Final Report Summary - MIDTAL (Microarrays for the detection of toxic algae)
Executive summary:
The MIDTAL project set out to develop a universal microarray to test for Harmful algal bloom (HAB) species and their toxins relevant to the European Community, to validate the array against field samples from around Europe and to seek to commercialise this technology for wider application. Early in the project the detection of species and toxins were disaggregated to two separate microarrays that were concurrently developed.
Species microarray: Three generations of improving microarrays were developed and tested during the project. The final generation includes 163 probes at various taxonomic hierarchies covering all the major harmful algal species of current interest in the European Union (EU). The origin of the initial species probes for the microarray was from previously developed technologies (e.g. FISH) that had to be tested (and in a number of cases redesigned) for the changed detection platform. Key features of the microarray design are the inclusion of internal controls (universal eukaryotic probe; Dunaliella probe) and the taxonomic hierarchy of probes that can be used to aid interpretation of results. Developments to the microarray detection system have included lengthening the probes (from 18 to 25 bp) to improve specificity; extending the attachment of the probes to the microarray slide to improve accessibility by target molecules and developing the microarray methodology to improve stringency and signal strength.
Two years of monthly field samples have been collected from key sites across the EU. The first year of these samples were tested against the second generation chip and the second year samples have been tested against the third generation chip. Samples from both years have been tested with the toxin chip.
The ribosomal ribonucleic acid (rRNA) content of different species in different phases of the growth cycle and grown under different environmental stresses was tested and shown to be reasonably consistent with no dramatic reductions in stationary phase. Calibration data generated for different species has allowed development of algorithms to convert microarray signal to cell numbers. Results so far indicate that cell counts and microarray hybridisations generally show reasonable agreement with the observation that usually the microarray records more species than the cell counts, especially in respect of small flagellate cells that are particularly difficult to quantify by light microscopy and Pseudo-nitzschia spp, which can only be reliably identified with electron microscopy.
Toxin microarray: Antibodies to cover the three main toxicity syndromes in the EU (PSP, ASP, DSP) have been incorporated into a microarray format employing plasmon surface resonance detection. The challenge of such testing is to be able to carry it out at environmentally relevant concentrations. Over the course of the project extensive work has been carried out to optimise extraction of toxins from cells and to increase the sensitivity of detection. Two years of field samples were tested with the toxin microarray and compared to Enzyme-linked immuno sorbent assays (ELISA). The latter were more sensitive but required extended preparation time and three different assays. For samples of significant toxicity both assays demonstrated good correlation of results and with cell counts and species microarray detection of relevant species.
The species microarray is now the subject of a patent application and commercial discussions with relevant companies to spot chips (Scienion, Germany), produce testing kits (Kreatech, UK) and sell and distribute kits (Microbia, France). The toxin chip commercialisation is still under discussion.
Project context and objectives:
Blooms of toxic or harmful microalgae, commonly referred to as HABs, represent a significant threat to fisheries resources and human health throughout the world. These phenomena manifest themselves in many ways, ranging from high phytoplankton biomass that discolours seawater with consequent loss of water quality to low density, yet highly toxic, populations, which can contaminate shellfish. The aquaculture industry in most European countries is a valuable resource and has been estimated to be worth approximately 60 million annually In Ireland alone. Monitoring programmes are a necessity because of the potential dangers to human health and the significant economic impacts of contaminated seafood posed by harmful events. In Europe, this requirement for monitoring is established in a series of directives in which monitoring of coastal waters for the presence of potentially harmful phytoplankton is mandatory (Council Directive 91 / 492). In MIDTAL, existing rRNA probes (18S, 28S) and antibodies for algal toxins were adapted and optimised for microarray format in order to develop a monitoring technique that strengthens our ability to monitor bulk water samples for toxic algae. The purpose was to provide a rapid test to aid national monitoring agencies by providing new rapid tools for the identification of toxic algae and their toxins so that they can comply with EC directive 91 / 492 / CEE that can be converted to cell numbers.
Traditionally phytoplankton monitoring is carried out by identification and enumeration using light microscopy. It has been recognised for some time that this technique requires a high degree of skill on behalf of the operator, and is time-consuming. Furthermore, the morphological similarity between different species within or even across phytoplankton genera has meant that light microscopy alone is often insufficient to assess the potential toxicity of a water sample. A variety of identification methods based on the sequencing of nucleic acids has been developed over the past decade or so that have considerably improved our ability to accurately identify organisms to species level. Deoxyribonucleic acid (DNA)-based molecular probe methods, such as Fluorescent in-situ hybridisation analysis (FISH), have been developed that can identify and quantify specific species in complex phytoplankton communities. Further advances have led to the development of DNA-biosensors for electrochemical detection of phytoplankton and their toxins and quantitative polymerase chain reaction (qPCR) techniques, which can provide accurate and reproducible quantification of gene copies.
MIDTAL has developed what will be the first potential commercial universal microarray (phylochip), capable of rapidly detecting the presence of specific harmful algal species before they develop into HABs and capable of quantifying them as the bloom develops. This microarray is based on fluorescent detection of labelled RNA hybridised to immobilised oligonucleotides probes. A microarray for phycotoxins based on plasmon surface resonance detection has also been developed. The project objectives were to:
1. test and optimise existing rRNA probes for toxic species and antibodies for toxins for their application to a microarray;
2. design and test the specificity of any new probe needed;
3. to construct a universal microarray from the probes tested and optimised by all of the partners for the detection of harmful algae and their toxins;
4. to provide national monitoring agencies with a rapid molecular tool to monitor toxic algae and to validate or replace traditional methods for monitoring for toxic algae; and
5. to integrate European efforts to monitor coastal waters for toxic algal species.
We achieved all of these objectives and are progressing towards commercialisation of the species microarray.
Project results:
For objective 1 of designing and testing the specificity of the probes, this was achieved in three steps representing three generations of microarray chips. In the first 18 months of the project, we reassessed all existing toxic algal FISH probes for their applicability and specificity in a microarray format. We designed new probes for those toxic algal species where none existed previously. In the first generation, we used oligonucleotides that were 18 bp in length but there were substantial cross-reactions. In the second generation chip, all probes were increased in length to 25 bp. This improved the signal to noise ratio and eliminated most but not all of the false positives. To prevent the occurrence of false positives, the hybridisation protocol was changed and the washing steps were conducted at a higher temperature and the hybridisation mixture was heated to 95 degrees of Celsius before being applied to the array. This increase in stringency was highly effective. The chip was also blocked with Kreablock, which substantially reduced the background of the slide for improved signal to noise ratio. For the third generation chip, additional thyamine nucleotides were added above the six carbon spacer to raise the probe even higher above the surface to facilitate binding to target molecules. Presently a total of 15 additional thyamine nucleotides plus six carbons are included in the synthesis of the oligonucleotides. A total of 163 hierarchical probes, consisting of modified and newly designed probes are now spotted on the third generation array (annex table 1), which enabled us to produce a universal microarray for the detection of toxic algae (objective 2). Together with the microarray for toxins, we have provided the national monitoring agencies with new tools to monitor toxic algae and to validate or replace traditional methods for monitoring for toxic algae.
Field collections were carried out in five countries and total RNA has been extracted from these for analysis by the microarray form two years of field data. Validation of the field material is by traditional cell counts made on the same material from which the RNA has been extracted. Additional validation was done with qPCR for selected species where qPCR primers were already available. Year 1 samples were hybridised with second generation chips and year 2 samples were hybridised with third generation chips. Results of year 1 analysis with the generation 2 chip were reported in month 36 and we report here the results of year 2 analysis with the generation 3 chip. Results indicate that the microarray signals are in general in agreement with the cells counts with the microarray indicating that we have achieved our goal of developing a species microarray for the detection of toxic algae species that was more sensitive that their detection using cell counts. The presence of the toxic species was further confirmed with the detection of their toxins using our toxin microarray. A comparison between toxins detected, species detected by the phylochip and species detected by cell counts in presented in annex table 2 for both years. Using the generation 3 chip, we were able to infer cell numbers from the microarray signal and a comparison between microarray inferred cell numbers and cell numbers counted can be seen in annex table 3.
Comparison of year 1 and 2 by partner
Partner 1 (Marine biological association of the United Kingdom): The first year data had problems with the RNA extraction of all cells (especially prorocentrum). Prorocentrum was detected via cell counting in all three samples of the first year but no signal was detected with the microarray. With an adapted RNA protocol for the second year, we were now able to break open all cells (especially prorocentrum), which can be seen in the signal of prorocentrum probes. Prorocentrum was detected in the second year for two samples in cell counting and three times with the microarray. The handling of the microarrays, the pre-hybridisation of the microarrays and the hybridisation itself resulted often in a high background, smears on the array and weak signals of the first year data. The use of the blocking procedure instead of the pre-hybridisation and the addition of KREAblock prior hybridisation solved the problem of high background and weak signals. The use of a new hierarchy-file of all probes implemented in the new GPR-analyser 1.27 helped to identify and exclude non-specific signals (cross reactions). Both years found more toxic species than those detected in microscopic counts. Pseudo-nitzschia cell counts show consistent agreement with both microarrays (second and third generation). In the first year, a few cells of Karenia brevis and K. mikimotoi had been detected in cell counts but not with the mircoarray. The genus level probe for alexandrium and dinophyta has been detected in all five samples of the 3rd year with the microarray but not always with cell counts. The reason for that is most likely the larger volume used for filtration (approximately 3 L) than for counting (1 L transect one microscope field of the counting chamber).
Partner 2 (Stazione Zoologica Anton Dohrn of Naples, Italy) has collected 24 samples (one per month) along two years, from 2009-09-08 to 2011-08-02 at its sampling site LTER MareChiara (Gulf of Naples). The RNA for hybridisation on microarrays was extracted from all the environmental samples and the corresponding fixed samples were counted using the Utermöhl method, in order to validate microarray data. While samples from the first year (2009-09-08 to 2010-08-03) were hybridised on the second generation of microarray slides, samples from the second year (2010-09-14 to 2011-02-08) were hybridised on the third generation of slides. Accordance between LM counts and microarray data were highly variable, depending both on the species and the type of slides.
With regard to the target genus of partner 2 (Pseudo-nitzschia), the signals of species-specific probes were fitting reasonably well with cell counts for most of the dates from the first and second year. This is generally the case for P. pseudodelicatissima and P. delicatissima species complex, although for few dates from the second year, the signals were strong even when only few cells were found in the net samples. For P. galaxiae, probes did not give a detectable signal also when the species was observed in cell counts.
The probes designed on P. fraudulenta often gave false positive signals because of cross-reactions even on the third generation slides. The strongest probe for the latter was thus reconsidered as partial genus level probe. Anyway, the analysis of false positives and negatives in field samples was complicated by the presence of many cross-reacting probes on the microarrays. For example, the probe PcaserausmultD02_25_dT reacted against P. calliantha and P. delicatissima, among the species found in the Gulf of Naples.
For the other HAB genera, we found that the probes for Alexandrium worked quite well for both years, with few false positives (A. tamarense for first year and A. minutum for the second year). The behaviour of the genus probes for dinophysis, karenia and prorocentrum varied with the different generations of slides: basically, on the third generation they did not switch on even if species of the genus were found either in counts - as in the case of prorocentrum - or in the net samples -as in the case of Karenia. The latter could depend on the presence of too few cells in the sample, falling under the detection threshold of the microarray. The species Gymnodinium catenatum not known to occur in the area - was not found in LM counts, but one probe was highlighted in a couple of occasions in both years. The Prymnesiophyceae probes from the two years were always switched on and we know from other observations that species of this family are always present in the samples from the Gulf of Naples. Raphidophyceae (Pseudochattonella and H. akashiwo) were not detected in the net samples in the few occasions where probes were switched on.
Partner 3 (Linnaeus University, Sweden): There seem to be a good agreement between cell counts and calculated cell numbers from normalised microarray signal, at least for the Dunaliella probes (note that not all species have the correct calibrations for calculating cell numbers). The signal intensity and especially signal / noise ratio has improved, i.e. background levels are much lower than before. Most notably, even though there are some false positives in year 2 samples, they seem to be fewer than in year 1. Also, some of these false positives were removed by the hierarchical file. However, there still seem to be some false negatives, in particular with Pseudo-nitzschia and Dinophysis genus level and some species level probes.
Partner 4 (IEO, Spain): There was observed a good qualitative agreement in Year 2 samples between detection of toxic species of dinophysis targeted by microarray probes (D. acuminata and D. acuta) and its presence in field samples. However, quantitative agreement was not optimal, despite that good calibration curves for at least D. acuminata species were obtained. The microarrays highlighted several potential toxic genera not detected in cell counts, like the dinoflagellates Azadinium and Karenia, the raphidophytes Heterosigma and pseudochattonella, and the haptophyte prymnesium.
Partner 5 (NUIG, Ireland): There was a marked increase in, and quality of, the labelling efficiencies and reduction of background noise on the microarray in samples taken by NUIG following the alterations made to the RNA extraction and hybridisation protocol for the third generation microarray for the second year samples as compared to those of the first year ones. An increase in specificity along with a marked decrease in cross reactivity and false positives was also observed in results of sample analysis. Some of the more interesting results from the sample series using the third generation chip site came from a bloom of alexandrium minutum in Cork Harbour (1 000 cell L-1 determined from microscopy in a sample taken on 23.06.2011) with the higher group and species level probe (AminuS01_25_dT; 5.4 s / n ratio) being highlighted on the third generation chip. This same sample when previously hybridised to the 2nd generation chip gave a false negative result of the A. minutum species level probe (1.45 s / n ratio). This exemplifies the increase in signal and specificity with the new and improved 3rd generation protocols. Correlations between microarray results and conventional cell counts in general was good, although the microarray appeared to overestimate numbers of Pseudo-nitzschia but this may still be a problem of sample volume.
Partner 6 (University of Oslo, Norway) sampled a site in the outer Oslofjord, which was not subject to major toxic bloom events during the sampling period, although several potentially toxic species were observed at low concentrations. Using the latest generation microarray and protocol (year 1 and year 2), results corresponded to those of cell counts, especially with larger species, such as Dinophysis, where best correlations were observed, but were less sensitive with respect to smaller species with low cellular RNA contents (e.g. pseudo-nitzschia or prorocentrum), especially in the presence of high quantities of other algae. For Pseudochattonella, a fish-killing alga difficult to detect in fixed samples using light microscopy, microarrays were successfully validated by qPCR. Poor correlations were mainly observed for Pseudo-nitzschia spp, where both false negatives and false positives were obtained using the arrays: here sensitivity and specificity still need to be further improved to be able to reliably detect and distinguish different species of this genus. In comparison, hybridisations using the initial hybridisation protocol without blocking and generation 2 chips (year 1 only) resulted in high numbers of false positives compared to light-microscopic cell counts.
Partner 7 (University of Westminster, United Kingdom): Year 1 and year 2 analyses of cell counts and microarray data from the second and third generation chip, respectively, are now complete. Overall, there was a good agreement between cell counts and array data in both years. However, year 2 saw distinct improvements in the signal strength and reducing background. Introduction of a heterokont competitor probe has distributed the heterokont signal for evenly around the probes. In some summer months, the microarray appeared more sensitive identifying more species this was perhaps caused by the larger volumes of seawater analysed. Cell counts for both dinophysis and alexandrium matched well over the course of the year with the microarray data in year 1. In year 2 Karenia mikimotoi was more prevalent cumulating in an autumn bloom, which was detected by both the microarray and counts.
Partner 9 (INTECMAR, Spain) has detected the toxic / harmful species gymnodinium catenatum, alexandrium minutum, Dinophysis spp., notably D. acuminata, D. acuta and D. caudata, pseudo-nitzschia spp., prorocentrum minimum, azadinium spp., karenia spp., karlodinium spp. and heterosigma akashiwo among many other species, in the counts of the concentrated samples. In comparing the counts to the microarray analyses, they obtained very good correlations for samples with Paralytic shellfish poison (PSP)-producing species: The microchip detected G. catenatum with no false positives or false negatives in 19 samples, 9 of which had cells and cysts. Alexandrium minutum was also detected even with very high accuracy in the estimating of number of cells. During year two, there was a good correlation of detection of pseudo-nitzschia at genus level and cells counts, especially during the occurrence of an ASP episode in the Ria de Muros but with the microarray detecting more species than were done by light microscopy.
Partner 10 (University of Rhode Island, USA) applied for funding in the United States but was never successful. Nevertheless, a student Angelica Herrera, a PhD student at CIBNOR in Mexico, co supervised by Medlin at the MBA travelled to URI in spring 2012 to work in Maranda's laboratory to continue the development of probes for prorocentrum spp. and to standardise the FISH step. These probes have been added to our microarray.
Partner 11 (Queens University of Belfast, United Kingdom) was successful in the production of a toxin microarray for the detection of PSP toxins, okadaic acid and DTXs, and domoic acid in seawater and algal culture samples. Year 1 and year 2 filter samples were analysed and compared by ELISA. The ELISA assays demonstrated better sensitivity but their use required an overnight incubation step and the use of three different assays compared to a single analysis. For samples of significant toxicity, both assays demonstrated good correlation in results. There is also a correlation of these samples with cell counts and RNA microarray data produced.
Exploitable foreground: The phylochip (species microarray) will enable monitoring agencies to check for the presence of toxic algae in their waters at an early stage and to take the proper measures to prevent contamination and the economic losses associated with this. They will also be able to check for toxins, which will allow screening of samples and reduction in the need for the mouse bioassay. The MIDTAL phylochip for toxins and toxic species is expected to reduce the health risk for humans who eat farm-raised fish and shellfish and even those who collect shellfish personally because warning notices not to collect shellfish can be posted earlier. We have filed for a patent for the species microarray and identified a commercial company (Scienion, Berlin Germany) to spot the chips, another commercial company (Kreatech, UK) to produce the kit with the chip and a further company (Microbia, Banyuls sur Mer, France) to sell and distribute the kits. Several monitoring agencies have contacted us to run a field trial of the chip.
Potential impact:
Socio-economic Impact of the MIDTAL project
Wider societal implications of the project
The public perception of HABs is often that of eutrophication, where increased nutrient loading results in large biomass blooms which have a consequent effect on water quality (anoxia) and the ecosystem through substantial alterations in both the benthic and pelagic ecology. All too often, events which directly or indirectly relate to increased nutrients such as dead zones in the northern Gulf of Mexico, a huge clean-up operation following an Enteromorpha bloom off Qingdao in time for the Olympic sailing competition in 2008, or the knock on effects of hydrogen sulphide toxicity due to rotting vegetation following extensive growth of both macro (Ulva, Brittany coast of France) and micro (Phaeocystis, North Wales, United Kingdom) algae, find substantial room in the popular press.
What is less well known is that HABs caused by toxin producing algae have not only a far more serious impact on the quality of life, but also they are entirely natural. Anthropogenic impacts on toxic HABs are limited to either their global spread through potential transport vectors, such as ballast water of ships, or because of alterations in the coastline resulting from the construction of harbours and marinas, particularly around the Mediterranean coast, which promotes the development of chronic dinoflagellate blooms, such as the PSP toxin producer alexandrium.
The global scale of toxin producing micro-algae should not be underestimated. For example, the most serious would be the numbers of human intoxications with ciguatera, caused by the dinoflagellate gambierdiscus, is currently estimated at some 50 000 per year. Every year, 1-2 human deaths are linked to the ingestion of PSP toxins. Although these problems are restricted to the tropical / warm temperate sphere of the globe, it demonstrates the urgent need to be able to monitor and prevent toxic HAB events. In Europe, this is the effect of a series of directives that require coastal member states to monitor water for toxin producing species and their toxins in shellfish. Starting with the EU 'Shellfish hygiene' directive 91 / 492 / EEC, a series of directives was issued to include newly discovered toxins, and stipulating the methods of analysis and maximum permitted levels in shellfish. The most important of these would be 2002 / 225 / EC and 2074 / 2005 (pertaining to toxin levels and analysis and methods) and more recently 15 / 2011 (analysis methods). The natural occurrence of toxin producing algae, and the continual human demand for shellfish consumption, means that their monitoring is here to stay.
The cost of this monitoring of plankton and toxins is enormous. Although there is limited hard?information on the economic impact of HABs, a relatively recent study in the US (Anderson et al., 2000) has estimated, on a national basis, that:
- the cost of monitoring is equivalent to 5 % annual shellfish industry turnover;
- the cost of lost harvest and damaged product due to contamination with biotoxins is 5 % of industry turnover;
- the public health costs due to lost working days, hospitalisations etc. add another 5 % of annual turnover.
In Europe, similar information is also difficult to uncover, but the context is well set in that if one takes the case of Ireland, the shellfish aquaculture production currently runs at?47 million annually (Bowne et al., 2007) and the budget for the Irish National Biotoxin and Toxic Phytoplankton monitoring programme, carried out under the auspices of the Food Safety Authority of Ireland, and operated through the Irish Marine Institute, is 1.7 million, representing approximately 3.5 % of annual industry turnover. Similarly, Scottish shellfish production is valued at approximately GBP 20 million, the most part of which is through culture of the edible mussel mytilus edulis, and the monitoring programmes, run by the food standards agency in Scotland, has a budget of just under GBP 2 million. Figures for the cost of harvest closures and destroyed stock are unavailable, but the parity with the monitoring cost with that of the US allows a confidence in the data provided by Anderson et al (2000).
Clearly the development of an industry that is both natural and sustainable, but which has such a heavy financial burden, requires all the assistance possible in order to overcome such a natural hazard as toxic HABs, as the (natural) problems caused by toxicity will never go away. The project MIDTAL has been an endeavour to improve national monitoring capabilities for toxin producing plankton species, and their toxins, in water samples.
Approximately 2 000 water samples are analysed annually in Ireland as part of the National monitoring programme (NMP). This requires a staff of 4 people, augmented slightly during the busy summer months. Most samples are scanned for toxic / harmful species but samples from 10 sites (out of a total of approximately 60) are analysed for their total phytoplankton community. Light microscopy is the routine analysis method, each sample requiring approximately 2 hours on average to examine. Occasionally, some samples for pseudo-nitzschia are analysed using FISH. Comparable figures for other monitoring programmes are annual throughputs of 1 000 samples (Scotland), 5 000 samples (REPHY, France), and 6 000 samples (Galicia, Spain). These figures reflect a work rate of processing some 20 samples per week per person. The number of man-hours involved in the monitoring process is clearly enormous. The microarray technique would reduce the number of inevitable mistakes due to human error that is an ever present facet of this type of work.
Although this sample analysis throughput rate is unlikely to be significantly increased from use of the microarray in any laboratory, the benefits of increased accuracy would far outweigh any required capital investment. Of particular relevance are the situations with respect to pseudo-nitzschia, which cannot be identified to species level using light microscopy, and alexandrium, another genus with which it is also virtually impossible to identify accurately to species using this technique. With respect to the latter, a case in point would be the situation in Cork Harbour, where for over a decade it was recognised that alexandrium tamarense was the PSP producing species (FAO, 2004). Introduction of the use of molecular techniques quickly rationalised the fact that A. minutum was in fact the toxin producer, and that the co-occurring A. tamarense was the non-toxic Gp III (western European) form. This is a level of accuracy essential if predictive models are to be produced that can forecast toxic blooms and that can then (and only then) allow their mitigation. Invariably, these models are produced using data derived from monitoring programmes.
As regards to toxin-testing capability, during the course of the project the European Commission published a new directive (15 / 2011) that changed the official reference method for the analysis of biotoxins in shellfish from the mouse and rat bioassay to a chemical method. The validated technique of Liquid chromatography (LC) mass spectrometry (MS) should now be applied as the reference method for the detection of lipophilic toxins and used as matter of routine, both for the purposes of official controls at any stage of the food chain and own-checks by food business operators. Of particular relevance is the statement that LC MS / MS is the reference method for okadaic acid and derivatives, yessotoxins, pectenotoxins and azaspiracids. The project MIDTAL does not, at present, seek to alter this legislation: the toxin analysis capability of the microarray can still be used satisfactorily as a method for end-product testing.
WP4 - Dissemination
A final workshop was held at month 44 and was very well received by the people attending the workshop, some of whom were from monitoring agencies. A manual for the hybridisation of the MIDTAL chip will be published by Koeltz, Germany.
Database: We have developed a database storage site to store the GPR files from each hybridisation. From this website, graphs can be generated to compare the performance of probes over time at various sites. A link to the database appears on the project website and well as on the MBA website
Task 1 - Website
www.midtal.com with 12629 hits as of 31 May 2012. The internal page has all reports, talks and protocols formulated to date.
Task 2 - Reports
All internal reports have been sent to the scientific officer on a 2-3 month basis. Minutes of each consortium meeting have also been sent.
Task 3 - Dissemination
Talks and posters and other presentations have continued throughout the lifetime of the project. A summary of all dissemination activities is on the ecas website.
A webinar program was held on 13 June and sponsored by Marine TT. Here is the link to a recorded version of that programme: www.marineTT.eu
List of Websites: www.midtal.com www.mba.ac.uk/midtal
The MIDTAL project set out to develop a universal microarray to test for Harmful algal bloom (HAB) species and their toxins relevant to the European Community, to validate the array against field samples from around Europe and to seek to commercialise this technology for wider application. Early in the project the detection of species and toxins were disaggregated to two separate microarrays that were concurrently developed.
Species microarray: Three generations of improving microarrays were developed and tested during the project. The final generation includes 163 probes at various taxonomic hierarchies covering all the major harmful algal species of current interest in the European Union (EU). The origin of the initial species probes for the microarray was from previously developed technologies (e.g. FISH) that had to be tested (and in a number of cases redesigned) for the changed detection platform. Key features of the microarray design are the inclusion of internal controls (universal eukaryotic probe; Dunaliella probe) and the taxonomic hierarchy of probes that can be used to aid interpretation of results. Developments to the microarray detection system have included lengthening the probes (from 18 to 25 bp) to improve specificity; extending the attachment of the probes to the microarray slide to improve accessibility by target molecules and developing the microarray methodology to improve stringency and signal strength.
Two years of monthly field samples have been collected from key sites across the EU. The first year of these samples were tested against the second generation chip and the second year samples have been tested against the third generation chip. Samples from both years have been tested with the toxin chip.
The ribosomal ribonucleic acid (rRNA) content of different species in different phases of the growth cycle and grown under different environmental stresses was tested and shown to be reasonably consistent with no dramatic reductions in stationary phase. Calibration data generated for different species has allowed development of algorithms to convert microarray signal to cell numbers. Results so far indicate that cell counts and microarray hybridisations generally show reasonable agreement with the observation that usually the microarray records more species than the cell counts, especially in respect of small flagellate cells that are particularly difficult to quantify by light microscopy and Pseudo-nitzschia spp, which can only be reliably identified with electron microscopy.
Toxin microarray: Antibodies to cover the three main toxicity syndromes in the EU (PSP, ASP, DSP) have been incorporated into a microarray format employing plasmon surface resonance detection. The challenge of such testing is to be able to carry it out at environmentally relevant concentrations. Over the course of the project extensive work has been carried out to optimise extraction of toxins from cells and to increase the sensitivity of detection. Two years of field samples were tested with the toxin microarray and compared to Enzyme-linked immuno sorbent assays (ELISA). The latter were more sensitive but required extended preparation time and three different assays. For samples of significant toxicity both assays demonstrated good correlation of results and with cell counts and species microarray detection of relevant species.
The species microarray is now the subject of a patent application and commercial discussions with relevant companies to spot chips (Scienion, Germany), produce testing kits (Kreatech, UK) and sell and distribute kits (Microbia, France). The toxin chip commercialisation is still under discussion.
Project context and objectives:
Blooms of toxic or harmful microalgae, commonly referred to as HABs, represent a significant threat to fisheries resources and human health throughout the world. These phenomena manifest themselves in many ways, ranging from high phytoplankton biomass that discolours seawater with consequent loss of water quality to low density, yet highly toxic, populations, which can contaminate shellfish. The aquaculture industry in most European countries is a valuable resource and has been estimated to be worth approximately 60 million annually In Ireland alone. Monitoring programmes are a necessity because of the potential dangers to human health and the significant economic impacts of contaminated seafood posed by harmful events. In Europe, this requirement for monitoring is established in a series of directives in which monitoring of coastal waters for the presence of potentially harmful phytoplankton is mandatory (Council Directive 91 / 492). In MIDTAL, existing rRNA probes (18S, 28S) and antibodies for algal toxins were adapted and optimised for microarray format in order to develop a monitoring technique that strengthens our ability to monitor bulk water samples for toxic algae. The purpose was to provide a rapid test to aid national monitoring agencies by providing new rapid tools for the identification of toxic algae and their toxins so that they can comply with EC directive 91 / 492 / CEE that can be converted to cell numbers.
Traditionally phytoplankton monitoring is carried out by identification and enumeration using light microscopy. It has been recognised for some time that this technique requires a high degree of skill on behalf of the operator, and is time-consuming. Furthermore, the morphological similarity between different species within or even across phytoplankton genera has meant that light microscopy alone is often insufficient to assess the potential toxicity of a water sample. A variety of identification methods based on the sequencing of nucleic acids has been developed over the past decade or so that have considerably improved our ability to accurately identify organisms to species level. Deoxyribonucleic acid (DNA)-based molecular probe methods, such as Fluorescent in-situ hybridisation analysis (FISH), have been developed that can identify and quantify specific species in complex phytoplankton communities. Further advances have led to the development of DNA-biosensors for electrochemical detection of phytoplankton and their toxins and quantitative polymerase chain reaction (qPCR) techniques, which can provide accurate and reproducible quantification of gene copies.
MIDTAL has developed what will be the first potential commercial universal microarray (phylochip), capable of rapidly detecting the presence of specific harmful algal species before they develop into HABs and capable of quantifying them as the bloom develops. This microarray is based on fluorescent detection of labelled RNA hybridised to immobilised oligonucleotides probes. A microarray for phycotoxins based on plasmon surface resonance detection has also been developed. The project objectives were to:
1. test and optimise existing rRNA probes for toxic species and antibodies for toxins for their application to a microarray;
2. design and test the specificity of any new probe needed;
3. to construct a universal microarray from the probes tested and optimised by all of the partners for the detection of harmful algae and their toxins;
4. to provide national monitoring agencies with a rapid molecular tool to monitor toxic algae and to validate or replace traditional methods for monitoring for toxic algae; and
5. to integrate European efforts to monitor coastal waters for toxic algal species.
We achieved all of these objectives and are progressing towards commercialisation of the species microarray.
Project results:
For objective 1 of designing and testing the specificity of the probes, this was achieved in three steps representing three generations of microarray chips. In the first 18 months of the project, we reassessed all existing toxic algal FISH probes for their applicability and specificity in a microarray format. We designed new probes for those toxic algal species where none existed previously. In the first generation, we used oligonucleotides that were 18 bp in length but there were substantial cross-reactions. In the second generation chip, all probes were increased in length to 25 bp. This improved the signal to noise ratio and eliminated most but not all of the false positives. To prevent the occurrence of false positives, the hybridisation protocol was changed and the washing steps were conducted at a higher temperature and the hybridisation mixture was heated to 95 degrees of Celsius before being applied to the array. This increase in stringency was highly effective. The chip was also blocked with Kreablock, which substantially reduced the background of the slide for improved signal to noise ratio. For the third generation chip, additional thyamine nucleotides were added above the six carbon spacer to raise the probe even higher above the surface to facilitate binding to target molecules. Presently a total of 15 additional thyamine nucleotides plus six carbons are included in the synthesis of the oligonucleotides. A total of 163 hierarchical probes, consisting of modified and newly designed probes are now spotted on the third generation array (annex table 1), which enabled us to produce a universal microarray for the detection of toxic algae (objective 2). Together with the microarray for toxins, we have provided the national monitoring agencies with new tools to monitor toxic algae and to validate or replace traditional methods for monitoring for toxic algae.
Field collections were carried out in five countries and total RNA has been extracted from these for analysis by the microarray form two years of field data. Validation of the field material is by traditional cell counts made on the same material from which the RNA has been extracted. Additional validation was done with qPCR for selected species where qPCR primers were already available. Year 1 samples were hybridised with second generation chips and year 2 samples were hybridised with third generation chips. Results of year 1 analysis with the generation 2 chip were reported in month 36 and we report here the results of year 2 analysis with the generation 3 chip. Results indicate that the microarray signals are in general in agreement with the cells counts with the microarray indicating that we have achieved our goal of developing a species microarray for the detection of toxic algae species that was more sensitive that their detection using cell counts. The presence of the toxic species was further confirmed with the detection of their toxins using our toxin microarray. A comparison between toxins detected, species detected by the phylochip and species detected by cell counts in presented in annex table 2 for both years. Using the generation 3 chip, we were able to infer cell numbers from the microarray signal and a comparison between microarray inferred cell numbers and cell numbers counted can be seen in annex table 3.
Comparison of year 1 and 2 by partner
Partner 1 (Marine biological association of the United Kingdom): The first year data had problems with the RNA extraction of all cells (especially prorocentrum). Prorocentrum was detected via cell counting in all three samples of the first year but no signal was detected with the microarray. With an adapted RNA protocol for the second year, we were now able to break open all cells (especially prorocentrum), which can be seen in the signal of prorocentrum probes. Prorocentrum was detected in the second year for two samples in cell counting and three times with the microarray. The handling of the microarrays, the pre-hybridisation of the microarrays and the hybridisation itself resulted often in a high background, smears on the array and weak signals of the first year data. The use of the blocking procedure instead of the pre-hybridisation and the addition of KREAblock prior hybridisation solved the problem of high background and weak signals. The use of a new hierarchy-file of all probes implemented in the new GPR-analyser 1.27 helped to identify and exclude non-specific signals (cross reactions). Both years found more toxic species than those detected in microscopic counts. Pseudo-nitzschia cell counts show consistent agreement with both microarrays (second and third generation). In the first year, a few cells of Karenia brevis and K. mikimotoi had been detected in cell counts but not with the mircoarray. The genus level probe for alexandrium and dinophyta has been detected in all five samples of the 3rd year with the microarray but not always with cell counts. The reason for that is most likely the larger volume used for filtration (approximately 3 L) than for counting (1 L transect one microscope field of the counting chamber).
Partner 2 (Stazione Zoologica Anton Dohrn of Naples, Italy) has collected 24 samples (one per month) along two years, from 2009-09-08 to 2011-08-02 at its sampling site LTER MareChiara (Gulf of Naples). The RNA for hybridisation on microarrays was extracted from all the environmental samples and the corresponding fixed samples were counted using the Utermöhl method, in order to validate microarray data. While samples from the first year (2009-09-08 to 2010-08-03) were hybridised on the second generation of microarray slides, samples from the second year (2010-09-14 to 2011-02-08) were hybridised on the third generation of slides. Accordance between LM counts and microarray data were highly variable, depending both on the species and the type of slides.
With regard to the target genus of partner 2 (Pseudo-nitzschia), the signals of species-specific probes were fitting reasonably well with cell counts for most of the dates from the first and second year. This is generally the case for P. pseudodelicatissima and P. delicatissima species complex, although for few dates from the second year, the signals were strong even when only few cells were found in the net samples. For P. galaxiae, probes did not give a detectable signal also when the species was observed in cell counts.
The probes designed on P. fraudulenta often gave false positive signals because of cross-reactions even on the third generation slides. The strongest probe for the latter was thus reconsidered as partial genus level probe. Anyway, the analysis of false positives and negatives in field samples was complicated by the presence of many cross-reacting probes on the microarrays. For example, the probe PcaserausmultD02_25_dT reacted against P. calliantha and P. delicatissima, among the species found in the Gulf of Naples.
For the other HAB genera, we found that the probes for Alexandrium worked quite well for both years, with few false positives (A. tamarense for first year and A. minutum for the second year). The behaviour of the genus probes for dinophysis, karenia and prorocentrum varied with the different generations of slides: basically, on the third generation they did not switch on even if species of the genus were found either in counts - as in the case of prorocentrum - or in the net samples -as in the case of Karenia. The latter could depend on the presence of too few cells in the sample, falling under the detection threshold of the microarray. The species Gymnodinium catenatum not known to occur in the area - was not found in LM counts, but one probe was highlighted in a couple of occasions in both years. The Prymnesiophyceae probes from the two years were always switched on and we know from other observations that species of this family are always present in the samples from the Gulf of Naples. Raphidophyceae (Pseudochattonella and H. akashiwo) were not detected in the net samples in the few occasions where probes were switched on.
Partner 3 (Linnaeus University, Sweden): There seem to be a good agreement between cell counts and calculated cell numbers from normalised microarray signal, at least for the Dunaliella probes (note that not all species have the correct calibrations for calculating cell numbers). The signal intensity and especially signal / noise ratio has improved, i.e. background levels are much lower than before. Most notably, even though there are some false positives in year 2 samples, they seem to be fewer than in year 1. Also, some of these false positives were removed by the hierarchical file. However, there still seem to be some false negatives, in particular with Pseudo-nitzschia and Dinophysis genus level and some species level probes.
Partner 4 (IEO, Spain): There was observed a good qualitative agreement in Year 2 samples between detection of toxic species of dinophysis targeted by microarray probes (D. acuminata and D. acuta) and its presence in field samples. However, quantitative agreement was not optimal, despite that good calibration curves for at least D. acuminata species were obtained. The microarrays highlighted several potential toxic genera not detected in cell counts, like the dinoflagellates Azadinium and Karenia, the raphidophytes Heterosigma and pseudochattonella, and the haptophyte prymnesium.
Partner 5 (NUIG, Ireland): There was a marked increase in, and quality of, the labelling efficiencies and reduction of background noise on the microarray in samples taken by NUIG following the alterations made to the RNA extraction and hybridisation protocol for the third generation microarray for the second year samples as compared to those of the first year ones. An increase in specificity along with a marked decrease in cross reactivity and false positives was also observed in results of sample analysis. Some of the more interesting results from the sample series using the third generation chip site came from a bloom of alexandrium minutum in Cork Harbour (1 000 cell L-1 determined from microscopy in a sample taken on 23.06.2011) with the higher group and species level probe (AminuS01_25_dT; 5.4 s / n ratio) being highlighted on the third generation chip. This same sample when previously hybridised to the 2nd generation chip gave a false negative result of the A. minutum species level probe (1.45 s / n ratio). This exemplifies the increase in signal and specificity with the new and improved 3rd generation protocols. Correlations between microarray results and conventional cell counts in general was good, although the microarray appeared to overestimate numbers of Pseudo-nitzschia but this may still be a problem of sample volume.
Partner 6 (University of Oslo, Norway) sampled a site in the outer Oslofjord, which was not subject to major toxic bloom events during the sampling period, although several potentially toxic species were observed at low concentrations. Using the latest generation microarray and protocol (year 1 and year 2), results corresponded to those of cell counts, especially with larger species, such as Dinophysis, where best correlations were observed, but were less sensitive with respect to smaller species with low cellular RNA contents (e.g. pseudo-nitzschia or prorocentrum), especially in the presence of high quantities of other algae. For Pseudochattonella, a fish-killing alga difficult to detect in fixed samples using light microscopy, microarrays were successfully validated by qPCR. Poor correlations were mainly observed for Pseudo-nitzschia spp, where both false negatives and false positives were obtained using the arrays: here sensitivity and specificity still need to be further improved to be able to reliably detect and distinguish different species of this genus. In comparison, hybridisations using the initial hybridisation protocol without blocking and generation 2 chips (year 1 only) resulted in high numbers of false positives compared to light-microscopic cell counts.
Partner 7 (University of Westminster, United Kingdom): Year 1 and year 2 analyses of cell counts and microarray data from the second and third generation chip, respectively, are now complete. Overall, there was a good agreement between cell counts and array data in both years. However, year 2 saw distinct improvements in the signal strength and reducing background. Introduction of a heterokont competitor probe has distributed the heterokont signal for evenly around the probes. In some summer months, the microarray appeared more sensitive identifying more species this was perhaps caused by the larger volumes of seawater analysed. Cell counts for both dinophysis and alexandrium matched well over the course of the year with the microarray data in year 1. In year 2 Karenia mikimotoi was more prevalent cumulating in an autumn bloom, which was detected by both the microarray and counts.
Partner 9 (INTECMAR, Spain) has detected the toxic / harmful species gymnodinium catenatum, alexandrium minutum, Dinophysis spp., notably D. acuminata, D. acuta and D. caudata, pseudo-nitzschia spp., prorocentrum minimum, azadinium spp., karenia spp., karlodinium spp. and heterosigma akashiwo among many other species, in the counts of the concentrated samples. In comparing the counts to the microarray analyses, they obtained very good correlations for samples with Paralytic shellfish poison (PSP)-producing species: The microchip detected G. catenatum with no false positives or false negatives in 19 samples, 9 of which had cells and cysts. Alexandrium minutum was also detected even with very high accuracy in the estimating of number of cells. During year two, there was a good correlation of detection of pseudo-nitzschia at genus level and cells counts, especially during the occurrence of an ASP episode in the Ria de Muros but with the microarray detecting more species than were done by light microscopy.
Partner 10 (University of Rhode Island, USA) applied for funding in the United States but was never successful. Nevertheless, a student Angelica Herrera, a PhD student at CIBNOR in Mexico, co supervised by Medlin at the MBA travelled to URI in spring 2012 to work in Maranda's laboratory to continue the development of probes for prorocentrum spp. and to standardise the FISH step. These probes have been added to our microarray.
Partner 11 (Queens University of Belfast, United Kingdom) was successful in the production of a toxin microarray for the detection of PSP toxins, okadaic acid and DTXs, and domoic acid in seawater and algal culture samples. Year 1 and year 2 filter samples were analysed and compared by ELISA. The ELISA assays demonstrated better sensitivity but their use required an overnight incubation step and the use of three different assays compared to a single analysis. For samples of significant toxicity, both assays demonstrated good correlation in results. There is also a correlation of these samples with cell counts and RNA microarray data produced.
Exploitable foreground: The phylochip (species microarray) will enable monitoring agencies to check for the presence of toxic algae in their waters at an early stage and to take the proper measures to prevent contamination and the economic losses associated with this. They will also be able to check for toxins, which will allow screening of samples and reduction in the need for the mouse bioassay. The MIDTAL phylochip for toxins and toxic species is expected to reduce the health risk for humans who eat farm-raised fish and shellfish and even those who collect shellfish personally because warning notices not to collect shellfish can be posted earlier. We have filed for a patent for the species microarray and identified a commercial company (Scienion, Berlin Germany) to spot the chips, another commercial company (Kreatech, UK) to produce the kit with the chip and a further company (Microbia, Banyuls sur Mer, France) to sell and distribute the kits. Several monitoring agencies have contacted us to run a field trial of the chip.
Potential impact:
Socio-economic Impact of the MIDTAL project
Wider societal implications of the project
The public perception of HABs is often that of eutrophication, where increased nutrient loading results in large biomass blooms which have a consequent effect on water quality (anoxia) and the ecosystem through substantial alterations in both the benthic and pelagic ecology. All too often, events which directly or indirectly relate to increased nutrients such as dead zones in the northern Gulf of Mexico, a huge clean-up operation following an Enteromorpha bloom off Qingdao in time for the Olympic sailing competition in 2008, or the knock on effects of hydrogen sulphide toxicity due to rotting vegetation following extensive growth of both macro (Ulva, Brittany coast of France) and micro (Phaeocystis, North Wales, United Kingdom) algae, find substantial room in the popular press.
What is less well known is that HABs caused by toxin producing algae have not only a far more serious impact on the quality of life, but also they are entirely natural. Anthropogenic impacts on toxic HABs are limited to either their global spread through potential transport vectors, such as ballast water of ships, or because of alterations in the coastline resulting from the construction of harbours and marinas, particularly around the Mediterranean coast, which promotes the development of chronic dinoflagellate blooms, such as the PSP toxin producer alexandrium.
The global scale of toxin producing micro-algae should not be underestimated. For example, the most serious would be the numbers of human intoxications with ciguatera, caused by the dinoflagellate gambierdiscus, is currently estimated at some 50 000 per year. Every year, 1-2 human deaths are linked to the ingestion of PSP toxins. Although these problems are restricted to the tropical / warm temperate sphere of the globe, it demonstrates the urgent need to be able to monitor and prevent toxic HAB events. In Europe, this is the effect of a series of directives that require coastal member states to monitor water for toxin producing species and their toxins in shellfish. Starting with the EU 'Shellfish hygiene' directive 91 / 492 / EEC, a series of directives was issued to include newly discovered toxins, and stipulating the methods of analysis and maximum permitted levels in shellfish. The most important of these would be 2002 / 225 / EC and 2074 / 2005 (pertaining to toxin levels and analysis and methods) and more recently 15 / 2011 (analysis methods). The natural occurrence of toxin producing algae, and the continual human demand for shellfish consumption, means that their monitoring is here to stay.
The cost of this monitoring of plankton and toxins is enormous. Although there is limited hard?information on the economic impact of HABs, a relatively recent study in the US (Anderson et al., 2000) has estimated, on a national basis, that:
- the cost of monitoring is equivalent to 5 % annual shellfish industry turnover;
- the cost of lost harvest and damaged product due to contamination with biotoxins is 5 % of industry turnover;
- the public health costs due to lost working days, hospitalisations etc. add another 5 % of annual turnover.
In Europe, similar information is also difficult to uncover, but the context is well set in that if one takes the case of Ireland, the shellfish aquaculture production currently runs at?47 million annually (Bowne et al., 2007) and the budget for the Irish National Biotoxin and Toxic Phytoplankton monitoring programme, carried out under the auspices of the Food Safety Authority of Ireland, and operated through the Irish Marine Institute, is 1.7 million, representing approximately 3.5 % of annual industry turnover. Similarly, Scottish shellfish production is valued at approximately GBP 20 million, the most part of which is through culture of the edible mussel mytilus edulis, and the monitoring programmes, run by the food standards agency in Scotland, has a budget of just under GBP 2 million. Figures for the cost of harvest closures and destroyed stock are unavailable, but the parity with the monitoring cost with that of the US allows a confidence in the data provided by Anderson et al (2000).
Clearly the development of an industry that is both natural and sustainable, but which has such a heavy financial burden, requires all the assistance possible in order to overcome such a natural hazard as toxic HABs, as the (natural) problems caused by toxicity will never go away. The project MIDTAL has been an endeavour to improve national monitoring capabilities for toxin producing plankton species, and their toxins, in water samples.
Approximately 2 000 water samples are analysed annually in Ireland as part of the National monitoring programme (NMP). This requires a staff of 4 people, augmented slightly during the busy summer months. Most samples are scanned for toxic / harmful species but samples from 10 sites (out of a total of approximately 60) are analysed for their total phytoplankton community. Light microscopy is the routine analysis method, each sample requiring approximately 2 hours on average to examine. Occasionally, some samples for pseudo-nitzschia are analysed using FISH. Comparable figures for other monitoring programmes are annual throughputs of 1 000 samples (Scotland), 5 000 samples (REPHY, France), and 6 000 samples (Galicia, Spain). These figures reflect a work rate of processing some 20 samples per week per person. The number of man-hours involved in the monitoring process is clearly enormous. The microarray technique would reduce the number of inevitable mistakes due to human error that is an ever present facet of this type of work.
Although this sample analysis throughput rate is unlikely to be significantly increased from use of the microarray in any laboratory, the benefits of increased accuracy would far outweigh any required capital investment. Of particular relevance are the situations with respect to pseudo-nitzschia, which cannot be identified to species level using light microscopy, and alexandrium, another genus with which it is also virtually impossible to identify accurately to species using this technique. With respect to the latter, a case in point would be the situation in Cork Harbour, where for over a decade it was recognised that alexandrium tamarense was the PSP producing species (FAO, 2004). Introduction of the use of molecular techniques quickly rationalised the fact that A. minutum was in fact the toxin producer, and that the co-occurring A. tamarense was the non-toxic Gp III (western European) form. This is a level of accuracy essential if predictive models are to be produced that can forecast toxic blooms and that can then (and only then) allow their mitigation. Invariably, these models are produced using data derived from monitoring programmes.
As regards to toxin-testing capability, during the course of the project the European Commission published a new directive (15 / 2011) that changed the official reference method for the analysis of biotoxins in shellfish from the mouse and rat bioassay to a chemical method. The validated technique of Liquid chromatography (LC) mass spectrometry (MS) should now be applied as the reference method for the detection of lipophilic toxins and used as matter of routine, both for the purposes of official controls at any stage of the food chain and own-checks by food business operators. Of particular relevance is the statement that LC MS / MS is the reference method for okadaic acid and derivatives, yessotoxins, pectenotoxins and azaspiracids. The project MIDTAL does not, at present, seek to alter this legislation: the toxin analysis capability of the microarray can still be used satisfactorily as a method for end-product testing.
WP4 - Dissemination
A final workshop was held at month 44 and was very well received by the people attending the workshop, some of whom were from monitoring agencies. A manual for the hybridisation of the MIDTAL chip will be published by Koeltz, Germany.
Database: We have developed a database storage site to store the GPR files from each hybridisation. From this website, graphs can be generated to compare the performance of probes over time at various sites. A link to the database appears on the project website and well as on the MBA website
Task 1 - Website
www.midtal.com with 12629 hits as of 31 May 2012. The internal page has all reports, talks and protocols formulated to date.
Task 2 - Reports
All internal reports have been sent to the scientific officer on a 2-3 month basis. Minutes of each consortium meeting have also been sent.
Task 3 - Dissemination
Talks and posters and other presentations have continued throughout the lifetime of the project. A summary of all dissemination activities is on the ecas website.
A webinar program was held on 13 June and sponsored by Marine TT. Here is the link to a recorded version of that programme: www.marineTT.eu
List of Websites: www.midtal.com www.mba.ac.uk/midtal