Evaluation of a unique, newly upgraded flow cytometer
Sara Engström
Degree project in biology, Master of science (2 years), 2009 Examensarbete i biologi 30 hp till masterexamen, 2009
Biology Education Centre and Department of Molecular Evolution, Uppsala University
Table of content
S
UMMARY3
1 I
NTRODUCTION4
1.1 To measure particles 4
1.2 Aims 5
2 R
ESULTS6
2.1 Initial flow cytometry tests 6
2.2 In depth flow cytometry tests 8
2.3 Alignment of the instrument 11
2.4 Limitations of the instrument 11
2.5 Apogee Histogram Software 1.98 evaluation 12 2.6 Examining the content of the samples 14
2.6.1 Transmission electron microscopy 14
2.6.2 Fluorescence microscopy 15
3 D
ISCUSSION18
3.1 Instrument set up and settings 18
3.2 Quality of samples 19
3.3 Staining characteristics and procedures 20
3.4 Future prospects 20
4 M
ATERIALS ANDM
ETHODS22
4.1 Strains 22
4.2 Flow cytometer 24
4.3 Flow cytometry sample preparation 24
4.3.1 Storage 24
4.3.2 Diluting the samples 25
4.3.3 Fixation 25
4.3.4 Freezing 25
4.3.5 Chromogen preparation and incubation 25
4.4 Transmission electron microscopy 26
4.5 Fluorescence microscopy 26
4.5.1 Counting cells 27
A
CKNOWLEDGEMENTS28
R
EFERENCES29
Summary
In this report an upgraded version of the A 40-MiniFCM flow cytom eter has been tested to determine if it is sensitive enough to detect viruses, either by light scatter or by fluorescent labelling. The instrument has been tested with plastic calibrati on beads, cells and viruses of various kinds, and the samples have been microscopically analysed, both by electron and light microscopy.
By these m ethods it has been re vealed that it is highly unlikely that the instrum ent can detect even large viruses. The fact that the in strument did not detect the presence of cells or viruses in some samples is with all likelihood due to both the sensitivity of the flow cytometer and im purities in the sa mples. If samples are to be exam ined by the f low cytometer it is important that other particles of similar size are as few as possible. Also, the presence of cells in one of the primary tests has been confirmed, which implies that some of the tests performed by the manufacturer might be misleading.
Further, these tests have led to the introduction of a new chromogen to the lab, which will affect the future work of the group. An interesting genom e distribution of Pyrobaculum calidifontis has been observed, as well as fluores cent staining of bacteriophage T4 genom e.
Finally the software of the instrument has been evaluated, allowing for future improvements.
1 Introduction
1.1 To measure particles
It is all in the details. All sorts of life, multicellular as well as unicellular, are built up by small compartments. If one desires to study cells, one needs to use highly sensitive equipment. If one extends beyond cells, and pursue informa tion about viruses, one need even finer instruments.
Flow cytometry is an ef fective method for c ounting cells and other m icroscopic particles.
Usually a c ell is sta ined with a f luorescent dye. The cells pass throug h a laser beam , which excites the dye, and causes it to em it light at a certain wave length, w hich is registered by a detector. Each stained cell gives rise to one signal (flash of light) that is proportional to the amount of dye that has attached to the cell. Each cell that passe s through the laser beam also scatters lig ht. The intensity of the sca ttered l ight i s de pendent on t he s ize, for m a nd transparency of the cell. The scatte red ligh t is registered by a separate detector (Shapiro 1995).
In this specific instrument the sample is pushed through a narrow tube together with sheath fluid (milliQ water), and the sam ple becomes a thin pillar in the centr e of the sheath f luid, thereby only allowing one partic le at the time to pass th rough the laser beam . Separate detectors register light that scatters off the particle, as well as light emitted by the excited dye, as seen in figure 1.
Figure 1. Detection of particles in the flow cytometer. The sample (indicated by light blue arrow) is surrounded by sheath fluid (indicated by dark blue arrows) and is pushed through a tube. The pillar of sample becomes so thin than the particles pass through the laser beam one at the time. When the laser beam meets a particle, it excites the dye that the particle is stained with.
The dye will then emit light that can be registered by detectors FL1 and FL2. The light from the laser is also scattered of the surface of the object, and can be detected by detectors LS1 and LS2.
In the flow cytometer used in this study it is possible to measure the scattered light and the fluorescent signal at the same time. Thus, the cell cycle characteristics of a number of archaea have been revealed by co mparing cell size with the DNA content of the cell (Lundgren et al.
2008).
The data are visualised by a graph that presents the num ber of counted particles on the Y-
axis and th e light inte nsity of the partic les o n the x-axis. The resolution of the x-axis is
dependent on the num ber of channels that the axis is divided into. The m ore channels, the
more variations in data can be visualised. If a low resolution is chosen, e.g. one tenth of the
maximum resolution available, the mean value of ten channels is calculated, and depicted into
one channel. In the graph the detector used to collect the data is stated above the channel numbers.
In all flow cytometry measurements there will be some levels of noise. There are two types of background noise. The first orig inates from the m achine itself, and the second originates from the sam ple, which is neve r entirely pure. One aim when trying to optim ise the instrument construction is to reduce the noise fr om the machine. The second type of noise is minimized when preparing the sam ple, by m aking sure th at it con tains the p articles which should be measured, and as few other things as possible.
By using a function in the so ftware called gating, one can c hoose to specify a num ber of channels in e.g. light scatter, and show the fluorescence only from t he particles that are included in the gating, i.e. partic les in the chosen light scatter channels. This function allows for filtering away noise that is only present in either light scatter or fluorescence, in a cer tain region of interest. Another function in the software , thresholds, allows the user to collect data only from particles that exceeds a given intensity in light scatter or fluorescence.
The instrument has two solid state lasers ( 405 nm and 488 nm ), two light scatter detectors (LS1 and LS2) that are located at different angles co mpared to the sam ple, and two fluorescence detectors (FL1 and FL2) that have different filters that allows them to registe r light of certain wave lengths.
The upgrade was built on a new light scatter configuration unique to this instrument; hence the instrument tested in this evaluation is the only exam ple of t his prototype. The reconfiguration included a new, narrower angl e on light scatter detector LS2. A new laser (488 nm) was also included, enabling the use of additional fluorescent dyes. Even though the fluorescence sensitivity of the in strument was not upgraded, it is of great interest since a signal in th e fluores cent chann els can be u sed to find a hidden signal in the ligh t sc atter channels via gating or threshold methods.
The initial tests perform ed by the manufacturer of the m achine were done on calibration beads, a sample of the archaeal SSV4 virus, from the same source as the SSV4 sample used in this study, and a sam ple of hum an pox virus. The preliminary results indicated that the flow cytometer was able to detect and cou nt the content of these samples. Detection at this level is well beyond the standard instruments, and therefore a promising result.
1.2 Aims
In this report I have evaluated a newly upgrad ed flow cytom eter, not yet on the market, to establish whether or not the detectio n limit is enhanced, and if so, how much? There were a number of questions that would be addressed if the flow cytometer were able to de tect virus- sized particles. Two interesting topics conc erning virus and light scatter were: how s mall particles could the instrum ent detect? W hich type of m orphology, if any, rendered a m ore intense light scatter? The following properties concerning fluorescence were also of interest:
how s mall genom es could the instrum ent dete ct? W hich dye and staining procedure were most efficient? Were different dyes optim al for different viruses? In cells, th e fluorescence intensity is fairly proportional to the genome size; is this true also for viruses?
A number of calibration beads, cells and viruses were used to evaluate this instrum ent, as
well as a number of different dyes and staining procedures. The dyes were all chosen to stain
DNA and R NA. To be able to interpret the results of the flow cytom eter measurements, it is
important to dete rmine what the sa mples conta in. Thus, an extensive pa rt of th is p aper is
devoted to procedures used to examine the actual content of the samples.
2 Results
2.1 Initial flow cytometry tests
Fluorescently labelled beads of 0.1 µm, 0.75 µm and 1.5 µm in diameter (Polysciences) were used as positive controls. The flow cytometer could detect all of these beads distinctly using both light scatter and fluorescence detectors, even though there was a loss of quality for the smallest beads (figure 2). As negative controls, buffers with the adequate dye concentration were used.
Initial staining tests of cells and viruses were conducted with fi ve dyes: SYBR Green, SYBR Gol d, SYTO Green, acridine orange, and a m ix between ethidium brom ide and mithramycin A. The mix is the standard dye used in the group before the upgrade.
Figure 2. Calibration beads. A number of beads of different size were measured by the flow cytometer. Signal intensity is indicated on the x-axis and is measured in channels. The number of channels in use is indicated by black numbers to the right of the x-axis. If all the channels in use are currently not displayed, the number of displayed channels appears above. Number of counts is indicated on the y-axis. Low intensity noise appears at the left end of the x-axis. A) 1.5 µm diameter beads, registered by the LS2 detector. The large peak represents single beads and the smaller peak to the right is probably beads which have clustered together. B) 0.75 µm diameter beads, registered by the LS2 detector. C) 0.11 µm diameter beads, registered by the LS2 detector. The signals detected from the beads partially overlaps the noise signal. D) 0.75 µm beads registered by the FL1 detector.
These data are from the same measurement as the data presented in figure 2B, but represents fluorescence instead of light scatter.
The archaeon Sulfolobus acidocaldarius was used as a standard organism throughout the evaluation, since the group has perform ed pr evious work on this organism , and its characteristics and appearance in the flow cytom eter and light m icroscopy are well known.
Sulfolobus acidocaldarius was detected by both light scatter and fluorescence, giving rise to well-defined peaks in light sc atter detectors LS1 and LS2. W hen s tained with ethidium bromide and m ithramycin A, S. acidocaldarius gave well-defined peaks in the FL2 detector (spectrum above 575 nm), but a less defined peak with the “green dyes” (SYBR Green, SYBR Gold, SYTO Green and acridine orange) which were registered by the FL1 detector (517.5 - 552.5 nm spectrum), as seen in figure 3A-C. Of the green dyes, SYTO Green rendered the highest resolution.
The viruses tested in this evaluation were c hosen in order to represent different types of genomes, genome sizes and m orphology. Moreover, viruses that in fect bacteria were chosen as well as those infecting archaea or eukaryotes. Thus, viruses a ssociated to all three domains of lif e wer e repr esented. During the initia l te sts of the viral sam ples, only th e sam ple containing SSV4 gave rise to a well-defined si gnal. The others sam ples all resulted in noisy signals that were dif ficult to interpr et. The noi se in the s amples covered more chann els than the background noise in the negative control. H ence, the samples contai ned particles of some sort which were counted, even though they lacked unity.
Besides the viral samples, Nanoarchaeum equitans was used in this study. It is the smallest extracellular organism known, thus its light scatte ring abilities are thought to be low. It also contains an extremely small genome, due to genome reduction. N. equitans lives as a parasite on larger archaeal host cells and thus has lost genes encod ing function s that it no longer performs itself (Waters et al. 2003). The samples that were used in this paper were grown on the host Ignicoccus ho spitalis, which has a larger volum e than both N. equitans and S.
acidocaldarius, but also has an unusually sm all ge nome for a free-living cell (Podar et al.
2008). If N. equitans can be stained and detected by the flow cytom eter, it opens up the possibility to examine the life cycle of that or ganism, in the sam e way as have been done for other archaea (Lundgren et al. 2008). Since nearly nothing is known about the cell cycle of N.
equitans, this would provide new information about a unique organism. Including N. equitans and I. hospitalis also provides a range of genom e sizes in between the ones of viruses and S.
acidocaldarius, improving the chances of pinpointing th e detection level of the instrum ent.
However, neither N. equitans nor I. hospitalis gave a well-defined peak in the light scatter or fluorescence detectors.
Since these initial tests showed such poor results, several different approaches were tried
with th e in tention to lo cate the pe aks, or e xplain their ab sence. In o rder to optim ize the
detection of the green dyes (SYBR Green, SY BR Gold, SYTO Green an d acridine orange) a
new filter for the FL1 detector was obtained and tested. The new f ilter collects a wider range
of the spectrum (507 – 583 nm ). This filter clea rly enhanced the signal for all the green dyes
when tested on S. acidocaldarius (figure 3D). As with the old filter, out of the green dyes, the
highest resolution was obtained with SYTO Green.
Figure 3. Flow cytometry measurements of S. acidocaldarius. The light scatter of S.
acidocaldarius was measured as well as genome distribution. The two connected peaks seen in the FL2 channels (figure 3B-D) represent two populations in the sample; cells with one (the left peak) or two (the right peak) copies of the genome. Cells that are currently dividing render a signal in between the two peaks. A) Light scatter registered by the LS1 detector. B) DNA stained with ethidium bromide and mithramycin A, registered by the FL2 detector (above 575 nm). C) DNA stained with SYTO Green, registered by the FL1 detector (narrow filter, 517.5-552.5 nm). D) DNA stained with SYTO Green, registered by the FL1 detector (wide filter, 507-583 nm).
2.2 In depth flow cytometry tests
Further tests were carried out by varying different param eters in the protocols, in order to enhance staining. A number of viral sam ples we re fixed in a final concentration of 0.5%
glutaraldehyde (G7651 Sigma-Aldrich), as recommended by Bru ssaard (2003). In agreem ent with Brussaard’s research I exp erimented with freezing the viral samples in liquid nitrogen before staining. I also varied the temperature fo r incubation of viruses with the fluorescent dye between room temperature and 80ºC. These procedures were tested separately as well as in com bination with each other, but did not result in an enhanced signal as far as the instrument could measure.
When dealing with N. equitans and its host cell, the sam ples were thawed in either ethanol or salt solution, to examine whic h solution preferably allow ed the cells to stay intact. W hile ethanol fixes the sam ple, the salt solution m ight be gentler to the c ells, e specially to I.
hospitalis f or which the optim al NaCl level is 2% (Huber et al . 2000), thereby preventing them from breaking due to osmotic pressure.
The signal in the SSV4 sam ple was detected only through light scatter (figure 4A-B). This
signal was present in all SSV4 m easurements, no matter which dye or stai ning protocol were
used. Gating from the light scatter peak gave no well-defined signal in the fluorescence channels, i.e. m ost of the part icles counted in th is peak em itted no, or low levels of excited light. The position of the peak lay near the background noise and repres ented scattered light of lower intensity than that of S. acidocaldarius.
The other viruses (AFV1, HHPV-1, HRPV-1, SIRV2, MS2, T4, CfMV and HaPV) gave rise to dif ferent va riants of noisy hills and plateaus, conf irming that the sam ples were not empty. However, no homogenous population of any sort could be detected. Figure 4C-D is a representative example of what most of the virus measurements looked like.
Of the viruses in this study T4 is the one with the largest genom e and therefore the one most likely to appear in the fluorescence measurements, and fairly likely to appear in the light scatter m easurements. Therefore a num ber of additional tests were perform ed on the T4 sample. Among them, T4 was run in different concentrations together with a constant number of 0.75 µm diameter beads, as an internal control. The beads rendered the same signal in each measurement, but only noise was detected from the T4 phages. However the noise levels were higher, the more concentrated the samples were, indicating the presence of particles.
In one m easurement of bacteriophage T4, a p eak appeared in the light scatter channels, when they were se t in logar ithmic mode (f igure 4E-F). However, th is result cou ld not b e repeated, which should be taken into consider ation when assessing the reliability of this measurement. Also it is im portant to be cautious of the logarithm ic mode, since it p resents data in a way that make the counts look more like a peak. Examples are shown in figure 4.
In addition to the ab ove, the two archaea Sulfolobus solfata ricus and Pyrobaculum calidifontis were also analysed by fl ow cytometry in order to generate m ore data from different genome sizes and cell m orphologies. For exam ple P. calidifontis has a genom e of 1800 kbp (Lundgren et al. 2008) and is rod-shaped, which should lead to different light scattering properties co mpared to s pherical cells, such as th e other arch aea examined in this evaluation.
Something interesting w as discovered when examining the fluorescence of Pyrobaculum calidifontis cells. In exponential phase P. ca lidifontis cells either lump together, or has divided the genom e addition al times, without splitting the cells, since evenly distributed, decreasing peaks are seen (figure 5). The signa l was stronger than the genom e size indicated, assuming that the relationship between signa l strength and genom e size is linear. This relationship, as well as param eters that can affect the linearity between signal and genom e size, are presented in section 2.4 Lim itations of the instrum ent. The fact that there are exceptions from the linearity should be rem embered when viewing th e calibration curve in this evaluation.
.
Figure 4. Linear versus logarithmic scale. Three samples have been plotted on linear as well as logarithmic scale on the x-axis. The coloured fields represent the same data. A) SSV4 sample plotted on a linear scale. B) SSV4 sample plotted on a logarithmic scale. C) T4 phage plotted on a linear scale. D) T4 phage plotted on a logarithmic scale. E) AFV1 virus plotted on a linear scale. F) AFV1 virus plotted on a logarithmic scale.
Figure 5. P. calidifontis multiple peaks. A sample of P. calidifontis was stained with ethidium bromide and mithramycin A, and measured by flow cytometry. A) DNA content registered by FL2 detector. The multiple peaks indicate the presence of multiple genomes. B) Light scatter registered by LS1 detector. The light scatter pattern reveals a group of particles with similar light scattering properties and a distribution of particles with gradually increased light scattering properties. These can be interpreted as larger cells, lumps of several cells, or cells which has scattered light from a different angle than the others. P. calidifontis are rod-shaped cells, which make the angle with which the particle passes the laser beam of greater importance than if the cells would have been coccoid.
2.3 Alignment of the instrument
To determine if the lack of peaks was due to lo w quality of the samples or limitations in the instrument, a num ber of tests were perform ed. First the a lignment of the instru ment was tested. In ev ery series of samples measured, beads of the sam e size were run as the first and last s ample, and their c oefficient of varia tion (cv) exam ined. This was standard procedure throughout the evaluation. If the cv of these two runs is increa sing over tim e, it is a first indication that alignm ent is failing. The beads, however, ke pt their cv throughout the experiments. A new set of m easurements were performed with beads of different sizes and compared with the beads run directly after the upgrading. Th ey were sim ilar; hence the alignment did not seem to have declined.
The flow cytometer was then exam ined physically by m aking sure that the particle stream in the flow cell was centred in relation to the laser beam. All the above tests indicated that the instrument had kept the alignment.
2.4 Limitations of the instrument
If nothing is awry with the sam ples, then th e sensitivity of the instrum ent m ight not be sufficient. In order to establish whether or not certain genomes are likel y to be distinguished from the n oise in th e fluorescen t channe ls, a calibration curve was m ade using S.
acidocaldarius and S. solfataricus.
The assumption behind this experiment was that for cells, the relationship between genome
size and amount of dye binding to the genome is linear. Hence the intensity recorded for a cell
should be linearly related to the genome size of the cell. GC content and packing of a genome
can to som e extent affect how m uch dye binds to the genome (Lundgren et al. 2008), but in
general the relationship is fairly linear. It is not known whethe r or not this holds true for
viruses and sm all genomes, but it should give an indication of wher e the genom e might end
up. The two cell types were stained with th e standard m ix of ethidium brom ide a nd
mithramycin A, and exam ined by the flow cytom eter. From these two sam ples four points of
intensity were measured, the intensity of one and two genome copies for each of the two cell types. A linear fit was calculated and, using this equation, values for I. hospitalis, N. equitans, T4 and SSV4 were extrapolated (figure 6).
Figure 6. Extrapolated intensity of small genomes in the FL2 channels. By measuring the intensity of S. acidocaldarius and S. solfataricus single and double genomes, the expected intensity of smaller genomes can be calculated. The measurements were done at 32768 resolution with a mix of ethidium bromide and mithramycin A. The extrapolated values placed SSV4 at channel 83, T4 at channel 102, N. equitans at channel 142 and I. hospitalis at channel 243. According to this linear fit the genome size of 0 bp cuts the x-axis at 81 channels, thereby making it impossible to draw conclusions about smaller genomes from this equation.
At these settings, Tris-EDTA buffer (TE buffer) (10 m M Tris-HCl, 1 mM EDTA, pH 7.9) with ethidium bromide and mithramycin A always blocks the 30 first ch annels, and often up to 55 channels. Running sheath fluid without any chromogen as a sample blocked the first 20- 25 channels, which should be considered th e m inimum background noise that is always present. Thus, it is theoreti cally possible to see signals fr om all extrapo lated genom es, however a peak lying near noise will be diffi cult to distinguish. Also, a genom e size of 0 kb renders 81 channels, which is very close to the SSV4 value.
The background increased significantly when a biological sam ple was run. In the measurements of S. acidocaldarius, at the settings u sed for the calibration cu rve, the background was to som e extent present in the 2 50 first channels, while it was present in the 200 first channels in the P. calidifontis measurements at the same settings.
2.5 Apogee Histogram Software 1.98 evaluation
The purpose of the software is to v isualize the da ta and to allow th e user to adjust different
settings that affect both how the instrum ent handles the incom ing signals, and how the data
are p resented. In gen eral the so ftware perform ed well. There were, however, a number of
points where the software connected to this specific prototype may be improved.
The peaks had a tendency to slide sideways on the x-axis during, and in between the runs, i.e. the sof tware gave the im pression tha t the intensity of the signal was changing.
Comparisons between the first and last run of control beads confirm ed this. This problem might be due to other things than the software, as for example the amount of dye used in each individual measurement. There might also be a mechanical reason for the sliding of the peaks.
The program has frozen, or shut down a nu mber of tim es during a run, for no apparent reason.
At some settings gaps and overlaps in data appeared. Often the gap was accom panied by the overlap so that a cut in the graph was followed by a p eak of similar area nearby. The incoming signal is processed differently deepen ing on if the intensity is low or high. At channel 3500 (of 65535 channels) a switch in the processing is made, creating a gap which the instrument is programm ed to elim inate (personal com munication Oliver John Kenyon).
Sometimes the compensation ended up rather far from the gap, as seen in figure 7A. One way of avoiding a gap or peak in th e area of interes t for a specifi c measurement, was to set th e switch for high or low gain to a nother region of the graph. Th e compensating peak could also be moved towards the gap by ente ring values in the service tab of the software, tuning the high and low gain calibration. If possible, one m ight look into alternative ways to handle the switching process, in order to completely eliminate this problem.
There were often counts behind the threshold lin e, even though that area should be empty, since the function of the threshold is to exclude all counts lower than the value entered for the threshold (f igure 7B). This pr oblem is due to digital to an alog an d analog to digital imperfections in the instrum ent and should, according to the m anufacturer, not cause any measurement errors.
Extra digital peaks app eared from time to tim e in the g raph. Usually they appeared when using threshold values f or light scatter and fluo rescent channels at the s ame time, and when threshold values were low. For ex ample a peak could arise at a certain poin t w hen only including particles that gave a strong signal in both light s catter and fluorescen t channels (figure 7C). However when looking at all data that appeared in eith er light scatter or fluorescence, there was no peak at this point.
When zooming in or out in the window, only the X scale bar (not the Y scale bar) changed, but the diagram itself changed in both X and Y direction. This made it difficult to determ ine the height o f a peak when it was zoom ed out (figur e 7D). This effect is due to the fact that groups of channels are merged and replaced by mean values in order to fit many channels into few channels when zoom ing out. One should be aw are that extrem e values m ight be lost in this process.
Some error messages concerning the fluid tank containing the sheat h fluid, occurred during the evaluation. T he errors caused the flow cytom eter to malfunction so that it was impossible to run samples in it. This error wa s however corrected, by the technical support of Oliver John Kenyon. The error em erged from a number of settings that had not been changed from the old settings to the new, upgraded ones, and are therefore not likely to cause problems in the future versions of the instrument.
During the last weeks of the evaluation the level of the cl eaning fluid tank had decreased
abnormally fast. No apparent site of leakage was identified, and the problem is at this time not
resolved.
Figure 7. Examples of software imperfections. A) A gap appears in channel 3500 (of 65535 channels), indicated by the black arrow. The compensating overlap is not centred above the gap, and therefore gives rise to a peak next to the gap, indicated by the red arrow. The sample consists of N.
equitans and I. hospitalis. B) Counts that are below the threshold value should not be visualised in the graph, nevertheless counts appear to the left of the dotted threshold line. In this example the threshold is set to FL2 = 726. The sample consists of P. calidifontis. C) Extra digital peaks are seen at some settings, usually when triggering for both light scatter and fluorescence. In this example thresholds are set to LS2 = 26 and FL2 = 5. The sample consists of SSV4, and should have the same appearance as the data presented in figure 4B. D) The same data as in figure 7B is presented, but zoomed out, showing all channels in use. In 7B it seems as if the tip of the largest peak is at 263 counts, while in 7D it seems as if the tip of the same peak is at 184 counts, since the Y scale bare is not changing, but the diagram is changing in the Y-direction.
2.6 Examining the content of the samples
2.6.1 Transmission electron microscopy
In order to examine the content of the SSV4, T4 and SIRV2 sa mples, they were prepared for transmission electron microscopy (TEM).
When prepared by the standard protocol a few T4 viruses were found (figure 8A), along with a number of particles that wer e similar to the heads of T4 pha ges concerning size and shape. These particles might be pieces of broken viruses (figur e 8B). The T4 sample seemed to contain less salt and debris compared to the SSV4 sample.
SSV4 prepared acco rding to the standard p rotocol as well as the prolonged pro tocol
showed something that appeared to be salt or protein aggregations. No cells or viruses could
be identified, since these formations covered most of the sample (figure 8C-D).
Several different concentrations of the sam ples were tested , but there were no noticeab le differences in appearance.
Figure 8. Transmission electron microscopy of T4 and SSV4 samples. A) T4 phage from sample prepared according to standard protocol. B) T4 phage-like particle from sample prepared according to standard protocol. C) SSV4 sample prepared according to standard protocol. The spots are aggregates that may be seen in all shapes and sizes throughout the sample. D) SSV4 sample prepared according to prolonged protocol. The small dark spots might be SSV4 virions. When examined at a larger zoom they became too diffuse to permit any conclusion to be drawn from them.
The light spherical structures are probably salt or protein aggregates, since they appeared in a large number of sizes all over the sample.
2.6.2 Fluorescence microscopy
To examine the quality of the viral sam ples and the N. equitans samples they were stained with 4'-6-diamidino-2-phenylindole (DAPI) and examined by light m icroscopy. In addition, the most promising samples (SSV4 since it produced a signal in the flow cytometer, and T4 as well as purified N. equitans, since they have comparatively large genomes) were stained with the green dyes, used in the flow cytometer, to enable direct comparison between the methods.
In several cases a sam ple was divided into two aliquots, one for the flow cytom eter and one for the m icroscope, to ensure that the same sa mple was tested in both m ethods, and thereby enabling a just comparison. As a positive control S. acidocaldarius was used, and as a negative control an amount of chromogen, equal to the one in the sam ple, was diluted in TE buffer.
The presence of N. equitans was confirmed, even though m any cells had attached to each other and there was a high level of debris, probably pieces of N. equitans and its host. N.
equitans was visib le in all dyes and appeared b oth as i ndividual s pots a nd a s c lusters wi th
each other and debris in the sam ple. Som e la rger and fainter spots ap peared which m ight represent the host. The larger spots were not seen in the purified samples.
Staining with SYTO Green gave th e brightest spots, even though all dyes emitted a strong light. Differences in intensity m ight be due to concentration of the dyes, and is therefore not to be taken as proof that one of the dyes bind better to the genomic content than the others.
Since N. equitans and I. hospitalis were grown at high salt concentrations it is possible that running them through a shell of milliQ water in the flow cytometer caused them to rupture. To examine this possibility a sample of N. equitans and a sample of N. equitans and I. hospitalis were diluted in sterile filter ed water instead of TE buffer and exam ined by light m icroscopy.
Purified N. equitan s, nonpurified N. equitans (i. e. I. hospitalis not rem oved) and S.
acidocaldarius diluted in TE, as well as a sa mple of S. acidocaldarius diluted in sterile filtered water acted as controls.
Cells were visible in all three samples when diluted in TE buffer, but in none of them when diluted in water. Since S. acidocaldarius did not stay intact in this test, but was present when measured in the flow cytometer, this test did not reveal whether or not N. equitans or its host stayed intact in the instrum ent. What the tes t showed was that the env ironment in the flow cytometer is not equivalent to sterile filtered water, since S. acidocaldarius endured the flow cytometer long enough to be detected. It should al so be taken into acc ount that a sam ple is only in contact with the sheath fluid for a short time in the instrument, while the cells viewed in the microscope were diluted in water for a couple of minutes.
In the T4 sam ple a number of bright spot s, probably the genom e of the phages, were clearly visible in the microscope (figure 9C).
In the SIRV2 sample vague blue spots were seen, which are probably genomes too small to emit an intense signal (figure 9D).
The AFV1 sample contained extremely vague spots which could alm ost not be discerned.
They m ight be stained genom es, but they were too dim for any definite conclusion to be drawn.
In the SSV4 sample (figure 9E-H) a number of spots were seen with much higher intensity than that of the T4 phages, but slightly lower than the one from S. acidocaldarius (figure 9A- B). When viewed in phase contrast, spherical structures were seen at the places of these spots.
In their appearance they were smaller, but otherwise identical to S. acidocaldarius cells, and therefore probable to be rem aining host cells of SSV4. The experim ent was repeated and the cells appeared each time, proving that they were present in the stock solution of SSV4 and not a contamination of a single sam ple during dilution. These cells were no t present in any other dilution of virus or the TE buffer control.
The other viral samples showed no more fluorescence than the negative control.
Impurities can be present in the sam ples, in the buf fer, on the slide and in the chromogen.
To get r id of the ones o n the s lides, the s lides were washed in 96% ethanol. The buffer was sterile filtered to m inimise contaminants. Dilution series were m ade on the chrom ogen and those viral samples which seemed to contain fl uorescent particles (T4, S SV4 and SIRV2), in order to trac k which par ticles originated from which source. The dilution series of the viral samples confirmed that the bright spots that appeared in the viral samples originated from the stock sample, since they were m ore numerous in more concentrated samples. These types of spots were not seen in any of the dilutions of the chromogen.
Another reason for the dilution series of chrom ogen was to determ ine the optim al
chromogen concentration as well as to monitor impurities. Series were made in TE buf fer as
well as in samples containing TE buffer and S. acidocaldarius. Too much chromogen in one
sample generated crystals of chromogen in TE buffer, while too little chrom ogen m ade it
difficult to detec t the presenc e of S. acidocaldarius . The optim al final dilution of the stock
solutions of the chromogens were DAPI 0.001 mg/ml, SYBR Green 0.5X, SYBR Gold 0.5X, SYTO Green 0.002 mM, acridine orange 10 mM.
Disadvantages of the green st ains was that stained part icles faded rapidly, m aking it difficult to capture photos of s mall fluorescent particles, which us ually dem anded a longer exposure time. The DAPI dye remained bright for a longer time, and was therefore preferred.
To determine if the peak in the SSV4 sam ple actually represented cells, the concentration of cells in the sam ple was estim ated manually by repeated counting in counting cham bers.
The cell concentration turned out to be about 7x10
4cells/µl, while the flow cytometry peak in the SSV4 sample contained the equivalent of 5x10
5cells/µl.
Figure 9. Light microscope images. S. acidocaldarius, SSV4, T4 and SIRV2 samples were stained with DAPI and examined through light microscopy. A) S. acidocaldarius phase contrast. B) S.
acidocaldarius stained with DAPI. C) T4 sample stained with DAPI. D) SIRV2 sample stained with DAPI. E) Cell and debris in SSV4 sample, phase contrast. F) Cell and debris in SSV4 sample, stained with DAPI. G) Cell in SSV4 sample, phase contrast. H) Cell in SSV4 sample, stained with DAPI. All photos were captured with 100x objective with oil and 2.5x optical zoom.
3 Discussion
The lack of a well-defined signal for m ost of the samples can be the symptom of two things:
either the flow cytom eter cannot detect the signal, or there is no signal to detect. If the fault lies within the instrum ent it can either be th at the upgrade is not se nsitive enough for these levels of light scatter, or that the filters in the f luorescent channels might not b e optimal for the dyes.
For the green dyes the s ensitivity was greatly improved by switching f rom a narrow f ilter (517.5 - 55 2.5 nm) to a wider one (507 – 583 nm), thereby collecting more of the em itted light. SYTO Green has an emission peak at 509 nm for DNA and 514 nm for RNA. This peak is ra ther clo se to th e sh ortest wavelengths th at are co llected by the wide f ilter. Hence, the sensitivity might be even m ore improved for S YTO Green with a f ilter collecting light down to e.g. 490 nm . FL2 has a filter which reco rds everything above 575 nm , and ethidium bromide has an em ission m aximum at 600 nm . Mo st of the light from that dye should be detected, but it is possible that also this filter may be improved.
If the sam ple does not produce a signal, the reasons can be m any. The staining m ight not be effective (the staining pro cedure is not optim al or the type of dye is not suitable). The shape and transparency m ight not allow enough light to scatter off the particle. The sam ple might also not contain what is expected. The virus abundance might be low or there m ight be other particles obscuring the signal.
The final lowest genome de tection level was 1800 kbp ( P. calidifontis) and the s mallest particle that was detected by light scatter was 0.11 µm beads. The sm allest cells detected by light scatter were S. acidocaldarius (800-1000 nm in diameter).
Both S. acidocaldarius and P. calidifontis peaks were fairly close to the noise peak when measuring fluorescence from their genomes. This means that a smaller genome probably ends up even closer to the noise. Even if the viru ses in a sample renders a more intense signal than the noise, it might be difficult to distinguish if the signal ends up too close to the noise.
3.1 Instrument set up and settings
Since fluorescently labelled particles could be se en in the light microscope in the SSV4, T4, N. equitans and SIRV2 samples, it is established that these samples contained something that could be stained with the dyes used in this evalua tion. Yet none of thes e samples showed a peak in the fluorescent detecto rs. Hence, ther e is som ething in the sam ple that was not registered by the instrument, or the signal was obscured by noise from the sample.
In the cas e of N. equitans there was so m uch debris and c ells that had attached to each other that it is not surprising that the signal m ight have been lost. In the T4 sa mple however, there were few other particles ap art from the bright spots, which are a ll of approximately the same strength. Some of them seemed stronger than the others in figure 9C. The reason for this is that the virions were on different levels on the slide.
If the flow cytom eter is sensit ive enough to detect th e intensity of these spots, then there
should be a peak in the fluorescent channels. This indicates that at least the T4 phage genomes
are too sm all to be detected. If the genom es have not disintegrated, then the phage heads
corresponding to the genomes should also be present in the sa mple. Since nothing apart from
noise could be detected in the light scatter chan nels, it seem s that the light s cattering is not
enough to be detected by the instrum ent. One r eason could be that a large fraction of the
phages m ight have been dam aged and created a debris, that obscured the signal from the
whole phage heads. However, it is likely that th e heads of the phages are the parts that reflect
light best. Therefore it is possible at least in th eory that the heads them selves would give rise
to a peak of their own if they were numerous enough.
The f low cytom eter was able to d etect the ca libration be ads both via light sca tter and fluorescence. However, the bead s are solid, en tirely spherical, and are likely to s catter light better because of that. They are also specifi cally labe lled with f luorescent m arkers that contribute to a strong f luorescent signal. In ad dition to th is, there is a good chance that the samples containing beads are more pure than th e biological samples, since there are no cells present to create debris. However, fact rem ains that beads of 0.11 µm dia meter can be detected by the instrument. This is an improvement since the upgrade.
3.2 Quality of samples
To perform good measurements the samples measured must be of good or at least fairly good quality. Above all it is important to know what quality the sam ples have. The sa mples that were kindly provided for this evaluation are of different ages and have been treared differently.
One reason to why nothing is seen in m ost of the viral samples could be that they are not stable at 4°C. They have not been kept frozen by the ones providing the sam ple, and not by me. Even if most of the viruses are dam aged, there might be some of them left stable enough to infect cells, which is usually wha t is required of them. For the flow cytom etry however, it is important that most of the viruses are intact, otherwise it is difficult to find the actual peak among all background noise from broken particles.
Even though the high tem perature m ight have led to degradation of som e pa rticles, microscopy has provided proof that there were s till a number of intact particles in the T4 sample, probably intact viruses in the SIRV 2 sam ples, and cells in the SSV4 sam ple.
Therefore broken particles cannot be the only answer.
It is understandable that N. equitans was not seen as a sharp p eak in th e flow cytometry light scatter or fluorescent channels m easurements since the preparation s contained s o much debris and gatherings of cell m aterial. The T4 sample, however, seem to be fairly free from impurities. Since th e obvious signals from the T4 phages, seen in the microscope, were no t visible in the fluorescent channels in the fl ow cytom eter it seem s that the flu orescence sensitivity is not enough to detect viral genomes.
The cell concentration in the SSV 4 sam ple was estim ated to 7x10
4per µl by m anual counting in counting chamber. This is to be considered a fairly good correlation with the peak in the flow cytom etry measurements, since it is within a f actor of eight. In this p rocedure, effort was taken to count only cells, and not de bris or other im purities. However, the flow cytometer will count everything that gives a strong enough signal. This can be the reason why the value of the peak was higher than the m anual value. Also, it is harder to distinguish particles that completely lack f luorescence, when using the m icroscope. Therefore cells th at have been completely emptied of their genome are more difficult to find.
The cells in this s ample are prob ably Sulfolobus host ce lls that ha d n ot been pro perly removed, since it is highly unlikely that a cont amination that has e ntered the tube af ter purification would have reached these high con centrations. Since the S SV4 sample used in this evalu ation shares the sam e origin as the sam ple used by the m anufacturer during the initial tests, and they bo th have the sam e appearance in th e instrument, it is like ly that both samples are suffering from the same contaminants.
The measurement in th e flow cyto meter indicated that the particle s scattered a good deal
less light th an S. acidocaldarius cells. Since no fluorescence was detected it also seem s that
the fluorescence levels were lower in these cells than in S. acidocaldarius. This was also seen
in the microscope. Lower inten sity in light sc atter and fluorescence would m ake sense if the
particles are viruses. However, it cannot be rule d out that the cells in the sam ple may have
been infected, which might change the optical properties and genome content.
3.3 Staining characteristics and procedures
A number of dyes and staining pr ocedures were tested, since different cells and viruses might differ in th eir su sceptibility of th e dyes. So me dyes m ight, for example, not be able to penetrate the capsid or m embranes well enoug h. The fluorescent signals m ight also vary depending on reasons mentioned in section 2.4 Limitations of the instrument.
No change in intensity was observed betwee n the variants of protocols described in materials and methods, which were largely built on the suggestions of Brussaard (2003). This might be because this evaluation deals with o ther organisms than Brussaard. The only shared virus is the T4 phage. It is also possible that the signal actually was strengthened, but still was too weak for the upgraded flow cytometer to detect any difference.
When examining the samples in the microscope, there was a defini te correlation between intensity of signal and genom e size. S. acidocaldarius (2226 kbp) was the brightest one, followed by the spots in the SSV4 sam ple, T4 (172 kbp), SIRV2 (35.8 kbp) and AFV1 (20.9 kbp), in that order. The remaining viral samples (all with genomes less than 20 kbp) could not be detected via th e light m icroscope. Therefore it is highly improbable that they could be detected by the fluorescence detectors in the flow cytometer.
The interesting m easurements of P. calidifontis indicate th at the cu lture contained either cells with multiple genomes or cells that had attached to each other.
The light s catter peak indicate d a low number of larger cells. Whether or not t hese represented several cells lum ped together, s ingle cells with la rger genom es or single cells passing the laser beam in different angles, might be answered by m icroscopy. It is an interesting discovery and an opening for further tests.
It is also interesting that these multiple peaks have not been seen in the old configuration of the instrument (only a low level of noise was detected beyond the first two peaks. A lso it has not been possible to find P. calidifontis by light scatter alone, with the instrument before the upgrade. P revious flow cytom etry studie s had to include fluorescence (personal
communications Erik Karlsson). Since this is po ssible with the new upgrade it is obvious that the light scatter configuration has improved the sensitivity of the instrument.
The light scatter peak was adjacent to, and ev en partially overlapped the noise peak, when only triggered on light scatter. A better separation of noise and peak was accomplished when triggering on fluorescence, as in figure 5B. It is questionable if something as small as viruses can be detected only by light scatter when a rod-shaped cell with a length of 1.5-10 µm barely can be detected. If the genom e of a virus is t oo small to be detected by fluorescence, then the chances of distinguishing that virus by light scatter probable were low.
3.4 Future prospects
If it is desirable to continue this research, the following advice should be considered. Use viral preparations that have been highly purified, pr eferably with a genom e of at least 20 kb, or a spherical or spindle m orphology, since this sh ape probably scatter m ore light. Check the purity of the sam ples and the presence of nucleic ac ids within the cell or virion by fluorescence microscopy. Check the presence of virus by TEM.
Purely cultured I. hospitalis should be an ideal candidate fo r this project. Since it has a large volume, but a sm all genome, it should be so mewhat easier to detect via light scattering.
Also the genome size is in between that of S. acidocaldarius (which is easily detected) and T4 (which was not detected). The genom es of pure cultured I. hospitalis have been detected previously, before the upgrade (unpublished data). The samples used in this evaluation were not harvested and purified specif ically for flow cytometry, which might be one reason that I.
hospitalis was not detected. It is also possible that the p resence of N. equitans obscured the
signal from I. hospitalis. A pure culture, prepared for flow cytometry should be the next step.
If I. hospitalis genome labeling or light scatter can be detected, it would provide an additional value for the calibration curve, which should allow for even m ore adequate extrapolations of smaller genomes. From there the g enome and ce ll size can be decreased in order to find the smallest detectable one.
The upgrade has improved the instrument, but not enough to detect the viral samples which were used in this evaluation. W ith the current version it is possible to m easure a num ber of things that was unavailable before the upgrade, e.g. 0.11 µm diameter beads, P. calidifontis by light scatter, and a higher resolution of standard organism s in the lab such as S.
acidocaldarius.
4 Materials and Methods
4.1 Strains
All strains used in the evaluation are described in table 1. The morphology of all cells and viruses used in this study is illustrated in figure 10.
Figure 10. Morphology of cells and viruses. The morphology of each cell and virus examined in this evaluation is depicted in this figure. Please note that the illustrations are not to scale. A) Ignicoccus hospitalis. B) Nanoarchaeum equitans. C) Pyrobaculum calidifontis. D) Sulfolobus acidocaldarius. E) Sulfolobus solfataricus P2. F) Acidianus filamentous virus 1. G) Haloarcula hispanica pleomorphic virus 1. H) Halorubrum pleomorphic virus 1. I) Spindle-shaped virus 4. J) Sulfolobus islandicus rod- shaped virus 2. K) Bacteriophage MS2. L) Bacteriophage T4. M) Cocksfoot mottle virus. N) Hamster polyomavirus.
Reference Huber et al. 2002, Podar et al. 2008 Huber et al. 2002, Waters et al. 2003 Amo et al. 2002, Lundgren et al. 2008 Boone et al. 2001, Chen et al. 2005 Boone et al. 2001, She et al. 2001 Prangishvili & Garrett 2004, Prangishvili et al. 2006 Personal communication Elina Roine Pietilä et al. 2009 Prangishvili et al. 2006, Peng 2008 Prangishvili et al. 1999 Valegård et al. 1990 Leiman et al. 2003 Mäkinen et al. 1995 Scherneck et al. 2000
Size (nm) 2000 (D) 400 (D) 1500-10 000 x 500-1000 800-1000 (D) 800-2000 (D) 900x24 ~ 55 (D) ~ 44x55 80x45 900x23 27 (D) 240x85 30 (D) 40 (D)
Morphology Coccoid Coccoid Rod-shaped Coccoid Coccoid Filamentous Pleiomorphic Pleiomorphic, elongated Single-tailed spindles Rigid rods Icosahedral head-tail-fibers Isometric Icosahedral
Genome size 1298 kbp 491 kbp 1800 kbp 2226 kbp 2992 kbp 20.9 kbp 8 kbp 7048 nt 15.1 kbp 35.8 kbp 3569 nt 172 kbp 4082 nt 5.4 kbp
Genome Circular dsDNA Circular dsDNA Circular dsDNA Circular dsDNA Circular dsDNA Linear ds DNA Circular dsDNA Circular ssDNA Circular dsDNA Linear dsDNA ssRNA dsDNA ssRNA Circular dsDNA
Type Archaea Archaea Archaea Archaea Archaea Archaeal virus Archaeal virus Archaeal virus Archaeal virus Archaeal virus Bacteriophage Bacteriophage Eucaryote virus Eucaryote virus
Abbreviation I. hospitalis N. equitans P. calidifontis S. acidocaldarius S. solfataricus AFV 1 HHPV-1 HRPV-1 SSV4 SIRV2 MS2 T4 CfMV HaPV
Table 1. Strains Name Ignicoccus hospitalis Nanoarchaeum equitans Pyrobaculum calidifontis Sulfolobus acidocaldarius Sulfolobus solfataricus P2 Acidianus filamentous virus 1 Haloarcula hispanica pleomorphic virus 1 Halorubrum pleomorphic virus 1 Spindle-shaped virus 4 Sulfolobus islandicus rod-shaped virus 2 Bacteriophage MS2 Bacteriophage T4 Cocksfoot mottle virus Hamster polyomavirus (D) = Diameter S. acidocaldarius (DSM 639), S. solfataricus (DSM 1617) and P. calidifontis (VA1, provided by Haruyuki Atomi (Kyoto University)) have previously been cultured in the lab of the group. The AFV1, SSV4 and SIRV2 samples were kindly provided by David Prangishvili (Pasteur Institute). HHPV-1 and HRPV-1 samples were kindly provided by Elina Roine (University of Helsinki). N. equitans and I. hospitalis were kindly provided by Harald Huber (University of Regensburg). T4 phages were kindly provided by Karin Carlson (Uppsala University). MS2, CfMV and HaPV were kindly provided by Lars Liljas (Uppsala University).
