The use of fluorescence microscopy and image analysis for rapid detection of non-producing revertant cells of Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002
© Schulze et al.; licensee BioMed Central. 2015
Received: 3 November 2014
Accepted: 31 March 2015
Published: 17 April 2015
Ethanol production via genetically engineered cyanobacteria is a promising solution for the production of biofuels. Through the introduction of a pyruvate decarboxylase and alcohol dehydrogenase direct ethanol production becomes possible within the cells. However, during cultivation genetic instability can lead to mutations and thus loss of ethanol production. Cells then revert back to the wild type phenotype.
A method for a rapid and simple detection of these non-producing revertant cells in an ethanol producing cell population is an important quality control measure in order to predict genetic stability and the longevity of a producing culture. Several comparable cultivation experiments revealed a difference in the pigmentation for non-producing and producing cells: the accessory pigment phycocyanin (PC) is reduced in case of the ethanol producer, resulting in a yellowish appearance of the culture. Microarray and western blot studies of Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002 confirmed this PC reduction on the level of RNA and protein.
Based on these findings we developed a method for fluorescence microscopy in order to distinguish producing and non-producing cells with respect to their pigmentation phenotype. By applying a specific filter set the emitted fluorescence of a producer cell with a reduced PC content appeared orange. The emitted fluorescence of a non-producing cell with a wt pigmentation phenotype was detected in red, and dead cells in green. In an automated process multiple images of each sample were taken and analyzed with a plugin for the image analysis software ImageJ to identify dead (green), non-producing (red) and producing (orange) cells.
The results of the presented validation experiments revealed a good identification with 98 % red cells in the wt sample and 90 % orange cells in the producer sample. The detected wt pigmentation phenotype (red cells) in the producer sample were either not fully induced yet (in 48 h induced cultures) or already reverted to a non-producing cells (in long-term photobioreactor cultivations), emphasizing the sensitivity and resolution of the method.
The fluorescence microscopy method displays a useful technique for a rapid detection of non-producing single cells in an ethanol producing cell population.
KeywordsPCC6803 PCC7002 Genetic instability Ethanol producer 3D fluorescence scan Phycocyanin Absorption spectra Fluorescence microscopy Image analysis
There has been a revamped interest in algae and cyanobacteria biotechnology in recent years, mostly due to the possible applications for biofuel production [1-4]. This study focuses on engineered strains of the model organisms Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002, which synthesize ethanol from pyruvate through the introduction of pyruvate decarboxylase (PDC) from Zymomonas mobilis and additional alcohol dehydrogenase (ADH) from Synechocystis sp. 6803. Both genes, contained within a plasmid vector, lead to a branching of fixed carbon towards ethanol production.
Recently the problem of cellular heterogeneity in ethanol producing phototrophic cultures has been recognized and has driven the development of new protocols to study the subpopulations in a photobioreactor (PBR). Even in clonal populations single cells may differ in terms of their genetic composition, physiology and biochemistry . This might have important practical consequences for the productivity and genetic stability of ethanol production in PBRs, as for example it influences the longevity of ethanol production and affects decisions on scale-up and culture management strategies. Internal research at Algenol has shown the mechanisms of the genetic heterogeneity within the ethanologenic vector cassette of an ethanol producing culture to include point mutations, insertions/deletions, and the presence of mobile genetic elements such as transposons. Mostly these genetic instabilities appear in the PDC gene of the ethanologenic cassette and lead to a non-functional PDC expression and therefore a stop in ethanol production.
In ethanol producing cells, fixed carbon is mainly directed into ethanol, leading to a typical phenotype with reduced biomass production, and in case of PCC6803- and PCC7002-based cell lines to a down regulation of the accessory pigment phycocyanin . Changes in the pigmentation of producer cells could be confirmed on RNA and protein level, where a 4-fold reduction in cpcB, which encodes the phycocyanin beta subunit, was measured, leading to a severe reduction in the amounts of phycocyanin subunits . As a result of inactivation of the PDC due to the mentioned mutations, the carbon metabolism is switched back to wild type (wt) conditions and the cells recover to a wt pigmentation phenotype.
However in induced cultures the non-producing cells, identified as “revertants”, have a selective advantage in regard to their much faster growth over producing cells and quickly overgrow the ethanol producing subpopulation resulting in loss of productivity. Consequently, the more revertant cells are present in scale-up cultures the earlier a decline in productivity in the reactors can be observed. The quantitative knowledge of reversions allows for pre-emptive measures before loss in ethanol productivity caused by an increasing population of reverted cells becomes crucial.
Today, absorption spectra are used to get an insight into the amount of reverted cells within a culture. Since the phycocyanin content is reduced in ethanol producing cells, an increase of phycocyanin absorption indicates the occurrence of reverting cells. However, when changes become visible within the absorption spectrum, a large amount of reverted cells is already present in the culture, thus resulting in a fast decline of the productivity. Measurements of the ethanol production rate similarly only detect problems when a large number of revertants has already spread through the population. Furthermore, loss in ethanol production can be also a consequence of other limitations, e.g. nutrient limitations, viability and productivity of cells, contaminations, etc.. Therefore the development of a quick and reliable method for quality control of scale-up and ethanol producing cultures that allows the early determination of revertant cells is of high importance.
In this paper we present a simple method for the distinction of the different phenotypes of ethanol producing and “revertant” cells via fluorescence microscopy. The approach is based on a previously developed fluorescence microscopic method  which has since also been adapted for high throughput application . It allows a simple and quick viability analysis for cyanobacteria single cells or viability in a mix. The method uses the red fluorescence of chlorophyll to distinguish vital cells from dead cells, which show an unspecific green fluorescence. Through the presented adaptations and further developments of this fluorescence-based cell viability assay, we have developed a protocol to differentiate between the distinct pigmentation phenotypes of producing and non-producing cells of Synechococcus and Synechocystis.
Synechocystis sp. PCC6803 - Wild type
Synechocystis sp. PCC6803 - ethanologenic strain #309 including plasmid construct pVZ325-PpetJ-PDC-synADH according to  and
Synechococcus sp. PCC7002 - Wild type
Synechococcus sp. PCC7002 - ethanologenic strain TK115 including plasmid construct pGEM-AQ4::smtB-PsmtA7002-PDC-PrbcL6803-synADH_deg according to 
Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002 wt strain and ethanol producer were grown in 200 ml bubbled reactors at 200 μmol photons*m−2*s−1 continuous illumination (fluorescent bulbs) and aerated with 0.5% CO2-enriched air for five days.
For Synechocystis 100 ml of autoclaved seawater BG-11 medium (63 g/l Instant Ocean®, 17.65 mM NaNO3, 0.18 mM K2HPO4, 0.03 mM Citric acid, 0.003 mM EDTA (disodium magnesium), 0.19 mM Na2CO3, 0.03 mM Ferric ammonium citrate and trace metals: 2.86 mg/l H3BO3,1.81 mg/l MnCl2-4H2O, 0.22 mg/l ZnSO4-7H2O, 0.39 mg/l Na2MoO4-2H2O, 0.08 mg/l CuSO4-5H2O, 0.05 mg/l Co(NO3)2-6H2O) were used with the addition of 5 × CuSO4-5H2O (1.6 μM Cu2+) for repression, and with 100 mg/ml gentamycin for the ethanologenic culture #309.
For Synechococcus the culturing medium was 100 ml autoclaved seawater BG-11 without Zn2+, with Vitamin B12 (0.004 mg/l), and for the ethanologenic culture TK115 with 200 mg/ml kanamycin.
Cultivation for ethanol production
For inoculation of the cultivation experiments the pre-cultures were centrifuged (4500 g, 10 min) and resuspended in induction media.
For Synechocystis sp. PCC6803 autoclaved seawater BG-11 without Cu2+ was used, since the promoter is induced by copper depletion. For the ethanologenic culture #309 100 mg/ml gentamycin were added to the media.
Synechococcus sp. PCC7002 was inoculated in autoclaved seawater BG-11 with 5 μM Zn2+ (in this case zinc induces the promoter) and Vitamin B12 (0.004 mg/L) and for the ethanologenic culture TK115 with the addition of kanamycin 200 mg/ml.
In order to exclude an impact of zinc addition and copper depletion the wt cultures were grown in the same induction media.
The cultures were cultivated in 250 ml bubbled reactors illuminated from one side with 100 μmol photons*m−2*s−1 in a temperature controlled chamber allowing a temperature profile with heating to a peak temperature of 38°C for 2 h and cooling to 25°C for 2 h per day. The cultures were aerated with a controlled air flow of 15 ml/min and 5% CO2 supply for day and night using a gas mass flow controller. Biomass accumulation (OD750nm), ethanol production and absorption spectra were measured daily.
Absorption spectroscopy and OD750 measurement
The spectrum of each culture was measured using the UV-Spectrophotometer UV-1800 (Shimadzu). The spectra of the cultures were recorded between 400 nm and 750 nm. The values were normalized to a relative absorption of 0.45 at the chlorophyll peak at 680 nm to allow a good overview of the ratio between chlorophyll and phycocyanin.
Three dimensional excitation/emission fluorescence spectroscopy
All cultures were washed with fresh media to eliminate signals from excreted substances and set to an OD750nm of 0.5. All cultures were exited in a range between 350 and 580 nm with a step size of 2 nm. The emission spectrum was recorded in the range of 450–750 nm for every excitation wavelength.
Sample preparation and fluorescence microscopy
In order to obtain a comprehensive data set of each culture the fluorescence of at least 500–1000 cells was monitored. For this 1 ml of the cultures from the different cultivation systems was transferred to a 1.5 ml reaction tube (OD750nm 1–2) and left with the cap open until microscopy. 30 μl of the cell suspension were transferred on a glass slide and the sample covered with a 24 mm × 24 mm cover glass. The suspension was fixed between the two glasses by light pressure to avoid movement of the cells but without destroying them. The edges of the covers were sealed using nail polish to avoid dry-out of the sample.
For microscopy the automated Olympus CX21 microscope with a 40× Plan Achromat objective was used. For each sample fluorescence images of 20 different positions that were predefined in the automated scanning tool of the Olympus Cell software, were recorded with an exposure time of 800 ms for every sample.
ImageJ Plugin for automated cell differentiation
The automated differentiation of the cell types was done with an ImageJ plugin that was based on previous works . Irregularities in the illumination of the microscope were corrected with the calculator plus function of ImageJ and an image without sample. To segment all cells from the background the fluorescent images were converted in an RGB-image stack. Afterwards, automated thresholding with the MaxEntropy method was used separately for both, the red and green channel. Both images were combined and the cells were registered using the particle analyzer function of ImageJ. Registered particles smaller than 10 pixel and particles touching the edge of the image were excluded from the analysis to eliminate the influence of artifacts. To enable a classification for each registered particle a histogram of the hue values within the particles was recorded and normalized. Additionally the mean particle brightness was used.
For the differentiation of the three cell types in cultures of Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002 in each case a neural network was trained separately with an independently created set of training data. For this, a simple feed forward network with one hidden layer based on the Encog java framework  was added to the plugin. Training was done with the combination of the resilient propagation algorithm and a genetic algorithm (when the change in the error rate was smaller than 1%) to an error rate of 0%. The trained neural network was used to differentiate the new data into the three classes wt cell without change in the phenotype (red signal), producer cell with changed phenotype (orange signal) and dead cell without photo pigments (green signal).
For the mixing experiment wild type and ethanol producing samples from Synechocystis sp. PCC6803 and Synechococcus sp. PCC7002 were each mixed in different ratios.
The Synechocystis sp. PCC6803 cultures were obtained after 48 hours of cultivation in induction medium as described under section “Cultivation for ethanol production”. The samples were adjusted to a cell density of OD750nm = 1 and mixed in the following percentages of wild type / producer: 0% / 100%, 10% / 90%, 25% / 75%, 50% / 50% and 0% / 100%. For each sample four replicates were measured.
The Synechococcus sp. PCC7002 cultures were from running 0.5 l photobioreactor cultivations in induction medium. The samples were adjusted to a cell density of OD750nm = 1 and mixed in the following percentages of wild type / producer: 0% / 100%, 5% / 95%, 10% / 90%, 25% / 75%, 50% / 50% and 0% / 100%. For each sample three replicates were measured.
Results and discussion
A prerequisite for the discrimination of non-producer and producer cells
Identification of optimal filter set for fluorescence microscopy
Set up of semi-automated fluorescence microscopy
Cultivation experiments of wt and ethanol producing strains to generate pigmentation phenotype
Mixing studies for a proof of concept
Identification of optimal filter set for fluorescence microscopy
Compared to the wild type both producer strains showed clearly reduced fluorescence signals in the range between 620 and 670 nm, highly likely caused by the reduced PC content in the cells. Additionally an unspecific green signal could be observed for all cultures at around 530 nm. The gray marked area in the 2D emission spectrum (500 – 680 nm, Figure 3) was chosen as the emission wavelength of interest to optimally observe changes in the fluorescence emission caused by a changed ratio between the unspecific green signal and the phycocyanin signal. The optimal filter set thus had an excitation wavelength of 388 nm and an emission of 510 nm long pass. A 680 nm long pass filter (Chroma Technology Corp and Semrock, USA) was used as a cut-off.
Automated fluorescence microscopy
Cultivation experiments of wt and ethanol producing strains to generate pigmentation phenotype and confirmation of the microscopic method
Biomass and ethanol productivity of wt and ethanologenic cultures of PCC6803 and PCC7002
Pigmentation and microscopical analysis of wt and ethanol producing cultures of PCC6803 and PCC7002
Mixing study as proof of concept
As a validation of the method we prepared mixed samples of wt and ethanol producing cells of PCC6803 and PCC7002. For PCC6803, the wt and ethanol producing samples were derived from induced cultures after 48 hours of cultivation in bubbled reactors. In case of Synechococcus sp. PCC7002 the samples derived from running lab-scale cultivation experiments in 0.5 l photobioreactors. Both showed ethanol production and the previously described difference in the pigmentation phenotype. Before mixing, both cultures were adjusted to a cell density of OD750nm = 1 and then mixed to aimed percentages of wt cells (see Methods). As reference pure samples of wt and ethanol producing cells were measured as well.
The cells in the wt strains that were misidentified as producers due to their orange emission, exhibited a changed phenotype not associated to the ethanol production. Probably these were cells in the process of dying and thus also showed a reduced content of photo pigments. With the method used, it is not possible to distinguish between these different causes for the changed phenotype. However, this is not a drawback for its practical application, as the focus of the method is the detection healthy revertant (wt) cells within a producer culture. Only these cells have the potential to overgrow the rest of the population due to their increased growth rate.
In case of the ethanol producing culture of PCC6803 and PCC7002, 76.3% and 81.0% cells were recognized as ethanol producing cells, but a portion of 17.7% and 7.7% still exhibited the pigmentation phenotype of wt cells (Figure 8). An explanation of wt cell detection in the ethanol producing population could be the presence of non-producing cells which either were not fully induced yet or exhibited a mutation in the PDC gene cassette resulting in loss of ethanol productivity. In both cases the PC / Chl ratio would be identical to that of wt cells. Here has to be noted that it is hardly possible to maintain an “ideal” 100% producer culture since it is never possible to induce all cells. Additionally the reversion of the cells can occur in any state of the cultivation process.
As a consequence of dead cells and the not 100% pure start samples the results of the mixing experiments of wt and ethanol producing cells were expected to deviate from the calculated optimal percentages. Therefore, based on the results for the pure samples, the expected percentages in the mixed cultures were adapted and showed a good correlation to the results that were obtained in the experiment (Figure 8).
Autofluorescence displays a useful parameter to investigate phototrophic single cells, for example for cell viability , but it also represents a very sensitive parameter in regards to changes in chlorophyll and phycobillisome content upon stress. In the present method we used the difference in the pigmentation phenotype to distinguish between non-producing and ethanol producing cells in a PBR on a single cell level.
The reduction of phycocyanin in ethanol producing cells of Synechocystis PCC6803 and Synechococcus PCC7002 could be evidently attributed to the cell’s response on ethanol production and not on ethanol itself. In both ethanol producing strains PC reduction has been observed on both levels of gene expression (transcript level and protein level). Those reductions resulted in different absorption spectra of ethanol producing cells versus non-producing wt and revertant cells and consequently also in different emission spectra which were visualized by using a specific filter set for fluorescence microscopy. Compared to the previously used method of measuring an absorption spectrum of the whole culture, the newly established analysis with the microscopical assay enables conclusions on the cell level, thus allowing a more precise and earlier detection of revertant cells. In addition the method allowed for the detection of dead or damaged cells, which show an unspecific green fluorescence. We performed cultivation experiments to confirm the applicability of this method as a simple and reliable quality control check for ethanol producing cultures of Synechocystis PCC6803 and Synechococcus PCC7002.
The assay might be applicable to other cyanobacterial expression systems as long as the cell pigmentation is affected, which is often expected as the redirection of nutrient flow causes starvation like conditions within the cell. Additionally the use of this new method within a FACS could enable a fast sorting of cultures before inoculation of production reactors and thus enable higher production rates.
We thank Michaela Kitschke, Katja Herrera-Glomm and Nadine Jurinke for their skillful support in the laboratory. We are grateful to Dan Kramer for providing cultures and laboratory facilities and Ulrich M. Tillich for helpful discussions on the research and reviewing of the manuscript.
- Mascarelli AL. Gold rush for algae. Nature. 2009;461:460–1.View ArticlePubMedGoogle Scholar
- Savage PE, Hestekin JA. A perspective on algae, the environment, and energy. Environ Progr Sustain Energ. 2013;32:877–83.View ArticleGoogle Scholar
- Haag AL. Algae bloom again. Nature. 2007;447:520–1.View ArticlePubMedGoogle Scholar
- Waltz E. Biotech’s green gold? Nat Biotechnol. 2009;27:15–8.View ArticlePubMedGoogle Scholar
- Brehm-Stecher BF, Johnson EA. Single cell microbiology: tools, technologies, and applications. Microbiol Mol Biol Rev. 2004;68(3):538.View ArticlePubMed CentralPubMedGoogle Scholar
- Dienst D, Georg J, Abts T, Jakorew L, Kuchmina E, Börner T, et al. Transcriptomic response to prolonged ethanol production in the cyanobacterium Synechocystis sp. PCC6803. Biotechnol Biofuels. 2014;7:21.View ArticlePubMed CentralPubMedGoogle Scholar
- Schulze K, Lopez DA, Tillich UM, Frohme M. A simple viability analysis for unicellular cyanobacteria using a new autofluorescence assay, automated microscopy, and ImageJ. BMC Biotechnol. 2011;11:118.View ArticlePubMed CentralPubMedGoogle Scholar
- Tillich UM, Wolter N, Schulze K, Kramer D, Brödel O, Frohme M. High-throughput cultivation and screening platform for unicellular phototrophs. BMC Microbiol. 2014;14:239.View ArticlePubMed CentralPubMedGoogle Scholar
- Dühring U, Friedrich A, Baier K, Enke H, Kramer D. Metabolically enhanced cyanobacterium with sequentially inducible production genes for the production of a first chemical compound. 2013. WO Patent App. PCT/EP2012/076,786.Google Scholar
- Heaton J. Programming neural networks with Encog 2 in Java. St. Louis: Heaton Research Inc; 2010.Google Scholar
- Rasband WS. ImageJ. Bethesda, Maryland, USA: U. S. National Institutes of Health; 1997–2014. [http://imagej.nih.gov/ij/]
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.