Ion channels expression and function are strongly modified in solid tumors and vascular malformations

Background Several cellular functions relate to ion-channels activity. Physiologically relevant chains of events leading to angiogenesis, cell cycle and different forms of cell death, require transmembrane voltage control. We hypothesized that the unordered angiogenesis occurring in solid cancers and vascular malformations might associate, at least in part, to ion-transport alteration. Methods The expression level of several ion-channels was analyzed in human solid tumor biopsies. Expression of 90 genes coding for ion-channels related proteins was investigated within the Oncomine database, in 25 independent patients-datasets referring to five histologically-different solid tumors (namely, bladder cancer, glioblastoma, melanoma, breast invasive-ductal cancer, lung carcinoma), in a total of 3673 patients (674 control-samples and 2999 cancer-samples). Furthermore, the ion-channel activity was directly assessed by measuring in vivo the electrical sympathetic skin responses (SSR) on the skin of 14 patients affected by the flat port-wine stains vascular malformation, i.e., a non-tumor vascular malformation clinical model. Results Several ion-channels showed significantly increased expression in tumors (p < 0.0005); nine genes (namely, CACNA1D, FXYD3, FXYD5, HTR3A, KCNE3, KCNE4, KCNN4, CLIC1, TRPM3) showed such significant modification in at least half of datasets investigated for each cancer type. Moreover, in vivo analyses in flat port-wine stains patients showed a significantly reduced SSR in the affected skin as compared to the contralateral healthy skin (p < 0.05), in both latency and amplitude measurements. Conclusions All together these data identify ion-channel genes showing significantly modified expression in different tumors and cancer-vessels, and indicate a relevant electrophysiological alteration in human vascular malformations. Such data suggest a possible role and a potential diagnostic application of the ion–electron transport in vascular disorders underlying tumor neo-angiogenesis and vascular malformations.

membrane de-polarization has a pivotal role at different stages of cell cycle progression and in various cell types [9]. Namely, endothelial cells hyper-polarization has been shown to contribute to cell division arrest [10], and channels are known to control migratory cellular properties in wound healing [11]. Further, Ca 2+ , K + and Cl − channels are essential regulators of cell proliferation and cancer development [12][13][14][15][16]. As recently further demonstrated, several ion-channels are directly involved in controlling tumor- [17] as well as non-tumor angiogenesis [18]. Expression and activity of TRPV4 channel have been found suppressed in tumor endothelium [19], and its activation has been found to normalize tumor vessels [20]; inhibiting Cl − channel activity has anti-angiogenesis and anti-glioma properties [21]; finally, human voltagedependent K + channel has been found associated with cancer aggressiveness and angiogenesis [22]. Therefore, ion-channels play a fundamental role in cancer progression as well as in angiogenesis.
Flat port-wine stains are non-tumor malformations of the skin capillaries [23]. Cutaneous capillary malformations are usually isolated finding. However, they may occasionally coexist with cerebral or ocular vascular malformations, constituting the rare sporadic neurocutaneous Sturge-Weber syndrome (SWS) affecting the cephalic microvasculature, or may represent signs of more aggressive vascular malformations or even vascular tumors.
The sympathetic skin response (SSR) is an alteration in skin electrical potential evoked by strong stimuli; it consists of a multineuronal reflex activated by a variety of afferent inputs where the efferent branches involve sympathetic sudomotor fibers. The electrodermal activity reflects sympathetic cholinergic sudomotor function which induces changes in skin resistance to electrical conduction. The response is mediated by ions flux via activation of receptor-coupled Ca 2+ , Cl − and K + channels [24]. Since SSR reflects peripheral C fibers activity, it is considered a reliable quantitative measure of sympathetic function and dysfunction as well as in polyneuropathies and dysautonomic disorders [25][26][27].
We have previously shown novel serum markers able to indicate cardiovascular diseases, [28,29] as well as soluble factors able to affect angiogenesis [30] and to discriminate infantile hemangioma from more aggressive vascular malformations or tumors [31]. We also investigated novel molecular markers of melanoma set-up and progression [32].
In the present study, we further addressed the issue of altered angiogenesis in tumor and non-tumor conditions. The expression level of 90 ion-channel genes was investigated in five different solid tumors, having a different histological origin. The expression level of several ion-channels genes was found to be strongly modified; we thus hypothesized that ion-transport may represent a measurable sign of the altered underlying angiogenesis. We, therefore, measured in vivo the ion-channel function in a human vascular malformation model, namely flat port-wine stains, as a model to test electrical-stimuli transport in a human vascular disorder accepted by the Ethic Committee.

Ion-channel gene expression investigation
Gene expression levels were investigated by accessing human cancer datasets available at Oncomine (http:// www.oncomine.org). The current Oncomine version contains several hundred different patients-datasets, referring to tumors biopsies obtained from almost any histological source; unfortunately vascular tumors (such as hemangioma, angiomas, hemangioendothelioma, angiosarcoma) are lacking from such database. In the current study 25, independent datasets from histologically different solid tumors were investigated, namely, bladder cancer, glioblastoma, melanoma, breast cancer, lung adenocarcinoma, as indicated in details in Table 1. Table 2 reports the detailed list of the 90 ion-channels and ion-channel related genes investigated in the present study. Briefly, different members were selected from 21 channel families, namely: amiloride-sensitive cation channels, calcium Channels voltage-dependent, cation channels sperm associated, FXYD domain containing ion transport regulators, gamma-aminobutyric acid (GABA) receptors, glutamate receptors ionotropic, potassium channels voltage gated subfamily, cholinergic receptors (Nicotinic), chloride channels, cyclic nucleotide gated channels, glutamate Receptors, sodium leak channels, purinergic receptors P2X, sodium-hydrogen exchanger regulatory factor 4, regulatory solute carrier proteins, sodium channels, glucose activated Ion channels, two pore segment channels, transient receptor potential cation channels, zinc activated ion channels, aquaporins.
Expression of such 90 genes was evaluated in oncomine database by setting "cancer vs. normal analysis" and choosing as cancer type: "superficial bladder cancer" within the bladder cancer group, "glioblastoma" within the brain and CNS cancer group, "breast invasive ductal carcinoma" within the breast cancer group, "lung adenocarcinoma" within the lung cancer group, and "melanoma".
Expression fold change (cancer vs. normal samples) and p values were reported for each analysis. Gene expression in tissue biopsies from 3673 patients was analyzed. Namely, 674 control-samples and 2999 cancer-samples were investigated.
Two additional datasets were identified and investigated within GEO database from NCBI (http://www. ncbi.nlm.nih.gov/gds), namely GSE41614 and GSE44115. Such datasets specifically refer to the vessels component within cancer samples; they were analyzed by means of the GEO2R interface available at the http://www.ncbi. nlm.nih.gov/gds site.

Patients recruitment
Several reports indicate that direct electrical stimulation may affect cell proliferation and dissemination in oncological setups and may induce other physiological effects [33][34][35][36][37][38][39]. Thus a clinical-study involving any electrical stimulation in cancer patients would be not acceptable by the Ethic Committee, according to the articles n. 14 and n. 16 of the Helsinki declaration on biomedical studies involving human patients, and to the article n. 16 of the Oviedo Convention.
For such reasons we submitted to the Ethic Committee the request to authorize the in vivo investigation in nontumor patients showing a pathological condition resembling, at least in part, the tumor neo-angiogenesis. Such request was authorized by the Ethic Committee and this explains way we investigated the SSR in angioma patients.
Fourteen patients (six males and eight females) affected by cutaneous flat port-wine stains were recruited at IDI-IRCCS, Rome. The study was approved by the institutional review board of IDI-IRCCS Hospital, Rome (IDI Ethic Committee 2011, n. 363). Patients with flat portwine stains diagnosis (age 18-70 years) undergoing no

Sympathetic skin responses SSR recording
SSR study was carried out according to the technical standards of the International Federation of Clinical Neurophysiology [40]. During the test, subjects were kept relaxed with comfortable light and temperature (26-28 °C); the test was started after 5 min of previous adaptation. The apparatus used was an electromyography and evoked potential equipment (MedelecSynergy, Viasys Healthcare, Madison WI USA). Recording electrodes consisted of a pair of superficial electrodes: recording was carried out on the glabrous skin on the flat port-wine stain, and the reference was placed 2 cm away from the lesion. The ground electrode was proximal to the recording electrodes. Electrical stimulation was applied through superficial electrodes over the right median nerve. The stimulus was strong but tolerable (not noxious). The electrical stimulus was applied four times at irregular intervals of 30-60 s (stimulus duration: 0.1 ms; intensity: 80 mA) to avoid habituation, and SSR waves were obtained. SSR recordings were carried out in quadruplicates at the angiomas lesion sites and onto a contralateral healthy skin region in each patient. The amplifier bandwidth was 0.1-100 Hz. Responses were recorded on the skin with an impedance <5 kΩ. The mean latency and peak-to-peak amplitude were calculated and used for the following analyses.

Statistical analysis
For gene expression data, the statistically significant threshold originally set at p = 0.05 was then corrected according to the Bonferroni correction for multiple comparisons, to take into account the multiple comparisons carried out. Therefore, the final corrected threshold was set at p < 0.0005 (value obtained from 0.05/90 comparisons, for each investigated dataset).
The expression level of any given gene was considered to be significantly altered vs control when the significant p value was matched in at least half-or in at least 1-of the specific databases investigated for each tumor type. Namely, regarding the "at least half datasets" stringency level, the significant p value had to be matched in at least 2 datasets out of the 4 investigated for bladder cancer; in at least 2 datasets out of the 4 studied for glioblastoma; in at least 2 datasets out of the 4 examined for melanoma; in at least 3 datasets out of the 5 investigated for invasive ductal breast carcinoma; in at least 4 datasets out of the 8 investigated for lung adenocarcinoma.
SSR was recorded four consecutive times for each patient on the diseased skin and four consecutive times on the healthy contralateral skin. Mean ± SE was computed; mean latency and mean peak-to-peak amplitude recorded at the diseased-skin level were compared to measures obtained on the contralateral healthy control site. Paired t Student test was carried out, and statistical significance was set at p < 0.05. Normal distribution of SSR data was tested according to the D' Agostino-Pearson

Expression level of ion-channels genes in tumors
The expression level of 90 ion-channels and ion-channels related genes (see Table 2 for the complete list) was analyzed in 3673 human biopsies of 5 histologically different solid tumors, namely: superficial bladder cancer, glioblastoma, melanoma, invasive ductal breast cancer, lung adenocarcinoma, as reported in Table 1 in more details. Table 3 indicates the ion-channel families showing a significantly altered expression (p < 0.0005). For each cancer type, two columns are presented: the left-hand column (lighten) reports genes indicating a significantly modified expression in at least 1-or less than half-of the investigated datasets. The right-hand column of each tumor type (shadowed) reports in bold all genes showing a significantly modified expression in at least half of the examined datasets (see criteria detailed in "Methods" section). All genes matching the shadowed columns criteria were found overexpressed. More in detail: • Within the calcium channel, voltage-dependent family, expression of CACNAD1 gene was found significantly modified in superficial bladder cancer datasets, namely in Dyrskjot dataset (p = 4 × 10 −6 ) and in Sanchez dataset (p = 5 × 10 −8 ), with an average 3.7-fold increase vs. ctrls; • Within the FXYD domain containing ion transport regulator family, expression of FXYD3 gene was found significantly modified in bladder cancer (Dyskjot dataset, p = 1 × 10 −6 , Lee dataset, p = 3 × 10 −6 and Sanchez dataset, p = 6 × 10 −21 ) with an aver- Within the shadowed columns, glioblastoma appears to have the highest number of modified ion-channel genes (namely: FXYD5, KCNE3, KCNE4, CLIC1, TRPM3).
Melanoma and breast cancer show no genes in the shadowed columns. However, within the lighten columns, breast invasive ductal cancer and melanoma show significantly modified ion-channel genes from 13 families and 6 families, respectively.

SSR recording in flat port-wine stains patients
The above reported analyses demonstrated that the expression level of several ion-channels is significantly altered in several human cancer biopsies. Such tumors are histological different. However, they all present an altered vascular tree, due to the tumor neo-angiogenesis. We hence hypothesized that vascular alterations observed in several different cancer types may harbor, at least to a certain extent, the observed ion-channels expression modifications. According to these findings, we hypothesized that measuring ion-channel transport may represent a non-invasive technique to investigate alterations in tumor-as well non-tumor altered angiogenesis. In vivo analysis of ion transport in a vascular malformation, the clinical model was carried out. In fact, measuring in vivo ion-transport in tumor patients was considered ethically not acceptable; we were then forced to identify a non-tumor clinical condition showing clear vascular anomalies. The flat port-wine stains clinical model was approved by the Ethic Committee as a safe model to investigate Sympathetic Skin Responses and electrical signal transport, as the less invasive approach possible in the current study. SSR depends on Ca 2+ , K +, and Cl − channels found to be altered in Table 3.
Therefore, 14 patients with flat port-wine stains diagnosis were consecutively recruited, namely eight female (mean age 27 years) and six male (mean age 29.1 years). SSR recordings were carried out in quadruplicates at the angiomas lesion sites and onto a contralateral healthy skin region in each patient. Statistical analysis was performed with paired tests as reported in Methods.  T  TPCN1  TPCN1  TPCN2   W   TRPM2  TRPM4  TRPV4   TRPM4  TRPV1  TRPV2   TRPM3 TRPM3   TRPC2  TRPM2  TRPM4  TRPV1  TRPV4 TRPC4 TRPM1 For each cancer type the lighten column reports ion-channels genes significantly modified in at least 1 dataset. The shadowed column reports ion-channels genes significantly modified in at least half of the investigated datasets. Channel family codes as in Table 2 Significance threshold: p < 0.0005 (see "Methods" section) Figure 1 shows the mean latency and means amplitude measured at diseased and healthy sites, expressed in mV. A strong and significant reduction in both latency and peak-to-peak amplitude signals was observed in the diseased site as compared to the healthy site, in the whole patients population. When a gender-specific analysis was carried out, both latency and amplitude were strongly reduced in female patients, while latency was reduced in male, although in a not-significant manner.

Ion channel expression in normal vessels as compared to tumor derived vessels
Data reported in Table 3 and in Fig. 1 led us to hypothesize that ion channels genes may play a relevant role in cancers as well as in other pathological conditions of altered angiogenesis. To further support this hypothesis we investigated two human samples datasets available at GEO database. The first dataset (GSE41614) reports transcriptional profiling of tumor-associated blood vessels in human invasive bladder cancer samples. In such dataset the expression data were obtained on laser capture microdissected vessels isolated from normal bladder tissue or from tumor bladder tissues. Data from ten samples were analyzed, i.e., five normal samples vs five cancer samples.
The second dataset (GSE44115) reports gene expression data in OCT frozen human angiosarcoma compared to OCT frozen normal mesenchymal tissues. In this case 18 angiosarcoma samples were compared to four controls (two from skeletal muscle normal uninvolved tissue and two pooled RNA from normal tissues). Tables 4 and 5 report the gene name, the log2 fold change, the p values adjusted according to the Benjamini and Hochberg false discovery rate methods, and the rank position (according to the by p value). Several genes, within the top 250 genes most significantly regulated, belong to ion channels families, either in bladder cancer blood vessels (9 genes, Table 4) and in angiosarcoma vessels (123 genes, Table 5). Such data indicate that several ion channels may play a key role in the vessels within cancer tissues.

Discussion
It is widely known that nerves and vessels follow similar anatomical paths. Often nerves and vessels show an overlapping anatomy with overlapping branches and ramifications. Several molecular factors are reported to control their respective patterns and growth in a coordinated manner [41], including semaphorin, netrin and slit [42], all strongly regulated by Ca 2+ , Na + , and Cl − channels [43,44]. Thus, an architectural or functional modification of the ones may affect the architecture or function of the others. We, therefore, argued that unordered angiogenesis occurring in tumors and vascular malformations may associate to a corresponding unordered nerve formation and therefore to a measurable alteration of the electric stimulus transport. We hypothesized that (i) ion channels (which are known to regulate nerve-and The expression level of members of several ion-channel families was found significantly modified in histologically different human cancer biopsies, according to the measures reported in Oncomine database, in an almost 4000 patients-vs-controls group. While several genes were found significantly modified (p < 0.0005) in at least one human dataset, we limit here the discussion to the genes reported in the shadowed columns of Table 3 selected according to highly stringent criteria, i.e. to the ion-channel genes found altered in at least half of the investigated databases of each tumor. We report in the current study that such genes show an increased expression in cancer vs. ctrls samples and have in most cases a definite role in vascular biology and/or cancer setup/progression or nervous system biology.
CACNA1D, a calcium related transporter gene, shows an average 3.75-fold increased expression in superficial bladder cancer datasets. No reports are present in  literature referring CACNA1D direct link to bladder cancer. Nevertheless, some studies relate its expression-and methylation-level to prostate cancer [45,46], indicating CACNA1D as a possible regulator of prostate cancer aggressiveness [47], or refer it to CNS disorders [48], diabetes [49] or calcium level within the vessels [50]. FXYD3 shows a 3.3 average fold increase expression in bladder cancer. It encodes a cell membrane protein regulating ion-pumps and ion-channels function and is known to have a role in tumor progression. Its activity is related to glucose and Cl − ions and has been indicated as a possible biomarker in bladder cancer [51,52] as well as other cancers including breast [53], colorectal cancer [54], endometrial cancer [55] and intrahepatic cholangiocarcinoma [56].
FXYD5 is a transmembrane auxiliary subunit of the Na + -K + -ATPase; it shows a 2.5-fold increase in glioblastoma. No direct link with glioblastoma has been reported to date, however it has been found up-regulated in adamantinomatous craniopharyngiomas in children [57] and other members of the FXYD family are known to be associated with different cancer types such as urothelial carcinoma [58], esophageal squamous cell carcinoma [59] and cholangiocarcinoma [60]. FXYD channels are involved in the anti-oxidative stress in vascular smooth muscle, thus controlling the vascular tone [61] and blood pressure [62]. Most interestingly, expression of FXYD2 and FXYD4 genes is modified in angiosarcoma vs control human samples ( Table 5).
The expression of serotonin receptor HTR3A was found increased by fivefold in lung carcinoma; no direct link has been reported between HTR3A and lung adenocarcinoma, yet; however, nucleotide polymorphisms of this gene have been related to opioid-or nausea/vomiting signaling pathways in cancer patients [63] and to bowel syndrome [64]. HTR3 is the only ligand-gated ion channel among the serotonin receptors, and it has been associated with neurological disorders such as depression [65] or schizophrenia [66]. Most interestingly about the current study, HTR3A acts as a ligand-gated ion channel neurotransmitter, and causes fast, depolarizing responses in neurons (http://www.genecards.org/cgi-bin/carddisp. pl?gene=HTR3A). It is up-regulated in rosacea, i.e., a chronic inflammatory skin disease often showing telangiectasias in the erythematotelangiectatic form (ETR) [67].
Within the potassium intermediate/small conductance calcium-activated channels, expression of 3 genes (namely KCNE3, KCNE4, KCNN4) has been found altered in the current study, namely in glioblastoma and lung adenocarcinoma. While no direct evidence relate KCNE3 or KCNE4 to glioblastoma, KCNE4 is most abundantly expressed in brain [68]; it exerts functions such as controlling the neuronal firing rate, the synaptic transmission [69] and the T-lymphocytes maturation [70] and it is known to regulate K-channels in vascular smooth muscle [71] and more in general neuronal excitability. Furthermore, KCNN4 single-nucleotide polymorphisms have been related to myocardial infarction [72]. Most interestingly, KCNN4 has been linked to vascular cells proliferation [73]. Notably, expression of several potassium channels is modified in bladder cancer vessels vs controls (KCNC4, KCNG4, KCNS2, see Table 4) and in angiosarcoma vs controls (namely KCNJ16, KCNQ2, KCNJ15, KCNJ12, KCNJ1, see Table 5).
Within the Cl-intracellular channels, CLIC1 expression was found altered in bladder cancer and glioblastoma. CLIC1 has been previously found up-regulated in glioblastoma [74], it is involved in different tumors and acts as an oncogene in pancreatic cancer [75], and has been indicated as a possible cancer biomarker [76]. Interestingly, CLIC1 has recently shown a key role in angiogenesis control in combination with integrins [77,78]. Interestingly, expression of one Cl − channel (namely CLCNKA) is strongly modified in angiosarcoma vs controls (see Table 5).
Within the transient receptor potential cation channels, TRPM3 gene expression was found increased in glioblastoma by 2.3-fold. Its expression has been previously found increased in glioblastoma [79] and is known to exhibit mechanosensitivity contributing to vascular and cardiac functions [80].
SSR is under the direct control of Ca 2+ , K − , Cl − ion channels and strictly depends on the sympathetic autonomous nerve function. In the current study, several ion channels related to the Ca 2+ , K +, and Cl − have relevant, and significantly increased expression in different human cancers and SSR was found strongly altered in human vascular malformations. SSR has been previously indicated as a possible useful diagnostic technique in CNS pathologies [81,82], as well as fibromyalgia [83] and diabetes [84]. Further, autonomic nerve development and function has been recently shown to play a key role in prostate cancer progression [85]. Thus, given the known role of several ion channels in the tumor, nerve, and vascular biology, we hypothesize that the altered SSR observed in vascular skin malformations, and the observed altered ion-channel gene-expression in several tumors may represent phenomena related to the unordered nerve-and vessel formation, common to tumorrelated and non-tumor-related vascular anomalies.
SSR recording gave more significant results in female while in male the trend was present but with no statistical significance. Such difference may be related to a different transport of electrical stimuli in the female skin, which appears evident at both healthy (white boxes in Fig. 1) and diseased sites (black boxes in Fig. 1), as compared to male. A gender-related difference in the skin thickness may underlie, at least in part, such observation likely associated with the skin thickness and water retention induced by the menstrual phases and the hormonal status in females.
In conclusion, data reported in the current study allowed us to conclude that ion-channel expression and function may be strongly affected in pathological conditions where vessels' (and nerves') architecture is altered. SSR measurement may thus represent a non-invasive useful tool to investigate vascular skin alterations in both tumor and non-tumor conditions.

Conclusions
The present study reports for the first time a detailed analysis of the expression level of 90 ion-channel genes in 3673 biopsies form humans affected by solid tumors and from healthy controls. At least ten genes from different ion-channels families were found firmly and significantly up-regulated in histologically different tumors having in common the underlying unordered tumor neo-angiogenesis (Table 3). Moreover, expression of at least 20 ion channels has been found to be strongly modified in cancer associated vessels vs controls (Tables 4, 5). The sympathetic skin responses (SSR), an electrical feature closely related to the ion-channels activity, was then measured in vivo and was found strongly modified in the skin of patients affected by the flat port-wine stains vascular malformation. The present study indicates that modified expression and activity of ion channels is likely related to vascular alteration, in both tumor and non-tumor conditions, and suggests the non invasive technique SSR as a simple, useful tool to investigate skin vascular anomalies.