OVERVIEW

In this area of research we seek to develop methods for the rapid identification and enumeration of microorganisms based on antibody-antigen interactions. We are pursuing several approaches with different sensitivities (and also different degrees of research complexity). The holy grail here is real-time, in situ operation. This is a highly desirable but as yet unrealized goal in aquatic microbial ecology.

APPROACH

We are studying new approaches to measure the abundances of specific microorganisms without the drawbacks of currently available DNA-based or immunological methods, which destroy the cells and tend to be slow and elaborate. We have taken several parallel approaches, all of which are immunological, i.e., based on antibody-antigen interaction. The first of these is aimed at improving current immunological methods by using flow cytometry. This work has reached sufficient maturity to be utilized routinely in the lab, as explained below. We also have compared the results obtained from flow cytometry with those from ELISA (Enzyme-Linked Immunosorbent Assay), and reached interesting conclusions, also discussed below.

We have also examined the feasibility of using a Quartz Crystal Microbalance (QCM) for the detection of harmful algae. This approach uses an antibody-functionalized crystal to detect A. anophagefferens cells in a liquid.

Research employing Atomic Force Microscope is aimed at detecting microorganisms using tactile sensing — in essence, by determining the force between an antigen on the organism’s cell surface (A. anophagefferens) and an antibody attached to the AFM tip. This approach is expected to be the most sensitive of the three and should be able to detect single cells.

In addition to these three methods, we decided to study a fourth one, which involves nanowire sensors. This approach appears to be more suitable for deployment in situ and may be as sensitive as the AFM methods. Sensing for cell identification and counting is such a crucial issue in our problem (and many others that involve cell detection) that we believe that understanding the capabilities and limitations of several approaches is essential for our progress.

For this work we have been using a harmful bloom-forming species of marine algae, A. anophagefferens (the cause of ‘brown tides’ in the NE U.S.) as the ‘target’ species. We have previously obtained a monoclonal antibody (MAb) that is highly specific for cell surface antigens in this alga, and we have been employing the antibody and the alga as a model system for testing the feasibility of our novel approaches to detecting the alga in a liquid environment.

SYSTEMS / EXPERIMENTS

Extension/improvement of existing biological detection methods for the brown tide alga

Present methodology for enumerating A. anophagefferens in natural water samples involves the application of a highly-specific monoclonal antibody (MAb) in an enzyme-linked immunosorbent assay (ELISA) (Caron DA, Dennett MR, Moran DM, Schaffner RA, Lonsdale DJ, Gobler CJ, Nuzzi R, McLean TI (2003) Development and application of a monoclonal antibody technique for counting Aureococcus anophagefferens, an algal causing recurrent brown tides in the mid-Atlantic United States. Applied and Environmental Microbiology 69:5492-5502).

Figure 19: Comparison of ELISA and flow cytometer counts of two A. anophagefferens cultures. R2=0.8972 and 0.8813 for the two cultures.

We have shown that counts of Aureococcus anophagefferens using the fluorescently-labeled MAb and flow cytometry can also be accomplished. However, differences between ELISA counts and counts by flow cytometry exist (R2=0.889, Figure 19) especially in the presence of grazing. The cytometric method relies on size characteristics of the intact cells, in addition to immunological reactivity, to discriminate target cells in a sample whereas the ELISA method relies solely on the binding of the MAb to antigens in the particulate fraction of a sample. This difference accounts for the limited correlation of the two sets of data (Figure 19).

Comparison of algal abundances derived using the flow cytometer method were compared directly with microscopy-based counts using a hemacytometer to help resolve discrepancies between ELISA and flow cytometry. Use of the hemacytometer allows quantitive cell counts of cultures. This method does not incorporate the use of the monoclonal antibody, and thus is independent of any bias introduced by use of the MAb. A pure culture of A. anophagefferens was serially diluted in preserved, filtered seawater to obtain a range of samples between 1.5x103 and 1.4x106 cells/ml. Samples were further diluted 1:10 in PBS for analysis on the flow cytometer and stained with FITC-conjugated MAb in a 1:1000 dilution. The hemacytometer counts required no dilution and were performed at 400x total magnification. We obtained excellent agreement between the two methods (R2=0.99, slope of 0.93; Figure 20) indicating that the flow cytometric method was producing an accurate estimate of the abundance of A. anophagefferens in pure cultures of the alga.

Figure 20: Standard curve of A. anophagefferens concentrations measured using the flow cytometer or a hemacytometer. Samples for analysis on the flow cytometer were diluted 1:10 in PBS to facilitate antibody binding while hemacytometer counts required no further dilution.

Figure 21: Comparison of flow cytometer and hemacytometer methods for enumerating A. anophagefferens in natural samples. The estimated concentration is based on a precise hemacytometer count of the culture used to spike the natural sample that was serially diluted.

This comparison was also carried out using natural seawater samples (which had no detectable A. anophagefferens cells present) spiked with known concentrations of A. anophagefferens. Samples were taken from the Los Angeles harbor, a region in which A. anophagefferens is absent, preserved, and filtered through 10 ?m netting. This pore size was chosen to allow for A. anophagefferens cells (2-3 ?m in diameter) to pass through but exclude larger cells that could clog the flow cytometer fluidics. A natural sample was spiked with a known concentration of A. anophagefferens from a laboratory culture and then serially diluted into other natural samples to obtain a range of samples from <200 to 6.7x105 cells/ml. These concentrations were obtained based on a precise count of the laboratory sample using a hemacytometer, and then diluted accordingly. Once again, abundance of A. anophagefferens determined by flow cytometry were in good agreement with the abundances of the alga estimated by microscopical methods. Algal abundances were strongly correlated in this culture-based study across a range of 200-6.7x105 cells/ml (R2=0.99, slope of 0.92; Figure 21).

The close agreement that we observed between algal abundances determined by flow cytometry and direct microscopical counts indicated that the flow cytometric method employing our MAb yielded accurate estimates of the alga. We speculated that the differences observed between abundance estimates obtained by the ELISA method and the flow cytometric method (Figure 19) were a consequence of binding of the MAb to cell debris (lysed cells) that retained antigenic character. To examine this possibility, an experiment was performed in which A. anophagefferens and a known predator of the brown tide alga, Pedinella hexacostata, were introduced together and the concentration of A. anophagefferens was monitored using bothe the ELISA and flow cytometric methods.

A. anophagefferens was grown to high abundance in laboratory cultures, and then the predator, P. hexacostata, was added to the culture. The ELISA method detects all occurrences of the A. anophagefferens antigen, whether present on whole cells or as cell fragments. The flow cytometric method measures physical characteristics of cells as they pass through a laser beam such as size, internal granularity, and fluorescence, as well as the fluorescent label emitted by the labeled antibody. A population of A. anophagefferens will have very specific size characteristics (the cell is about 2-3 ?m in diameter) as well as a noticeable signal in the FL3 fluorescence region due to the red autofluorescence of chlorophyll a. When bound to the fluorescently-labeled antibody, A. anophagefferens also exhibits a distinct FL1 signal due to the green emission of the FITC conjugate. These parameters combined to allow the user to gate (i.e. differentiate) a specific population and enumerate the cells therein. Thus, appropriately gating allows the detection of intact cells (by size, autofluorescence) that also label with the MAb (FL 1 signal).

Figure 22: Changes in the abundance of A. anophagefferens (determined by the ELISA and flow cytometric methods) in a culture exposed to predation by the herbivorous protist Pedinella hexacostata.

Pedinella hexacostata is an efficient grazer of A. anophagefferens, so the abundance of A. anophagefferens decreased as Pedinella consumed the cells. We anticipated that estimates of the abundance of A. anophagefferens in the culture would differ between the ELISA method and the flow cytometric method if digested/egested algal cell debris retained antigenic quality. The results indicated exactly that (Figure 22).

When cultures of A. anophagefferens were consumed by P. hexacostata, the ELISA method overestimated the concentration of A. anophagefferens remaining in the cultures (Figure 22). Because the ELISA method did not differentiate between whole cells and cell fragments that retained antigenic character, estimates of the brown tide alga decreased during predation, but decreased to a lesser extent than estimates obtained by flow cytometry (approximately one order of magnitude difference by the end of the experiment. Abundances obtained by flow cytometry were confirmed by microscopical counts of the brown tide alga. These results are significant to the study of natural samples of A. anophagefferens which are usually under predation pressure and, therefore, may contain cell fragments which could artificially inflate the concentration using the ELISA method.

The immunocytometric technique for the identification and enumeration of Aureococcus anophagefferens has also recently been applied to natural samples of seawater from areas affected by brown tides. Samples from inland waters of Maryland were taken by staff of the Maryland Department of Natural Resources, preserved with 1% glutaraldehyde, and analyzed for A. anophagefferens using both the ELISA and immunocytometric methods. Data from the Public Landing and Taylor's Landing sampling sites show the abundance of A. anophagefferens at which experienced varying intensities of Brown Tide (Figure 23A). An obvious difference in the concentration of cells as determined by the two methods w (Figure 23B). As we have previously theorized, this discrepancy may be attributable to the presence of cell fragments retaining antigenic properties in these natural samples.

Figure 23: Immunocytometric enumeration of A. anophagefferens in natural samples from Maryland, Summer 2004. The concentration of A. anophagefferens at Public Landing and Taylor's Landing (A), as analyzed by both ELISA (dashed line) and immunocytometric (solid line) techniques. The relationship between immunocytometric and ELISA enumeration of A. anophagefferens shows a consistent overestimation of cell abundance (B, dashed line represents a 1:1 relationship).

Additional samples are being analyzed by both methods for comparison with each other as well as comparison with an older polyclonal immunofluorescent technique. Taken together, this data will contribute to our overall understanding of the advantages and limitations of the three methods commonly used to monitor A. anophagefferens.

QCM Experiments

We have explored the use of the Quartz Crystal Microbalance (QCM) as a novel approach to the detection and identification of specific microorganisms in natural samples. QCM can detect target molecules based on changes in a crystal's resonance frequency with the addition of mass to the crystal surface. The use of the QCM method to detect marine microorganisms was examined via the functionalization of a quartz crystal with antibodies specific to the target organism (A. anophagefferens).

Antibodies were immobilized onto the thiol-reactive gold electrode surface of a quartz crystal using a wide variety of crosslinkers which target different reactive groups on the antibody. The general structure of an antibody is shown in Figure 24. The regions in pink are the antigen-binding fragments (Fab) while the gray regions are the crystallizable fragments (Fc) that do not directly participate in antigen recognition. Note the variety of functional groups available for reaction with a crosslinker, including amines (NH3), disulfides (S-S), carbohydrates (CH3) and carboxylic acids (COO-).

Figure 24: Generalized structure of an antibody molecule (from http://pim.medicine.dal.ca/abstruc.htm, Tim Lee).

The ideal functionalization scheme targets the antibody molecule in a region on the Fc fragment. This prevents the crosslinker from obstructing the Fab fragment and therefore preserves antibody function. We tested three crosslinkers and functionalization schemes: N-Succinimidyl 3-(2-pyridyldithio) propionate (SPDP); 3-(2-pyridyldithio) propionyl hydrazinde (PDPH); and EDC/NHS (via mercapto-propionic acid) (Figure 25).

Figure 25: MAb immobilization schemes onto gold surface. A) Reaction with SPDP introduces a cleavable disulfide group to amine groups on the MAb. B) Reaction with PDPH (following oxidation of the MAb with sodium periodate) introduces a cleavable disulfide group to carbohydrates on the MAb. C) A self-assembled monolayer (SAM) of mercapt-propionic acid (MPA) introduces a carboxyl group to the gold surface which, on further reaction with EDC/NHS, is introduced to amine groups on the MAb.

The SPDP and EDC/NHS functionalization schemes involved immobilization via the MAb's amine groups. The PDPH scheme immobilizes the MAb via carbohydrates which are abundant in the Fc region.

Initial data from crystals functionalized using these schemes have been variable due to mechanical problems with the crystal reaction chamber. One of the more reliable sets of data was obtained from the run of a crystal functionalized with SPDP (Figure 26). A noticeable change in frequency was observed with the addition of A. anophagefferens at 35 minutes, followed by a slight desorption following rinses with PBS at 95 minutes. Due to recurring issues with drift of the signal, and the relatively high abundances of the target alga required to create sufficient signal, further work in this area are not planned at this time.

Figure 26: Frequency curve for crystal functionalized with MAb using SPDP. The three different curves are from three overtones measured by the QCM. Note the decrease in frequency upon addition of A. anophagefferens followed by a slight desorption after the PBS rinse.

AFM Immunosensing

The Atomic Force Microscope (AFM) is a highly sensitive instrument capable of detecting single molecules based on their affinity for a functionalized tip. Identification of single-cell microorganisms such as the brown tide alga (BTA) by measuring forces with an AFM is conceptually very simple. The microorganisms are attached to a flat surface, and the tip of the AFM is functionalized with monoclonal antibodies (MAbs) to the organisms. The tip is moved towards the cell surface until it touches it, and then retracted. A force-distance (f-d) curve is obtained by measuring the deflection of the cantilever that occurs as the tip moves towards and away from the surface. If there is an antibody-antigen (Ab-Ag) binding event, this bond must be broken for the tip to retract to its original position, and this can be found from the f-d curve. If a MAb-Ag bond is broken, the Ag must be present on the cell surface, and therefore the cell must be a brown tide alga. (Note that this approach works for arbitrary cells, provided that monoclonal antibodies to them are available.) To implement this detection scheme requires that we attach cells to a substrate surface, and that we functionalize an AFM probe tip with Abs. We will first report on the current state of the project, and then summarize the (somewhat tortuous) road that took us there.

To attach the brown tide alga, A. anophagefferens (BTA) to a substrate surface we coated Si/SiO2 with poly-ethylene imine (PEI) followed by the monoclonal antibody to the BTA (Scheme 1).

Addition of the BTA cells led to incomplete coverage (not a full monolayer) on the Si/SiO2 surface. To determine the strength of the attachment, TM (Tapping Mode) AFM imaging was carried out under PBS conditions. We were encouraged by the stability of the attachment since the same cell could be imaged more than 20 times without being detached from the surface.

Scheme 1: Attachment of BTA cells onto Si/SiO2 surface.

To functionalize the silicon nitride tip we attached the antibody using a number of different approaches, including incorporating a linker group between the tip and the mAb. These approaches used 3-amino-propyl triethoxy silane (APTES) to generate reactive functional groups on the tip needed to couple to the antibody. Treatment with APTES, however, often leads to multilayer formation on the tip which is not desirable in single molecule studies. We later decided to attach the antibody using the protocol in Scheme 2 since this chemistry is known to produce only a monolayer of antibodies on the tip, and produces more reproducible results than APTES.

This procedure did not incorporate a spacer group between the tip and the antibody. A flexible linker seems to be important when studying interactions between a single Ab and Ag, presumably because it allows the Ab-Ag pair to explore various orientations until the correct one is reached. However, for our problem of cell detection, we had better results without a linker, and with many Abs on the tip.

Now we were ready to carry out the force-distance (f-d) analysis. Since there was not a dense coverage of the cells on the surface, an optical microscope was used to position the tip over one of the cells. Next, the parameters such as applied force and dwell time were established. During the course of one experiment, the tip was moved from one part of the cell to another after about 30-40 f-d runs, and even placed over another cell. This process was continued until more than 100 f-d runs had been recorded.

Sometimes, only one unbinding event was observed while other force curves revealed two unbinding events-see Figs. 27a and b. The latter may be due to the fact that the antibody has two recognition sites and may reflect unbinding of the two sites from the antigen.

Scheme 2: Legend: a) Ethanolamine-HCl in DMSO, r.t., 12 h; b) Glutaraldehyde in phosphate buffer, pH 7, r.t., 1 h; c) mAb in phosphate buffer, pH 7, r.t., 1 h.

Figure 27: F-d curves with one (a) and two (b) unbinding events.

More than 100 force-distance curves were usually recorded in one experiment. Histograms of these data are shown below in Figure 28.

Figure 28: Force-distance curves for algal cells (top), and blocked cells (bottom).

The mean adhesion value is approximately 250 pN (Figure 28, top) between MAb and a BTA cell surface. This value is consistent with other antigen-antibody interactions known. More importantly, blocking of the antigen on the cell surface with antibody resulted in a loss of this specific adhesion force (Figure 28, bottom). This indicates that we are able to discriminate between antibody-antigen bonds and non-specific interactions. We also performed control experiments using bare tips without any antibody to BTA on both the cell surface and the polymer coated silicon surface. Both experiments revealed small adhesion forces due to non-specific interactions (data not shown).

Previous Attempts: We tried to attach the antibody to an AFM tip by a number of procedures including self-assembly of thiols on gold-coated silicon probes as well as traditional APTES treatment of silicon tips. Also, we explored the role of a PEG linker between the tip and antibody to allow greater flexibility for the antibody to locate its antigen on the cell surface. These experiments typically resulted in adhesion forces that were much higher (data not shown) than those using the protocol shown above. The reproducibility of these results was generally poor.

Several different surfaces were used to immobilize brown tide algae cells. Since the chemical composition of the cell surface is unknown, it was decided that a flat surface would be a good starting point. Mica was chosen because it presents a flat surface and is relatively inexpensive and easy to use. We also tried poly-L-lysine coated mica since it presents a surface charge that is opposite to that of mica. Also, SAMs of thiols terminating in various functionalities (methyl, carboxyl) on gold were employed to determine if they could promote cell adhesion. All of the above procedures led to incomplete and weak attachment of the BTA on the substrate.

In short, attaching BTA cells to a surface proved to be a much more difficult problem than we had expected, and took us a considerable amount of effort to come up with the current, successful approach.

Discussion:

We have shown that we can discriminate between a BTA cell and the substrate surface by using the force information acquired with the AFM. The interaction is mediated by a MAb-Ag pair and therefore should be highly specific to the BTA. Further experimental confirmation of specificity is desirable-see below.

A force-distance curve is acquired in a few seconds, and 50-100 curves are usually needed to average out noise and construct histograms such as those shown above. Therefore a cell can be identified in minutes. (We have not attempted to optimize the procedure; it is likely that much shorter times can be achieved.) It is easy to imagine a system in which first an image is acquired to determine where are the cells in a scan (this takes on the order of a couple of minutes), and then the tip is directed to each of the cells and f-d curves are acquired. Assuming there are some 20 cells in a 100x100 m scan, the analysis would take about an hour. This competes favorably with current laboratory techniques, and provides single-cell sensitivity, which is not readily available otherwise. For abundance measurements it would be necessary to calibrate the system by relating the number of cells on the sample substrate to their concentration in the water. We have not attempted to do this but it seems feasible.

The tip functionalization procedure is relatively complex and time consuming, but can be performed off-line and therefore is not a limiting factor. The major problem with the AFM approach is sample preparation. Currently this takes a few hours in a chemistry lab. However, we think that it may be possible to attach the BTA (and other microorganisms that may be present in the sample) to a surface such as a polycarbonate filter in a much simpler manner. This would make the AFM approach much more attractive.

Nanowire Sensors

Algal detection has significant implications for environmental studies and hence lies at the heart of our CENS program. Sensing for identifying and counting the BTA of interest is a critical issue, and we are studying various approaches to accomplish it. Recently, we began developing a new generation of algal sensors based on nanowire / nanotube transistors, and we have made significant progress. We have successfully synthesized single-walled carbon nanotubes and a variety of novel nanowires based on In2O3, SnO2 and CdO. These materials are subsequently employed to construct nanowire / nanotube transistors and have been demonstrated to work as sensors in both the gas phase and an aqueous environment. Our current work is focused on the demonstrating the operation principle and testing the reliability and detection limit of these sensors.

Figure 29: Schematic diagram of the nanowire algae sensor under development.

The operation of our future nanosensors is depicted in Figure 29. Electrical leads will be patterned to contact both ends of a semiconducting nanowire. The silicon substrate, separated from the nanowire by a SiO2 dielectric layer, will be used as a gate electrode. This nanowire transistor will be placed into a fluid cell and algae in either sea-water or buffer solutions will be flown through the system. Attachment of algae to the nanowire surface will occur through antibody-assisted interaction. Such attachment will effectively work as a chemical gate for the nanowire and lead to variation in the nanowire carrier concentration, which will be recorded via measurement of the nanowire conductance.

We have carried out preliminary tests to confirm the capability of our nanotube transistors to detect algae cells (Figure 30). A fluid cell has been built with PDMS to guide the algae suspension to flow across a nanotube transistor, which has been functionalized with the algae antibody, as shown in Figure 30B. We have monitored the current going through the nanotube device in real time. The device was first stabilized in the buffer solution to establish a reference point, and then algae were added into the fluid channel. A drop in conduction was observed, as shown in Figure 30C. Our observation thus confirms that the conduction of carbon nanotubes can be modulated by exposure to algae. However, more in-depth studies will need to be carried out to determine the conduction modulation mechanism. The current hypothesis is that when algae cells attach to the antibodies on the nanowire surface, the dielectric environment is changed, which could lead to a variation in the nanowire conductance.

Fig. 30: (A) Algae cells injected into the micro channel for detection by the carbon nanotube. (B) Photograph of the microchannel. (C) Preliminary data of algae sensing.

ACCOMPLISHMENTS

• Extension/improvement of existing biological detection methods for the harmful brown tide alga (Aureococcus anophagefferens): Immuno-based flow cytometry.
  • Comparison of ELISA (Enzyme-Linked Immunosorbent Assay) and flow cytometry approaches, showing that flow cytometry has several advantages.
  • Demonstration of correlation between flow cytometry and microscopy-based enumeration methods for A. anophagefferens.
  • Demonstration of the usefulness of flow-cytometry for routine laboratory detection and counting of brown tide algae in medium to high concentrations.
• Test bed experiment: Artificial stimulation of brown tide in a thermally stratified column, and demonstration of predation effects in the test bed.
• Methods for attaching A. anophagefferens-directed antibodies to surfaces. This is a generic capability, useful for functionalizing the tips of AFMs (Atomic Force Microscopes) that are used in force sensing, and also for other applications, such as functionalization of nanowires. These methods can be generalized to other types of microorganisms for which monoclonal antibodies are available.
• Demonstration of discrimination between brown tide algae immobilized on a surface and the surface itself, by using force-distance curves obtained with an AFM with a functionalized tip. This detection technique is sensitive to a single cell (brown tide alga). In principle, it can be generalized to any cell for which monoclonal antibodies are available.
• Fabrication of nanowire and carbon nanotube devices that will serve as ultra-sensitive sensors for brown tide cells.
• Demonstration of nanowire sensing principles for biomolecules and other materials.
• Demonstration of the operation principle of using nanosensors to detect brown tide algae, where the presence of the algae is directly converted into electrical signals.

FUTURE DIRECTIONS

• Growth of a more mobile organism in the test bed (water column) to assess the behavior in the vertical dimension in response to nutrients, light intensity, and temperature.
  • Use of the photosynthetic dinoflagellate Symbiodinium sp.
• Directed sampling of an A. anophagefferens bloom in the column using a chlorophyll sensor and gradient-detection algorithms.
• Ramp-down the AFM force-sensing work and conclude it within the year. Investigate a simple method for attaching brown tide cells to a substrate surface by using polycarbonate membranes. Further demonstrate specificity of the sensing mechanism by comparing force-distance curves for brown tide algae with their counterparts for other algae.
• Demonstrate selectivity of the AFM detection by running experiments with other types of algae.
• Development of a new generation of algae sensors based on nanowire and carbon nanotube transistors. These sensors exploit recent advances in the field of nanotechnology, and are particularly attractive because they are cheap to fabricate, can be deployed in large numbers to achieve unprecedented spatial resolution, can provide access to environments unreachable until now, and—importantly—they are much more sensitive than their macroscopic counterparts. Our next step is to further develop this nanosensing technique and study the reliablity, reproducibility and field deployability of these novel nanosensors.
• Launch exploratory research on the application of nanosensor networks to studies of cell cultures and, eventually, in vivo cells. This work might later be spun off as a separate application area within CENS, or be shifted to other funding sources. (It represents a direct evolution of the research within the M3 thrust, and therefore it is reasonable to start it from within this area.)

PEOPLE

FACULTY

Prof. David Caron
Prof. Ari Requicha
Prof. Chongwu Zhou

STAFF

Dr. Mrinal Mahapatro
Carl Olberg
Beth Stauffer

STUDENTS

Alex Lee
Chao Li