Measurement

Abstract

In order to achieve defined protein levels in a cell, promoters of defined strength are an obvious requirement. Such promoters can only be valuable in synthetic biology if they are well characterized. For future eucaryotic devices that require [http://www.partsregistry.org/PoPS PoPS (Polymerase per second)] as an input, our promoters will be very suitable since they deliver PoPS as an output. PoPs is the standard unit of synthetic biology, but it is very difficult to measure directly. For bacteria, relative measurements (relative promoter units, RPU) are most commonly used and it has been shown how to convert them to PoPs [1]. We identified several challenges to achieve the same in mammalian cells, suggest solutions and provide easy-to use relative measures for application in mammalian cells - one based on RNA levels (Relative Mammalian Promoter Units, RMPU), the other based on folded protein levels (Relative Expression Units, REU). We apply those measurements to the characterization on CMV, an existing promoter from the registry. Finally, we discuss how to transform these units into PoPS.

Introduction

The need for standardized measurements of promoter activity in vivo has been widely accepted across the synthetic biology community [1]. Only if a part is well characterized initially, function of an engineered device or system can be predicted reliably. Most work of synthetic biology has focused on bacteria, especially Escherichia Coli, as a model system. Novel tasks in synthetic biology, especially for medical applications, will require synthetic biology of mammalian cells. Mammalian systems are the most complex biological systems, and therefore, little work has been done in the field of mammalian synthetic biology, leaving a huge potential for future research.

Box 1: Calculation of PoPs, where γM is the mRNA degradation rate, a is the GFP maturation rate, γi is the degradation rate of immature GFP, ρ is the tranlation rate of immature GFP from mRNA and n is the number of copies of promoter per cell. [1]

For mammalian cells, no such unit has been reported, nor is there a kit for measuring promoter activity. Mammalian cells have higher levels of complexity - the process from a gene to a functional protein involves transcription (regulated by the promoter itself and chromatin structure) , RNA modification, RNA splicing, RNA export, RNA degradation, protein folding and protein degradation [5]. All of these processes are highly regulated, both generally and specifically for the individual gene.

A frequently used system is the use of a dual luciferase assay. We dismissed the idea of using luciferase as it is an invasive technique - it requires killing the cells before measuring [2]. This makes time-course experiments impossible. On the other hand, GFP can be detected non-invasively. Its expression can be measured with a wide variety of methods. As outlined below, simple fluorescence measurements by fluorimeters cannot be the method of choice for measuring promoter activity in mammalian systems.

For the quantification of reporter gene expression in every cell of the transfected population we used a dual assay with GFP and mCherry. Our cells are co-transfected with a reference plasmid, in our case a plasmid containing our reference promoter JeT in front of mCherry. The fluorescence intensities were measured by flow cytometry and fluorescence microscopy with ImageJ analysis.

On the other hand, we attempted to measure the promoter strength through quantification of gene expression on mRNA level. Hereby we applied the novel technique real-time RT-PCR, which has been developed tremendously during the last 20 years. The method consists of 2 steps, a reverse transcription, which synthesize cDNA from extracted mRNA, and a real-time PCR, which is based on normal PCR, but collects the data throughout the PCR process, thus in real-time. Signal detection is achieved using fluorescent dye, which can either be added to mastermix (SYBR Green approach), or be labeled to probe. The accumulation of PCR products over cycles results in the increase of fluorescence. The cycle number where the fluorescence passes a certain threshold is known as Ct value. This values correlates directly to initial amount of mRNA presented after RNA extraction. Through comparing of Ct values detected by different sample, the difference in their expression profile can therefore be illustrated. In our approach to measure promoter strength using real-time RT-PCP, promoters of interest were cloned into same backbone containing eGFP as reporter gene. Through quantification of GFP expression driven by the test promoters, the difference in promoter strength can be therefore easily illustrated. This approach measures directly on mRNA level, which means, translation and mature of the GFP molecule don't have to been taken into account as in the FACS approach.

Results

Identification of challenges unique to higher eukaryotes

We identified the following challenges which complicate promoter characterization in eukaryotes and detail how we proceeded so as to avoid these problems:
Variance of transfection efficiencies.

Figure 1. Transfection efficiency for vectors containing different inserts varies. By Flow cytometry, we identified transfected cells and found transfection efficiencies to vary

Commonly used are dual assay measurements using co-transfection as a control for transfection efficiency [2]. Such assays can be used as an application to relative characterizations, but does not provide accurate normalization. In order to resolve that issue, we recommend a method which allows for relative characterizations but by itself is able to distinguish between transfected and untransfected cells (flow cytometry or image analysis). For experiments aiming at absolute numbers, stable integration of the measurement plasmid into the mammalian genome containing exactly one integration site is required. Then, copy number per cell is one .

Regulation of promoter activity by chromatin structure.

Considering stable integration of a promoter into the mammalian genome, another level of complexity is added: Chromatin structure affects transcription, though mostly not affected by the promoter[10]. Therefore, integration at different sites of the genome will not result in the same expression strength.

A cell line having a FRT site stably integrated into the genome is required. Such a cell line involves the recombination of sequences between short Flippase Recognition Target (FRT) sites by the Flippase recombination enzyme (FLP or Flp) derived from the 2µ plasmid of the baker's yeast Saccharomyces cerevisiae.[3]. We started the development of such a cell line.

RNA processing.

Many RNAs are spliced after transcription. Also, RNAs are processed before being exported from the nucleus, and can be retained in the nucleus both dependent and independent of processing [5], [6]. Therefore, several mRNA populations exist: Unprocessed, Processed and retained in the nucleus, and functional mRNA outside of the nucleus. These effects are a current research focus of molecular cell biologists and barely understood even in model organisms as simple as yeast.

Splicing does not present a problem for reporter gene constructs, as these are cDNAs which are not spliced. Nevertheless, this means that the ratios between RNA levels and protein levels will change for a single promoter if a different output gene is linked to the promoter. We therefore suggest to assume a black box between RNA and protein, and the introduction of two independent units.

Variety of cell lines

TOP10 or DH5α cells have been widely accepted as chassis systems by synthetic biologists working in bacteria. For the work with mammalian cells, no such consensus exists; also, it would not be sensible to limit synthetic mammalian biology to a small number of cell lines, as every cell line is suited for a special applications. Scientists working on breast cancer virotherapy by synthetic promoters would choose a breast cancer cell line, whereas scientist working on Stem cell therapies to myocardial infarction would choose a cardiomyocyte cell line. Cell lines differ greatly, even in expression strength of constitutive promoters (shown below).

Figure 2: CMV and JeT strength changes depending on conditions We characterize GFP expression from [http://partsregistry.org/wiki/index.php?title=Part:BBa_K203112 JeT] and [http://partsregistry.org/Part:BBa_I712004 CMV] fluctuate dependent on condition (Erolimus induces extreme starvation). Measured by Flow cytometry 20 hours after transfection (unless specified otherwise).

Lack of truly constitutive promoters in mammalian cells

Fig. 2 shows measurements of GFP expression from CMV und varying conditions (relative to GFP expression from JeT under the same condition). This result demonstrates that every promoter in mammalian cells underlies regulation, and therefore, is not truly constitutive. We analyzed the Sequence of CMV by [http://www.gene-regulation.com/pub/databases.html TRANSFAC Professional] and found it to contain two NF-κB binding sites, two CREB-binding sites, and single Ap1, RFX1 and SRF binding sites. Of NF-κB, CREB and Ap1, we know (compare Synthetic promoters) that they have a high constitutive activtiy, but neveretheless, they underlie regulation. For example, NF-κB is induced by inflammation conditions, whereas CREB is activated by the second messenger cAMP[12] and thus responds to many hormones, starvation conditions etc. This hardens comparison of promoters in different conditions. We discuss how this can be achieved.

A promoter measurement kit for use in mammalian systems

We separated JeT's core promoter from its proximal promoter by a HindIII site; it can therefore be used for screening of synthetic proximal promoters or for modifying the strength of a promoter by core promoter swapping. In addition, it contains a FRT site which will allow for stable integration into mammalian cells also containing a FRT site. Thus, it provides the possibility to characterize the promoter in a defined genome and in this way helps to avoid some of the challenges outlined above. For the same reason, it also contains a mammalian selection marker (hygromycine). For the generation of the plasmid, please see our Notebook. As a reporter gene, it contains GFP, which is followed by a SV40 mammalian terminator. We generated another plasmid pSMB_REFERENCE, which contains mcherry instead of GFP. It can be used for normalizations to transfection efficiency in Flow cytometry and image analysis.

Two units for promoter activity in mammalian cells

We define two units: 1 relative mammalian promoter unit (RMPU) is defined in analogy to[1], but mRNA-based. It is the amount of total mRNA (that is, processed and unprocessed mRNA inside and outside of the nucleus) generated by a promoter x, relative to the amount of total mRNA generated by the JeT promoter j ([http://partsregistry.org/wiki/index.php?title=Part:BBa_K203112 BBa_K203112])[7] in steady state. It is directly proportional to PoPs, as we show. In steady state, change of mRNA levels is 0, thus:

Where M' is the change in total mRNA level, γM is the mRNA degradation rate and n is the number of promoters per cell (adapted from [1]. Therefore

Since both promoters are cloned seperatly into the same backbone, they generate identical mRNA. As the most important determinant for mRNA stability is the 3' untranslated region, and mRNA stability is generally not affected by promoter structure [11], we are able to assume the same mRNA degradation rate γM for both promoters.

We also expect the same n for both promoters as this is normalized for during the measurements as shown below. We then define RMPU, and achieve a cancellation of terms:

On the other hand, we define Relative Expression Units (REU). 1 relative expression unit (REU) is the amount of total folded protein generated by a promoter, relative to the amount of folded protein generated by the JeT promoter ([http://partsregistry.org/wiki/index.php?title=Part:BBa_K203112 BBa_K203112]) under the same cellular condition. It is not directly proportional to PoPS as too many levels of regulation, such as post-transcriptional modifications, enhanced splicing and nuclear shuttling/transport lie between transcription and protein [5], and REUs would strongly depend on conditions which affect RNA. Still, it is a very useful measure, as for most applications such as metabolic pathway engineering, protein level, not mRNA levels, are of importance. Also, it can be measured with a much easier experimental setup than RMPU. See discussion for ideas on how to compare REU measurements taken at different conditions.

Measuring RMPU by qRT-PCR

To measure the RMPU (Relative Mammalian Promoter Units), HeLa cells were transfected with plasmid containing the promoter of interest. HeLa Cells transfected with plasmid containing JeT promoter served as the reference. The transcription efficiency was checked using FACS. After 20 h and 50 h expression, total RNA was isolated, followed with real-time RT-PCR, where the mRNA amount of GFP was quantified. To get comparable results of different samples, we used multiple housekeeping genes as internal control, while using non-transfected HeLa mRNA as plate-to-plate correction. The program was performed with ABI StepOnePlus real-time PCR system. Data was collected from the associated software. GFP expression was normalized against housekeeping genes using own written MatLab-script. The cmv promoter was about 2.89 times stronger than JeT after 20 h and 2 times stronger after 50 h.

Measuring REU by flow cytometry and image analysis

The GFP reporter expression is examined by two-color flow cytometry as well as fluorescence microscopy with subsequent image analysis. To account for potential differences in transfection efficiency, we have used the fluorescent protein mCherry coupled to the reference promoter JeT, which is co-transfected together with GFP coupled to the promoter of interest. By comparing transfection/expression levels of mCherry, we could exclude samples with low cell numbers or transfection efficiencies. The relative fluorescence of the CMV promoter coupled to GFP was measured in three different cell lines: HeLa, MCF-7, and U2-OS. As one would expect, the Relative Expression Units (REU) vary slightly between the different cell lines. Overall the CMV promoter is very strong relative to our reference promoter JeT with value of 5.52, 6.76 and 9.73 in HeLa, MCF-7 and U2-OS respectively (Figure 4).

Different core promoters result in different expression strength

We cloned the CMV core promoter in front of the JeT proximal promoter to obtain JeT_CMV. We characterized this construct in three different cell lines, HeLa, MCF-7, and U2-OS and found it to have 0.5-0.6 percent of JeT's activtiy depending on the cell line (Fig. 5). The fact that variations in the core promoter can be used to vary expression strength of a certain promoter of interest comes in useful if the transfer function of an existing promoter is to be altered, and it can be used to further diversify the synthetic promoters we created.

Fig. 4: Flow cytometry and microscopy measurement data of CMV (REU) in different cell lines. The three cell lines MCF-7, U2-OS, and HeLa were cotransfected with the CMV promoter coupled to GFP and a reference plasmid including the promoter JeT coupled to mCherry. The relative fluorescence (REU) of GFP is measured 20 hours after transfection. The standard error of the mean (SEM) is indicated by the error bars.

Figure 5: Flow cytometry measurement data of JeT_CMV (REU) in different cell lines. The three cell lines MCF-7, U2-OS, and HeLa were cotransfected with the JeT_CMV promoter coupled to GFP and a reference plasmid including the promoter JeT coupled to mCherry. The relative fluorescence (REU) of GFP is measured 20 hours after transfection. The standard error of the mean (SEM) is indicated by the error bars.

A stable cell line for promoter measurement

Coming soon...

Discussion

Flow Cytometry/Fluorescence Microscopy

Our flow cytometer measurements of CMV and JeT-CMV in different cell lines (Fig.3 and 4) show that the strength of these promoters varies in different cell lines. Also, the ratio between the two promoters varies slightly, but overall CMV is much stronger than JeT, while JeT-CMV is consistently lower than JeT. Our measurements were reproducible considering the strength ratios between our standard promoters. The standard error of the mean shows minimal fluctuations in the measurements. These can be due to several factors: the experimental conditions, different persons performing transfection and cell culture, the exact time schedule of the experiment has to be kept equal in each experiment, which may sometimes not be possible and cells may behave differently when they reach a large passage number. Our microscopy measurements were used to support our data obtained by flow cytometry. Out initial microscopy measurements show similar results, the fluctuations can be explained by the different preparation of the cells. In flow cytometry we measure live cells while the cells have to be fixed and imaged the next day in microscopy. Also we were only able to perform one set of measurements with microscopy, because microscopy is very time consuming. Apart from having a validation of our results, fluorescence microscopy is going to gain importance in the future where we plan to construct a stable cell line which expresses fluorescent proteins with different colors in various compartments of the cell according to which promoter is activted. Read more about our future plans in the outlook section.

qRT-PCR

The results of the real-time RT-PCR measurement show that the promoter strength varied significantly over time. This could be due to the fact, that we used transient transfection, e.g. the plasmid containing promoter can't be replicated while the cells are proliferating. This could possibly result in lost of plasmid, thus reduced copy of promoter presented in the cell. It could also be possible that the cells containing the promoter construct kept dying due to changes in cell physiology. ??? discuss the variance 12 replicates?

Characterization of promoters under different condtions

Considering what we learned about the lack of absolute, non-changing reference standards, characterizing promoters under different conditions becomes inherently difficult. All parameters affecting REU and RPMU are altered by a change in conditions. For the assumptions on which REU and RMPU are based to remain valid, promoters must only be compared in one set of conditions. In most scientific work to this date, transcriptional activity is assessed by comparing protein levels between unrelated conditions (see[8] for a typical example). Although giving sufficient information as to whether a promoter is up- or downregulated, this standard cannot be used for an absolute characterization, as if a condition induces, for example, a system-wide increase in translation, it will result in a ratio between the promoters which is not proportional to the change in transcription. If, on the other hand, measurements are related to a reference promoter at each condition individually, this does not represent true levels of change if the reference promoter itself is up- or downregulated by that condition. Providing absolute measurements to compare between different conditions will be a major challenge for synthetic mammalian biology in the next years. We suggest, for now, always to give promoter strength relative to the strength of JeT under the condition regarded.

Towards PoPs

The concept of RMPU can easily be converted into PoPs, at least for constitutive promoters. RNA degradation equals RNA synthesis in steady state. Therefore we suggest to block RNA synthesis by applying Actinomycine D, which works in minutes (Karsten Weis, personal correspondence), and then measuring RNA degradation. According to equation 2, PoPS will then equal mRNA degradation, divided by copy number (which is 1, if the experiment is conducted in a stable cell line). Of course, Actinomycine is toxic to cells and might affect mRNA turnover rates system-wide. Accounting for this effect will present a major challenge.

Concluding remarks

We reported two new units for promoter characterization in mammalian cells, one of which being directly proportional to PoPS. We showed the advantages and limitations of the concept. Future work will have to focus on imprving comparisons between different conditions.

Method details

Flow cytometry

Quelle für die gesmte Beschreibung des Flowcytometers

Flow Cytometry is a commonly used method for the measurement of fluorescence intensity levels. It uses the principles of light scattering, light excitation, and emission of fluorochrome molecules to generate specific multi-parameter data from particles and cells in the size range of 0.5 µm to 40 µm diameter . One unique feature of flow cytometry is that it measures fluorescence per cell or particle. In contrast to high-resolution microscopy, flow cytometric measurements do not reveal intracellular distribution of the measured signal, but give an integrated signal of each cell. However, flow cytometry allows fast, multi-parametric measurements of high numbers of cells (>1000 cells/sample) in a quantitative manner. A flow cytometer consists of the fluidics system where cells are hydro-dynamically focused in a sheath of fluid; one or several laser which is the source of focused light for fluorescence and scatter; the optics which gather and direct the light emitted from the cells; up to 8 detectors which receive the light and convert it into an electrical signal, which is finally delivered to a computer system to analyze the signals of each cell. The first detector (Forward Scatter; FSC) is in line with the light beam; its signal is proportional to the size of the cell. The larger the cell, the more light is scattered and the higher the electric signal in the detector. Perpendicular to the light beam are several other detectors, including the Side Scatter (SSC). The Side Scatter represents the granularity of the cell and this can be used to further distinguish different cell populations and exclude apoptotic cells. A two-dimensional scatter plot of Forward versus Side Scatter can give a more detailed overview over the nature of the cells in the sample. This plot is used to separate dead cells and cell debris from healthy cells. Following the Side Scatter, the light is directed through a series of mirrors and filters, so that particular wavelengths are delivered to the correct detector. The fluorescence signal is then converted to a voltage signal and can be visualized as a histogramm, showing frequencies of cells with different fluorescence levels (usually distributed over 1024 channels). We have used a Beckman Coulter FL500 MPL, equipped with a 488 nm and 561 nm laser, thereby allowing simultaneous and quantitative measurement of GFP and mCherry fluorescence. Cells were prepared in 96-well format with 10^4 cells/well and transfected with the promoter of interest and if necessary induction drug was added for inducible promoters. Before measurement, the medium was removed, the cells were washed with 1xPBS and trysinized with 60μl of Trypsin per well. After 10 minutes incubation at 37°C 1xPBS + 1% BSA were added up to a volume of 200 μl per well. When starting the flow cytometry measurement we first adjusted the gates and the background signal to the negative control (Fig. ?). Figure blablablu shows an exapmle of a gated positive control (JeT), where we can see a second normal distribution peak indicating the gated GFP positive cells.

Figure : Analysis of flow cytometry measurement concerning a negative control in HeLa. The number of events is plotted against the fluorescence (log) of GFP. This figure shows a negative control in the cell line HeLa, where the background signal is set under 10 and the area under the curve is colored in green.

Figure : Analysis of flow cytometry measurement concerning a positive control (JeT) in HeLa. The number of events is plotted against the fluorescence (log) of GFP. This figure shows a a positive control of the reference promoter JeT in HeLa, where the area under the curve is colored in green.

Microscopy and Image Analysis

For further characterization of our promoters we decided to use fluorescence microscopy. Our cells are co-transfected with a reference plasmid, which contains JeT in front of mCherry. Analysis of reporter gene expression can be restricted to cells that express the reference plasmid and are therefore likely to have been successfully co-transfected with the reporter plasmid. This is particularly useful when transfection efficiency is low. We applied this approach to microscopy measurement to minimize the experimental error. The promoter strength was evaluated in relative expression units (REU) relative to the JeT promoter which was set to 1. Microscopy images of our samples fixed with 4% formaldehyde were taken using the Nikon Eclipse 90i upright automated widefield microscope in the Nikon Imaging Center at the University of Heidelberg. Each image was taken in the GFP and mCherry channel and the exposure time had to be adjusted for over- or underexposed images. The conversion factor was experimentally determined by imaging a fluorescent plate with exposure time varying from 10 to 70 ms (the estimated range of our promoters) and plotting the mean grey values of these images versus time. A linear relationship between the exposure times and grey values was observed for this time frame and the conversion factor was determined to be 0.998. For image analysis we used the ImageJ (Image Processing and Analysis in Java) software measuring the mean grey values of each cell containing the promoter of interest. ImageJ is an open source software developed at the National Institutes of Health (NIH (http://rsbweb.nih.gov/ij/). The measurement is performed by setting the boundaries around the cells using the threshold in the mCherry image and generating a mask that is then applied to the image of the GFP channel in the following steps:

• Open image and select split channels
• Select mCHerry image (all images should be 8-bit)
• Go to Image> Adjust> Threshold
• Set lower bar to maximum and adjust upper bar so that the cells are still visible
• Remove unwanted spots such as dead cell using the drawing tools
• Apply the threshold (image should appear in all white now with black spots indicating the cells)
• Go to Process> Math> Divide and divide the image by 255 (image should be all white now)
• Next go to Process> Image Calculator and multiply the GFP channel with the mCherry channel pictures.
• Now select the result image and adjust the threshold so that the lower bar is set to the maximum (255) and the upper bar is set to 1 (only GFP containing cells should be visible through the mask)
• Go to Analyze> Set Measurement and make sure the measurement is limited to the threshold and the result image of the multiplication is selected.
• Go to Analyze> Measure.
• Mean grey values should show up in the 'Results' window

Alternatively the images can be analyzed using the ROI manager where the cell have to be selected by hand using the drawing tools and added to the ROI list. In comparison the threshold method is much quicker and the results are comparable. We preselect the mCherry positive cells which are very likely to contain GFP. The mean grey values of 5-10 images per sample were averaged and adjusted to the same exposure time. The results were given in REU relative to JeT, which was always prepared as a reference sample for each measurement.

Real-time RT-PCR

Real-time RT PCR is a specific and sensitive technique for detecting and quantification of gene expression on RNA level. In order to obtain reliable results and reduce the variance during PCR procedure, we did the measurement for each promoter with 12 replicates. Prior to PCR, we extracted the total RNA using Qiagen RNAesy mini Kit (????reference???). The extraction was performed using the QiaCube robot to avoid possible mistake and ??variance??. We quantified the expression of GFP reporter gene together with other 5 housekeeping genes for normalization. In order to perform the PCR, primers and probes were designed according to guidelines of Qiagen (??Ref??): when possible, The target PCR product spans over exon-exon boundary and has an intron in between to exclude amplification of genome DNA which may be presented in the extraction (????improvement needed??). The primers have fluorophor and Queching at 5 3 end, repectively. During amplification, probes bind specifically to the target gene. The polymerase hydrolyzes the probe during elongation due to its exonuclease activity. The signal of the released fluorophor can then be detected and correlates therefore to the amount of PCR product. ?how to set Ct? ?how data normalized?

References

[1] Kelly, JR et al in Journal of Biological Engineering 3 (2009): "Measuring the activity of BioBrick promoters using an in vivo reference standard"
[2] http://www.promega.com/tbs/tm058/tm058.html
[3] Zhu, XD and Sadowski, PD in J Biol Chem 270 (1995), p. 23044-23054: "Cleavage-dependent Ligation by the FLP Recombinase".
[4] Ducrest, A-L et al. in Nucleic Acids Res. 30 (2002), p. e65: "Detection of promoter activity by flow cytometric analysis of GFP reporter expression"
[5] Alberts, B. et al: „Molecular Biology of the Cell“ (Book. 5th edition, 2008. Garland Science) Chapter 6
[6] York JD et al in Science 285 (1999), p. 96-100: "A phospholipase C-dependent inositol polyphosphate kinase pathway required for efficient messenger RNA export."
[7] Tornoe, J. in Gene 297 (2002), p. 21-32: „Generation of a synthetic mammalian promoter library by modification of sequences spacing transcription factor binding sites.“
[8] Rushton, P.J. et al. in Plant Cell 14 (2002), p. 749–762 „Synthetic plant promoters containing defined regulatory elements provide novel insights into pathogen- and wound-induced signaling“
[9] Endy, D. et al. in Nature 438 (2005), p. 449-453: "Adventures in synthetic biology" (comic). Available online at [http://www.nature.com/nature/comics/syntheticbiologycomic/index.html nature.com]
[10] Alberts, B. et al: „Molecular Biology of the Cell“ (Book. 5th edition, 2008. Garland Science) Chapter 7, p. 467-476
[11] Alberts, B. et al: „Molecular Biology of the Cell“ (Book. 5th edition, 2008. Garland Science) Chapter 7, p. 492
[12] Alberts, B. et al: „Molecular Biology of the Cell“ (Book. 5th edition, 2008. Garland Science) Chapter 7, p. 905-908