A Mechanistic Study of Carbonic Anhydrase-Enhanced Calcite Dissolution

Abstract

Carbonic anhydrase (CA) has been shown to promote calcite dissolution (Liu, 2001, https://doi.org/10.1111/j.1755-6724.2001.tb00531.x; Subhas et al., 2017, https://doi.org/10.1073/pnas.1703604114), and understanding the catalytic mechanism will facilitate our understanding of the oceanic alkalinity cycle. We use atomic force microscopy (AFM) to directly observe calcite dissolution in CA-bearing solution. CA is found to etch the calcite surface only when in extreme proximity (~1 nm) to the mineral. Subsequently, the CA-induced etch pits create step edges that serve as active dissolution sites. The possible catalytic mechanism is through the adsorption of CA on the calcite surface, followed by proton transfer from the CA catalytic center to the calcite surface during CO₂ hydration. This study shows that the accessibility of CA to particulate inorganic carbon (PIC) in the ocean is critical in properly estimating oceanic CaCO₃ and alkalinity cycles.

Key Points

• Carbonic anhydrase induces etch pit formation when in contact or in extreme proximity (~1 nm) with calcite
• The catalysis is likely via the adsorption of carbonic anhydrase on the mineral, followed by proton transfer to the calcite surface
• The accessibility of the enzyme to particulate inorganic carbon is vital in properly estimating oceanic CaCO₃ and alkalinity cycles

Plain Language Summary

Calcite dissolution in aqueous solution is one of the most biogeochemically important reaction and has a significant impact on atmospheric CO₂ regulation and climate change. Carbonic anhydrase (CA) is an enzyme that regulates pH and CO₂ balance inside or surrounding organisms in aqueous and terrestrial environments and is shown to promote calcite dissolution. In this study, we investigate why this ubiquitous enzyme enhances calcite dissolution rates in order to facilitate the proper estimate of dissolution fluxes in natural environments. When observed at atomic scale, we find that the enzyme etches the mineral only in extreme proximity. This observation provides important clue for the catalysis mechanism and reveals that the accessibility of CA to CaCO₃ in the ocean is critical in properly estimating the natural fluxes.

1 Introduction

The enzyme carbonic anhydrase (CA) promotes CO₂(aq) equilibration with the rest of the ambient dissolved inorganic carbon (DIC) species. Biological processes, which require either rapid production or consumption of CO₂, need enzymatic catalysis of the CO₂ hydration reaction, and CA has been found at the site of biomineralization in many marine calcifiers (Miyamoto et al., 1996; Moroney et al., 2001; Rahman et al., 2008; Tambutté et al., 2007). In addition to its effects in vivo, CA has also been experimentally shown to catalyze the dissolution kinetics of calcium carbonate, in freshwater karst environments far from equilibrium (Liu, 2001; Liu et al., 2005) and in seawater near equilibrium (Subhas et al., 2017). The elevated carbonate dissolution rates in the presence of CA may have critical, yet underestimated, impact on the rate of chemical weathering (Liu et al., 2005; Thorley et al., 2015; Xie & Wu, 2014) and the rate at which alkalinity is recycled in the ocean (Subhas et al., 2019). In fact, the substantial amount of calcium carbonate dissolution in middepths (600 to 900 m) of the water column in the North Pacific (Berelson et al., 2007; Feely et al., 2002) likely occur within confined acidic environments such as diffusively limited marine snow aggregates or the guts of zooplankton/metazoan (Dong et al., 2019), and this leaves open the possibility that CA produced by marine organisms can catalyze the dissolution of calcium carbonate in these confined settings in situ (Subhas et al., 2019).

Isoforms of CA can be found intracellularly and extracellularly, bound to outer cell walls or membranes, or untethered in organisms (Elzenga et al., 2000; Martin & Tortell, 2008; Mustaffa et al., 2017). For laboratory experiments in solution, the quantification of CA catalysis on carbonate dissolution was measured as a function of the equivalent activity of bulk CA (bovine) concentration (Liu et al., 2005; Subhas et al., 2017). The catalysis effect was attributed to the removal of CO₃²⁻ as the production of H⁺, therefore maintaining undersaturation (Dreybrodt et al., 1996), or the increase in H₂CO₃ availability (Subhas et al., 2017). Recent experiments in benthic chambers, however, showed no obvious catalysis in reef sediment dissolution when CA was added to the overlying seawater (Kessler et al., 2020). As a result, whether the catalysis requires direct interaction of CA and the carbonate surface or occurs by altering local water chemistry, and if the latter, how “local” is the effect, become significant in identifying where and how much does CA promote carbonate dissolution in natural environments.

This study uses atomic force microscopy (AFM) to investigate how CA catalyzes calcite dissolution in seawater. The catalytic mechanism of CA is discussed based on the observed dissolution patterns in CA-free and CA-bearing seawater, and these results both deepen our understanding of CA catalysis mechanisms and shed light on the role of localized calcite dissolution in the context of the global carbon and alkalinity cycles.

2 Materials and Methods

Calcite (104) surfaces were obtained by using a razor blade to cleave a large crystal of optical-quality Iceland spar. The experimental solution was standard reference Dickson seawater (https://www.nodc.noaa.gov/ocads/oceans/Dickson_CRM/batches.html), acidified to the desired saturation states by adding HCl. The undersaturated seawater was prepared in and stored in gas-impermeable bags with no headspace. DIC and alkalinity were measured to determine the saturation state ( , where K^*_sp is the stoichiometric solubility product), using carbonate equilibria parameters and formulations described in Dong et al. (2018) as prescribed in the worksheet CO2SYS (https://www.nodc.noaa.gov/ocads/oceans/CO2SYS/co2rprt.html). In CA-bearing seawater, lyophilized CA from bovine erythrocites purchased from Sigma-Aldrich (C2624) was added to seawater with the total [CA] = 0.04 mg ml⁻¹. This concentration was one of the higher values used by Subhas et al. (2017) who demonstrated a large effect at this level.

Experiments were conducted at 21°C and 1 atm. In situ fluid cell imaging was conducted using an Asylum Research Cypher ES Environmental Atomic Force Microscope. The commercially available AFM probes we used were Arrow UHFAuD from Asylum Research (https://afmprobes.asylumresearch.com/arrow-uhfaud.html) and SNL-10 from Bruker (https://www.brukerafmprobes.com/p-3693-snl-10.aspx). Because the Cypher AFM had a gas headspace in the fluid cell (~3 ml), the headspace was manually adjusted by adding CO₂-controlled gas that was in equilibrium with the solution. Alkalinity and DIC measurements of the influent and effluent solution confirmed that Ω remained constant throughout the experiment. All experiments were conducted at a flow rate of 15 ml hr⁻¹. At this flow rate, water was in contact with the mineral surface for ≤1 min (Dong et al., 2020).

Step retreat velocities were measured as half of the etch pit-widening rates to eliminate the potential artifact of drifting between images. Because opposing steps in an etch pit are different step types (acute vs. obtuse), our pit-widening velocity is the average of acute and obtuse step velocities. Uncertainty in step velocity was determined as the standard error of step velocities at 1–8 different etch pits and at 2–6 different time periods, with the time periods averaging a few minutes per interval. In experiments that had no etch pits, step retreat velocity was determined as the movement of edge fronts when no obvious drifting was detectable between images. More details of the experimental materials and methods can be found in Dong et al. (2020).

3 Results

The average step velocities of acute and obtuse edges in dissolution etch pits are similar in dissolved CA-bearing seawater (0.04 mg ml⁻¹) compared to CA-free seawater (Figure 1a and supporting information Table S1). Because the step velocity is dependent on surface properties of the crystal (e.g., defect densities), which may vary between samples, and even different regions of the same sample (scatter in Figure 1a), we conducted three experiments at Ω ~ 0.86, 0.73, and 0.51 during which we used the same Iceland spar sample and flowed CA-bearing seawater immediately after CA-free seawater to eliminate the potential effect of surface properties on dissolution rate (Figure 1b). The slight difference in Ω between the CA-free and CA-bearing seawater was corrected for by normalizing the rates to the same Ω using the rate-undersaturation correlation in Figure 1a. The ratios of step velocity with and without dissolved-phase CA are not significantly higher than 1 for all three experiments (Figure 1b and Table S1). According to Subhas et al. (2017), CA enhances the dissolution rate of calcite across all saturation states, and the effect is most pronounced close to equilibrium. In Subhas et al. (2017), a rate increase of ~2.5 orders of magnitude at [CA] = 0.04 mg ml⁻¹ is observed at Ω = 0.85; whereas far from equilibrium, the enhancement is a factor of 2–3; yet we saw no difference in step retreat velocities.

Figure 1

(a) Average step velocity of acute and obtuse edges of Iceland spar in seawater versus undersaturation, with and without CA. Trend lines are fittings to all “no CA” rates. Each data point is an individual dissolution experiment using a new calcite crystal. (b) Ratios of step velocity with dissolved-phase CA and without CA, by flowing the CA-bearing solution immediately after the CA-free solution onto the same calcite crystal. The absolute step velocities at Ω ~ 0.86 and 0.74 are not included in panel (a), because the step velocities are not the average of acute and obtuse step retreat velocity.

During experimental solution preparation, we noted that CA-bearing seawater became very surfactant-like and coagulated particles were seen in the solution. We filtered the solution (Ω = 0.78, [CA] = 0.04 mg ml⁻¹) through a 0.2 μm (pore size) membrane disk both immediately after the solution was made and after 2 days of shaking to determine the amount of CA aggregate in the seawater. Even though the Stokes radius (the hydrodynamic radius of a macromolecule with a collection of multiple subparticles) of bovine CA is only 2.1 nm (La Verde et al., 2017), our experiment shows that 0.6% of CA is captured on the 0.2 μm filter during filtration immediately after the solution is made, whereas after 2 days of shaking, up to 16% of CA in this solution becomes flocculants that are captured on a 0.2 μm filter.

During one dissolution experiment in CA-bearing seawater at Ω = 0.83, a CA aggregate (the irregular-shaped white regions in Figure 2a) appeared in the observation window of the AFM and landed on the calcite surface. Within 3 min, a massive amount of dissolution was observed underneath the CA aggregate (dark trenches indicated by the yellow arrows in Figure 2), whereas no obvious change was found at an etch pit not in contact with “solid” CA (indicated by the black arrow in Figure 2a). Because the shape of the CA aggregate changed slightly and small aggregates were free floating as seawater flowed through the AFM fluid cell, it was difficult to identify from the images whether the dissolution sites adjacent to the main CA aggregate was covered by a CA aggregate at some point within the 3 min. Nevertheless, it is clear that CA-promoted dissolution is an extremely local process where particulate CA touches or is close enough (~1 nm according to the color scale of the images) to the calcite surface, dissolution is strongly enhanced. Height profiles along a horizontal transect indicate that the rate at which the “etching” happens at the CA-calcite interface is on the order of 1 monolayer (0.3 nm) per minute (Figure 2b).

Figure 2

(a) CA aggregate (irregular white regions) in contact with calcite surface and its effect on dissolution at Ω = 0.83. The straight horizontal dark stripes near the CA aggregate are artifacts in the AFM imaging process that attempt to balance the total gray scale in individual rows and are not trenches on the surface. The irregular-shaped dark areas, however, are real etched features (indicated by the yellow arrows). (b) Height profiles along the horizontal transect between the open black squares at 0 and 3 min. The transect at 3 min is not marked as a dashed line in panel (a), so that the etched morphology can be better seen. (c) Simple illustration of CA-induced dissolution on calcite surface. Note that some CA aggregates did not induce dissolution (e.g., the blob that appeared at 2 min next to the black arrow in panel a), likely because they did not touch the surface.

This phenomenon we describe as contact-associated dissolution was further confirmed by scanning with a CA-coated AFM tip. Specifically, we added a droplet of CA-free seawater at Ω = 0.56 on top of a calcite crystal. We then picked out a large coagulated CA aggregate from CA-bearing seawater with tweezers and placed it on the calcite surface without touching the seawater droplet. We scanned the calcite surface in CA-free seawater to obtain the initial morphology of an area. Then we moved the AFM probe to the CA aggregate and submerged it into the CA for ~10 s. Finally, we moved the AFM probe to the previous area and scanned again. CA attached to the probe immediately created nonrhombic etching patterns, which subsequently reshaped to rhombic pits when CA floated away from the imaging tip (Figure S1).

From these experiments, we conclude that the catalysis effect of CA on carbonate dissolution is attributed to the creation of “etch pits” when CA touches the carbonate surface. The removal of one calcite monolayer that is in contact with CA happens within seconds (<1 min). These observations provide supporting information for the bulk rate—undersaturation correlation with and without CA reported in Subhas et al. (2017) and also discussed by Dong et al. (2018) and Naviaux et al. (2019). Far from equilibrium (orange area in Figure 3), etch pits can form homogeneously on calcite surfaces in the absence of CA, whereas near equilibrium (blue area), the chemical potential of the solution is smaller than the free energy needed for nucleation of an etch pit, and dissolution only occurs at existing edge fronts as step retreats. Between these two states, dissolution is dominated by the formation of etch pits at defect centers. For uncatalyzed dissolution rates in seawater, the transition of dominating dissolution mechanism is indicated by the change of slope in the log rate versus log (1 − Ω) correlation, and dissolution near equilibrium falls below the trend line extended from low Ω because the generation of new edges (dissolution-active sites) is not occurring close to equilibrium. Our key new understanding is that CA in contact with calcite surfaces immediately opens new etch pits for enhanced dissolution. The effect of this etch pit generation far from equilibrium is to enhance dissolution rate by only a factor of 2–3 at [CA] = 0.04 mg ml⁻¹, because the CA-created etch pits are adding to those already generated from the bulk solution chemistry. The addition of CA near equilibrium, however, increases dissolution rate substantially by creating proportionally many more new edge fronts for subsequent dissolution when dissolution rate is limited by the formation of etch pit due to low chemical potential of the solution. As a result, the addition of CA removes the kink in the rate-undersaturation correlation.

Figure 3

The relationship between saturation state, CA concentration, and calcite dissolution rate in bulk dissolution experiments in seawater reported in Subhas et al. (2017). The background colors indicate the transition of dominating dissolution mechanisms from step retreat (blue), to defect-assisted (white), to homogeneous etch pit spreading (orange) as a function of (1 − Ω).

Our new observation advances the theoretical analysis of the physical-chemical parameters reported in Subhas et al. (2017), based on the crystal growth and dissolution theory in Dove et al. (2005). Subhas et al. proposed that the addition of CA increases either the density of nucleation sites (n_s) or the rate of step retreat (β) and meanwhile decreases the free energy barrier to etch pit nucleation (α) by increasing the concentration of carbonic acid at defects on the calcite surface near equilibrium. Our new results show that CA increases the density of nucleation sites n_s by etching the mineral surface locally but does not change the step retreat velocity β. The decrease of α cannot be fundamentally demonstrated by the experiments but is likely due to the production of protons at the CA active center and the subsequent proton transfer to the mineral surface, making it easier to form etch pits compared to the dissolution process without CA.

Based on the observation that calcite dissolution is only enhanced where CA is in close proximity with the mineral surface, a mechanistic model of this local process is provided below. First, the hydrophilic sites on CA that are electrically charged interact with the calcite surface and “attach” the enzyme to the mineral. Semiempirical quantum-mechanical calculations were able to show the net charges of different side-chain atoms in bovine CA II (BCA II) at pH 7.5 (Saito et al., 2004), and the interactions between these partially charged side-chain atoms and the Ca²⁺/CO₃²⁻ on the calcite surface likely drive the adsorption of CA to calcite. The amino acids with electrically charged side chains (arginine, histidine, lysine, aspartic acid, and glutamic acid) in the CA ribbons “grab” the mineral surface and serve as the premise of enhanced dissolution. After the CA molecule is attached to the calcite surface, protons produced during water protolysis are transferred from the CA catalytic center to the mineral and generate localized dissolution (Figure 4). The active site of CA contains a Zn^II ion with a bound hydroxyl group (Zn^II-OH) surrounded by three histidine residues held in a distorted tetrahedral geometry (Figure 4a). The Zn^II-bound hydroxy group attacks CO₂ to initiate hydroxylation and produce bicarbonate, which is displaced from the Zn^II ion by a molecule of water (Silverman & Lindskog, 1988). The Zn^II-bound water loses a proton to generate a new Zn^II-OH for another round of catalysis (Krebs et al., 1993; Krebs & Fierke, 1993) (Figure 4a). It is generally accepted that this proton is shuttled to the bulk solution by a series of intra and intermolecular proton-transfer steps (An et al., 2002; Ren et al., 1995) (Figure 4b). The transfer of a proton from Zn^II-bound water to the bulk solution is the rate-limiting step in catalysis (Becker et al., 2011; Silverman & Lindskog, 1988). The catalytic turnover (k_cat) for BCA III is 6,400 s⁻¹ (Krishnamurthy et al., 2008; Ren et al., 1988), and the upper limit of the CA-enhanced dissolution is set by this proton removal rate, as well as the CA concentration and its frequency of attachment to the calcite surface.

Figure 4

An illustration of a hypothetical mechanistic method of CA-promoted dissolution on calcite. (a) The proton generated during the catalysis of CO₂ hydration attacks calcite (104) surface and promotes dissolution. (b) The adsorption of CA on the calcite (104) surface and the intramolecular proton transfer from the catalytic center to the mineral surface.

Even though the binding energy between CA and the mineral surface was not experimentally determined, we believe that CA-mineral contact is necessary for the enzyme-catalyzed reaction. The high efficiency of enzymatic reaction is orientation specific, instead of random collision. Free CA in solution phase cannot achieve direction specificity, and hence, the reaction will be significantly less efficient without the CA-mineral contact. We further calculated the time that a free-floating aggregate would stay within the AFM observation window in a laminar flow boundary layer and compared it to how long the CA aggregate was observed experimentally. At our experimental conditions, the thickness of the laminar boundary layer above the calcite plate is about 1.3 mm. At 10 nm above the calcite surface (relevant to AFM observation), it takes only ~5 s to pass through the 4 μm wide observation window as a free-floating particle in the laminar flow, significantly shorter than the time that the CA aggregate stayed within the scope in Figure 2 (>3 min). Therefore, we think that the observed CA aggregate was attached to the mineral surface as the electrically charged side chains on CA interact with the calcite surface.

4 Geological Implications

Respiration of organic matter in sinking marine particles and coral reefs can produce locally acidic microenvironments and promote CaCO₃ dissolution in waters that are otherwise supersaturated (Jansen et al., 2002). Meanwhile, CA activity in these carbonate-rich environments would significantly enhance the rate at which alkalinity is cycled between solids and the undersaturated seawater (Eyre et al., 2018; Subhas et al., 2017). The observation in this study that extremely close proximity (<1 nm) is necessary for CA to promote CaCO₃ dissolution indicates that estimates for enhanced dissolution in natural environments should be based on the amount of CA readily accessible to carbonate surfaces in extremely local scales. This conclusion is likely responsible for the absence of CA catalysis in reef sediments in Kessler et al. (2020), as CA dissolved in the overlying seawater cannot access the carbonates in the sediment column.

CA exists both intracellularly (iCA) and extracellularly (eCA) and is often associated with the cell wall, plasma membrane, or periplasmic space (Badger, 2003; Burkhardt et al., 2001; Elzenga et al., 2000; Hopkinson et al., 2013; Moroney et al., 1985; Nimer et al., 1999; etc.). In natural seawater, eCA extracted from cell membranes is measured to be (3.0–20.1) × 10⁻⁶ mg CA g⁻¹ seawater in the Baltic Sea (Mustaffa et al., 2017) and (0–5.1) × 10⁻⁹ M in the North Pacific (Subhas et al., 2019). Subhas et al. (2019) also show that ~90% of the CA activity is externally bound in natural oceanic environments by showing similar CA activities between sonicated and vortexed samples. In the North Pacific, CA activity in sinking particles normalized to particulate organic carbon (POC) follows a single relationship (CA/POC = (1.9 ± 0.2) × 10⁻⁷ mol mol⁻¹), consistent with CA/POC values in plankton tow materials and picked pteropods during the same cruise, whereas suspended particles have generally lower and varying CA/POC values of (0–2) × 10⁻⁷ mol mol⁻¹. This ratio could potentially vary in other oceanographic provinces with different phytoplankton communities, as diatoms commonly have extracellular CA whereas cyanobacteria do not (Hopkinson et al., 2013). Scaled to the volume of individual pteropod shell and marine snow particles, Subhas et al. (2019) calculated localized CA concentration of 0.06–0.32 mg CA g⁻¹ seawater using the POC concentration and CA/POC ratio in these microenvironments. These CA concentrations are comparable to measured CA activities in extracellular diatom boundary layer and their internal compartments (0.09–6.8 mg CA g⁻¹ seawater, Hopkinson et al., 2011, 2013) and are higher than those documented to catalyze calcite dissolution in natural seawater (0.01–0.04 mg CA g⁻¹ seawater, Subhas et al., 2017). Despite the new progress made in quantifying eCA in natural seawater, whether the eCA associated with the cell membranes can contact and interact with the carbonate mineral is still an open question and requires further investigation. Overall, to what extent can CA enhance particulate inorganic carbon (PIC) dissolution rates, and thus the marine alkalinity cycle highly depends on local availability of CA to PIC, should be considered as an important yet still unconstrained aspect of oceanic carbon cycling.

5 Conclusions

From our novel use of AFM, no significant difference in step retreat velocity is observed between calcite dissolution in CA-free and CA-bearing seawater. Enhanced dissolution is only detected on a carbonate surface that is in extreme proximity with (~1 nm) CA aggregates. The dissolution pattern closely resembles the outline of CA aggregates that are observed sitting on the mineral surface. The interaction of CA and calcite generates irregular etching patterns that do not follow the rhombohedral morphology of normal calcite etch pits. The generation of etch pits by CA enhances calcite dissolution rate by a modest factor of 2–3 far from equilibrium when etch pits can form homogeneously at the saturation state without CA. Near equilibrium, when etch pits cannot form at the chemical potential of the solution, the edge fronts CA introduces can elevate dissolution rates up to 2.5 orders of magnitude. The possible catalytic mechanism for enhanced etch pit formation is through the adsorption of CA on the calcite surface, followed by proton transfer from the zinc ion to the calcite surface during the conversion of CO₂ to bicarbonate. These results imply that caution is required in estimating the rate enhancement of carbonate rock weathering and oceanic PIC dissolution with bulk CA concentrations in natural environments, because the key to elevated dissolution rates is the contact that occurs between CA and the mineral surface.

Acknowledgments

This work was supported by the National Science Foundation (NSF) Ocean Acidification grants (OCE1220600, OCE1220302 and OCE 1559004) and the University of Southern California (USC) Dornsife Doctoral Fellowship. We thank Adam V. Subhas for his helpful discussions in preparing the experiments and the manuscript.

Open Research

Data Availability Statement

Data set for this research is available in this in-text data citation reference: Dong, S., Berelson, W., Teng, H., Rollins, N., Pirbadian, S., El-Naggar, M., Adkins, J. (2020). Step velocities during calcite dissolution in seawater with and without carbonic anhydrase, version 1.0. Interdisciplinary Earth Data Alliance (IEDA). https://doi.org/10.26022/IEDA/111627. Accessed 2020-09-08.