Recent evidence indicates that in addition to the function of NHE1 in ion translocation and pH_i homeostasis, the exchanger also acts as a scaffold to assemble signaling complexes and as a plasma membrane anchor for the actin-based cytoskeleton [3,6]. To selectively impair only ion translocation by NHE1, we engineered an ion translocation-defective NHE1 containing an isoleucine substitution for glutamine 266 (NHE1-E266I). In cells expressing NHE1-E266I, the scaffolding and actin anchoring functions of NHE1 are retained, but ion translocation is absent [5,6]. Wild-type NHE1 (LAPN cells) and NHE1-E266I (LAPE cells) were stably expressed in NHE1-null LAP1 cells, which are derived from NHE1-expressing Ltk-mouse muscle fibroblasts [16,17]. As previously reported [5] NHE1 expression in LAPN and LAPE cells, as determined by immunoblotting, is similar and steady-state pH_i in the continuous presence of serum and HCO₃^- is ~7.35 for LAPN cells and ~7.10 for LAPE cells. The presence of HCO₃^- allowed the function of anion exchangers contributing to pH_i homeostasis in the absence of ion translocation by NHE1.

For DNA microarray analysis, significant regulation of genes in LAPE cells compared with LAPN cells was defined as a fold change > 1.5 with a p value of < 0.05 from five independent cell preparations and microarray hybridizations. Of the 6,500 probe sets, 198 or 3.05% were significantly different in LAPE cells. Two widely used approaches to analyze DNA microarray data include hierarchial clustering of genes with similar expression patterns [18] and grouping of biologically related genes into processes or pathways [19,20]. We used the latter strategy to group genes regulated by NHE1 activity into related biological pathways or processes. Genes were grouped according to key-words representing functional categories and GenMAPP, developed by the Conklin laboratory at the University of California, San Francisco [20,21], was used to visualize gene expression data on maps representing biological pathways. The advantage of a pathway-based analysis is that it provides a global perspective of functionally-related genes. Pathway-based grouping indicated a substantial number of differentially expressed genes associated with growth factor/hormone signaling and growth and oncogenesis (Fig. 1). The caveat of pathway-based analyses is that based on key-word representation, some genes are implicated in multiple biological processes. Hence, we listed all genes exhibiting significant changes in LAPE cells compared with LAPN cells (Table 1). Data in Table 1 are grouped according to biological function with absolute changes indicated.

The expression of a substantial number of genes encoding proteins related to growth factor/hormone signaling and growth and oncogenesis was altered in LAPE cells compared with LAPN cells. A schematic cascade of growth factor signaling indicates that a number of genes regulated in LAPE cells function in Ras-dependent signaling (Fig 2A). Activation of many of these signaling proteins, including serotonin (5HT) [22] and thrombin (PAR2) [23] receptors, Ras [13], Raf [24], PLCγ1 [25], and Erk [26,27] stimulates NHE1 activity. Increased NHE1 activity and a resulting intracellular alkalinization are thought to be necessary for oncogenic transformation [14] and tumor development [15]. Immunoblot analysis confirmed increased protein expression of Fyn, PLCγ1, and ERK2 in LAPE cells compared with LAPN cells (Fig. 2B), In LAPE cells, the global increased expression of a number of genes involved in growth factor and oncogenic signaling suggests a feedback response to acidic or osmotic stress. Alternatively, because the proliferative response is suppressed in LAPE cells [5] (Fig. 3), increased growth factor signaling could result from a feedback mechanism to maintain cell proliferation.

Expression of a number of transcription factor genes acting downstream of growth factor signaling was also differentially regulated in LAPE cells compared with LAPN cells (Fig. 2A). Fra1 and Fra2 are Fos proteins and components of the AP1 transcription factor. They form dimmers with Jun transcription factors to regulate a number of cell processes including differentiation, proliferation, and oncogenic transformation [28]. The regulated expression of two members of the C/EBP (CAAT/enhancer binding protein) family of transcription factors was confirmed. RT-PCR was used to confirm decreased expression of C/EBPδ (Fig. 2C), which dimerizes with C/EBPβ in response to Ras-ERK signaling to regulate adipocyte [29,30] and epidermal [31] differentiation. Increased expression of GADD153 (CHOP), which is a transcription factor in the C/EBP family, was confirmed by TaqMan analysis (Fig. 2D). GADD153 dimerizes with other C/EBP isomers to inhibit their binding to C/EBP binding sites in the promoters of a number of genes involved in differentiation and mitogenesis [32,33]. Expression of GADD153 increases in response to DNA damage [34], and oxidative stress [35,36], hence its role in cell cycle progression is also included in Figure 3.

NHE1 activity has a permissive effect in promoting cell proliferation [6,7] and cDNA microarray analysis indicated a number of genes with roles in DNA synthesis and cell cycle control had altered expression in LAPE cells compared with LAPN cells. We recently reported [5] that the proliferative rate of LAPE cells is ~3 to 4-fold less than that of LAPN cells and that LAPE cells lack a pH-dependent timing of cell cycle progression that is specifically associated with delayed G2/M entry and transition. Consistent with these findings there was an upregulation of genes associated with G2/M arrest and DNA repair responses (Fig. 3A). Decreased protein expression of GADD45 and FEN1 was confirmed by immunoblotting (Fig. 3B,3C). The decrease in FEN1 in LAPE cells was most marked when cells were synchronized by a double thymidine block, and released from the block for 3 to 9 hours (Fig. 3C). GADD45, which acts in DNA repair, is generally upregulated with cell cycle arrest [37,38], and FEN1 is thought to play an essential role in DNA replication and in base excision repair [39,40]. Although decreased expression of GADD45 and FEN1 in LAPE cells appears paradoxical, recent findings indicate that a decrease in GADD45 would contribute to p53 instability [38] and FEN1 is stimulated by proliferating nuclear antigen [41], which is decreased in LAPE cell (Fig. 3A). Additional growth arrest and DNA damage-inducible proteins, including Rad54 and GADD153, were upregulated. Rad54 functions in homologous recombination repair pathways to maintain telomere length [42] and it facilities chromatin remodeling [43], which correlates with increased histone H3 in LAPE cells (Fig. 2). As described above, increased GADD153 expression, which was confirmed TaqMan analysis (Fig. 5B), is induced by growth arrest, DNA damage, and environmental stress. Despite an established role for GADD153 in inducing apoptosis [44,45], and its increased expression in LAPE cells (~45-fold), there was no indication that LAPE cells have increased necrosis or apoptosis compared with LAPN cells.

Consistent with LAPE cells having delayed G2/M entry and progression [5], the array analysis indicated upregulation of genes negatively regulating G2/M, including 14-3-3 θ and Wee1 kinase. Increased expression of Wee1 kinase, which induces inhibitory phosphorylation of Cdc2 on tyrosine 15 [46], was confirmed by TaqMan analysis (Fig. 3D). Moreover, there was a downregulation of genes associated with promoting cell cycle progression, including CDC25 (Fig 3A) and CKS-2 (Fig. 3A), and involved in chromatin assembly, including histone H3 (Fig. 2A). Although cDNA array analysis of asynchronous LAPE cells did not indicate a change in cyclin B1 expression, we found that with a time-dependent release of cells from a double thymidine block, cyclin B1, as indicated by immunoblotting, was significantly downregulated in LAPE cells compared with LAPN cells (Fig. 3C). Hence, loss of NHE1 activity likely decreases the stability of cyclin B1 protein rather than decreasing cyclin B1 gene expression. Decreased cyclin B1 expression and increased Wee1 kinase expression is consistent with our previous finding that Cdc2 kinase activity is inhibited in LAPE cells compared with LAPN cells [5].

A global pattern of metabolic genes differentially regulated in LAPE cells would favor glycolysis and oxidative phosphorylation, possibly in response to reduced ATP (Fig. 4A). Paradoxically, expression of genes encoding enzymes that regulate substrate entry for glycolysis was decreased. Decreased expression of the glucose transporter Glut-4 was confirmed by TaqMan analysis (Fig 4B). Additionally, decreased phosphorylase kinase would favor decreased conformational change of phosphorylase b to phosphorylase a, and reduced glycogen breakdown to glucose 6-phosphate and decreased galactokinase would favor decreased utilization of galactose for glycolysis. Key regulators of glycolytic flux, however, including hexokinase, phosphofusctose kinase, and pyruvate kinase were upregulated in LAPE cells compared with LAPN cells. Increases in hexokinase and phosphofusctose kinase, which catalyze the first and second ATP ultilization steps of glycolysis, respectively, and in pyruvate kinase, which catalyzes the final reaction of glycolysis, would favor increased production of pyruvate. Phosphorylation of fructose 6-phosphate by phosphofusctose kinase is a rate-determining reaction and the activity of phosphofusctose kinase is stimulated by low AMP and inhibited by high ATP and by citrate. In LAPE cells, a decrease in citrate transport protein would likely decrease cytosolic citrate, a negative regulator of phosphofusctose kinase, and indirectly increase phosphofusctose kinase activity. Glycolytic flux and NADH would also be favored by decreased expression of fructose 1,6-bis-phosphate, which limits substrate recycling, and lactate dehydrogenase, which catalyzes the reduction of NADH by pyruvate to yield NAD⁺ and lactate. An established metabolic difference in oncogenic transformed cells compared with normal cells is increased lactic acid production [47] and decreased lactate dehydrogenase in LAPE cells correlates with increased NHE1 activity being necessary for oncogenic transformation [14] and tumor development [15].

Consistent with increased glycolysis and pyruvate production in LAPE cells, key enzymes favoring ATP production by electron transport and oxidative phosphorylation were increased (Fig. 4A). Increases in NADH-ubiquinone oxidoreductase and cytochrome c oxidase would favor oxidation of NADH. Increases in subunits A, C, and D of ATP synthase would increase endergonic synthesis of ATP. ATP synthase in the inner mitochondrial membrane is a proton translocator, extruding protons into the mitochondrial matrix. Whether decreased cytosolic pH in LAPE cells compared with LAPN cells alters the pH of the mitochondrial matrix and proton-electromotive force powering ATP synthesis remains to be determined. Collectively, the profile of gene expression in LAPE cells suggests equilibrium towards increased glycolytic flux and ATP production.

cDNA array analysis indicated that loss of NHE1 activity in LAPE cells was associated with the regulation of a number of genes involved in cytoskeleton organization, cell adhesion, and extracellular matrix assembly. The regulation of several genes correlates with reported effects of NHE1 activity and pH_i on cell shape determination [6], cell polarity [9], actin-filament bundling [9,48], cell-substrate adhesion [9,10] and cell migration and metastasis [9,11,14]. NHE1 acts as an anchor for actin filaments by binding directly members of the ERM (ezrin, radixin, moesin) family of actin binding proteins, and in LAPE cells, expression of NF2 (merlin), a tumor suppressor protein and member of the ERM family, was downregulated. Consistent with a role for NHE1 activity in cell polarity and actin dynamics, loss of NHE1 activity was associated with decreased expression of myosin regulatory light chain, and increased expression of gelsolin (Fig. 5A), a pH-dependent actin severing and capping protein. Although regulation of the microtubule-based cytoskeleton by NHE1 and pH_i have previously not been reported, a number of microtubule-related genes, including Clip 170, KIF4, kinesin light chain, and a dynein heavy chain, were upregulated in LAPE cells.

Consistent with NHE1-dependent cell adhesion and migration, LAPE cells had a marked (~26-fold) decrease in type IV collagenase (MMP-9) expression, and zymography confirmed that activity of MMP-9, but not activity of MMP-4, was selectively decreased in LAPE cells compared with LAPN cells (Fig. 5B). Correlating with a decrease in MMP-9 expression, LAPE cells also had similar marked (~24-fold) decrease in the expression of lipocalin 24p3, which was confirmed by TaqMan analysis (Fig. 5C). NGAL (Neutrophil Gelatinase Associated Lipocalin), the human homolog of mouse lipocalin, is covalently bound to MMP-9 and protects MMP-9 from degradation [49,50]. Lipocalins are transcriptionally regulated by C/EBPβ [51], which is likely suppressed by decreased expression of C/EBPδ (Fig. 2C) and increased expression of GADD153 (Fig. 2D). Moreover, increased expression of MMP-9 [52] and lipocalins [53] is associated with tumor cell growth and invasion, which correlates with a role for NHE1 activity in these processes [14,15].

The generation of LAPN and LAPE cells was as previously described [5]. In brief, NHE1-null LAP1 cells developed from parental Ltk-mouse muscle fibroblasts [16] were used for stable expression of wild-type NHE1 (LAPN cells) or expression of NHE1-E266I containing a single point substitution of glutamate266 for isoleucine that results in complete loss of ion translocation activity (LAPE) [5,6]. Cells were maintained in DMEM supplemented with 10% FCS in the presence of 25 mM NaHCO₃ and 5% CO₂. Total RNA was prepared from cells plated for 48 h by using Qiagen's RNeasy^® midi kit.RNA. RNA was collected from five independent cell platings and used for five separate DNA array hybridizations.

Total RNA was converted to double-stranded cDNA using the SuperScript Choice system (Gibco BRL), except that HPLC-purified T7-(dT)₂₄ oliomer (5'-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG – (dT)₂₄) was used instead of the oligo (dT) or random primers provided with the SuperScript Choice kit. Double-stranded cDNA was collected by ethanol precipitation. Biotinylated cRNA was then generated from the cDNA by an in vitro transcription (IVT) reaction using the ENZO BioArray™ HighYield™ RNA Transcript Labeling Kit. IVT products (cRNA) were collected by using Qiagen's RNeasyR mini kit, then ethanol-precipitated and quantitated. The cRNA was fragmented by alkaline treatment and hybridized to a GeneChip probe array from Affymetrix (Santa Clara, CA). The Affymetrix murine oligonucleotide array (Mu11KSubB) is complementary to ~6,500 murine genes and expressed sequence tags (ESTs). Each gene or EST is represented on the array by 16 – 20 feature pairs. Each feature pair contains a 25-bp oligonucleotide sequence, which is either a perfect match to the gene or a single central-base hommomeric mismatch control.

Affymetrix GeneChip analysis was performed using standard procedures [54]. The expression level of any particular transcript was calculated by subtracting he difference between the fluorescence intensities of the perfect match and mismatch feature pairs and then averaging over the entire probe set (Avg Diff). The Avg Diff value for each transcript was averaged over the five experiments for LAPN and LAPE cells; these average values obtained from five independent hybridizations were then used to calculate fold changes in LAPE cells relative to LAPN cells for each transcript. We did not use comparison algorithms supplied with the Affymetrix software.

Proteins from total cell lysates were separated by SDS-PAGE as previously described [5] and transferred to PVD nitrocellulose membranes for immunoblotting. Antibodies for immunoblotting included Fyn (NeoMarkers), PLCγ-1 and Erk 1–2 (Cell Signaling), GADD45, FEN1, and cyclin B1 (Santa Cruz Biotechnology), and actin (Sigma).

Confirmation of GeneChip data was accomplished using TaqMan™ chemistry with the ABI 7700 Prism real-time PCR instrument (ABI, Foster City CA). Custom primers specific to the genes of interest were synthesized by Life Technologies and TaqMan™ probes for each gene were synthesized by Integrated DNA Technologies, Inc. The forward and reverse primers for mouse GADD153 (GenBank accession no. X67083) were 5'-GAAACGAAGAGGAAGAATCAAAAAC-3' and 5'-ATCTGGAGAGCGAGGGCTTT-3', respectively, and the probe was 5'-FAM/ACCCTGCGTCCCTAGCTTGGCTGAC/TAM-3', corresponding to an amplicon of 122 bp. The forward and reverse primers for mouse Wee1 kinase (GenBank accession no. NM_009516) were 5'-TTGCTCTTGCTCTCACAGTCGT-3' and 5'-TGGGAAAGCACTTGTGGGAT-3', respectively, and the probe was 5'-FAM/CCTTCCCAGAAATGGAGAGCACTGGC/TAM-3', corresponding to an amplicon of 118 bp. The forward and reverse primers for mouse Glut4 (GenBank accession no. NM_009204) were 5'-TGGCCATCTTCTCTGTGGGT-3' and 5'-ATTGGCTAGGCCCATGAGG-3', respectively, and the probe was 5'-FAM/TATGCTGGCCAACAATGTCTTGGCC/TAM-3', corresponding to an amplicon of 138 bp. The forward and reverse primers for mouse 24p3 (GenBank accession no. W13166) were 5'-GGCAGCTTTACGATGTACAGCA-3' and 5'-TCTGATCCAGTAGCGACAGCC-3', respectively, and the probe was 5'-FAM/CATCCTGGTCAGGGACCAGGACCAG/TAM-3', corresponding to an amplicon of 111 bp. For each gene, PCR was conducted in triplicate with 50 μl reaction volumes of 1x PCR buffer A (Applied Biosystems, Foster City, CA), 2.5 mM MgCl₂, 0.4 μM each primer, 200 μM each dNTP, 100 nM probe and 0.025 u/μl Taq Gold (ABI, Foster City CA). For each experiment, a large master mix of the above components was made and aliquoted into each optical reaction tube. Each primer/probe set (5 – 10 μl) was then added, and PCR conducted using the following cycle parameters: 95°C 12 min × 1 cycle, (95°C 20 sec, 60°C 1 min) × 40 cycles. Data analysis was carried out using sequence detection software that calculates the threshold cycle (Ct) for each reaction which is used to quantitate the amount of starting template in the reaction. A difference in Ct values (ΔCt) was calculated for each gene by taking triplicate Ct values from three reactions and subtracting the mean Ct of the triplicates for the control gene, GAPDH, for each cDNA sample at the same concentration. An additional difference in Ct values (ΔCt) was calculated for each gene by taking the triplicate ΔCt values for each gene in the mutant LAPN1-E266I cells and subtracting the mean ΔCt of the triplicates for the wild-type LAPN cells. The relative expression levels were calculated as = 2 ^-ΔΔCt [55].

Relative quantitative RT-PCR was preformed using QuantumRNA™ 18S internal standards from Ambion, Inc. (Austin TX) that included 18S Primers and Competimers™. By optimizing the assay and choosing an appropriate18S Primer:Competimer ratio for each sample, the 18S signal was reduced to the same linear range as that identified empirically for the gene specific product. The amplicon for the 18S primers was 315 bp. Custom primers specific to the genes of interest were synthesized by Life Technologies. The forward and reverse primers for calpactin I light chain (GenBank accession no. M16465) were 5'-GTGGACAAAATAATGAAGGAC-3' and 3'-ACAAGAAGCAGTGGGGCAGAT-5', respectively, corresponding to an amplicon of 222 bp. The forward and reverse primers for CEBPδ (GenBank accession no. NM_007679.1) were 5'-ATACCTCAGACCCCGACAGCG-3' and 3'-CAAAAGTCTGTCGGAAATGTC-5', respectively, corresponding to an amplicon of 220 bp. Total RNA was isolated from LAPN1 and LAPN1-E266I cells in four separate experiments, using Qiagen's RNeasy^® midi kit. Reverse transcription of total RNA from each sample was carried out using random decamers and the RETROscript™ kit from Ambion. RT reactions were then subjected to PCR using the gene specific primers above (final concentration of 0.4 uM each), the appropriate 18S

Primer:Competimer ratio and 10uCi/ul [α-³²P]dCTP for labeling. PCR was conducted using the following cycle parameters: (94°C 30 sec, 57°C 30 sec, 72°C 30 sec) × 21 cycles for CEBPδ and 19 cycles for calpactin I light chain. Empirically derived 18S Primer:Competimer ratios were 1:18 for CEBPδ and 2:8 for calpactin I light chain.

Activity of type IV collagenase (MMP9) was determined by zymography, as previously described [56].