit specific protein kinases 23. In some instances, the PRD in substrate proteins may compete against the auto-inhibitory interaction between the SH3 domain and catalytic domain, by interacting with SH3 domain in protein kinases such as Src 24. Since the human MIXL1 contains a PRD immediately downstream of the residue Tyr20 (11 residues downstream from the residue Tyr20), we postulated that the PRD in human MIXL1 might be involved in tyrosine phosphorylation. Thus, we examined the tyrosine phosphorylation on PRD-deleted MIXL1 proteins. Since ΔCAD mutant shows stronger signals for tyrosine phosphorylation, a construct containing the mutant MIXL1 gene with the double deletion of PRD and CAD domains (ΔPC) was generated and transfected into 293T cells. The tyrosine phosphorylation of the mutant ΔPC was determined by immunobloting with the antibody 4G10. As shown in Fig. 5, the tyrosine phosphorylation on the mutant ΔPC was much weaker than that of the mutant ΔCAD, suggesting that the PRD, although not absolutely necessary for Tyr20 phosphorylation, may be important for either achieving full-scale phosphorylation or maintaining the phosphorylation state by preventing dephosphorylation. Future studies with antibodies specific for the Tyr20 phosphorylated protein will elucidate how and when the phosphorylation occurs. More importantly, it will be critical to examine phosphorylation of endogenous MIXL1 in hematopoietic tissues and breast cancer cells. Thus the present report is a first step in elucidating the potential function of the PRD and Tyr20, features unique to avian and mammalian MIXL1.

Since MIXL1 localizes to the predominantly to the nucleus (Guo and Nagarajan unpublished results), the kinase(s) responsible for tyrosine phosphorylation on MIXL1 is likely to be localized in the nucleus. Ten out of the 90 tyrosine kinases encoded in humans, are known to localize to the nucleus (reviewed by Cans et al 25). Thus the likely candidates are c-ABL1, Wee1, FRK, LYN, FES family (FES and FER) and JAK family (JAK1, JAK2, JAK3, and TYK2). Although the present studies were conducted in HEK293 cells, several of these kinases are expressed in the hematopoietic system. Additionally, since the tyrosine phosphorylation of MIXL1 was not examined in cytoplasm, we could not rule out that MIXL1 is tyrosine phosphorylated in the cytoplasm and translocated into the nucleus.

Unlike serine/threonine phosphorylation, tyrosine phosphorylation on homeodomain proteins is rarely reported to date. Hence the role of tyrosine phosphorylation in the regulation of homeodomain proteins largely remains unclear. The only reported case is the tyrosine phosphorylation of HoxA10 during interferon γ-induced myeloid differentiation. In this case, interferon γ-induced differentiation led to HoxA10 tyrosine phosphorylation in the myelomonocytic cell line U937, which decreased DNA binding of HoxA10 to Pbx-HoxA10 binding sites 26. However, SHP1 protein-tyrosine phosphatase (SHP1-PTP), which antagonizes myeloid differentiation, decreased tyrosine phosphorylation of HOXA10 homeodomain thereby enhancing HOXA10-mediated repression 27.

A tissue- or cell cycle-specific phosphorylation may alter MIXL1 activity. Future studies will elucidate whether phosphorylation mediated protein-protein interactions due to the unique PRD in human, mouse and chicken Mix-like proteins indeed substitutes for the functional diversity achieved by multiple members in Xenopus Laevis.