Final Report: Development and application of a fiber optic array system for detection and enumeration of potentially toxic cyanobacteria

EPA Grant Number: R833828
Title: Development and application of a fiber optic array system for detection and enumeration of potentially toxic cyanobacteria
Investigators: Anderson, Donald M. , Carmichael, Wayne W
Institution: Woods Hole Oceanographic Institution
EPA Project Officer: Klieforth, Barbara I
Project Period: June 1, 2008 through May 31, 2011 (Extended to May 31, 2013)
Project Amount: $508,494
RFA: Development and Evaluation of Innovative Approaches for the Quantitative Assessment of Pathogens and Cyanobacteria and Their Toxins in Drinking Water (2007) RFA Text |  Recipients Lists
Research Category: Drinking Water , Water


The overall project goal was to adapt and validate a rapid and accurate optical fiber-based technology for cyanoHAB cell detection and enumeration in both laboratory and field settings. Specific aims were to: (1) design ribosomal RNA (rRNA) signal and capture probes for the three most important toxic cyanobacteria (Microcystis, Cylindrospermopsis, and Anabaena) using published sequences; (2) design and test a second probe pair for each species, to incorporate redundancy into the array; (3) test these probes in the fiber-optic array format and determine detection limits, specificity, and dynamic range; (4) refine hybridization conditions to reduce processing time; (5) develop procedures to analyze multiple cyanoHAB species simultaneously using a single fiber bundle in a multiplexed format and validate it using mixed cultures and spiked and unspiked field samples; (6) work with individuals and agencies responsible for fresh- and brackish water management to determine desired detection limits, precision, new cyanobacteria species for future probe design, and operational characteristics for the assay and instrumentation that would be developed around it; and (7) prepare a detailed protocol for sample handling and processing for use with this method.

Summary/Accomplishments (Outputs/Outcomes):

In freshwater systems, blooms of cyanobacteria (cyanoHABs) represent the most significant threat to human health from harmful algae, as many species produce neurotoxic, hepatotoxic, dermatotoxic, or other bioactive compounds that have caused human illness, animal mortalities, and adverse ecosystem and economic impacts. Cyanotoxins in drinking water can cause gastrointestinal complications, liver damage, neurological symptoms, and even death. Exposure to cyanotoxins during recreational activities is also a problem, as skin contact, inhalation, or ingestion can lead to rashes, allergies, and gastrointestinal complaints.
The detection and management of cyanoHABs is challenging, as the blooms can take multiple forms, ranging from dense surface scums to dilute suspensions that can still cause harm. Many different species and strains co-occur, and strains of the same species can be toxic or non-toxic, or can vary dramatically in the amount of toxin produced under different conditions. Microcystis is the organism by which the World Health Organization provisional drinking water guidelines are based. The provisional drinking water guideline value for Microcystin-LR, a potent hepatoxin that induces liver failure, is 1 µg/L (corresponding to approximately 10,000 cyanobacterial cells/mL). The potential health effects resulting from cyanotoxin exposure range from low risk, corresponding to approximately 20,000 cells/mL, to adverse effects, which corresponds to scum formation.  Currently, the majority of assays for cyanobacteria detection monitor toxin production as opposed to cell proliferation. Such assays are either not suitable for large-scale screening or monitoring programs and are a secondary detection method rather than a primary monitoring approach to assessing potential risk. It is therefore vital to have a means for monitoring toxic strains prior to scum formation and toxin testing.
Cell detection using standard light microscopy is sometimes not feasible in monitoring programs that generate large numbers of samples, as distinguishing characteristics can be difficult to discern under the light microscope. As a result, managers need rapid, sensitive methods that can accurately identify and enumerate harmful species in a water body, yet to date, no methods exist that can be used to enumerate multiple species in the same sample and that can be deployed in small, bench-top instruments or on moorings for automated detection. This proposed project employed an innovative approach to harmful algal bloom (HAB) cell enumeration –fiber optic genosensors, building on prior studies on marine HAB species, for which this technology is well advanced. This technology is transferrable to a wide range of probe-based assay systems and autonomous monitoring instrumentation.
Probe development
The approach to probe development was the same for the three target taxa: CylindrospermopsisMicrocystis, and Anabaena. First, probes were designed by aligning 16S rRNA gene sequences of target and non-target species publicly available in GenBank and probe sequences identified. Candidate probes were checked for secondary structure using OligoCalc (Kibbe, 2007), and specificity and coverage tested using the probeCheck online database. The best candidates were then tested for cross-reactivity using cyanobacteria cultures (below).
Cross-reactivity testing was carried out using fluorescent in situ hybridization (FISH), or “whole cell” analysis with target and non-target cyanobacteria species. Previously, it has not been possible to use whole cell or FISH methods to detect or enumerate cyanoHABs due to interference from phycoerythrin autofluorescence. However, during the course of the probe design and testing, we developed a FISH technique that utilizes a combination of filters and fluorophores, which effectively blocks autofluorescence (Figure 1), thus allowing the utilization of fluorescent probes for targeted detection of toxic cyanobacteria. Published probes were tested first but either exhibited cross-reactivity or failed to detect one or more of the target strains, necessitating the design and testing of new probes. A total of 10 Microcystis (8 designed, 2 published), 5 Cylindospermopsis (all designed), and 10 Anabaena (6 designed, 4 published) probes were tested. Capture probe development and testing was successful for C. raciborskii and Microcystis spp.; howeverall Anabaena probes exhibited some degree of cross-reactivity or failed to detect all target strains tested. These difficulties are likely due to the complicated and unresolved taxonomy of Anabaena, which is polyphyletic and clusters with Nostoc,Aphanizomenon, and Cylindrospermopsis.
Therefore, only the probes for Cylindrospermopsis and Microcystis were successfully transitioned to the fiber optic microarray format. In addition to working with cyanobacteria cultures, we also applied our FISH assay to the detection of cyanoHAB taxa from freshwater systems affected by cyanobacteria blooms. One such field sample was collected during a Microcystis spp. bloom associated with a fish kill in a water retention pond in Hillsboro, OR. Using the probe designed for the microarray, we were able to successfully detect the presence ofMicrocystis cells (Figure 2), thus demonstrating the efficacy of the FISH assay for the identification of these taxa in field samples.

Text Box:  
Fig 2. Detection of Microcystis from a field sample during a cyanoHAB bloom (Microcystis spp.) in Hillsboro, OR. Note the many other species in the background.
Fiber optic microarray
The fiber optic microarray technology used in this research has been previously applied to a wide array of chemical and biological measurements, as well as the detection of marine HAB species. The advantages of this approach include the ability to detect and enumerate dozens of target species simultaneously with high specificity using the same, re-usable optical fiber bundle. Additionally, the simplicity of the procedure and the ability to re-use its sensor array hundreds of times without loss of sensitivity makes this technology a promising candidate for further development, including adaptation for use with cyanoHABs in laboratory, bench-top systems or on remote, moored instruments. Briefly, the microarray is comprised of optical fiber bundles, comprised of two types of glass: the inner glass core has a slightly higher refractive index than the outer ring of cladding glass. Because of the refractive index mismatch, light is transmitted through the core over long distances via total internal reflection. Wells are etched at one end of the fiber, into which polystyrene beads are loaded, coupled with a capture probe sequence (Fig 3, left). Thus, each well is optically “wired” for a particular taxa and can be individually interrogated. The resulting fluorescent signals from the hybridization are Text Box:   

Fig 3. Fiber optic microarray system used for the simultaneous detection of cyanoHAB taxa. The array consists of fiber bundles that are differentially etched to produce microwells, into which a bead slurry containing either a single bead type or multiple bead types are loaded (left). The resulting fluorescent signals following the hybridization with target molecules are transmitted through the fibers and observed using a CCD camera (right).
observed using a charge coupled device (CCD) camera (Figure 3, right).
Text Box:   
Fig 4.  Encoding images using both 3.1 µm (top, left) and 4.5 µm polystyrene beads (bottom, left).  The 3.1 µm beads were not successfully encoded; however, three bead types were achieved using 4.5 µm beads.  Each dye intensity provided uniform signal intensity that was distinguishable at 200 and 100 ms of exposure time.
The first step in developing the multiplexed array requires bead encodings that distinguish each bead type on the array.  Both 3.1 µm amine-functionalized and 4.5 µm carboxyl-functionalized beads were encoded with Europium (III) dye solution to obtain homogenous and distinguishable intensities. The 3.1 µm beads were not successfully encoded; however, the 4.5 µm polystyrene beads provided encodings that were uniform as required for the array.  The different capture probe sequences, and a negative control sequence (one that will provide background signal), each require different encoding intensities (Figure 4). 
Text Box:  
Fig 5. Amine-modified capture probe concentrations coupled to carboxyl polystyrene beads.  Each bead type provided relatively similar signal when accounting for standard deviations.  To avoid surface crowding, 1 µM of DNA was used for bead coupling without sacrificing assay sensitivity.
Primary optimization experiments tested varying concentrations of capture probe used in the coupling to determine the most efficient bead:DNA ratio and hybridization times needed for adequate target capture. For the coupling optimization, four concentrations (1 µM, 10 µM, 25 µM, 50 µM) of amine-modified DNA capture probe were coupled to carboxylated polystyrene beads. Each was incubated with 100 pM of synthetic target, after which fluorescence was determined. From the data collected (Figure 5), 1 µM (approximately 6-10 µg of amine-terminated DNA capture probe) was used for bead coupling. 
Generally, each amount of DNA provided relatively similar signal with slightly higher signal in 25 µM and 10 µM of DNA capture probe. The decision to proceed with 1 µM of capture probe was to reduce steric hindrance when transferring from 18-24 bp synthetic targets to 1,500 nt RNA targets. Since the 1 µM capture probe concentration provided comparable signal as the other three concentrations, with lower standard deviations, we proceeded with this coupling concentration without sacrificing sensitivity.
To optimize for hybridization time, incubations with 10 minute imaging intervals was performed using 100 pM of synthetic target (Figure 6). At higher concentrations the signal plateaued at around 30 minutes. However, with lower concentrations, the signal plateau was reached at around 60 minutes, indicating that a 1- to 2-hour hybridization would be adequate for target capture.
Text Box:  
Fig 6.  Time series of hybridizations in the presence of 100 pM of synthetic target.  Every 10 minutes signal intensities were obtained until a signal plateau was reached, indicating capture saturation.
Cross reactivity analysis was also performed using synthetic complements. Each bead type was incubated for 30 minutes in the presence of non-complementary target to determine the extent of non-specific signal (Figure 7).
Text Box:  

Fig 7.  Cross reactivity analysis with a 30-minute hybridization in the presence of non-complementary targets.  Each bead type was tested against its complement and non-complements.  Negligible signal is observed for Cylindrospermopsis and Microcystis in the presence of non-complementary targets.
Text Box:  
Fig 8.  Representative calibration curve with a dynamic range of four orders of magnitude.  The arrays were able to measure 10 pM of synthetic complement within 1 hour of hybridization.
Most of the observed nonspecific signal was seen with the Anabaena beads in the presence of non-complementary target. Once actual cyanobacterial cultures were applied to the microarray, Anabaena was removed from the array due to cross reactivity issues. The observed cross reactivity was expected due to the cross reactivity observed during the FISH testing.
Each capture probe was tested using the microarray format with synthetic targets to assess detection limits and establish the linear dynamic range. Net signal intensity, defined as the average hybridization signal minus the background signal, was used to distinguish positive from background signals. The threshold for a positive signal was set to three times the standard deviation of the background signal. Probes were first tested in the microarray format using synthetic targets labeled with a Cy3 fluorophore (direct hybridization). These tests showed that both capture probes were capable of detecting 10 pM of synthetic target, with a dynamic range of four orders of magnitude (Figure 8).
RNA extracts of cyanobacterial cultures were then tested in the sandwich hybridization assay format with a universal signal probe. In addition to testing the intact RNA, RNA was fragmented (60-200 bp in length) in effort to reduce the steric hindrance and stability of the secondary structure.  RNA fragmentation was performed using different target concentrations, fragmenting solution volumes, and incubation times, and tested using the array. However, RNA detection was unsuccessful in the sandwich assay format.  In an effort to identify the factors that potentially could contribute to the lack of signal, synthetic DNA and RNA were tested on the array. The array was capable of detecting 100 pM of synthetic target in the sandwich hybridization format within 30 minutes.   In addition, RT-PCR amplicons were tested (using the capture and signal probes as primers) on the array, and again this generated positive signal (Figure 9). 
Text Box:  

Text Box: Fig 9.  Polyacrylamide gel (left) showing the 66 base pair amplicon for an RT-PCR reaction performed with Cylindrospermopsis capture and signal probes as primers (these primers were closer in proximity, hence the 66 bp amplicon, however they are Cylindrospermopsis specific and cannot be used for the multiplexed array).  Sandwich assay of a 1 in 10 dilution of the RT-PCR product hybridized for 30 minutes generated a detectable net signal of 450 a.u. (below).
Based on these data, the lack of signal from the extracted RNA could potentially be due to 16s rRNA secondary structure and/or proximal placement of the universal signal probe in relationship to the capture probes.
As an alternative approach, the RNA was directly labeled using the Universal Linkage System (ULS) kit from Kreatech.  This step thus eliminates the secondary binding of the universal signal probes and still allows for fragmentation of the RNA as previously reported. Microarray testing and optimization using cyanobacteria cultures along with the direct labeling approach is currently underway. Preliminary results indicate that theMicrocystis bead based array is able to detect as few as ~ 4,000 Microcystiscells/mL and ~ 40,000 Cylindrospermopsis cells/mL (Figure 10). Note that these are the lowest concentrations that have been tested thus far, not the lower limits of detection.
Text Box:  

Fig 10.  3-D bar graph with the absolute fluorescent signal from 1 mL, 0.1 mL and 0.01 mL (corresponding to ~ 400,000 to ~ 4,000 cells/mL).  The above cultures were hybridized for one hour or two hours as indicated.  Samples of 0.01 mL for Cylindrospermopsis at 1 hour and 2 hours and 0.1 mL at 1 hour have not yet been tested.  Samples of 0.01 mL for Microcystis for 2 hours have not yet been tested.  C = Cylindrospermopsis. M = Microcystis.
Studies underway will further optimize the array for 16s rRNA targets, determine the cellular limits of detection, and demonstrate duplex detection forCylindrospermopsis and Microcystis in mixed field samples.
The final step in array validation is the detection of cyanobacteria from field samples.  For this, water samples were collected from Cullaby Lake, a coastal plain lake in northwestern Oregon that is characterized by high nutrient concentrations and summer algal blooms. Opportunistic sampling was carried out during blooms in July-Aug, 2013 and samples were preserved for cyanoHAB quantification using light microscopy, FISH, and the fiber optic array. Analyses are currently underway, and results will be included in the publication (in prep) describing this research. 


The overall project goal was to adapt and validate a rapid and accurate optical fiber-based technology for the detection and enumeration of potentially toxic cyanobacteria (CyanoHABs). We successfully designed molecular probes for two of the most prominent cyanoHAB taxa: Microcystis andCylindrospermopsis. Concurrently, we also developed a fluorescent in situ hybridization (FISH) technique that enables the utilization of these probes for targeted detection of toxic cyanobacteria via fluorescence microscopy, and successfully used this technique to detect the presence of cyanoHAB cells in field samples. The probes are also amenable to other detection approaches utilizing probe technology (e.g., sandwich hybridization) that are commonly used in HAB monitoring programs. The probes were subsequently transitioned to a fiber-optic microarray format, which provides high throughput detection with multiplexing capabilities, and tested against synthetic targets and RNA extracts from target and non-target species. Laboratory experiments with the microarray demonstrated the successful detection of target taxa, with a linear dynamic range of detection that spanned four orders of magnitude (established using synthetic complements). In summary, this research successfully developed novel fiber-optic genosensors for the detection of cyanoHAB species, and also produced technology that is transferrable to a wide range of probe-based assay systems and autonomous monitoring instrumentation to provide early warning of organisms that threaten public and ecosystem health.
Student participation
The extensive experimentation and analyses that Tufts University graduate student Shonda Gaylord carried out with the microarray will comprise part of her Ph.D. thesis. This project also supported two undergraduate interns, Robert Arnold and Eric Armstrong, who participated in this research through NOAA’s Hollings Scholarship program. The poster presented by Robert Arnold describing this research was awarded first prize at the conference for Hollings scholars in 2008. This project also benefited from the participation of Dr. Yunjung Park, a postdoctoral guest investigator in the Anderson Lab.  


Kibbe WA (2007). "OligoCalc: an online oligonucleotide properties calculator." Nucleic Acids Res. 35: May 25.

Journal Articles:

No journal articles submitted with this report: View all 16 publications for this project

Supplemental Keywords:

Health effects, ecological effects, human health, toxics, bacteria, ecosystem, aquatic, environmental chemistry, biology, ecology, genetics, limnology, monitoring, analytical, northeast, central, northwest, RFA, Scientific Discipline, INTERNATIONAL COOPERATION, Water, Environmental Chemistry, Health Risk Assessment, Environmental Monitoring, Drinking Water, Environmental Engineering, microbial contamination, microbial risk assessment, monitoring, real time analysis, gene microarray assay, aquatic organisms, other - risk assessment, early warning, drinking water contaminants, drinking water system

Progress and Final Reports:

Original Abstract
  • 2008
  • 2009 Progress Report
  • 2010 Progress Report
  • 2011 Progress Report