Overexpression of the G-Protein Inwardly Rectifying Potassium Channel 1 (GIRK1) in Primary Breast Carcinomas Correlates with Axillary Lymph Node Metastasis¹

Over the course of the last two decades, numerous genetic changes have been described that are associated with the incidence of cancer in many tissue types. The study of genetic changes associated with cancer generally involves two broad experimental approaches. The first approach is to identify mutations in the genomic DNA that may ultimately lead to a malignancy. The second approach involves investigating tumor-related changes in gene expression, known as expression profiling, by either measuring mRNA or protein expression directly.

Various germ-line mutations have been identified in malignant tissue,including breast cancer. Mutations in the gene that codes for the protein p53 are known to cause changes in cell cycle regulation(1, 2) and apoptosis (3), thus leading to uncontrolled cell growth and malignancy. p53 mutations affect these cellular processes, in part, by regulating the transcription of the genes Fas/APO1 (apoptosis), Tsp1(antiangiogenesis), IGF-BP3 (apoptosis), and others(reviewed in Ref. 4). Therefore, germ-line mutations can alter multiple downstream biological pathways that regulate tumor biology. Germ-line mutations in the genes BRCA1 and BRCA2 are associated with increased incidences of breast cancer, although the mechanism is less clear (5, 6). Although 45–50% of women with BRCA1/2 mutations present with breast cancer at some point in life, these mutations are associated with only 7% of all incidences of breast cancer(7).

Although identification of germ-line mutations is important to understand the underlying process of carcinogenesis, identification of“downstream” targets is crucial to understanding the biological pathway(s) and gene products that are involved in controlling the transition from normal cells to malignancy and finally to metastatic disease. A strategy of expression profiling, coupled with functional genomics, must be used to discover genes that may be exploited as biomarkers and therapeutic targets. Using this strategy, a panel of genes can be discovered that can be used to tailor therapies for a particular individual’s cancer, the modern approach to drug discovery and molecular medicine.

Numerous investigations have been performed to discover altered gene expression in primary breast carcinomas. Examples of these include the Her2/neu oncogene, cyclin D1, and bcl-2. Overexpression of Her2/neu has been shown to be a marker for many types of cancers, including breast cancer, and is a prognostic indicator for survival as well as a target molecule for adjuvant therapy (8, 9). Cyclin D1overexpression correlates with estrogen-receptor positive,well-differentiated tumors (10, 11) and lobular-type breast carcinomas (12). Bcl-2 overexpression is associated with a metastatic phenotype and may play a role in regulating apoptosis(13).

The primary focus of this study was to use an expression profiling strategy,RDA³(14), to discover genes that are differentially expressed in a primary, invasive breast carcinoma as compared with adjacent normal breast tissue from the same individual. We then sought to determine whether these differentially expressed genes were overexpressed in other primary breast carcinomas and to identify pathological characteristics that may be common among those breast cancers that overexpress these mRNAs. Using this experimental strategy,we identified an overexpressed mRNA that encodes a G-protein coupled inwardly rectifying potassium channel (GIRK1) that is overexpressed in ∼30% of invasive breast cancers and the expression of which significantly correlates with the presence of LNM.

Breast tissue, including both tumor and benign normal, were obtained from fresh surgical specimens after gross pathological examination. Surgical consent forms (approved by the Faulkner Hospital institutional review board) were signed by the patient before surgery to allow the use of a portion of the tissue for research purposes. The breast tissue was wrapped in foil, frozen on dry ice, and placed in liquid nitrogen for long-term storage. RNA isolation was performed using the RNAeasy Mini kit (Qiagen, Valencia, CA). Small pieces of tissue (∼10 mg) were cut from the frozen tissue sample, placed into lysis buffer, and homogenized using a Polytron tissue homogenizer (Kinematic, Basal,Switzerland), and total RNA was isolated according to the manufacturer’s standard protocol.

The RDA experiments were performed essentially as described by Lisitsyn and Wigler (14) with slight modifications using mRNA(cDNA) as the RDA targets. Double-stranded cDNA was synthesized from total RNA as described previously (15). cDNA from benign(“driver”) and malignant (“tester”) breast tissues were then digested with the restriction enzyme Sau3A (Life Technologies, Inc., Rockville, MD), ligated to a specific BamHI primer set (Primer Set I) using T4 ligase (New England Biolabs, Cambridge, MA), and amplified using Taq polymerase to produce amplicons. Primer Set I adapters were removed from the designated tester amplicons by Sau3A digestion and ligated to Primer Set II (specific BamHI primer set, unique in sequence compared with Primer Set I) as above. Tester cDNA amplicons (200 ng)were mixed with driver cDNA amplicons (10 μg), lyophilized to dryness, and resuspended in 4 μl of 3 × EPPS buffer(30 mm EPPS, 3 mm EDTA). Hybridization parameters were 98°C for 3.5 min and 67°C for 20 h. After hybridization, the RDA products were subjected to 10 cycles of PCR using Primer Set II (94°C × 15 s,70°C × 1 min; auto extension, 1 min/cycle) and then treated with mung bean nuclease (New England Biolabs) to eliminate single-stranded cDNA. A further round of PCR amplification for 19 cycles (94°C × 15 s, 70°C × 2.15 min; auto extension, 5 s/cycle) using Primer Set II was then performed. The first-round RDA products were then purified (Qiaquick PCR Spin Columns; Qiagen). A second round of RDA was performed by changing the adapter primers (Primer Set III) on the first-round RDA product using Sau3A and T4 ligase (New England Biolabs) as described above. The first-round RDA product (12.5 ng) was then hybridized to the driver amplicons (10 μg) as described above. PCR amplification of the second-round hybridization product was performed as above using Primer Set III.

Second-round RDA products were digested with Sau3A to remove the primer adapters, ligated into a plasmid cloning vector (pZErO-2;Invitrogen, Carlsbad, CA), transformed in TOP-10 competent Escherichia coli (One Shot kit; Invitrogen), and grown at 37°C on kanamycin-containing Luria broth agar plates. Clones were randomly selected and grown overnight in liquid Luria broth with kanamycin at 37°C. Plasmid DNA was then isolated (QIAwell 8 Ultra Plasmid kit; Qiagen), and cDNA inserts were PCR amplified and sequenced using an ABI 310 sequencer (Applied Biosystems, Foster City, CA) and subjected to BLAST searches for initial identification.

Initial screening of a sequenced gene fragment revealed differential expression of an EST, clone 30T2-7. We subsequently identified this fragment as part of a G-protein coupled inwardly rectifying potassium channel, GIRK1 (see below). Therefore, hereafter we will refer to the 30T2-7 fragment as GIRK1. We isolated total RNA from breast tissues as described above. Single-stranded cDNA was synthesized from total RNA using an oligo d(T) primer (Life Technologies) as described previously (15). To quantify GIRK1 expression in breast tissues, we initially used competitive PCR [PCR MIMICs (16)].

We synthesized primers that amplified a ∼400-bp fragment of β-actin and used a β-actin MIMIC to normalize the cDNA concentrations. Theβ-actin MIMIC was synthesized by first PCR amplifying an arbitrary 300-bp DNA fragment. This fragment was then PCR amplified using chimeric primers (β-actin sequences on the 5′ ends) to generate a larger β-actin amplicon as compared with the endogenous β-actin amplicon. PCR conditions were: 1× buffer, 2 mm MgCl, 200μ m deoxynucleotide triphosphates, 200 nmprimer, cDNA fragment, 0.5 unit Taq Gold. The PCR temperature profile for MIMIC amplification was 94°C × 15 s,54°C × 30 s, and 72°C × 30 s for 25 cycles. The MIMIC fragment was purified (Qiaquick;Qiagen) and quantified.

Next, we normalized the cDNA to β-actin expression. We titrated the volume of each tumor cDNA (constant β-actin MIMIC concentration) for each sample in multiple PCR reactions until the endogenous β-actin amplicon and the MIMIC β-actin amplicon were equally amplified in all cDNAs (data not shown). In subsequent reactions, the same volume of cDNA was used. Primer sequences for amplification of the endogenousβ-actin and β-actin MIMIC amplicon were forward 5′-GCGTTACACCCTTTCTTGACAAAACC-3′ and reverse 5′-AGCAATGCTATCACCTCCCCTGTG-3′.

We then synthesized a MIMIC for GIRK1 (5′ GIRK1primer sequences, 3′ β-actin sequences), as described above. This PCR reaction generated an amplicon larger than the endogenous GIRK1 amplicon, but with GIRK1 primer-specific sequences on both the 5′ and 3′ ends. The primer sequences for GIRK1 are forward 5′-TTGTCTCTTCCCCACTCTCCTTCTGG-3′ and reverse 5′-GACGGAACCAAAAACAAACTCTCCAAG-3′.

Breast tissue cDNAs were analyzed for GIRK1 expression in semiquantitative MIMIC PCR reactions using the volumes of cDNAs determined above in the β-actin MIMIC experiments. Pilot PCR reactions were performed to determine the concentration of GIRK1 MIMIC that showed equal amplification of GIRK1 and MIMIC in one tumor cDNA. The conditions for the competitive PCR reactions were as follows: 1× buffer, 2 mm MgCl, 200 μmdeoxynucleotide triphosphates, 400 nm forward and reverse GIRK1 primers, 0.25 unit Taq Gold, DNA, and 160 nm GIRK1 MIMIC, final concentrations. The PCR temperature profile was 94°C × 10 min to activate the Taq Gold, then 94°C × 15 s,54°C × 30 s, and 72°C × 30 s for 35 cycles.

PCR products were then fractionated at 140 V for 30 min on a 2%agarose gel containing ethidium bromide and analyzed using an Eagle Eye Gel Documentation System (Stratagene, Cedarville, TX). Tissues in which the GIRK1 PCR product relative band intensity was greater than the MIMIC PCR product band intensity were identified as overexpressing GIRK1 mRNA. Tissues in which only the MIMIC was amplified and little or no GIRK1 amplification was observed were identified as not overexpressing GIRK1 mRNA.

RACE reactions were performed to characterize the 30T2-7 clone further. RACE reactions were performed using a 5′ RACE kit (Life Technologies,Inc.). 3′ RACE fragments for GIRK1 were generated using a 5′primer corresponding to the original GIRK1 EST and an oligo d(T) primer (Life Technologies). After PCR amplification, the fragments were purified (Qiaquick; Qiagen) and sequenced as described above. Sequence analysis, including alignment and identification of contiguous, overlapping fragments, were performed using Sequencher software (Gene Codes Corp., Ann Arbor, MI).

To quantify GIRK1 expression in multiple breast tissues, we used an ABI Prism 7700 Sequence Detection System (Applied Biosystems)that uses TaqMan chemistry (17). Total RNA was isolated as above except that before RNA elution, DNase I (27 units) was added to the column to eliminate genomic DNA contamination (RNAeasy; Qiagen). cDNA was synthesized in a 50-μl reaction using a TaqMan Reverse Transcription kit, following the manufacturer’s recommended protocol(Applied Biosystems).

To normalize GIRK1 expression relative to tissue epithelial cDNA, we used primers and a TaqMan probe corresponding to cytokeratin 19. Expression of GIRK1 was measured relative to the epithelial cell-specific gene, CK19. CK19 has been shown to be specific for breast luminal epithelial cells (both normal and malignant; Ref. 18) and is used routinely as a marker for the presence of epithelial-derived metastatic carcinoma(19, 20, 21, 22). Because each experimental sample has an endogenous reference standard (CK19), correction for the amount of cDNA added to any individual reaction is not needed. Primers and the TaqMan probe to GIRK1 and CK19 were designed by using Primer Express software (Applied Biosystems) and synthesized by The Oligo Factory (Applied Biosystems). The GIRK1 amplicon was generated in the region of the original 30T2-7 fragment obtained in the initial RDA experiment. TaqMan probes were labeled with FAM on the 5′ end and TAMARA on the 3′ end.

Primer and probe sequences for GIRK1 are forward 5′-GGGATTTGGACATGGCTAAGTC-3′, reverse 5′-GGCCTGTTTTCATTCTCTTAACTGATAC-3′, and TaqMan probe 5′-FAM-AAGGCCATTGCTCACCATGAACACTGT-TAMARA-3′. Primer and probe sequences for CK19 are forward 5′-CCTCATATTGGCTTCGCATG-3′, reverse 5′-AGGTCAGTGTGGAGGTGGATTC-3′, and TaqMan probe 5′-FAM-CACTCAGGATCTTGGCGAGATCGGTG-TAMARA-3′.

Each 25-μl PCR reaction for GIRK1 consisted of 300 nm forward primer, 300 nmreverse primer, 125 nm TaqMan probe, cDNA (10 ng RNA equivalents), and 1× (final concentration) TaqMan Universal Master Mix (Applied Biosystems). Identical but separate TaqMan PCR reactions were performed for CK19 expression for each experimental tissue except that 900 nm forward, 900 nmreverse, and 125 nm TaqMan probe were used. Optimal primer, probe, and cDNA concentrations were determined in a separate set of experiments to insure that both CK19 and GIRK1 fragments were amplified with equal efficiency (data not shown). All reactions were done in duplicate. A no-template control PCR reaction was added to each experiment to insure that no amplification of fragments occurred in the absence of added cDNA. PCR parameters were 50°C × 2 min, 95°C × 10 min, and 40 cycles at 95°C × 10 s,60°C × 1 min.

We used the TaqMan software to calculate a Ct value for each reaction, where the Ct value is the point in the extension phase of the PCR reaction that the product is distinguishable from the background. The Ct value can be obtained from any portion of the logarithmic phase of the PCR curve; the ΔCtvalues will remain consistent if calculated from any part of the curve(23). We normalized the Ct values for GIRK1 amplification by subtracting the Ct value calculated for CK19 from the same tissue to obtain a ΔCt using the following equation: Ct_GIRK1 − Ct_CK19 = ΔCt.

To quantify GIRK1 expression in each experimental tissue and calculate fold changes in GIRK1 expression, a calibrator was used. By using GIRK1 and CK19 fragments of known concentrations, consistent measurements of GIRK1 expression can be obtained between experiments. We PCR amplified both the GIRK1 and CK19 fragments in a preparative PCR reaction using the forward and reverse primer sequences shown above. The GIRK1 and CK19 amplicons were purified and quantified.

For each TaqMan experiment, both GIRK1 and CK19 calibrator fragments were amplified in separate tubes. The calibrator ΔCt value was subtracted from the experimental ΔCt values to calculate the fold change in GIRK1 expression in each experimental tissue: ΔCt_experimental − ΔCt_calibrator = ΔΔCt. Fold increases in GIRK1expression were calculated by the following equation according to ABI User Bulletin #2:2^-ΔΔ^Ct = fold change in expression.

Descriptive statistics (mean, SD, and SE), fold change calculations, T statistics, and χ²analysis were performed using Excel software. A returned P < 0.05 was considered to be a significant statistical correlation between tumor characteristics and GIRK1 expression.

We performed the RDA procedure using a matched normal-tumor tissue pair obtained from the same patient undergoing surgical treatment for breast cancer at our institution. The RDA experiment yielded several mRNAs that were overexpressed in the primary tumor but not in the normal tissue (data not shown). We chose to focus on one clone in particular,30T2-7, because this fragment was initially identified as an unknown EST after undergoing initial subcloning and sequencing. To determine whether 30T2-7 was indeed a unique sequence or part of an unidentified fragment of a gene identified previously, we performed 5′ and 3′ RACE PCR to generate longer, overlapping fragments. 3′ RACE experiments identified a poly(A) tail 230 bp 3′ from the original 30T2-7 EST,indicating a putative 3′ end of a gene. 5′ RACE RT-PCR identified cDNAs that had overlapping sequence with the 3′ end of a previously cloned gene, a 501-amino acid G-protein coupled inwardly rectifying potassium channel (GIRK1). Additional RACE experiments identified a∼1.5-kb cDNA (mRNA) sequence 3′ to the original GIRK1(including the original 30T2-7 fragment) sequence published previously(Fig. 1; Ref. 24). We then sequenced cDNA fragments from breast tissue that corresponded to the deposited GIRK1 sequence obtained from human brain tissue (including the open reading frame) and confirmed its presence in breast tissue. In addition, we confirmed the presence of the breast-derived 30T2-7 fragment in human brain cDNA(data not shown).

Fig. 2 shows representative semiquantitative competitive RT-PCR data for GIRK1 expression in matched normal and tumor. Increased endogenous GIRK1 PCR product band intensity relative to the exogenous MIMIC product band intensity (see “Materials and Methods”) indicates increased GIRK1 expression in a particular tissue. The data indicate that GIRK1 is expressed at low levels in normal breast tissue (Fig. 2, Lane 3). GIRK1 is highly overexpressed in tumor tissue (Fig. 2, Lane 5) associated with LNM; GIRK1 is not overexpressed in matched adjacent normal tissue (Fig. 2, Lane 4). No overexpression of GIRK1 is measured in a tumor not associated with LNM (Fig. 2, Lane 7) nor in the matched, adjacent normal tissue (Fig. 2, Lane 6). PCR reactions with MIMIC or cDNA alone (Fig. 2, Lanes 1 and 2, respectively) indicate no contamination of MIMIC or GIRK1 product occurred in the PCR master mix. Further semiquantitative competitive RT-PCR experiments revealed that 6 of 6 normal breast tissues did not overexpress GIRK1, and 5 of 13 primary invasive breast tumors overexpressed GIRK1 (data not shown).

Upon closer inspection of the patient final diagnosis, we noticed a clear association between those tumors that overexpressed GIRK1 and those tumors associated with LNM. The cursory observation that overexpression of GIRK1 may be related to LNM led us to perform a more rigorous comparison between mRNA expression and pathological variables from 56 separate invasive breast carcinomas and 6 normal, nonmalignant breast tissues. We used TaqMan PCR to quantify the target gene, GIRK1, and the reference gene, CK19, in these additional tissues to identify those tumors that overexpress GIRK1. Ct values (see“Materials and Methods”) obtained for CK19 were consistent for each tissue type tested (normal Ct, 22.8 ± 1.2;tumor Ct, 23.3 ± 1.8), indicating no significant differences in amounts of CK19 mRNA between any of the tissue types (P = 0.30).

Because we did not have access to matched normal/tumor pairs for each tumor studied, we compared the ΔCt values(GIRK1-CK19) for each tissue tested (normal and tumor) to the ΔCt values of the calibrator fragments to maintain consistent results (see “Materials and Methods”). Cutoff values for GIRK1 overexpression in the tumor specimens were set at three SDs above the mean fold increase in GIRK1expression measured in the normal breast tissue (fold increase compared with the calibrator fragments, 3.95 ± 5.93; n = 6). Therefore, fold increases in GIRK1 expression of ≥21.75 in the tumor specimens were considered to be overexpressed. Shown in Table 1 is the measured GIRK1 expression, 95% confidence intervals for each data point, and the pathological characteristics for each tumor studied. Comparison of the TaqMan data and representative competitive MIMIC PCR data shown in Fig. 2 indicate good agreement between the two quantitative methods.

We then compared LNM, tumor size, tumor grade, and estrogen receptor expression with the presence and absence of GIRK1overexpression. Using χ² statistical analysis,we found that only the presence of LNM (P < 0.0029) significantly correlated with GIRK1 overexpression(Table 2). GIRK1 was overexpressed in 50% (13 of 26) of the lymph node-positive tissues tested. In contrast, GIRK1 was overexpressed only in 13% (4 of 30) of the lymph node-negative tissues tested (Fig. 3,A). Of the 19 tissues tested with >1 positive lymph node,58% showed overexpression of GIRK1. Shown in Fig. 3 B is the mean fold increase in GIRK1 expression(± SE) for all normal and tumor tissues (normal tissue, 3.95 ± 0.99; 0 lymph nodes, 10.4 ± 0.98; 1 lymph node,50.1 ± 3.7; >1 lymph node, 119.9 ± 3.3).

Total lymph nodes dissected from GIRK1-negative cases were not significantly different from GIRK1-positive cases(GIRK1-negative, mean 14.4 ± 0.44; GIRK1-positive, mean 12.7 ± 0.5; P = 0.19).

We measured GIRK1 expression in four primary tumors and an associated lymph node from each case. The data shown in Table 3 indicate that GIRK1 is overexpressed in the corresponding lymph node when the tumor overexpresses GIRK1; GIRK1 is not overexpressed in the corresponding lymph node when the primary tumor is negative for GIRK1 overexpression.

We show in this study that GIRK1 overexpression in primary invasive breast carcinomas correlates with LNM, and that the GIRK1 gene is comprised of an additional 1.5 kb of 3′ UTR. Although the mean fold change in GIRK1 expression is variable in the tumors tested (see Table 1), GIRK1overexpression follows a general trend of increasing expression if more than one node is involved (as compared with zero or one node).

We have shown data that indicate that GIRK1 is differentially expressed by measuring the expression of a portion of the 3′ UTR of the gene. We have also investigated the differential expression of a portion of the ORF, although not as extensively as the 3′-UTR. Preliminary investigations indicate that the expression of the ORF closely mirrors that of the 3′-UTR; tissues that overexpress the 3′-UTR overexpress the ORF as well (data not shown). These data would indicate that the entire gene is differentially expressed as opposed to just the ORF or the 3′-UTR.

The data shown in Table 3 indicate that GIRK1 is not overexpressed in fully replaced lymph nodes from patients whose primary cancer does not overexpress GIRK1. Although we have limited cases where the corresponding negative or positive lymph nodes were available for study, the data provide insight with regard to the absolute necessity of GIRK1 overexpression for LNM to occur and the localization of GIRK1 expression:

(a) This result would suggest that overexpression of GIRK1 alone is not sufficient for metastasis to occur in all breast cancers. However, whether GIRK1 plays a role in the metastatic mechanism of the subset of invasive tumors showing overexpression and nodal metastasis is speculative but intriguing,especially in view of the strong correlation between GIRK1overexpression and the presence of multiple lymph nodes positive for metastatic carcinoma.

(b) The data would indicate that GIRK1 expression may be confined to the epithelial cells of the breast carcinoma. As shown by careful pathological examination, the positive lymph nodes studied were fully replaced by metastatic tumor (i.e., only tumor epithelial cells were observed). Therefore, it is highly probable that GIRK1 could only be expressed by the epithelial component and not the stromal component of the breast. Additional studies aimed toward synthesizing an antibody to GIRK1 for immunochemistry studies or in situ hybridization will clarify this issue.

These data presented in this study indicate that the identification of GIRK1 may be useful as: (a) a biomarker for LNM of breast cancer; and (b) a therapeutic target for inhibition of further metastasis. Complications may occur as a result of axillary lymph node dissections, especially in conjunction with radiation therapy (25). An accurate, predictive test using a panel of biomarkers could possibly eliminate the need for axillary dissections in the majority of women with breast cancer. Numerous studies have identified genes, the function of which may be necessary for cancer metastasis, including integrins, cadherins, growth factors,and matrix metalloproteinases (reviewed in Ref. 26). Overexpression of the H-mts1 gene concomitant with underexpression of nm23, both of which alter cytoskeletal dynamics, has been shown to correlate with increasing nodal metastasis and nodal spread of the primary breast cancer (27). Identification and correlation of these and other markers with LNM, including GIRK1, may be a useful surrogate for lymph node dissections and have predictive value as to the likely occurrence of distant metastasis.

Further studies aimed toward determining whether GIRK1 plays a mechanistic role in LNM is warranted in light of these data to determine whether GIRK1 may indeed be exploited as a pharmaceutical target. GIRK1 is a G-protein coupled inwardly rectifying potassium channel that is part of a family of inwardly rectifying potassium channels identified in both brain and heart tissue that regulate membrane potential (24). Numerous studies have been performed to determine a function for this class of ion channel. These classes of potassium channels form heterodimers to become functional (i.e., GIRK1/GIRK4), act to conduct potassium ions into the cell rather than out of the cell(thereby hyperpolarizing the cell membrane; Ref. 28), and play a role in maintaining the resting membrane potential(29). Indeed, human breast carcinomas exhibit hyperpolarized membrane potential as compared with the surrounding,morphologically normal breast epithelium or nonmalignant, benign fibroadenomas (30).

GIRK1 may also play a role in cell proliferation. The GIRK family of channels have been shown to functionally interact with membrane-bound integrins that interact with the cytoskeletal matrix of the cell (31). In heart tissue, GIRK1 channel activity is directly activated by βΔsubunits of G-proteins after agonist stimulation of the m2 muscarinic acetylcholine receptor (32, 33). Muscarinic receptors, as well as other G-protein-coupled receptors, can induce cell proliferation and transform NIH3T3 cells via a ras-dependent pathway(34, 35, 36, 37, 38). Because >50% of the metastatic tumors we studied showed mRNA overexpression, GIRK1 could be a candidate for a pharmaceutical target, depending upon further functional studies.

We have shown the utility of the new model of target discovery and that a strategy of expression profiling by RDA can reveal the expression of genes that are related to specific traits of an individuals breast cancer, the first step in the process of functional genomics. Future studies designed for target validation will provide insight into the suitability of GIRK1 as a novel target for pharmaceutical intervention.

We thank other members of Faulkner Hospital Pathology Department, especially Drs. Stephen Pochebit and Becky Mosher, for contributions to this work, including the acquisition of human tissue specimens and medical expertise.