The Rate of Air-Sea CO2 Exchange: Chemical Enhancement and Catalysis by Marine Microalgae.

Appendix: Fluorescence Measurements of Carbonic Anhydrase

This appendix contains the details of fluorescence measurements of CA, the main conclusions of which were reported briefly in Section 3.5 . The principles behind this technique have already been introduced in Section 3.5 , and illustrated by the excitation-emission matrix spectra plotted in figure 3-7 .

A.1 Experimental Methods

A.1.1 Fluorescence measurement

Fluorescence was measured on a Perkin Elmer spectrofluorimeter. Cells were quartz, transparent down to 250nm if clean. Readings were taken at a matrix of pairs of excitation and emission wavelengths obtained by manually operating the monochromators. For investigating several different peaks in many samples, this is quicker than scanning linear spectra. There was a range of photomultiplier sensitivities: results reported here used sensitive settings, either "10" or "12". Setting "12" was found to give a reading 3.2 times higher than setting 10.

Readings were usually steady (to ± 0.1, except at 285/340nm: to ± 2.0) but occasionally large fluctuations occurred (up to 50%), particularly for solutions with a large background of light scattering. The needle usually returned to a consistent lowest reading, which is the one recorded. I made no attempt to compare fluorescence intensity with, for example, a quinine standard, or to calculate corrections for spectral variation of lamp intensity and photomultiplier response. This is only necessary for estimating quantum efficiencies. Slight temperature variation between samples did not seem to be very significant in this case. For batches of samples where fluorescence analysis took considerable time, initial samples readings were rechecked at the end, sometimes there was a slight loss of intensity, about 1-2%.

In theory, the sensitivity of fluorescence lies in detecting a different wavelength to that used for excitation, so apart from the narrow Rayleigh-Tyndall and Raman scattering which are well defined, there should be no "background" in the absence of fluorophores. However, because of the arrangement of the mirrors and monochromator gratings in the instrument, it is possible for light of the "wrong" wavelength to be detected if it approaches from an unusual angle. This occurs when there is multiple scattering by tiny particles present in the solution and seemed to be a problem in real seawater and microlayer samples.

A.1.2 Chemicals

Crystalline carbonic anhydrase (extracted from bovine erythrocytes and lyophilised) was obtained from Sigma. The label on the bottle indicated that it was 92% protein, with a catalytic activity of 6800 Wilbur Anderson Units per mg. Most Carbonic Anhydrases have a molecular mass of 30-40,000 per active site, or per zinc atom, and in the following text I have calculated concentrations on a molar basis using an assumed mass of 40,000 (the high value is compensated partially by the overweighing due to impurities in the crystals). The solid was stored in the refrigerator and its activity seemed to be stable over time, since in a test comparing CA solutions made up from a "new" bottle which had been stored for about a week with those made up from an "old" bottle which had been kept for three months, there was no significant difference in the CA-DA peak). The crystals were easily dissolved in distilled water and seawater, and produced a frothy solution. When shaken in a volumetric flask (between 2 and 20mg in 250ml), a few long stringy crystals sometimes reappeared, perhaps in the vortex created on shaking. These may not always have redissolved and constitute an possible error in concentrations. However, stock solutions for any one experiment can be considered consistent. New stock solutions were made up every two or three days.

Dansyl Amide was obtained from Aldrich. It is crystalline with a molecular mass of 250.38. Fluorescence measurements showed that solutions of DA kept in volumetric flasks at room temperature were stable over several weeks. DA is very sparingly soluble in distilled water or seawater, so dilute HCl was used to dissolve it for initial stock solutions, but this acid made a negligible contribution to the alkalinity of the final solutions in seawater (checked by pH measurement and calculation).

It was immediately apparent that the fluorescence of DA and of the CA-DA complex is very dependent on the solute used. Distilled water and seawater were compared. Seawater seems to be a suitable medium for CA-DA fluorescence, whereas in distilled water there was no sign of any energy transfer to make a CA-DA peak. The difference is probably mainly due to the difference in pH. Note that in distilled water, the DA fluorescence peak position was different from that reported below, but the solute did not seem to affect tryptophan fluorescence.

Therefore in all subsequent work, the bulk of all solutions for fluorescence analysis was a seawater medium. Initially, this was real seawater from the North sea, which had been collected about six weeks earlier and kept in the dark at room temperature. Later, to try to minimise any scattering effects and avoid any background fluorescence from protein or humic acid, artificial seawater was made up from salts (Na, Mg, Ca, K, Cl, SO4, HCO3, H3BO3) added to distilled water -the pH was checked and found to be about 8. Initially there was an inconsistency between samples which was traced to tryptophan fluorescence in the laboratory distilled water supply -thereafter MilliQ water was used. This reduced the background fluorescence but this was still large compared to the CA-DA peak for the lowest CA concentrations. Moreover this background fluorescence in "blank" samples seemed to vary with time -by up to 15% per day on one occasion.

A.2 Calibration experiments

Several calibration experiments were performed, in which a series of solutions was made up by varying either CA or DA concentrations, and the fluorescence recorded at the main peak positions and also for some "shoulder positions". The results of all these calibrations are shown in figure A-1 (a-k). The blue squares show the tryptophan peak, the red diamonds show the DA peak, and the magenta circles show the CA-DA peak. Other symbols show shoulders, minima and scattering. Figures (a)-(c) show real seawater, figures (d)-(k) show artificial seawater. Figures (a)-(g) show increasing CA concentration on the x-axis, figures (h)-(k) show increasing DA concentration. Due to the wide range of measurements various axis scales are used, so to check consistency between the sets of data, 3D CA/DA/fluorescence plots were also made (not shown here) combining all the artificial seawater data for each peak - generally they seemed consistent although there are a few strange points (e.g. figure (g)).

For high CA concentrations (20-100nM) there is an almost linear relationship between CA concentration and fluorescence at the CA-DA peak. However at CA concentrations below about 20nM the fluorescence at the CA-DA "peak" (285/460nm) is less than at the "shoulder" at 325/460nm and the minima at 285/380nm -i.e. the CA-DA "peak" is just a shoulder of the DA peak and background tryptophan fluorescence and scattering. This CA-DA "peak" still shows a smooth curve with increasing CA concentration, and an increase of only 5nM CA can be clearly detected. However the shape of the curve for the CA-DA peak is puzzling - it levels off at low CA concentrations, whereas we might expect either a straight line down to an intercept representing background fluorescence/scattering, or instead a curve levelling off at high CA concentrations due to fluorescence quenching and competition for DA. This is frustrating since we are most interested in the low concentrations. It might be explained by some process which reduces CA activity in the solution, by an amount which is relatively independent of the total CA concentration -for instance reaction with molecules present in limiting quantities in the seawater, consumption by microorganisms, or adhesion to the container.

As expected, the height of the DA peak is independent of CA concentration (except due to overlap from the CA-DA peak when this is very high) whereas the tryptophan fluorescence is almost linearly related to increasing CA concentration, except for a similar levelling off at the lowest concentrations -also suggesting that some of the protein is being lost. With DA concentration is 10-7 M the slopes of the tryptophan and CA-DA curves are similar, but with ten times more DA (10-6 M) the CA-DA slope is much greater, so the reaction has not yet gone to completion at 10-7 M.

Looking at the effect of increasing DA concentration, firstly we observe that the height of the DA peak responds linearly as expected, demonstrating that DA is always in excess. Fluorescence at the CA-DA "peak" increases with low concentrations of DA, reaches a maximum at about 20 times the CA concentration, and then falls off gradually, presumably due to fluorescence quenching by the much more abundant DA (as noted by Chen and Kernohan 1967). Fluorescence at the tryptophan peak declines sharply as the first DA is added. This is also to be expected since as the CA-DA complex is formed, direct emission from tryptophan is correspondingly reduced. According to Chen and Kernohan (1967), adding excess DA should reduce the tryptophan peak to 25% of its original height. However, their curve (using pH 7.4 buffer as a solute) shows a much more complete reaction than mine, and in that case the stoichiometry of the reaction was 1:1. In my case I need 20 times more DA than CA (in molar terms) to achieve optimum fluorescence., and similarly, the titration of bovine CA by DA by Newman and Raven (1993) (using pH 8.2 buffer as a solute) required 10 times more DA than CA. The difference may be related to pH. On the other hand we should not use to great an excess of DA as the "background" fluorescence (which is indicated by the measurements at 285/380nm, 285/520nm, 325/460nm, and 325/350nm) also increases gradually with DA concentration.

A.3 Measurements in seawater collected from an algal bloom in the North Sea

During June 1994 an intense spring bloom of Phaeocystis arose in the North Sea off Lowestoft. On 7th June I took the opportunity to collect some fresh seawater from a fishing boat at a location about 5 miles offshore, depth 25m, wind force 3-4, where Phaeocystis colonies could be seen by eye, about 200 "colonies" per litre. Subsurface samples were collected using a bucket and microlayer samples were collect using a metal mesh "Garret screen" of 80x80cm (see Section 2.4.2 ). About 100ml was collected per dip of the screen, which suggests a film thickness of less than 200mm. Samples were kept in plastic bottles and were not filtered. Some measurements were made immediately after returning to the laboratory (about 3 hours after sampling), while others were made the next day, the samples being kept in the cold store. Test solutions were made up of 80% sample, plus various amounts of CA and DA stock solutions. Figure A-2 shows the results of the fluorescence measurements at the three main peaks. Each group of three bars shows the effect of adding DA, the second and third sets in each group show the effect of added CA spikes.

Generally, the signal to noise ratio was much poorer in this real seawater, due both to a higher background fluorescence and rapid variation during measurement -this is indicated by the "error bars" in the figure and may be due to movement of Phaeocystis colonies within the fluorescence cell. Strangely, the CA-DA and tryptophan fluorescence seems to be about 25% greater in the samples analysed the following day -possibly because the colonies had begun to break up, raising the background protein fluorescence and scattering.

At the DA peak, there is a linear response to added DA, there is little effect of the CA spikes, and little difference between the samples -this is all as expected since DA is in excess. The tryptophan peak is more interesting -despite the large variation shown by the error bars. There is not much effect of added DA as expected, and only a small effect of the added CA spikes, but it seems that there is about twice as much protein in the microlayer samples compared to the bulk. Comparison was also made of tryptophan fluorescence (without any added DA or CA) of microlayer and subsurface samples from two other locations closer to the shore: at the site closest to the shore the microlayer reading was 17% higher than the subsurface, and at the middle site it was 63% higher, compared to 100% higher at the outermost site. Much greater enrichment of protein in the microlayer has been reported in the literature (see Section 2.4.4 ), but the enrichment depends on the thickness of the microlayer being sampled. For example if all the extra protein were in the top few micrometres then if a 200m m sample show enrichment by a factor of 2 (as here), a 20m m sample would show enrichment by a factor of 10. This problem applies to all methods of sampling the microlayer, as discussed in Section 2.4.2 .

On the other hand this apparent enrichment of protein in the microlayer sample may just be due to a scattering phenomenon -which would affect the tryptophan peak, close to the Raman scattering line, much more than the other peaks. Another possible explanation is that the "Garret screen" was catching subsurface Phaeocystis colonies larger than its mesh size as it was lifted up horizontally out of the water -hence the extra protein did not actually come from the microlayer.

What about the CA-DA peak? There is clearly an effect of adding DA in the samples spiked with extra CA, but the magnitude of this response varies between samples, which is strange since the amount of CA in the spikes (5 or 10nM) was the same in each case.

Therefore the response to added DA in the unspiked samples, which is very small, must be interpreted cautiously - the signal to noise ratio is too poor to state whether there was any CA present in the unspiked seawater, but if there was a little CA present then the concentration was lower than that of the lower spike (5 nM). Comparing the microlayer and bulk samples, the CA-DA peak is about 17% higher on average in the microlayer, but this may just be due to an increase in background fluorescence and scattering. Note that measurements were also made at the shoulder at 325/460nm: here the fluorescence was fairly constant but nearly twice as high as at the CA-DA peak, and was about 15% higher in the microlayer sample compared to the bulk.

Attempts were made to reduce the background fluorescence and scattering by removing small particles and Phaeocystis colonies. Centrifuging the samples at 3600rpm had no effect at all. Filtering the microlayer sample reduced the "tryptophan" peak considerably, by about 60%, and reduced the 325/460nm shoulder by about 20% but had little effect on the rest of the spectra. The noise was not reduced and I still couldn't see a small spike of CA with DA present.

Note that CO2 gas exchange measurements were made using water from a similar spring bloom of Phaeocystis two years later, as described in Section 8.3.5 . There might have been a slight catalysis by CA in the microlayer, but these experiments were similarly inconclusive.

A.4 Measurements of microlayer enrichment in a laboratory tank.

Even if the technique is not useful in real seawater due to background noise, it might be useful to investigate microlayer enrichment of added bovine CA in artificial seawater in a laboratory tank. For this purpose, the microlayer sampling method of Harvey and Burzell (1972) was used, in which a glass plate is slowly pulled out of the water vertically, and the sample collected by wiping off the adhering film. This method can collect a much thinner microlayer sample than a Garret screen, and it does not require a large tank. However many dips are needed to collect a sufficiently large sample for fluorescence measurement -about 30ml is needed altogether since to estimate the CA concentration solutions containing various combinations of CA and DA spikes must be made up from each sample. I used a tank of 30 cm x 45 cm x 10cm deep, and a 30cm square glass plate dipped to 10cm and wiped both sides with a rubber blade, the drips were collected in a small bottle. Dipping the plate 25 times gave 30 ± 3ml (the presence of CA didn't make much difference to the volume adhering, despite the visible lowering of surface tension in 10-7M solutions). Thus the microlayer sampled was about 20 m m thick, using about 1.5 square metres of surface film per sample. This is greater than the surface area of the tank but the microlayer should be at least partially regenerated at least between dips. For one sample, air was bubbled through the water before each dip of the plate to aid the process of surface enrichment.

From the calibrations above it seems that we can only reliably measure CA against a variable background at concentrations of at least 100nM. This is equivalent to 54mg of bovine CA in this volume of water. This was added as a solid directly into the artificial seawater in the tank, to reduce any losses from adhesion on volumetric flasks. On mixing a "froth" formed on the surface - to avoid sampling this froth the first ten dips were kept separately but analysis showed that it contained little more CA than the subsurface sample.

Microlayer and bulk samples were taken both before and after adding CA to the tank, and then a further sample was taken with bubbling to aid surface enrichment. The remaining results are shown in figure A-3. As before, the sets of three bars show the effect of increasing DA concentration. CA spikes of 20 nM were also added to each sample - the thin bars show the extra fluorescence due to the added CA spike. Note that these look similar to the "error bars" in figure A-2 but have a different meaning! The left three columns show fluorescence at the CA-DA, tryptophan, and DA peaks, the other columns show shoulders and a measurement at 325/350nm which is the "wrong" side of the Raman ridge and indicates variation in scattering. The vertical scales vary between the peaks but not between the samples. All the fluorescence readings for the CA+ bubbling microlayer sample containing 078 mM DA and no CA spike seem anomalously high compared to the no DA, 3.9 mM DA and CA spiked solutions -so there was probably some experimental artefact affecting this solution.

Comparing the microlayer and subsurface samples before adding CA, there is only a slight difference, tryptophan and scattering peaks seem slightly higher in the microlayer sample. After adding CA there is a clearer difference -the tryptophan peak is about 25% higher in the microlayer and the scattering is more than doubled. There is not a significant difference for the CA-DA peak without bubbling, but after bubbling the height of the CA-DA peak seems to have doubled for the microlayer sample. Bubbling causes even greater increases in the tryptophan and scattering peaks from the microlayer sample, but seems to have no effect on the subsurface sample, for any of the peaks. Not surprisingly the DA peak is least affected by added CA and microlayer enrichment. The shoulder at 285/520nm seems to mimic the response of the DA peak, and the response at 325/460nm is a combination of the response of the DA peak and the CA-DA peak.

Interpretation of these results is difficult because we do not know to what extent the increased scattering in the microlayer sample -which after bubbling is four times greater than the subsurface sample, contributes to an increased "background" fluorescence reading at the other peak positions. This scattering is clearly due to something in the surface film, not generated after or during dipping, since the bubbling makes so much difference.

It is possible that the laboratory air used for bubbling contained dust which accumulated at the surface, or that particles already in the water were driven to the surface by the bubbling.

Scattering should be reduced as the difference between excitation and emission wavelengths increases, but nevertheless we cannot be certain that it is not responsible for the apparent increase in tryptophan and CA-DA fluorescence in the surface microlayer. Moreover, there is considerable noise which is indicated by the effect of the added CA spike, which ought to be the same for all samples, microlayer and bulk, but which clearly varies considerably. However, a best guess, attempting to subtract the background noise from the CA-DA peak, is that the microlayer probably contained about twice as much CA as the bulk.

For a 20m m thick microlayer, we might expect a greater enrichment, and this would be necessary for the hypothesis that CA might significantly catalyse the global air-sea CO2 flux to be realistic (see Section 3.3 ). However, there are several reasons why this experiment does not adequately mimic conditions at sea. Firstly, the surface activity of bovine CA may be very different from that of the extracellular CA of marine algae. Such "extracellular" CA, whose presence is required to explain the physiology of many types of algae, will be likely to be particularly surface active and may also be bound to lipids of the plasmalemma (see discussion in Section 2.3.2 ). Secondly, the current understanding of the real sea-surface microlayer is that it is a complex organic web of molecules- protein, carbohydrate and humic acids - some of which may be created in situ by pkytoplankton (see Section 2.4.3 ). This would not be reproduced in this laboratory tank containing artificial seawater and no other organic molecules than the CA itself. Although in the real sea-surface microlayer there would be more "competition" for the actual surface, there would also be many other molecules to trap a molecule of CA in this web. Finally, I was using concentrations far higher than I expect in the real bulk seawater. The surface enrichment factor is probably greater at lower concentrations.

A.5 Decline in CA-DA fluorescence over time

There are several reasons why carbonic anhydrase activity might be expected to decline fairly rapidly in seawater samples. Firstly, CA is a large protein molecule whose activity depends not only on its chemical structure but also on its macromolecular structure - which is relatively fragile. Secondly, it may stick to sample bottles -indeed this would explain the observed levelling off of the calibration curves at low CA concentrations. Thirdly, it may be destroyed by protease enzymes produced by bacteria, even if all microorganisms have themselves been removed from the sample. The lack of any catalytic activity in seawater reported by previous workers (Goldman and Dennet 1983, Williams 1983) may be simply due to using seawater which had been stored for too long. To investigate this, some old samples containing CA and DA, whose fluorescence had already been measured, were checked again later.

The oldest samples had been kept for three weeks after adding CA and DA, and contained real but not fresh seawater (see earlier) stored in glass volumetric flasks at room temperature. The DA peak (325/520nm) was about 10% depleted in some samples, and virtually unchanged in others. On the other hand, less than 10% remained of the CA tryptophan peak (some of which is scattering), and less than 5% remained of the CA-DA peak. Other samples, based on the same original seawater but stored for about 10 days in plastic bottles, again showed no change in the DA peak. For these samples, the CA-DA peak, with background fluorescence subtracted (measured in a similar sample without added CA), was reduced to about 2% of its former value. The tryptophan peak, with background subtracted, was reduced to about 5% of its former value. In another sample with only 1/10 as much CA initially, more seemed to have survived: about 40%, although the signal:noise ratio was poor at this low concentration.

In three-week old samples stored in small plastic centrifuge tubes, the CA-DA peak decreased by a similar magnitude as above. Surprisingly, these samples also seemed to have lost some DA (up to 28%) and gained some tryptophan (up to double the background). However the background at 325/460nm had also increased, so this may just be a scattering effect.

This observed decline in CA activity over time was later confirmed by the experiments measuring CO2 air-water fluxes in the gas exchange tank, reported in

A.6 Potential improvements

In summary, fluorescence measurement of carbonic anhydrase using dansyl amide was possible at reasonably high concentrations of CA in old or artificial seawater (over 20nM), but in real seawater and in microlayer samples the signal to noise ratio became too great to make sufficiently accurate CA measurements at such low concentrations. Some useful conclusions were nevertheless drawn from these experiments, and the results have already been summarised in Section 3.5.3 .

Although it was decided not to continue using this approach, there are some possible ways this technique might be improved. Critically, it would be important to find out why the reaction does not seem to be going to completion, and what causes the levelling off of the calibration curves at low CA concentration. An improved optical system might reduce scattering, and it might even be possible to measure fluorescence in the microlayer in situ by passing the light along optical fibres (the tank would have to be in darkness). This would remove the need for sampling the microlayer, which may have been removing CA activity and introducing other contaminants.

Note: If you jumped into this page from elsewhere, you may find it more convenient to go to the "frames" version of this thesis, alternatively you can use the links below to jump straight to a particular chapter:

Title Page Abstract Contents Figures Overview Chapter1 Chapter2 Chapter3 Chapter4 Chapter5 Chapter6 Chapter7 Chapter8 Chapter9 Chapter10 Appendix References Acknowledgements Links Download