Therapeutic, Clinicopathological, and Molecular Correlates of PRKACA Expression in Gastrointestinal Cancers

Abstract

Background/Objectives: PRKACA alterations have clear diagnostic and biological roles in the fibrolamellar variant of hepatocellular carcinoma and a potential predictive role in that cancer type. However, the roles of PRKACA have not been comprehensively examined in gastric and colorectal cancers (GC and CRC). This study, therefore, sought to investigate the roles of PRKACA expression in GC and CRC. Methods: The clinico-genomic data of 441 GC and 629 CRC cases were analyzed for therapeutic, clinicopathological, and biological correlates using appropriate bioinformatics and statistical tools. Furthermore, the deregulation of PRKACA expression in GC and CRC was investigated using correlative and regression analyses. Results: The results showed that PRKACA expression subsets were enriched for gene targets of chemotherapeutics, tyrosine kinase, and β-adrenergic inhibitors. Moreover, high PRKACA expression was associated with adverse clinicopathological and genomic features of GC and CRC. Gene Ontology Enrichment Analysis also showed that PRKACA-high subsets of the GI cancers were enriched for the biological and molecular functions that are associated with cell motility, invasion, and metastasis but not cell proliferation. Finally, multiple regression analyses identified multiple methylation loci, transcription factors, miRNA species, and PRKACA copy number changes that deregulated PRKACA expression in GC and CRC. Conclusions: This study has identified potential predictive and clinicopathological roles for PRKACA expression in GI cancers and has added to the growing body of knowledge on the deregulation of PRKACA in cancer.

1. Introduction

Gastrointestinal (GI) cancers remain important public health concerns despite the huge number of resources that have been expended in research efforts to decipher the clinical, molecular, and biological characteristics of these cancers [1,2,3]. Gastric and colorectal cancers (GC and CRC), as a group, comprise the commonest malignancies worldwide ahead of the female breast, lung, and prostate cancers, accounting for over 3.0 million (15.33% of all malignancies) new cases [4,5] and constitute the 2nd commonest cause of cancer deaths with 1,703,966 (16.76% of all malignancies) new deaths [4,5]. An improved understanding of the tumor biology of GI cancers will inform innovative and enhanced strategies for cancer management [6,7,8]. Hence, new research that illuminates GI carcinogenesis and progression is warranted.

This study investigated the potential therapeutic significance of PRKACA expression in GC and CRC, as well as its clinicopathological and molecular correlates. PRKACA encodes the catalytic subunit-α of PKA, a kinase enzyme that functions downstream of the β-receptors in the β-adrenergic signaling pathway [9,10,11,12]. PKA phosphorylates multiple downstream targets in the β-adrenergic signaling pathway, including transcription factors ATF, CREB, GATA1, STAT3, Src, and BARK [9,11,12]. However, the roles of PRKACA in cancer are incompletely understood, inasmuch as the cyclic adenosine monophosphate (cAMP)-PKA signaling has been shown to possess both tumor-suppressive and oncogenic roles in different tumor types and contexts [12]. A preclinical study demonstrated that crotonylation of PRKACA enhances the activity of PKA and thereby promotes colorectal cancer development via the PKA-FAK-AKT pathway, whereas PRKACA is secreted by prostate cancer cells [11]. In clinical cancer, PRKACA participates in a fusion event with DNAJB1 in up to a hundred percent of fibrolamellar hepatocellular carcinoma and has therefore been identified as a suitable diagnostic marker for and a driver of oncogenic activities in that cancer [13,14,15]. In addition, PRKACA was found to be upregulated in the serum and tumor of gastric and colorectal cancer patients but did not show any clinical correlates in a study that was markedly limited by sample sizes [16]. The therapeutic significance of PRKACA expression in cancer had previously been suggested by a breast cancer study, which showed that PRKACA mediates resistance to anti-HER2 therapy in breast cancer cell lines via inactivation of BAD and Bcl2-associated death promoter [17]. Indeed, elevated PRKACA expression was found in trastuzumab-resistant breast cancer patients, evidence that PRKACA is activated in trastuzumab-resistant breast cancer cases [16]. With respect to therapy, PRKACA could be a target for the treatment of cancer. For example, it has been suggested that the PAK signaling and PRKACA activities could be inhibited upstream using β2 receptor blockade with propranolol as a strategy for the treatment of invasive cancer [9]. In addition, Schalm et al. and Bauer et al. [14,18] demonstrated that PRKACA—in the DNAIJ-PRKACA fusion context—is a viable target for the treatment of the PRKACA-dependent fibrolamellar variant of hepatocellular carcinoma. However, the biological and therapeutic significances of PRKACA expression in clinical GC and CRC have not been thoroughly interrogated. Therefore, the aims of this study include (i) to interrogate the potential therapeutic significance of PRKACA in clinical GC and CRC, (ii) to investigate the clinicopathological and genomic correlates of PRKACA expression in GC and CRC, (iii) to investigate the biological significance of PRKACA expression in GC and CRC, (iv) to investigate the mechanisms of PRKACA deregulation in GI cancers.

2. Results

2.1. Enrichment of Gene Targets of Multiple Drug Agents in PRKACA-Low and -High Cancers

2.1.1. Gastric Cancer Cohort

Normalized PRKACA expression values were dichotomized using the mean PRKACA values for each cohort. There were 206 PRKACA-low (50.0%) and 206 PRKACA-high (50.0%) GC cases. GSEA followed by drug set annotations on Enrichr (https://maayanlab.cloud/Enrichr/enrich, accessed on 25 July 2024) revealed enrichment of genes that are targets of multiple drugs, including tyrosine kinase inhibitors, β-blockers, and conventional chemotherapeutic agents, among other drug classes, in both the high and low PRKACA subsets, and an adjusted p value of ≤0.05 and false discovery rate (FDR) of ≤0.05 (Figure 1). Gene targets enriched in the PRKACA-high subset include those for conventional chemotherapeutic agents such as paclitaxel, irinotecan, camptothecin, ifosfamide, and methotrexate; tyrosine kinase inhibitors sorafenib, dasatinib, axitinib, gefitinib; β-blockers such as atenolol, and metoprolol; and calcium channel blockers, such as dexverapamil, verapamil and diltiazem (Supplementary Materials Drug-targets_PRKACA_high_GC). PRKACA-low GC subset, on the other hand, shows significant enrichment of targets that are downregulated following treatment with chemotherapeutic agents such as irinotecan, camptothecin, and daunorubicin (Figure 1). Furthermore, the PRKACA-low subset was enriched for gene targets associated with tyrosine kinase inhibitors such as momelotinib, ponatinib, sunitinib, dasatinib, sorafenib, vandetanib, among many others (Supplementary Materials Drug-targets_PRKACA_low_GC). However, it was observed that one phenotype (PRKACA-high, for example) might be associated with both upregulation and downregulation of the gene set of the same drug, depending on the library that was interrogated.

2.1.2. Colorectal Cancer Cohort

Normalized PRKACA expressions were dichotomized into PRKACA-low (N = 268, 50%) and PRKACA-high subsets (N = 268, 50%). Significant enrichment of gene targets of multiple drugs was observed in both PRKACA subsets in the CRC cohort. Similar to the GC cohort, PRKACA expression in the CRC cohort exhibited enrichment of genes that are associated with β-blockers, conventional chemotherapeutics, and tyrosine kinase inhibitors (Figure 1). Furthermore, as was observed in the GC cohort, the pattern of upregulation or downregulation of drug targets differs with the gene sets or libraries interrogated. For example, in the PRKACA-high subset, irinotecan gene sets showed both downregulated or upregulated targets depending on the library that was interrogated. This finding may have resulted from the differences in the biology of the cancer cell lines (e.g., breast cancer (MCF7) vs. leukemia (HL60) cell lines) that were utilized to generate the gene set libraries (Supplementary Material Drug-targets_PRKACA_high_CRC and Supplementary Material Drug-targets_PRKACA_low_CRC).

2.2. Clinicopathological and Molecular Features of PRKACA Expression

2.2.1. Gastric Cancer Cohort

The clinicopathological and molecular correlates of PRKACA expression were interrogated using the Chi-square test. The results showed that in the GC cohort, high PRKACA expression was significantly associated with the 60-years-and-below age group, higher nodal stage, and late disease stage but not with gender, histological tumor type, pathological tumor stage, or distant metastasis (

Table 1. Clinicopathological correlates of PRKACA expression in gastric and colorectal cancers.

PRKACA Expression
Clinicopathological Features		Low PRKACA	High PRKACA	Total	X2 *	p Value	Adjusted p Value
Gastric cancer cohort
Age Group	30–60 yrs	52	79	131	7.630	0.006	0.016
	61–90 yrs	150	126	276
	Total	202	205	407
Gender	Male	137	128	265	0.857	0.355	0.441
	Female	69	78	147
	Total	206	206	412
Race/Ethnicity	White	126	133	259	2.609	0.112	0.179
	Native Americans	1	0	1
	Black	10	2	12
	Asians	31	56	87
	Total	168	191	359
Pathological tumor stage	pT1	13	9	22	0.753	0.386	0.441
	pT2	38	49	87
	pT3	85	96	181
	pT4	63	51	114
	Total	199	205	404
Pathological nodal stage	N0	46	77	123	9.775	0.002	0.008
	N1	54	55	109
	N2	48	31	79
	N3	46	36	82
	Total	194	199	393
Pathological metastasis stage	M0	183	181	364	0.280	0.596	0.596
	M1 and above	15	12	27
	Total	198	193	391
TNM stage	Early Stage	45	81	126	13.226	<0.001	0.002
	Late Stage	152	123	275
	Total	197	204	401
Histological type of gastric cancer	Diffuse type Adenocarcinoma	10	3	13	3.892	0.049	0.098
	Intestinal type Adenocarcinoma	196	203	399
	Total	206	206	412
Colorectal cancer cohort
Age Group	31–60 yrs	70	103	173	9.295	0.002	0.007
	61–90 yrs	198	165	363
	Total	268	268	536
Gender	Male	141	142	283	0.007	0.931	0.931
	Female	127	126	253
	Total	268	268	536
Race/Ethnicity	White	116	167	283	3.876	0.049	0.106
	Asian/Native American	7	6	13
	Black	34	29	63
	Total	157	202	359
Pathological tumor stage	pTis and pT1	5	10	15	0.120	0.729	0.790
	pT2	55	33	88
	pT3	172	196	368
	pT4	36	29	65
	Total	268	268	536
Pathological nodal stage	N0	170	134	304	7.160	0.007	0.018
	N1	55	76	131
	N2	43	56	99
	Total	268	266	534
Pathological metastasis stage	M0	202	192	394	0.272	0.602	0.711
	M1	35	38	73
	Total	237	230	467
TNM stage	Early Stage	153	129	282	3.956	0.0467	0.106
	Late Stage	115	137	252
	Total	268	266	534
Histological type of colorectal cancer	Adenocarcinoma NOS	225	240	465	3.452	0.063	0.117
	Mucinous Adenocarcinoma	40	26	66
	Total	265	266	531
Colonic tumor site 1	Right colon	164	92	256	23.553	<0.001	<0.001
	Left colon	92	128	220
	Total	256	220	476
Primary tumor site	Colon	268	177	445	108.404	<0.001	<0.001
	Rectum	0	90	90
	Total	268	267	535
Vascular Invasion (VI)	VI absent	180	174	354	1.746	0.186	0.269
	VI present	48	62	110
	Total	228	236	464
Lymphovascular Invasion	LVI absent	153	146	299	0.513	0.474	0.616
	LVI present	87	95	182
	Total	240	241	481
Perineural Invasion (PNI)	PNI absent	75	95	170	1.884	0.170	0.269
	PNI present	20	39	59
	Total	95	134	229

* Pearson Chi-square test for 2 × 2 tables, and Linear-by-linear association test for >2 × 2 tables; TNM = Tumor, Node and Metastasis.

). PRKACA expression did not show any association with a prognosis on Kaplan-Meier analysis (Log Rank X2 = 1.832, p = 0.176). Furthermore, high PRKACA expression was associated with the genome-stable and Epstein-Barr Virus subtypes of GC and with tumors having low aneuploidy and high global methylation scores (

Table 2. Molecular correlates of PRKACA expression in gastric and colorectal cancers.

PRKACA Expression
Molecular Correlates		Low PRKACA	High PRKACA	Total	X2 *	p Value	Adjusted p Value
Gastric cancer cohort
Molecular subtypes	CIN	120	102	222	7.231	0.007	0.014
	MSI	42	31	73
	GS	18	32	50
	EBV	11	19	30
	POLE	2	5	7
	Total	193	189	382
MSS vs. MSI	MSS	140	138	278	1.192	0.275	0.275
	MSI	42	31	73
	Total	182	169	351
Aneuploidy score	Low Aneuploidy Score	88	110	198	5.563	0.018	0.024
	High Aneuploidy Score	112	87	199
	Total	200	197	397
Global methylation score	Low Global methylation score	128	74	202	28.455	<0.001	<0.001
	High Global methylation score	77	131	208
	Total	205	205	410
Colorectal cancer cohort
Molecular subtypes	CIN	112	178	290	21.901	<0.001	<0.001
	MSI	42	21	63
	GS	39	16	55
	POLE	3	4	7
	Total	196	219	415
MSS vs. MSI	MSS	154	198	352	11.260	0.001	0.002
	MSI	42	21	63
	Total	196	219	415
Mutation Count	Low Mutation count	93	120	213	6.168	0.013	0.017
	High Mutation count	136	110	246
	Total	229	230	459
Fraction of Genome Altered	Low Fraction Genome Altered	151	112	263	10.592	0.001	0.002
	High Fraction Genome Altered	102	136	238
	Total	253	248	501
MANTIS Score	Low MANTIS Score	109	116	225	0.724	0.395	0.395
	High MANTIS Score	135	123	258
	Total	244	239	483
MSI Sensor	Low MSI Sensor Score	106	151	257	19.578	<0.001	<0.001
	High MSI Sensor Score	150	96	246
	Total	256	247	503
Aneuploidy Score	Low Aneuploidy Score	120	89	209	8.537	0.003	0.005
	High Aneuploidy Score	92	121	213
	Total	212	210	422
Tumor Mutational Burden	Low TMB	83	108	191	3.701	0.054	0.061
	High TMB	115	102	217
	Total	198	210	408

* Pearson Chi-square test for 2 × 2 tables, and Linear-by-linear association test for >2 × 2 tables; CIN = chromosomal instability, MSI = microsatellite instability, EBV = Epstein-Barr Virus, GS = genome stable, POLE = polymerase e, MSS = microsatellite stable, MANTIS = Microsatellite Analysis for Normal Tumor InStability (computational tool for assessing the extent of MSI in cancers), MSI Sensor = an alternative computational tool for assessing MSI in cancers.

).

2.2.2. Colorectal Cancer Cohort

In the CRC cohort, high PRKACA expression was associated with younger age groups (particularly the 60-years-and-below age group), left-sided colon cancer, rectal cancer, and nodal stage, but not with gender, race/ethnicity, pathological tumor stage, metastasis, TNM stage, lymphovascular and vascular invasion, and histological type (Table 1). As with the GC cohort, PRKACA expression did not show any association with overall or progression-free survival (OS: Log Rank X² = 0.039, p = 0.844; DFS: Log Rank X² = 0.054, p = 0.816). Moreover, high PRKACA expression was associated with the chromosomal instability and microsatellite stable (MSS) subtypes of CRC and with tumors having low mutation count, low MSI Sensor score, and high Fraction Genome Altered and Aneuploidy scores, but not with MSI MANTIS score and tumor mutational burden (TMB) (Table 2).

2.3. Biological Significance of PRKACA Expression

2.3.1. Gastric Cancer Cohort

GSEA showed significant differential enrichment of HALLMARK_MYOGENESIS, HALLMARK_APICAL_JUNCTION, HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION, and HALLMARK_COAGULATION gene sets in the PRKACA-high subset of GC cohort (Supplementary Material GC_GSEA.PRKACA_high). GO Enrichment Analysis showed that the enriched genes identified have a biological function in Cell-Matrix Adhesion (GO:0007160), Regulation of Cell Migration (GO:0030334), Actomyosin Structure Organization (GO:0031032), Contractile Actin Filament Bundle Assembly (GO:0030038), Regulation of Cell Motility (GO:2000145), etc., functions which are involved in cell motility, migration and metastasis (Figure 2 and Supplementary Material GOEA_GC_PRKACA_high). On the other hand, the PRKACA-low GC subset showed differential enrichment of the HALLMARK_G2M_CHECKPOINT, HALLMARK_MTORC1_SIGNALING, HALLMARK_PROTEIN_SECRETION and HALLMARK_E2F_TARGETS gene sets (Supplementary Material GC_GSEA.PRKACA_low). GO Enrichment Analysis identified the core enrichment genes in these gene sets to have such biological functions as Mitotic Cell Cycle Phase Transition (GO:0044772), Mitotic Spindle Organization (GO:0007052), DNA-templated DNA Replication (GO:0006261), etc., all of which are associated with cell replication, proliferation and tumor growth (Figure 2 and Supplementary Material GOEA_GC_PRKACA_low). The results of the analyses suggest that PRKACA may be a pro-migration/invasion and anti-proliferation gene in gastric carcinogenesis and gastric cancer progression.

2.3.2. Colorectal Cancer Cohort

GSEA showed differential enrichment of the HALLMARK_MYOGENESIS, HALLMARK_APICAL_JUNCTION, HALLMARK_KRAS_SIGNALING_DN, HALLMARK_COAGULATION, HALLMARK_ANGIOGENESIS, HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION, HALLMARK_APICAL_SURFACE, HALLMARK_HEDGEHOG_SIGNALING, HALLMARK_NOTCH_SIGNALING, and HALLMARK_IL6_JAK_STAT3_SIGNALING gene sets in the PRKACA-positive CRC cohort (Supplementary Material CRC_GSEA.PRKACA_high). GO Enrichment Analysis showed that the enriched genes identified in the GSEA have biological functions, which included Cell-Matrix Adhesion (GO:0007160), Actomyosin Structure Organization (GO:0031032), Positive Regulation of Cell Migration (GO:0030335), Positive Regulation of Cell Motility (GO:2000147), etc. These biological functions are associated with cell motility, invasion, and metastasis, as noted above in the GC cohort (Figure 3 and Supplementary Material GOEA_CRC_PRKACA_high).

The PRKACA-low CRC subset did not show differential enrichment of any of the HALLMARK gene sets associated with cell proliferation.

2.4. Differential Core Enrichment Gene Sets between GC and CRC PRKACA-High Subsets

A total of 281 genes comprising core enrichment components from the five Hallmark gene sets were found for the GC PRKACA-high subset, while 503 genes were enriched in the CRC PRKACA-high subset (Supplementary Material Core.Enrichment_GeneSets). A comparison of the core enrichment gene sets in the GC and CRC PRKACA-high subsets showed that only two enriched genes, EGFR and LAMP2, overlapped between the GC and CRC subsets, whereas the GO enrichment analysis found that the enriched biological processes, molecular functions, and cellular components subserved by the enriched genes were similar between the GI cancers. This is evidence that PRKACA or the PKA signaling may regulate diverse genes/genetic pathways in different cancer types to accomplish the same biological processes and molecular functions.

2.5. Deregulation of PRKACA Expression in Gastrointestinal Cancers

2.5.1. Gastric Cancer Cohort

PRKACA deregulation was investigated using copy number, methylation, miRNA, and transcription factor expression. PRKACA copy number alterations were obtained from the copy number segment data of the TCGA GC cohorts using the segment mean threshold of 0.3 (for gain/amplification) and −0.3 (for deletion). Using this threshold, the study identified 31/412, 375/412, and 6/412 PRKACA deletions, neutral, and gain/amplification, respectively. There was a significant association between PRKACA copy number changes and expression (X² = 23.609, p < 0.001; Figure 4). A total of 20 PRKACA promoter methylation loci and their beta values were retrieved from the gastric cancer methylation data. Bivariate correlation analysis demonstrated that 4/20 methylation loci were correlated with PRKACA expression (see Supplementary Material GC_PRKACA_Methylation_loci). Gene enrichment analysis with DEseq2 identified the top 40 miRNA species that were differentially enriched in the PRKACA-high and -low cases in the GC cohort. Bivariate analysis showed that the expression of 20/40 miRNA species showed correlations with PRKACA expression (see Supplementary Material GC_PRKACA-targeting miRNA). A total of 56 PRKACA-targeting transcription factors were retrieved from the transcription factor database TF2DNA_DB. Bivariate analysis showed that the expression of 27/56 transcription factors was correlated with PRKACA expression in the GC cohort (see Supplementary Material GC_PRKACA_TranscriptionFactors).

PRKACA copy number changes, methylation loci beta values, PRKACA-targeting transcription factors, and the differentially enriched miRNA values were incorporated into a multiple regression analysis. The results showed that multiple regulatory mechanisms independently predicted PRKACA expression in the GC cohort (

Table 3. Deregulation of PRKACA expression in gastric cancer *.

Gastric Cancer
R	R2		Adjusted R2	S.E. of Estimate
0.742	0.550		0.526	366.808
Coefficients
	Unstandardized Coefficients		t	p
	B	S. E.
(Constant)	2453.432	240.529	10.200	<0.001
hsa-mir-490	0.586	0.130	4.514	<0.001
PRKACA CNV	450.825	69.936	6.446	<0.001
ZNF644 Expression	−60.541	15.418	−3.927	<0.001
ZNF557 Expression	79.896	24.840	3.216	0.001
cg17818798	3007.832	745.072	4.037	<0.001
cg26613742	−2233.574	373.632	−5.978	<0.001
cg01596520	1031.285	338.217	3.049	0.002
HEY2 Expression	71.020	18.869	3.764	<0.001
CDC5L Expression	−8.872	4.218	−2.103	0.036
TFEB Expression	21.175	7.597	2.787	0.006
cg19621460	447.432	160.599	2.786	0.006
hsa-mir-7641-1	13.596	4.805	2.830	0.005
SP8 Expression	−61.200	21.970	−2.786	0.006
ZNF101 Expression	45.030	16.502	2.729	0.007
ZKSCAN3 Expression	128.247	46.163	2.778	0.006
ZNF451 Expression	−97.305	34.028	−2.860	0.005
cg01010868	−29,345.397	12879.446	−2.278	0.023
ANOVA
	df		F	p
Regression	17		22.777	<0.001
Residual	317
Total	334

* Multiple linear regression analysis.

, Figure 5).

2.5.2. Colorectal Cancer Cohort

In the CRC cohort, there were 10/480, 455/480, and 15/480 PRKACA copy number deletions, neutral, and gain/amplification, respectively. Furthermore, there were significant associations between PRKACA copy number changes and expression (X² = 5.039, p = 0.007; Figure 4). The bivariate analysis identified 11/20 methylation loci whose beta values exhibited a significant correlation with PRKACA expression in the CRC cohort (see Supplementary Material CRC_PRKACA_Methylation_loci). Only one methylation locus, cg17818798, exhibited a shared PRKACA methylation-expression correlation between the CRC and GC cohorts. The expression of 25/40 miRNA species identified in the miRNA gene enrichment analysis of the CRC cohort showed correlations with PRKACA expression on bivariate analysis (see Supplementary Material CRC_PRKACA-targeting miRNA). The expression of seven (7) miRNA species, including hsa-let-7c, hsa-mir-1-1, hsa-mir-1-2, hsa-mir-133b, hsa-mir-143, hsa-mir-7641-1, and hsa-mir-99a were found to be commonly correlated with PRKACA expression in the GC and CRC cohorts. Of the 56 PRKACA-targeting transcription factors retrieved from the TF2DNA_DB, the expression levels of 26 showed a significant correlation with PRKACA expression (see Supplementary Material CRC_PRKACA_TranscriptionFactors). The expression levels of 16/56 PRKACA-targeting transcription factors retrieved from the TF2DNA_DB were found to be commonly correlated with PRKACA expression in both the GC and the CRC cohorts. These included HOXA1, ZNF451, TFEB, TSHZ3, ZNF644, ZNF557, DHX34, HEY2, ZNF235, KLF15, HAND2, E2F7, TGIF1, ZBTB7A, MAFK, and CHAMP1.

Multiple linear regression analysis showed that a combination of regulatory mechanisms, including PRKACA copy number alterations, transcription factor expression levels, promoter methylation status, and miRNA expression levels, independently predicted PRKACA expression in the CRC cohort (

Table 4. Deregulation of PRKACA expression in colorectal cancer *.

R	R2			Adjusted R2	S.E. of Estimate
0.773	0.598			0.561	259.799
Coefficients
		Unstandardized Coefficients		t	p
		B	S. E.
(Constant)		−520.078	360.444	−1.443	0.150
ZBTB7A Expression		31.065	4.177	7.437	<0.001
hsa-mir-143		0.001	0.000	2.461	0.015
cg15814923		−6980.818	2832.849	−2.464	0.014
hsa-mir-577		−0.505	0.196	−2.578	0.011
TSHZ3 Expression		92.833	19.104	4.859	<0.001
DHX34 Expression		18.602	6.649	2.798	0.006
cg17818798		−996.038	497.031	−2.004	0.046
cg19586199		−610.229	186.442	−3.273	0.001
ZNF451 Expression		−124.101	27.985	−4.434	<0.001
ZNF442 Expression		324.096	163.862	1.978	0.049
ZKSCAN8 Expression		68.663	15.481	4.435	<0.001
MAFK Expression		−15.686	4.159	−3.772	<0.001
HOXA1 Expression		80.336	23.028	3.489	0.001
ZNF516 Expression		−111.062	23.760	−4.674	<0.001
cg17119568		2626.810	942.395	2.787	0.006
hsa-mir-216a		20.726	7.769	2.668	0.008
ZNF140 Expression		−37.664	20.633	−1.825	0.069
ZNF557 Expression		89.689	34.613	2.591	0.010
hsa-mir-1-2		−1.278	0.643	−1.986	0.048
cg20110535		846.904	368.866	2.296	0.023
hsa-mir-5092		−78.546	31.141	−2.522	0.012
ZNF235 Expression		−162.614	78.187	−2.080	0.039
ANOVA
	df			F	p
Regression	22			16.538	<0.001
Residual	245
Total	267

* Multiple linear regression analysis.

; Figure 5). The methylation locus cg17818798 and ZNF451 and ZNF557 expression were the predictors of PRKACA expression common to both the GC and CRC cohorts.

3. Discussion

PRKACA, the α-catalytic subunit of PKA, is an essential component of PKA signaling that catalyzes the downstream targets of PKA [9,12]. This study demonstrated that PRKACA expression in GC and CRC displays enrichment of genes that are associated with multiple conventional chemotherapeutic agents, FDA-approved tyrosine kinase inhibitors, and β-adrenergic blockers. The study findings support the role of PRKACA as a potential biomarker for multiple tyrosine kinase inhibitors, chemotherapeutic drugs, and β-adrenergic blockers. The findings also lend credence to the Cole and Sood study [9], which suggested that the inhibition of β-adrenergic signaling could possibly be utilized for the treatment of invasive cancer. This is even more so with the demonstration that high PRKACA expression was associated with the tumor hallmark of invasion or migration, rather than proliferation, in the GC and CRC cohorts. Moreover, the enrichment of gene targets of multiple drugs in PRKACA-high cases highlights the concept of an integrated, multitargeted approach to cancer therapy [19], wherein a single biomarker could at once be predictive of response to multiple drugs with varying mechanisms of action, on the basis of the association of that biomarker with a global gene expression network that incorporates the gene targets for those drugs [20]. The advantages of such an integrated approach to cancer therapy would include the augmentation of the therapeutic options for the oncologist and the reduction of the risks of drug toxicity and resistance without the violation of the tenets of personalized medicine [21,22]. To reiterate, the pro-migration/invasion characteristics of PRKACA, as found in this study, support a proposal for targeting PRKACA directly [14,18] and/or the β-adrenergic pathway [9] for treating tumor invasion.

The tumor-promoting activities of the PRKACA and PKA signaling have been described for classic and fibrolamellar variants of hepatocellular carcinoma, and breast, prostate, ovarian, and non-small cell lung cancers, glioblastoma, astrocytoma, and leukemia [11,12,13,14,15]. The association of PRKACA expression with some adverse clinicopathological features of gastrointestinal cancers, such as advanced age, tumor site, tumor histotype, nodal stage, and TNM stage, are in keeping with the tumor-promoting activities in the above-mentioned studies. Furthermore, this study showed that PRKACA expression is associated with poor prognosis molecular features and subtypes of gastrointestinal cancers. To the best of our knowledge, these findings have not been previously elucidated in any gastrointestinal cancer cohorts.

Interestingly, GSEA and Gene Ontology Enrichment Analysis demonstrated that while PRKACA-high tumors exhibited differential enrichment of genes that regulate cell motility, migration, and metastasis in the GC and CRC cohorts, there was enrichment of genes that regulate cell proliferation and growth in the GC, but not the CRC, PRKACA-low subset. The differential enrichment of actin filament organization, actin polymerization, actomyosin structure, focal adhesion structure, and membrane organelle organization are in keeping with the study by McKenzie et al. [23], which found that the PKA signaling is locally and rapidly activated by mechanical stretch in an actomyosin contractility-dependent manner, thus establishing the activated PKA as an effector of cellular mechanotransduction and a regulator of mechanically guided cell migration. Our findings also concur with the study by Tonucci et al. [24], which demonstrated that phosphorylation of CIP4 by PKA promotes the formation of functional invadopodia and, thus, confirms PKA phosphorylation of CIP4 as a regulator of the metastatic phenotype in cancer cells. Furthermore, the gene enrichment analysis results are in agreement with the Cheng et al. study [25], which demonstrated that PKA targets focal adhesion kinase for the promotion of cancer cell metastasis by cAMP. Moreover, Duan et al. [26] demonstrated that the phosphorylation of TPI at serine 58 by PRKACA enhances its enzymatic activity and glycolysis and, thus, its promotion of cancer growth and metastasis. Overall, our findings suggest that the primary function of PRKACA in cancer may be pro-migration/invasion rather than cell proliferation in GI cancers and may explain the seemingly paradoxical roles that have been described for PRKACA in different cancers [12]. For example, whilst PKA signaling has been demonstrated to regulate actomyosin contractility and cell migration, as well as membrane deformation, actin polymerization, cancer cell invasion, and metastasis, through phosphorylation of CIP4 and FAK [23,24,25,26], activating PRKACA mutations have been associated with smaller tumor sizes of adrenocortical adenomas [27,28,29,30]. PRKACA mutations in these adenomas have been demonstrated to have a limited impact on cell proliferation and tumor growth [30].

The concept of a proliferation-invasion dichotomy in cancer cells is not new in oncology [31,32,33,34]. In many tumor types, individual cancer cells exist either as proliferating or invading cells within the tumor body, using various molecular mechanisms to switch between proliferative and invasive phenotypes [31,32,33,34]. Conversely, genes and genetic pathways that possess pro-metastasis, but not proliferation or even anti-proliferation, functions in tumorigenesis and tumor progression are not uncommonly activated in cancer. The TGFB pathway is a ready example of a pathway that has seemingly paradoxical roles in cancer. The TGFB pathway, an anti-cell proliferation pathway, is known to induce the epithelial-mesenchymal transition and cell migration in cancer cells [35]. Furthermore, the cell cycle regulator p16 inhibits tumor cell proliferation through impedance of the G1-S phase progression [36,37]. However, p16 has been demonstrated to be upregulated or overexpressed in the frontiers of invading tumor cell nests and in tumor buds, evidence that it is essential for tumor cell invasion and metastasis [38].

The most common deregulatory mechanisms of PRKACA expression include structural variation or gene fusion [13,14,15], somatic mutations [10,27,28,29,30], and copy number alterations [27,39]. Whilst no structural variation or somatic mutations of PRKACA were sought in this study, PRKACA copy number alterations were demonstrably a significant mechanism of PRKACA expression regulation in our gastrointestinal cancer cohorts. However, the predominant regulatory mechanisms that better predicted PRKACA expression in our gastric and colorectal cancer cohorts are epigenetic, transcriptional, and microRNA regulatory mechanisms. These aforementioned mechanisms are the most common gene regulatory mechanisms in cancers generally [40].

In conclusion, this study has demonstrated that a subset of GC and CRC exhibit elevated expression of PRKACA, which is associated with the overrepresentation of gene targets of multiple FDA-approved tyrosine kinase inhibitors and β-adrenergic signaling inhibitors. PRKACA-high tumors were also associated with adverse clinicopathological and molecular characteristics of GI cancers. Furthermore, the study showed that tumors with high PRKACA expression were enriched for biological and molecular functions associated with cell motility, cell invasion, and metastasis. The deregulatory mechanisms of epigenetics, miRNA, and copy number alteration mechanisms were found to predict PRKACA expression in GC and CRC.

4. Materials and Methods

4.1. Study Cohorts

This study retrospectively analyzed the clinicopathological and genomic data of the cancer genome atlas (TCGA) colon, rectal, and gastric cancer cohorts [41,42,43,44]. All the data were retrieved from the Genome Data Commons (GDC) and cBioPortal for Cancer Genomics databases. Transcript quantification was accomplished with RNASeq (mRNA) and miRNASeq (miRNA). Methylation beta values were generated using a methylation array on the Illumina Human Methylation 450 platform, while the masked copy number segment data were generated using the Affymetrix SNP 6.0 genotyping array.

4.2. Data Processing

The clinico-genomic data of the GI cancer cohorts were retrieved from the GDC and CBioPortal databases using Linux-based scripts and codes that were written in the Windows-based Ubuntu 20.04 environment. Linux-based codes and scripts were also utilized to prepare the gene expression datasets in accordance with the Molecular Signature Database (MSigDB) [45,46] and DESeq2 Gene Enrichment Analyses requirements (https://cloud.genepattern.org/, accessed on 25 July 2024) [47], whereas Excel spreadsheet was utilized in the generation of the phenotype and derivative gene set files (see below), following which these were converted to cls and grp files, respectively. The gastric cancer cohort comprised 441 primary cases with clinicopathological (including prognostic data), RNASeq, chromosomal copy number segment, methylation, and somatic mutation data. The following data were available for this cohort: clinicopathological (between 380 and 441 of 441 cases for each clinicopathological indices; Table 1); mRNA expression (415/441 cases); chromosomal copy number segment (441/441 cases); methylation (between 393 and 440 of 441 cases for individual methylation loci); microRNA expression (441/441 cases) data.

The colorectal cancer cohort included 629 primary cases with the following amount of data: clinicopathological (between 545 and 629 of 629 cases for each clinicopathological indices; Table 1); mRNA expression (534/629 cases); chromosomal copy number segment (512/629 cases); methylation (between 331 and 524 of 629 cases for individual methylation loci); microRNA expression (506/629 cases) data.

The PRKACA expression data from these cohorts were converted into normally distributed data by using the method described by Templeton [48] before their utilization for statistical analyses.

4.3. Study Approach

First, GSEA and drug-target annotation with Drug Signature Database (DSigDB) gene set libraries in the Enrichr environment (https://maayanlab.cloud/Enrichr/enrich, accessed on 25 July 2024) [45,46,49,50,51] were utilized to infer the potential for PRKACA expression to predict susceptibility or resistance to drug agents. Then, the relationship between the clinicopathological indices and PRKACA expression in each of the GI cancers was sought in the cohorts using the appropriate statistical tests. In addition, the biological significance of PRKACA expression in GI carcinogenesis was confirmed using GSEA with the GC and CRC gene expression datasets and the MSigDB Hallmark gene sets [45,46]. The list of genes in the core enrichment of the significant gene sets was applied to gene ontology analysis in the Enrichr environment [45,46,49,50] to verify the biological, molecular, and functional significance of the enriched genes. Moreover, the mechanisms of altered PRKACA expression were sought in each of the GI cancer cohorts using the copy number segment, structural variation, methylation (beta values), transcription factor, and miRNA expression data. PRKACA-targeting transcription factors were retrieved from the transcription factor database TF2DNA_DB (https://www.fiserlab.org/tf2dna_db/search_genes.html, accessed on 25 July 2024) [52], and their expression values were correlated with PRKACA expression in either cancer cohorts using bivariate analyses. Significantly correlated transcription factors were then selected for regression analyses. Differential miRNA enrichment was sought between PRKACA-low and PRKACA-high cases using the online DESeq2 software v.3 on the GenePattern computing environment [47]. Gene enrichment analyses by DESeq2 were used to identify the top 40 miRNA species that differentially expressed in the PRKACA-low and -high subsets in the GC and CRC cohorts. Significantly enriched miRNAs, which also showed correlations with PRKACA expression by bivariate analyses, were then incorporated into a regression analysis, together with the methylation, PRKACA-targeting transcription factors, structural variation, and PRKACA copy number indices, to infer their roles in the deregulation of PRKACA expression.

4.4. Statistical Analyses

The enrichment analysis in the GSEA and DESeq2 software was performed using the software’s default parameters. GSEA was performed as a phenotype permutation. Gene ontology enrichment and drug-target association annotations were performed in the Enrichr environment using thresholds nominal p-value ≤ 0.05 and FDR ≤ 0.05. The clinicopathological and genomic data of the cancer cohorts were input into SPSS version 29. The Chi-square (or Fisher) test was used to probe for significant associations between categorical variables, while bivariate correlative analysis was utilized to test the correlations between continuous variables. Multivariate analysis of PRKACA expression correlation was investigated with multiple linear regression and binary logistic regression analyses. The one-way ANOVA test was used to measure the mean differences of continuous variables between discrete groups. The prognostic significance of PRKACA expression was defined using Kaplan–Meier and Cox regression analyses. A p-value of <0.05 was used as the threshold for significant tests, while the Benjamini–Hochberg correction was used to correct for multiple testing at an FDR of <0.05.