Functional interaction between the ser/thr kinase PKL12 and N-acetylglucosamine kinase, a prominent enzyme implicated in the salvage pathway for GlcNAc recycling

or in GlcNAcK protein levels. These results suggest that PKL12/GlcNAcK association is not involved in regulation of cellular GlcNAcK activity.


Introduction
PKL12 (protein kinase expressed in day 12 fetal liver), also known as Krct (kinase related to cerevisiae and thaliana), EDPK (embryo-derived protein kinase) and MPSK1 (myristoylated and palmitoylated serine-threonine kinase-1), has recently been isolated from several sources and partially characterized (1)(2)(3)(4). All correspond to the same mammalian gene, human or murine, for which the denomination STK16 has been proposed (International Committee for Human Nomenclature).
PKL12 protein appears to be the first mammalian member of a new ser/thr kinase subfamily not closely related to those previously reported (1). This subfamily includes the putative homologues from Saccharomyces cerevisiae and Arabidopsis thaliana, forming a group of four sequences close in size to a theoretical minimal catalytic domain. It has therefore been proposed that PKL12 may be the catalytic subunit of a more complex holoenzyme composed of catalytic and regulatory subunits (1). E. coli-expressed PKL12, his-tagged or as a GST-fusion, and a FLAGtagged PKL12 protein have been shown to have functional kinase activity, able to phosphorylate exogenous substrates (1)(2)(3) and to promote autophosphorylation (1,2) with the sequence-predicted ser/thr specificity (1,3).
PKL12 mRNA appears to be broadly distributed, both in murine fetal stages (E 6.5-E 18.5) (2,3) and in adult tissues, at low levels in skeletal muscle, heart and spleen, and with high expression in liver, testis and kidney (1)(2)(3). Despite its ubiquitous distribution, in situ analysis showed that PKL12 mRNA is preferentially expressed within specific cellular types in several adult tissues, with predominant expression in epithelial compared to mesenchymal compartments (2).
Further analyses have confirmed the broad distribution of the PKL12 protein in murine tissues and cell lines, although a lack of correlation between mRNA and protein levels was reported (Ligos et al., 2001), suggesting post-transcriptional regulation (2). hPKL12 is acylated by myristic acid at glycine residue 2 and by palmitic acid at cysteines 6 and/or 8 (4). It has been proposed that hPKL12 membrane localization via a myristoylationdependent mechanism is required for the subsequent palmitoylation modification, as demonstrated by guest on  http://www.jbc.org/ Downloaded from PKL12-GlcNAck interaction 4 in other models (4,5). A membrane-associated protein kinase, PKL12, must therefore play a role in intracellular signalling, with a general, highly conserved cellular function (Ligos et al., 2001).
Subcellular localization analysis has shown that PKL12 is a Golgi-resident enzyme. Transient overexpression of PKL12 in NIH-3T3 cells promotes its accumulation in structures related to filopodia and lamellipodia, inducing redistribution of focal contacts and disorganization of the actin cytoskeleton, but no marked alterations in Golgi (Ligos et al., 2001). A regulatory role is thus hypothesized for PKL12 in the control of extracellular matrix-cell adhesion, mediating the dynamic equilibrium of organization/disorganization of focal adhesion structures and actin cytoskeleton (Ligos et al., 2001). Concurring with this proposal, high level forced expression of PKL12 protein in adherent cell lines appears to be incompatible with their survival, whereas PKL12 can be overexpressed in non-adhesion-dependent cell lines without disturbing growth and survival parameters (Ligos et al., 2001).
Based on a two-hybrid analysis, we have identified and demonstrated functional interaction in vitro and in vivo between PKL12, a Golgi-resident ser/thr kinase, and a recently cloned enzyme (6) of amino sugar metabolism, N-acetylglucosamine kinase (GlcNAcK). Although GlcNAcK is not a substrate of PKL12, nor does PKL12 influence GlcNAcK activity either in vitro or in vivo, we have found that both enzymes co-localize in vivo upon overexpression, being a functional GlcNAcK capable to influence PKL12 kinase activity on exogenous substrates. These results indicate a potential in vivo role for GlcNAcK in PKL12 translocation and a tentative regulatory role for PKL12-mediated phosphorylation on substrate proteins. BA/F3 stable clones (K3 and K8) overexpressing PKL12 protein have been described (Ligos et al., 2001).

Yeast two-hybrid analysis
The yeast two-hybrid protein interaction screen was carried out essentially as described (7) using the HF7c yeast strain (Clontech, Palo Alto, CA). Murine PKL12 was fused to the GAL4 DNA-binding domain in the pGBT8 vector (kindly provided by Dr. M. Serrano), generating the pGBT8-PKL12 vector. The coding sequence for the murine PKL12 orf was amplified by PCR from the pcDNA3-PKL12 plasmid, using specific primers (sense primer: 5´-GATGTCGAATTCT TAATGGGCCACGCACTG-3´; antisense primer: 5´-CGACTGAGATCTTCAGATTGGGTGG TGTG-3´) for simultaneous elimination of the PKL12 orf initiation codon and creation of an An NIH-3T3 cDNA library generated in the pGAD424 vector (provided by Dr. M. Serrano) was used in the screening. Yeast stably transformed with the pGBT8-PKL12 vector, and thus able to grow in tryptophan-free medium, were obtained. Stable expression of GAL4bd-mPKL12 protein was confirmed by western blot (not shown). Yeast were then transformed with the NIH-3T3 cDNA library in the pGAD424 vector and selected for growth in minimal medium without tryptophan, leucine or histidine. As positive controls, pGBT8-p16 and pGAD424-cdk4 plasmids were cotransfected in yeast; as negative control the pGBT8 vector plasmid was cotransfected with the pGAD424-cdk4 plasmid. Yeast clones obtained after selection in tryptophan-, leucine-and histidine-free medium were also tested for β-galactosidase expression.
After isolation of pGAD424-derivative plasmids from clones positive for both selective criteria, they were re-confirmed by direct cotransfection with the pGBT8-PKL12 vector in HF7c yeast, and inserts were directly sequenced using the vector primers.

mSIP16(GlcNAcK) full-length cDNA cloning
Specific oligonucleotides (sense primer: 5´-AGGCGACACAGGGGCGAGAGA-3´; antisense primer: 5´-GAAAGCGGTGCCTCAACTCCTC-3´) were synthesized (Isogen, Maarssen, The Netherlands) based on the 5´ and 3´ sequences of the mSIP16 obtained from the ESTs data bank and our data. Total mRNA was prepared from NIH 3T3 cells and used to obtain cDNA.
PCR was carried out on cDNA derived from NIH 3T3 cells with the primers indicated above (94 o C, 1 min; 60 o C, 1 min; 72 o C, 1,5 min; 30 cycles), using Taq/Pow as above. The fragment obtained, of the predicted size, was cloned in the pGEM-T (Promega) plasmid. Several clones were obtained and fully sequenced. The final clone selected was termed pGEM-GlcNAcK.

E. coli expression and purification of histidine-and GST-tagged proteins
Histidine-tagged PKL12 was expressed in E. coli strain M15 and purified to homogeneity using Ni-NTA agarose resin (Qiagen), as described (1); GST-SIP16 protein expression and purification was performed in a similar manner. The coding sequence for the orf corresponding to murine SIP16 was PCR amplified from the pGAD424-SIP16 plasmid using specific primers Several independent induction/purification experiments rendered an almost pure fraction of GST-SIP16 protein, with two protein bands (see Results, Fig. 3). Both bands were recognized by anti-GST antibody and the specific anti-SIP16 antiserum; we thus concluded that in E. coli or during the purification process, the SIP16 protein is proteolyzed at a specific carboxy-terminal point, rendering both products. These bands are not present in negative control purified fractions.
Control GST and GST-SP (a fusion of GST with a protein that is a substrate for the kinase activity of the PKL12 protein; (Ligos et al., 2001) proteins were expressed and purified essentially as described for GST-SIP16. Protein concentration of purified fractions was determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA) with a bovine serum albumin standard.

Rabbit antiserum generation and testing
Purified GST-tagged SIP16 protein was prepared as above. Outbred New Zealand rabbits were injected intradermally in multiple sites using 250 µg of purified protein emulsified with an equal volume of Freund's complete adjuvant. Two 125 µg intramuscular boosts of the same material in incomplete adjuvant were given 4 and 7 weeks later. Sera were collected 7 and 10 days after the last injection and tested in ELISA; the IgG serum fraction was purified as described (1).

Western blotting
Protein (20 µg) was separated in SDS-PAGE and transferred to nitrocellulose membrane (Bio-Rad). The membrane was blocked for 1 h in Tris-buffered saline (25 mM Tris; TBS) plus 0.1% Tween, with 5% non-fat dry milk, followed by incubation with primary antibody for 2 h and secondary antibody for 40 min. Western blots were developed using the ECL system (Amersham, Aylesbury, U.K.). The polyclonal anti-PKL12 and anti-GlcNAcK antisera were used at 1:2000 and 1:3000 dilution, respectively.

Immunoprecipitation assays
Cells were transfected as above, cultured for an additional 24 h and lysed in 100 µl of IP In vitro GlcNAcK activity was determined as described (7). In brief, GlcNAcK assays were performed in a final volume of 225 µl containing 60 mM Tris-HCl pH 7.5, 20 mM MgCl 2 , 5 mM GlcNAc, 10 mM ATP (disodium salt), 10 mM phosphoenolpyruvate, 2.5 U pyruvate kinase, 50 nCi [1-14 C] GlcNAc, and variable amounts of protein extract. Incubations were carried out (37 o C, 2 h), and the reaction terminated by adding 350 µl ethanol. Radiolabeled substrates were separated by descendent paper chromatography and measured by liquid scintillation analysis.

Amino acid sequence comparison
Initial comparative sequence searches were performed using the BLAST algorithm (9) on a non-redundant database (EMBL NRDB). Additional remote homologue searching was done by HMM (hmm search, default parameters) over protein nrdb database (10). Sequences were aligned with CLUSTALW software (11

Interaction between PKL12 and GlcNAcK in yeast two-hybrid
We used the yeast two-hybrid protein interaction screen to search for proteins able to associate with the murine PKL12 (STK16) ser/thr kinase and to delineate signalling pathways and/or regulatory circuits in which this kinase participates. PKL12 is expressed ubiquitously and NIH-3T3 cells are a study model (1-3, Ligos et al., 2001). HF7c yeast was stably transformed with the GAL4bd-PKL12 plasmid (pGBT8-PKL12); a NIH-3T3 library fused to the GAL4 activator domain in the pGAD424 plasmid was then transformed and screened in minimal medium. Several clones were selected that grew under the selection conditions and also expressed lacZ, the second transactivatable marker gene. pGAD424 plasmids harbored by the positive yeast clones were isolated; positive interaction was confirmed by individual cotransformation of HF7c yeast with pGBT8-PKL12 plasmid or the negative controls pGBT8-p16 (12) and pGBT8. Positive clones were isolated and fully sequenced. Several positive clones (representing 23% of sequences obtained) were found containing partial sequences corresponding to the same gene. The clone containing the longest ORF corresponded to a 334-amino-acid protein (38 kDa) in phase with GAL4db. This clone, pGAD424-SIP16∆, was selected for some of the further experiments.
This protein was denominated SIP16 (STK-16 interacting protein), as sequence homology analysis showed that it has a yet-undescribed function. Homologous EST sequences from several murine and human tissues, and a C. elegans (WO6B4.2) protein (unknown function) showed significant similarity to the SIP16∆ encoded orf. Based on the data bank-annotated EST sequences, we identified the putative 5´ UTR sequence and the transcription initiation codon of the mSIP16 orf. We designed specific oligonucleotides to clone the full-length cDNA, which was obtained from NIH-3T3 cells. mSIP16 protein consists of 343 amino acids (approximate M r 40 kDa) and hSIP16 was electronically assembled from the data bank EST sequences. Similarity searches using human and murine sequences (92% similarity) revealed the presence of ortholog sequences in several other eukaryotic organisms (Fig. 1

), including Drosophila melanogaster and
Streptomyces coelicolor (43% and 27% similarity, respectively), and partial EST sequences in pig,  (13). Figure 1 shows the amino acid sequence comparison between eukaryotic enzymes and the most representative members of the prokaryotic family, including the YHCI, YAJF, YCFX and ALSK enzymes from E. coli, and SCRK from B. subtilis.
These data strongly suggested that SIP16 is an eukaryotic sugar kinase. When preparing the first version of the manuscript, a final search in the data banks revealed that new sequences were included (NP_062415; NP_060037; CAB61849) with high similarity (98%) to mSIP16. These sequences corresponded to the murine and human GlcNAcK proteins (6); a PKL12 (STK16) interaction with GlcNAcK was therefore suggested.

Functional interaction of PKL12 and GlcNAcK in vitro and in vivo
The specificity of the PKL12-GlcNAcK interaction was first studied in a cell-free system.
A fusion protein (GST-SIP16∆) consisting of glutathione S-transferase fused to the specific sequence contained in the pGAD424-SIP16∆ clone isolated from the two-hybrid screening, or the control GST proteins were expressed in bacteria and crude extracts incubated with purified his-PKL12 protein (1). GST-SIP16∆ protein was recovered from the mixtures on glutathione-Sepharose beads; the resulting precipitates were resolved in SDS-PAGE and analyzed by western blot with rabbit polyclonal anti-PKL12 antiserum. The specific PKL12 band was observed only when bacterial extracts containing the GST-SIP16∆ protein were included in the incubation mixture with purified his-PKL12 protein (Fig. 2A).
The specificity of the PKL12/GlcNAcK interaction was also studied in mammalian cells. GlcNAcK protein (Fig. 2B). These results confirm the interaction of PKL12 and GlcNAcK revealed by the two-hybrid system.

GlcNAcK is a ubiquitously expressed cytoplasmic protein in cell lines and murine tissues
Northern blot analysis showed broad distribution of GlcNAcK mRNA in cell lines and murine tissues (our unpublished results; 6). To confirm this at the protein level, we generated a rabbit antiserum specific for GST-GlcNAcK. The anti-GlcNAcK antiserum detected both purified GST-GlcNAcK expressed in E. coli (Fig. 3A) and the endogenous murine protein (Fig. 3B) in several cell lines. Antibody specificity in western blot was confirmed by competition for the signal with an excess of purified GST-GlcNAcK protein (Fig. 3A). The double band revealed by the anti-GlcNAcK antiserum corresponded to the presence of two protein species in the purified GlcNAcK fraction, probably due to carboxy-terminal proteolytic processing of the intact enzyme in E. coli.
The endogenous GlcNAcK protein levels were variable in the cell lines analyzed, with a surprising near-absence of GlcNAcK protein in the A20 pro-B cell line. This result implies that GlcNAcK activity is not absolutely required for cell viability.

GlcNAcK is not a substrate of the PKL12
We analyzed whether the GlcNAcK protein could be used as a substrate by the intrinsic kinase activity of the PKL12 protein (1). Purified his-PKL12 was incubated with purified GST-SIP16∆ protein in the standard assay conditions (see Experimental Procedures). The autoradiograph obtained after SDS-PAGE resolution of the products of this in vitro kinase assay is shown in Fig. 5A; controls are described in the figure legend. No significant phosphorylation was detected in the GlcNAcK band, indicating that at least in vitro, GlcNAcK is not a substrate for the PKL12 kinase activity. None of these bands was observed in the negative control fraction (hiscontrol). In addition, no significant alteration was noted in the PKL12 autophosphorylation band (further tested; see Fig. 6). Similar results were obtained when GST-GlcNAcK or purified rat GlcNAcK proteins were used in the assay (not shown).

GlcNAcK is able to regulate PKL12 kinase activity on exogenous substrates
Once established that GlcNAcK is neither a substrate for PKL12 kinase nor influences its autophosphorylation capacity, we tested whether GlcNAcK affects PKL12 activity on an exogenous substrate (GST-SP) protein. Full-length GlcNAcK fused to GST (GST-GlcNAcK) was compared with the previously described GST-SIP16∆ construct. Both purified proteins were tested for their intrinsic kinase activity (see Experimental Procedures) in comparison with purified rat GlcNAcK. As predicted, GST-GlcNAcK showed a specific activity (20 U/mg) very similar to that of the rat enzyme. The GST-SIP16∆ protein was completely inactive, due to the lack of highly conserved amino acids of the ATP-binding motif shared by all sugar kinases (13). GST-SIP16∆ protein in the kinase assay, however GST-GlcNAcK appears to downregulate GST-SP phosphorylation. In conclusion, although GlcNAcK protein interacts with PKL12, this association is not sufficient for the regulatory effect observed. GlcNAcK must be completely functional to be able to influence the phosphorylation capacity of PKL12 on exogenous substrates.

PKL12 does not influence in vitro and in vivo GlcNAcK activity
The influence of PKL12 on the activity of GST-GlcNAcK was tested in vitro.

DISCUSSION
Using the yeast two-hybrid system, we identified a protein (SIP16) that interacts with the ubiquitous ser/thr kinase PKL12, recently isolated by our group and others (1)(2)(3)(4), which appears to have an important role in the control of cell adhesion (Ligos et al., 2001). PKL12 is a small ser/thr kinase with no detectable regulatory domain in its primary structure; it was thus proposed that it could be the catalytic subunit of a more complex holoenzyme (1). The detection of the SIP16 interacting protein challenged this hypothesis. SIP16/PKL12 interaction was confirmed both in vitro and in vivo, and co-localization of the proteins was described, partially in the Golgi area of interphase cells and, more prominently, following overexpression in NIH-3T3 cells (Fig. 4).
Preliminary sequence homology searches in data banks revealed several human and murine ESTs clearly corresponding to SIP16, which allowed us to assemble the virtual human full-length cDNA; it was also distantly related to orf from C. elegans and D. melanogaster. These sequences did not show notable similarity with other characterized eukaryotic proteins, although they showed a clear relationship with bacterial sugar kinases (Fig. 1). Murine and human sequences for GlcNAcK were recently included in data banks (NP_062415, NP_060037, CAB61849), and subsequently published (6) demonstrating that mSIP16 is nearly identical (98%) to murine GlcNAcK.
GlcNAc is a major component of complex carbohydrates, found in glycoproteins (14), glycolipids (15) and proteoglycans (16). GlcNAcK catalyzes GlcNAc phosphorylation at carbon atom 6. It is the key enzyme in the salvage pathway for recycling GlcNAc derived from lysosomal degradation of oligosaccharide moieties or nutritional sources, into GlcNAc 6-phosphate. GlcNAc 6-phosphate can then enter a catabolic route that links hexosamine metabolism with the glycolytic pathway (17), or may enter an anabolic pathway leading to UDP-GlcNAc formation (18).
De novo UDP-GlcNAc synthesis follows the sequence of intermediates: fructose 6phosphate → glucosamine 6-phosphate → N-acetylglucosamine 6-phosphate → Nacetylglucosamine 1-phosphate → UDP-GlcNAc (18). The extent to which redundancy in UDP-GlcNAc generation is used by cells to respond to distinct external stimuli is unclear, as is the manner in which activity of the respective enzymes is regulated (19). and UDP-GlcNAc de novo and salvage synthesis appear to be interconnected; it has been proposed that UDP-GlcNAc displays a cellular sensor of energy availability, modulating gene expression in response to nutrient availability (17,20). The effects of UDP-GlcNAc are mediated via modification of the O-GlcNAcylation status of critical proteins as a counterpart of ser/thr phosphorylation (reviewed in 21). Alterations in the natural equilibrium between the metabolic pathways have been implicated in diabetes pathogenesis (22).
We thus examined a potential regulatory role for the GlcNAcK-PKL12 interaction. We analyzed whether GlcNAcK is a substrate for PKL12 kinase activity, mediating in the regulation of GlcNAcK catalytic activity. In vitro kinase assays using the recombinant GST-SIP16∆ protein as exogenous substrate showed that GlcNAcK contains no substrate site for PKL12 activity (Fig. 5).
These results were confirmed using GlcNAcK purified from rat liver (not shown). In addition, we studied the effect of PKL12 overexpression on cellular levels of GlcNAcK protein and activity, both in transient transfection of adherent cells and in stable PKL12 clones generated in nonadhesion-dependent cell lines (Fig. 7). We detected no significant differences in cellular GlcNAcK activity or protein mass in either model, independent of the amount of PKL12 expressed. We conclude that there is no regulation of GlcNAcK and consequently, no direct regulation of the UDP-GlcNAc pool by PKL12.
GlcNAcK does not appear to regulate the in vitro kinase autophosphorylation activity of PKL12 either positively or negatively (Fig. 5, 6). A negative regulatory effect was nonetheless shown on the PKL12-kinase activity over the exogenous E. coli recombinant GST-SP substrate when a fully active GST-GlcNAcK was included in the assay (Fig. 6). The regulatory effect of GlcNAcK on PKL12 is not only due to binding of the two proteins, since addition to the kinase assay of the truncated GlcNAcK (GST-SIP16∆), which binds to PKL12 (Fig. 1) regulation of PKL12 activity by full-length GlcNAcK, a structurally intact, rather than an enzymatically active protein may be necessary. Point mutations of GlcNAcK, which only affect the enzymatic activity, will clarify this in the future.
We next studied subcellular localization of endogenous GlcNAcK in NIH-3T3 cells (Fig.   4). The enzyme accumulates mainly in two areas, in the perinuclear region and, predominantly, at the cell leading edge in membrane-associated structures similar to filopodia. This contrasts clearly with the subcellular localization of glucosamine 6-phosphate acetyltransferase (EMeg32), an enzyme involved in de novo synthesis of UDP-GlcNAc (23). Since nucleotide sugar synthesis is believed to take place in the cytoplasm, it was therefore consistent that the enzyme associates with the cytoplasmic face of various intracellular membranes (Golgi and late endosome/lysosome). It was thus tempting to assume that membrane association of EMeg32 to the cytoplasmic leaflet of Golgi and other intracellular membranes may facilitate local cytoplasmic release of GlcNAc-6-P; no further factors would be needed to make this compound available for the last two steps leading to synthesis of UDP-GlcNAc, which is finally transported to ER and Golgi by specific proteins (24).
Here overexpression in adherent cells appears to be incompatible with viability, as it causes disorganization of actin cytoskeleton and focal adhesion contacts, detaching cells from the substrate (Ligos et al., 2001). We interpret these results as indicating that GlcNAcK may have a role as a docking or scaffold molecule that could recruit PKL12 to its site of physiological action following activation. Such a dual role also has been suggested for EMeg32, which co-purifies with the cdc48 homologue p97/VCP, a protein implicated in mitotic membrane fusion events (19). A regulatory or scaffold function for the ATPase activity of p97/VCP has also been proposed (19). The other potential function of the PKL12/GlcNAcK translocation mechanism is GlcNAcK transport to filopodia and lamellipodia. This would establish UDP-GlcNAc biosynthesis independent of the de novo pathway in a specific subcompartment, as discussed above. Taken together, our results strongly favor novel roles for PKL12 and GlcNAcK, completely different from previously described properties of these enzymes, including a regulatory role for GlcNAcK on PKL12mediated kinase activity over certain specific substrates.