CrossTalk proposal: Physiological CO₂ exchange can depend on membrane channels

Introduction

Since the discovery that CO₂ passes through aquaporin‐1 (AQP1; Nakhoul et al. 1998; Cooper & Boron, 1998), the importance of channel‐ vs. lipid‐mediated gas transport has often been portrayed as an either/or issue. However, depending upon physiological context, the role of channels may be insignificant or dominant.

In a landmark study, Mitchell (1830) examined gas permeation across barriers of natural rubber or animal tissue, rank‐ordered the ‘relative facility of transmission’ of several gases, and recognized that these move independently of one another in a mechanism dependent upon ‘infiltration’ (i.e. solubility) in the organic molecular barrier – the first statement of ‘solubility theory’. Later, Graham showed that permeation across rubber membranes depends on not only solubility but also diffusion through the barrier (Graham, 1866) – the first statement of ‘solubility–diffusion theory’. Meanwhile, Fick proposed his law of mass diffusion, which Wroblewski combined with Henry's law to produce our modern transport equation (Wroblewski, 1879):

$$J=\frac{Ds}{l}\Delta p$$

Here J is flux, D the diffusion constant in the barrier, s solubility in the barrier, l barrier thickness and Δp trans‐barrier partial pressure difference.

In the late 1890s, Overton used solubility theory to develop his revolutionary hypothesis that the boundary layer of the cytoplasm (‘Grenzschicht des Protoplasten’) – now termed the plasma membrane – is impregnated with lipids. However, modern reference to ‘Overton's rule’

$$\begin{array}{ccc}& & \text{True}\text{membrane}\text{permeability}\hfill \\ & & \text{to}\text{substance}X\left(\mathrm{i}.\mathrm{e}.,P_{M,X}\right)\propto s\hfill \end{array}$$

is problematic because (1) the ‘solubility’ rule is really Mitchell's and ignores both (2) D and (3) membrane proteins, which themselves impact J in three ways. First, in the plane of the membrane, proteins impermeable to X displace and organize nearby lipids, reducing P _M,X (Wang et al. 2007; Boron, 2010). Second, transporters and channels carry a wide range of lipid‐soluble solutes (Al‐Awqati, 1999). Thus, lipid solubility does not prove permeation via membrane lipid. Third, exomembranous portions of integral membrane proteins can almost completely cover some membranes (Takamori et al. 2006). Boron proposed the ‘access–solubility–diffusion–egress theory’ to account for the resistance of these proteins, and the ordering of water near charged lipid head‐groups, to the entry of a substance into (or exit from) membrane lipid.

The first clear experimental data opposing the solubility–diffusion theory was the demonstration that gastric gland apical membranes have no measurable $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ (Waisbren et al. 1994). In artificial lipid bilayers, a major determinant of $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ may be membrane lipid cholesterol content (C _M,chol; Itel et al. 2012; Kai & Kaldenhoff, 2014). In one study, raising C _M,chol from 0 to 20% lowered $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ by ≥10‐fold; raising C _M,chol from 20 to 70% decreased $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ by an additional 10‐fold (Itel et al. 2012), predominantly due to a decrease in D rather than s. The CO₂‐impermeable plant aquaporin NtPIP2;1 reduces $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ of artificial membranes (Kai & Kaldenhoff, 2014) by displacing lipids, further reducing ‘background’ CO₂ permeability. True membrane permeability depends on two parallel pathways that sum like parallel electrical conductances: $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ = $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ + $P_{\mathrm{Channel},\mathrm{C}{\mathrm{O}}_2}$. In series with the membrane are unconvected layers (ULs; Fig. 1 A) that reduce the measured apparent membrane permeability ($P_{\mathrm{M}{\prime }_{,}\mathrm{C}{\mathrm{O}}_2}$):

$$\begin{array}{ccc}\hfill \frac{1}{P_{\mathrm{M}\prime ,\mathrm{C}{\mathrm{O}}_2}}& =& \frac{1}{P_{\mathrm{UL},\mathrm{C}{\mathrm{O}}_2}}+\hfill \\ & & \frac{1}{P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}+P_{\mathrm{Channel},\mathrm{C}{\mathrm{O}}_2}}\hfill \end{array}$$

Figure 1. Models of CO₂ movement across membranes of artificial lipid bilayers (A), as well as plasma membranes with low cholesterol content (B), moderate cholesterol content without (C) and with (D) gas channels, and high cholesterol content without gas channels (E) .

Membrane composition has dramatic effects on CO₂ permeability. Red arrows represent CO₂ flux. The density of CO₂ molecules on upper side of membrane (exaggerated by ∼5‐fold, assuming a membrane thickness of 6 nm and 5% CO₂ at room temperature, for illustrative purposes) is the same in all examples; the densities above and below the membranes represent snapshots during the early moments following CO₂ addition to extracellular solution. The density of membrane proteins is less than indicated for synaptic vesicles by Takamori et al. (2006). In A, the artificial lipid bilayer contaminated with decane has an extremely high permeability to CO₂. The square brackets UL indicate the unconvected layers either side of the membrane. B–E show biological membranes with integral proteins and increasing levels of cholesterol, both of which reduce ‘background’ CO₂ permeability (i.e. not due to channels). In B, the presence of gas‐impermeable proteins and low cholesterol concentrations (∼25% of membrane lipid) lowers CO₂ permeability and flux. In C, further increasing membrane cholesterol content (∼40%) dramatically decreases CO₂ permeability. In D, membranes (e.g. from RBCs) with the same content of integral membrane proteins and cholesterol as in C have a much higher CO₂ permeability because specialized channels such as AQP1 and the Rh‐associated glycoprotein (RhAG) augment the passage of CO₂. In E, membranes (e.g. apical membranes of proximal colon) with extremely high cholesterol content (∼77%) are expected to have a very low CO₂ permeability.

Thus, channels likely contribute more when ULs are thin and $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ is low (as in red blood cells (RBCs)), but less when ULs are thick and/or $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ is high (as in some solid tumour models).

Options in cell membrane design

$P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ can span many orders of magnitude. At one end of the spectrum are artificial lipid bilayers, typically loosely packed lipids containing as much as 30% of the solvent n‐decane (Fig. 1 A). Here, $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ = $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$, high enough to make measurement difficult.

Further along the spectrum are plasma membranes from cells with modest C _M,chol and abundant integral membrane proteins (Fig. 1 B). Examples include MFC7 breast tumour cells and ascites tumour cells, with C _M,chol of ∼25% (Haeffner et al. 1984; Todor et al. 2012). With such a low C _M,chol, $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ could be sufficiently high that, even without channels, ascites tumour cells could accommodate their modest ${\dot{V}}_{\mathrm{C}{\mathrm{O}}_2}$ of 0.12 ml g⁻¹ min⁻¹ (Warburg, 1956), according to Endeward's analysis (2014).

Even further along the spectrum are Madin‐Darby canine kidney (MDCK) cells, with an intermediate C _M,chol (37%), no known gas channels, and a low $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ (Itel et al. 2012) that presumably represents $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ and is sufficient to meet a low metabolic demand (Fig. 1 C). Interestingly, depleting MDCK cells of cholesterol (i.e. raising $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$) or expressing AQP1 (i.e. raising $P_{\mathrm{Channel},\mathrm{C}{\mathrm{O}}_2}$) raises $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$.

Falling into the same cholesterol content category (∼40%) as MDCK cell membranes are RBC membranes (Fig. 1 D). Although RBCs have a low $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$, a high gas‐channel content gives them a high $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ (see below).

At the far end of the spectrum are proximal colon apical membranes, with C _M,chol of 77%, consistent with the observed low CO₂ permeability (Fig. 1 E; Endeward et al. 2014). Apical membranes of gastric glands have no measurable $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ (Waisbren et al. 1994).

Gas channels

In 1998 Boron's laboratory identified the first family of gas channels by showing that CO₂ moves through AQP1, heterologously expressed in Xenopus oocytes (Nakhoul et al. 1998; Cooper & Boron, 1998). Both p‐chloromercuriphenylsulfonic acid (pCMBS) (Cooper & Boron, 1998) and DIDS (Endeward et al. 2006) significantly reduce AQP1‐dependent $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$; because pCMBS but not DIDS reduces water permeability, these agents act via different pathways. AQP1 also conducts NH₃ (Nakhoul et al. 2001) and NO (Herrera et al. 2006).

Ripoche et al. (2004) and Khademi et al. (2004) identified another gas‐channel family, the Rhesus (Rh) proteins, by demonstrating permeability to NH₃. Work with human RBCs showed that the Rh complex also conducts CO₂ (Endeward et al. 2008). Further studies show that each AQP and Rh protein exhibits a characteristic selectivity for CO₂ vs. NH₃, with some AQPs being impermeable to CO₂, NH₃, or both (Musa‐Aziz et al. 2009; Geyer et al. 2013 b,c).

Recently, Boron's group identified a third gas‐channel family: the urea transporter UT‐B, which is permeable to NH₃ (Geyer et al. 2013 a). The known gas‐channel families consist of physiologically active monomers surrounding a central structure that, for AQPs and Rhs, is a lipophilic pore. We hypothesize that some of these central pores conduct CO₂ or other dissolved gases, and that other families of gas channels remain to be identified – known proteins with previously unappreciated gas‐selective pathways. Thus, arguing that particular membranes lack particular channels ignores the possibility of yet‐to‐be‐discovered channels.

Physiological role

The first demonstrated physiological role for channels in gas transport was CO₂ uptake (driven by an exceeding small gradient) via NtAQP1 in tobacco plants during photosynthesis (Uehlein et al. 2003). Studies on human RBCs show that DIDS plus the genetic deletion of either AQP1 (Colton‐null) or Rh complex reduces $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ by ≥90% (Endeward et al. 2006, 2008), leaving, at most, 10% for CO₂ pathways through lipid. Thus, channels almost certainly make a physiologically important contribution to CO₂ exchange in pulmonary and systemic capillaries, particularly during conditions of short transit time, such as exercise (Endeward et al. 2014).

Channels could make significant contributions in other systems with high CO₂ fluxes and high AQP levels. Examples include alveolar type I pneumocytes (AQP5; Verkman et al. 2000), astrocytic endfeet/blood–brain barrier (AQP4; Nagelhus et al. 2004) and renal proximal tubules (AQP1; Schnermann et al. 1998).

In conclusion, channels contribute to $P_{\mathrm{M},\mathrm{C}{\mathrm{O}}_2}$ on a sliding scale that depends on the balance of $P_{\mathrm{Channel},\mathrm{C}{\mathrm{O}}_2}$ vs. $P_{\mathrm{Back},\mathrm{C}{\mathrm{O}}_2}$ and ULs. Moving the field forward, and knowing where cells sit on the sliding scale, will require understanding this balance by examining, in simple systems, how physiological ULs and altered membrane composition (i.e. the nature and number of lipids, non‐channel proteins, channels) affect gas fluxes.

Call for comments

Readers are invited to give their views on this and the accompanying CrossTalk articles in this issue by submitting a brief (250 word) comment. Comments may be submitted up to 6 weeks after publication of the article, at which point the discussion will close and the CrossTalk authors will be invited to submit a ‘Last Word’. Please email your comment, including a title and a declaration of interest, to jphysiol@physoc.org. Comments will be moderated and accepted comments will be published online only as ‘supporting information’ to the original debate articles once discussion has closed.

Additional information

Competing interests

None declared.

Funding

This work was supported by grants to WFB from the Office of Naval Research (N00014‐11‐1‐0889, N00014‐14‐1‐0716, N00014‐15‐1‐2060), the NIH (U01‐GM111251), and by the Meyer/Scarpa Chair.

Biographies

Gordon Cooper is a Senior Lecturer in Biomedical Science at the University of Sheffield. He has a background in renal and epithelial physiology and his research has focused on the transport of small solutes and gases across biological membranes.

Rossana Occhipinti is Adjunct Instructor in the Department of Physiology and Biophysics at CWRU. Her PhD is in applied mathematics and her current research focuses on mathematical models of CO₂ and acid–base transport.

Walter Boron is the Myers/Scarpa Professor and Chair of Physiology and Biophysics at CWRU. He has had a longstanding research interest in transport across cell membranes, especially as it pertains to acids and bases and the regulation of intracellular pH. This work unexpectedly led to the first descriptions of CO₂‐impermeable membranes and gas channels.