Potential Shortcomings of Genomic Database: The Case of Na_v1.5 Expression in Breast Cancer

In the current post-genomic era, it is convenient and fashionable to question how the expression of a certain gene might relate to disease progression and/or patient survival and, for this, publicly available genomic databases are frequently used. Such questioning is also true for ion channels, which are at the heart of bioelectricity. Thus, in a recent major review, Liu et al. (2024) examined the involvement of voltage-gated sodium channels (VGSCs) in the cancer process.¹ In particular, several aspects of the role of the VGSC subtype Na_v1.5 in breast cancer were examined in considerable depth. Overall, the evidence supported the notion that functional VGSC expression accelerates metastatic progression. However, in the bioinformatic account given by Liu et al.,¹ data from The Cancer Genome Atlas (TCGA) suggested that expression of SCN5A (the gene coding for Na_v1.5) correlated positively with the survival of breast cancer patients. This is in sharp contrast to substantial in vitro and in vivo evidence, extending to humans, showing that Na_v1.5 expression/activity promotes metastasis, the main cause of death from cancer.² The evidence is even stronger when VGSCs are considered as a whole.³

There are several reasons for this discrepancy, as follows:

• 1.The gene may not necessarily translate to protein. Indeed, for a gene to give rise to a functional protein, it has to be translated and put through several stages of post-translational processing.
• 2.The “mature” protein may not be trafficked to plasma membrane and fail to be anchored along with subunit(s) and/or other associated proteins in a functional macro-molecular complex. For example, non-metastatic human breast cancer (MCF-7) cells contain plenty of Na_v1.5 channel protein but these remain internalized and non-functional; this occurs even after over-expression (unpublished observations).
• 3.Na_v1.5 expressed in breast cancer is mainly an embryonic splice variant.^4,5 It is not clear that this aspect is represented in the data bank (although the gene is still SCN5A).
• 4.In vitro, only VGSC-expressing MDA-MB-231 breast cells can demonstrate Matrigel invasiveness; non-expressing cells remain above the transwell filter.⁶ In vivo, this would mean that only VGSC-expressing cells will be able to escape from the primary tumor and metastasize and the remaining cells will have a low level or no VGSC. In such a case, the primary tumor of even the most aggressive cancers may actually appear to contain lower amount of protein/gene. This could then lead to a negative correlation between VGSC/Na_v1.5 expression and metastasis and, hence, survival, as indicated by the TGCA database.

Thus, in the first instance, correlating gene expression directly with patient output is not straightforward and may fail because there are several steps between transcription and its pathophysiological consequence(s) (Fig. 1).

FIG. 1.

The basic steps leading from gene expression to mature protein synthesis, signaling and physiological function. In particular, there are over 100 different ways in which RNA may be modified.⁷ Relevant and emphasized here is the protein to signaling progression being bioelectric.

Unlike genomic data, however, protein-based and functional studies support the notion that Na_v1.5 expression is negatively correlated with breast cancer progression disease recurrence and overall survival.⁸ Most recently, Leslie et al.⁹ have shown that (i) tumor size, (ii) lymph node reactivity, (iii) “Nottingham Prognostic Index,” (iv) distant metastases and (v) metastasis-free survival are significantly higher for breast cancer patients expressing high vs. low level of Na_v1.5 protein.

More generally, one major reason why bioelectric states and the information they carry for tissues cannot be predicted from proteomic or transcriptomic states is that they function as reprogrammable circuits.¹⁰ For example, in the brain, where bioelectric signaling is best understood, cognitive processes via action potentials propagating throughout neural networks do not require changes in ion channel gene/protein levels—they operate on a very fast timescale because ion channels (and electrical synapses known as “gap junctions”) can open and close post-translationally on a time scale of milliseconds. The very same channel protein can be driving cell voltage in different ways by regulating its conductance in place, as a function of prior physiological states such as voltage or pH. This gives the nervous system historicity (memory, sensitivity to past events), much more rapid signaling than protein expression changes, and the ability to propagate different patterns (computations) on the precisely same molecular hardware. The same is true for all body cells. The same ion channel can contribute to different levels of membrane potential (V_m) depending on its conductance state at any time. Likewise, the same V_m state can be produced by the action of many different channels and pumps, which is very advantageous in bioelectric approaches to biomedical intervention. And finally, the cellular voltage is the sum total of numerous channels and pumps. Thus, there is no 1:1 mapping between the kinds and amounts of channel protein at the membrane (or in the mRNA pool, or in the genome) and the bioelectric states, which drive downstream changes of cell behavior and transcription (Fig. 2). Ultimately, four-dimensional physiomic measurements and computational modeling of propagating bioelectric patterns will be needed to complement the snap-shot omics of biochemical components in cells.

FIG. 2.

The dissociation between expression levels and bioelectrical state. Two “model” cells can have very different expression levels of electrogenic proteins (here, Na⁺/K⁺-ATPase and V-ATPase) and thus appear quite distinct in omics plots of Principal Component Analysis (A). However, these cells may have the same physiological state because the different complements of channels and pumps drive the same V_m (B). Conversely, the exact same channel expression can result in different bioelectric states in different cells if their history or associated signaling causes the channels to be in different open/closed state, which is not apparent from biochemical omics.

In conclusion, and moving on from Liu et al.,¹ it would follow (i) that more carefully designed experiments/analyses, as well as computational modelling of the physiology, are needed to correlate ion channel gene expression with pathophysiological (as well as physiological) consequences, and (ii) that publicly available genomic data banks, such as TCGA, must be used with caution and in the right context.

Author Disclosure Statement

M.B.A.D. holds shares in Celex Oncology Innovations Ltd., which aims to develop ion channel drugs against cancer. M.L. is supported by Astonishing Labs and Morphoceuticals, companies, which apply bioelectric strategies in various areas of biomedicine.

Funding Information

No funding was received for this article.