Reply from L. P. Cid, T. J. Jentsch and F. V. Sepúlveda: intestinal electrolyte and fluid secretion – a model in trouble?

In a letter to the editor Dr Michael Lucas manifests his disagreement with ‘with all the conclusions drawn regarding intestinal secretion’ in a recently published paper of ours (Julio‐Kalajzić et al. 2018).

In the paper in question we assumed the widely supported hypothesis that anion secretion across the intestinal epithelium occurs when certain agonists or toxins activate CFTR Cl⁻ channels though cAMP (or cGMP) intracellular signalling. As CFTR channels have an apical membrane location and the anion is above its equilibrium potential in the cells, efflux ensues. A variety of approaches, which would be too long to enumerate here, support the idea that this Cl⁻ secretion is the main driving force behind intestinal fluid secretion and its derangement in pathological conditions is responsible for secretory diarrhoea and cystic fibrosis. Continued Cl⁻ secretion requires the simultaneous activity of basolateral K⁺ channels to maintain cellular ionic homeostasis and membrane potential. In the paper under discussion we explored the identity of these channels in the mouse using a single or combined knock‐out animal approach and measured response in short‐circuit current across mouse colon after pharmacological blockade of electrogenic Na⁺ absorption as a simple way to assess electrogenic Cl⁻ secretion (for reviews see Barrett & Keely, 2000; Frizzell & Hanrahan, 2012). Our conclusion is that there are a variety of K⁺ channels that can fulfil the role of supporting electrogenic Cl⁻ efflux and we identify the K_2P K⁺ channel TASK‐2 as one of these. The persistence of a secretory activity despite multiple genetic ablations, we think, suggests that further K⁺ channel(s) not yet identified underpin the intestinal anion secretory process.

Lucas's objections to our reasoning arise from a belief that the basic model of intestinal fluid secretion is flawed and that changes in short‐circuit current are ‘coincidental phenomena resulting from interrupted fluid absorption’. The evidence for this is purported to come from experiments from his laboratory that appear to show that E. coli STa enterotoxin does not evoke intestinal fluid secretion in rat small intestine in vivo. Additionally it is pointed out that the K⁺ channel blockers glibenclamide and clotrimazole, and the activator cromakalim, do not affect STa‐mediated fluid absorption inhibition, and hence K⁺ channels do not contribute to the process. The inference from these experiments is that intestinal secretion in general is not explained by fluid secretion from the enterocyte and is unrelated to short‐circuit current changes evoked by the toxin.

The results of Lucas et al., and therefore the conclusions thereof, fly in the face of a large body of clinical observations and experimental results. We will briefly comment on those related to STa. E. coli toxins have long been known to cause diarrhoea and fluid secretion in humans and animals (see for instance Evans et al. 1973; Giannella, 1981). Fluid secretion elicited under the influence of STa has been repeatedly shown in vivo using different experimental designs. Just to cite the most recent such demonstration we are aware of, Cil et al. (2017) used a jejunal closed loop approach in the mouse to show that STa evokes an important intraluminal accumulation of fluid that is measured as increase in loop weight at the end of the experiment. To convince oneself that this is genuine increase in intraluminal fluid, and is not due to ‘increased extracellular volume’, it might be enlightening to examine the photographs of the intestinal loops under the various conditions in Fig. 3 of the paper by Cil et al. (2017). These show quite clearly that a marked increase in the volume of fluid within the intestinal lumen has taken place after treatment with STa.

STa binds to a cell surface receptor in the intestine, which subsequently leads to an activation of guanylyl cyclase. The rise in cGMP then stimulates fluid and ion secretion mainly by direct effect on chloride channels (Field et al. 1978; Guerrant et al. 1980). Intestinal specific guanylyl cyclase C (GC‐C) is the receptor of STa, with the toxin causing marked elevations of cGMP upon binding and activation of GC‐C (Schulz et al. 1990). Guanylin and uroguanylin are endogenous GC‐C peptide ligands produced in intestinal epithelium and secreted into the lumen from where they promote cGMP production in a paracrine manner. The cGMP‐dependent kinase type II (PKG II) is physically associated to CFTR and produces its activation through CFTR phosphorylation (French et al. 1995; Vaandrager et al. 1996). In humans, loss of function mutations activating GC‐C cause non‐CF meconium ileus consistent with an important role of GC‐C in neonates (Romi et al. 2012). Conversely, gain of function mutations affecting GC‐C lead to familial chronic diarrhoea syndrome (Fiskerstrand et al. 2012). The genetic ablation of PKG II in mice, on the other hand, has been shown to abolish the intestinal effects of STa (Pfeifer et al. 1996). Interestingly this paper shows that STa‐dependent increases in short‐circuit current consistent with CFTR activation and fluid secretion into the intestinal lumen measured in vivo are abolished in mice deficient in PKG II.

As is the case for cholera toxin, part of the STa effect is to decrease fluid absorption through cGMP‐mediated inhibition of NHE3 and decrease in NaCl absorption, which is electrically silent (Steinbrecher, 2014; Thiagarajah et al. 2015). The relative weight of this inhibition in a situation of toxin‐induced secretory diarrhoea is not clear. Nevertheless, a mouse with its NHE3 removed by genetic modification shows only slight diarrhoea (Schultheis et al. 1998), consistent with a preponderant role of CFTR in fluid secretion. The contribution of this NHE3 inhibition also appears minor judged by experiments in which the increase in intraluminal fluid accumulation of mouse jejunum under the effect of STa, very much as that induced by cholera toxin, is abolished using a highly specific inhibitor of CFTR (Cil et al. 2017). The same inhibitor abolishes the increase in short‐circuit current evoked by STa across the jejunum in vitro. These results are entirely consistent with the hypothesis, developed originally for intestinal effect of cholera toxin and cAMP, that intestinal fluid secretion driven by CFTR‐mediated chloride net flux into the lumen accounts for the hypersecretory effect of STa.

Concerning the ‘mismatch between current measurements in vitro and fluid movement in vivo’ in our paper, the two‐ to threefold discrepancy calculated by Lucas would not be surprising considering that different intestinal segments are being used for the short‐circuit current and fluid accumulation experiments, respectively.

One of the open questions in our paper is the reason behind the lack of evidence for alterations in the luminal water accumulation in the different K⁺ channels knockout models examined. Various possible explanations were advanced in our Discussion section and will not be rehearsed here. The most obvious one is that the K⁺ channel activity remaining, or arising, after genetic ablation is enough to support CFTR‐mediated Cl⁻ efflux. Acute experiments using wild‐type animals, the correct K⁺ channel inhibitors and the administration protocol for in vivo assay of basolaterally acting compounds might help to solve this problem.

As to the crumbling of the intestinal Cl⁻ secretion model assuredly announced by Lucas, we think that it is, just as the premature announcement of the demise of Mark Twain, ‘grossly exaggerated’.

Additional information

Funding

This work was funded by the FONDECYT‐Chile grants 1120743 and 1160043. The Centro de Estudios Científicos (CECs) is funded by the Base Financing Programme of Conicyt grant PFB01.

Competing interests

The authors declare that they have no competing interests.