Technical Comments

Response to Comment on “Enhanced water permeability and tunable ion selectivity in subnanometer carbon nanotube porins”

See allHide authors and affiliations

Science  30 Mar 2018:
Vol. 359, Issue 6383, eaaq1241
DOI: 10.1126/science.aaq1241


Horner and Pohl argue that high water transport rates reported for carbon nanotube porins (CNTPs) originate from leakage at the nanotube-bilayer interface. Our results and new experimental evidence are consistent with transport through the nanotube pores and rule out a defect-mediated transport mechanism. Mechanistic origins of the high Arrhenius factor that we reported for narrow CNTPs at pH 8 require further investigation.

At this point, high water permeability of small-diameter carbon nanotubes is hardly a surprise. The unitary permeability of 0.8-nm-diameter carbon nanotube porins (CNTPs) that we report for water transport at pH 7.8 (1), 6.8 × 10−13 cm3/s, is comparable to the 5.1 × 10−13 cm3/s value predicted by molecular dynamics simulations for a carbon nanotube of similar diameter (2). Horner and Pohl (3) themselves predicted that transport through a narrow carbon nanotube pore should be faster than transport through aquaporins and they have estimated the water diffusion coefficient in these pores to be on the order of 3 × 10−5 cm2/s (4), which is comparable to the values of 0.9 × 10−5 cm2/s and 4.4 × 10−5 cm2/s that we report for water transport at pH 7.8 and 3.0, respectively (1). Barati Farimani and Aluru (5) predicted a diffusion coefficient of 1.2 × 10−5 cm2/s for (7,7) carbon nanotubes, which also accommodate a single-file arrangement of water; again, this is close to our reported values. High water transport efficiency has been measured for larger-diameter carbon nanotubes (6, 7) and for 1.36-nm stacked graphene nanochannels (8). Thus, our reported unitary CNTP water permeabilities are consistent with the broad range of fast transport predictions and observations in the literature.

Horner and Pohl suggest that water transport through the carbon nanotubes can be verified by using a species that can obstruct the water transport through the pore. To follow this suggestion, we measured water transport through 0.8-nm-diameter CNTPs in the presence of Ca2+ ions. We previously reported that Ca2+ ions obstruct proton transport through the narrow CNTP pore by binding to the negatively charged groups at the CNTP mouth (9); thus, it was reasonable to expect that the same mechanism could slow down water transport. Indeed, the measured unitary water permeability of 0.8-nm-diameter CNTPs decreased by a factor of >5 in the presence of 10 mM CaCl2 (Fig. 1A). We also note that our previous measurements showed that wider 1.5-nm-diameter CNTPs can be partially obstructed by single-stranded DNA molecules (10), with the magnitude of the blockade matching the value expected for a 1.5-nm channel; this finding again supports the idea that molecular transport occurs through the interior of the nanotube.

Fig. 1 Water transport obstruction by Ca2+ ions in CNTPs, and topography of CNTP in lipid bilayers.

(A) Stopped-flow kinetics of osmotically driven size change of vesicles containing 0.8-nm-diameter CNTPs in the presence and absence of 10 mM CaCl2. Inset shows the full range of transport kinetics in the presence of CaCl2. Experiments followed the methodology described in (1). (B) HS-AFM image of CNTPs functionalized with a neutral moiety in a lipid bilayer supported on mica. Experiments followed the methodology described in (11).

The suggested mechanism of transport mediated by the pH-dependent defects in the lipid bilayer created by fully or partially buried nanotubes in the bilayer interior is also incompatible with a number of observations reported in our past and present work. First, larger-diameter nanotubes should create more (and larger) defects; however, we observed substantially smaller unitary permeability for wider 1.5-nm-diameter CNTPs. This observation is again consistent with predictions that single-file water transport in subnanometer carbon nanotube pores should be faster than the more bulk-like transport in larger-diameter nanotube channels. Second, the defect-mediated transport mechanism is largely incompatible with the ionic conductance properties that we reported (1). An increased number of defects at a lower pH, as suggested by Horner and Pohl, would simply lead to an increase in overall conductance at all ionic strengths. Instead, we observed a qualitative transition from a saturating conductance characteristic at pH 7.8 to a nearly linear conductance characteristic at pH 3.0, which is fully consistent with ion transport through a narrow pore with charged groups at the entrance that neutralize at a lower pH. Our previous measurements of proton transport showed that Ca2+ ions are less effective in blocking the transport through 1.5-nm CNTPs than through smaller 0.8-nm CNTPs (9), again consistent with transport through the CNTP interior.

We have previously studied the configuration of CNTPs in lipid bilayers by means of cryo–transmission electron microscopy (cryo-TEM) (10) and high-speed atomic force microscopy (HS-AFM) (11). HS-AFM data show that CNTPs protrude above the bilayer plane, and cryo-TEM data indicate that the nanotubes are tilted with respect to the normal axis of the bilayer by an average of only 15°; furthermore, we do not find experimental evidence of a prevalent highly tilted configuration, as suggested by Horner and Pohl. Figure 1B shows a HS-AFM image of a supported bilayer containing these CNTPs functionalized with neutral methylamide groups; in preliminary experiments, they display water permeability and an activation energy at pH 8 comparable to those of unfunctionalized CNTPs at pH 3. Such findings indicate that CNTPs are still protruding from the bilayer plane, with an average surface density matching the expected CNTP loading of the vesicles. These observations do not support the suggestion by Horner and Pohl that increased permeability observed at pH 3 is due to complete burial of the nanotubes within the bilayer and creation of associated lipid packing defects. Taken together, these data strongly argue that transport in our system occurs through the carbon nanotube pores, not through defects in the nanotube-bilayer interface.

Horner and Pohl also raise an issue with the method we used to analyze our stopped-flow kinetics. We relied on a widely used approach by the groups of Zeidel (12) and Agre (13) and have reproduced literature-reported water permeability and activation energy values for aquaporin pores and background lipid bilayer water permeability in control experiments (1). We have also tried to use the analytical solution suggested by Horner and Pohl (4) and found that it fits our data poorly.

A wider survey of the literature shows that the relationship between unitary water permeability of biological channels and activation energy is more complicated than what a simple transition state theory approach suggests; for example, the permeability of the GlpF protein, 1.1 × 10−12 cm3/s (3), is more than 5 × 103 times the value expected on the basis of its activation energy of 7 kcal/mol (14). In 2001, de Groot and Grubmüller also commented that for water transport in protein channels, “considerable mismatch is seen between measured rates and corresponding activation energies”; they suggested that “observed high water-permeation rate is achieved through the highly collective motion of water molecules” (15). Similar collective motion of water molecules has also been observed in water transport simulations in carbon nanotubes (2). Simulations and recent experiments also reported unusual temperature-dependent effects and structural transitions for water in carbon nanotube pores (16, 17), although corresponding effects on activation energy are much less well understood. Even though carbon nanotube pores are often compared to aquaporins, they are not identical. Simulations show that the aquaporin structure is engineered to offer “replacement interactions” by two specialized amino acid motifs whenever the water-water bonds are broken (18). There is no reason to believe that the more primitive structure of the carbon nanotube rim can lower the activation energy to the same extent. Instead, fast transport is enabled by the smooth graphitic nanotube walls (2, 19). We believe that unraveling the details of the energy barriers for water transport through small-diameter carbon nanotubes requires additional studies, and we are actively pursuing them.

References and Notes

Acknowledgments: Additional experiments performed for this response were supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering. Author contributions: A.N., R.H.T., and Y.Z. designed additional experiments, which were performed by R.H.T. (transport measurements) and Y.Z. (HS-AFM). All authors participated in writing and editing the response.
View Abstract

Navigate This Article