Team:Freiburg/Project/effector

From 2013.igem.org

Revision as of 22:39, 4 October 2013 by PhilippS (Talk | contribs)


Effectors

We engineered the CRISPR/Cas9 system that is relying on a protein-RNA-DNA interaction to generate a DNA binding protein. For this we used a catalytically inactive Cas9, termed dCas9. As we wanted to manipulate gene expression with our system we fused different effector domains to the dCas9. By doing so we created proteins that are able to activate, repress and alter the epigenetic state of genes.

A main feature of our uniCAS system is the so-called RNAimer, a plasmid coding for the required RNAs. By using several RNAs mutliple targeting is an immense advantage in comparison to other DNA-binding proteins. To highlight the specificity of our system we conducted the experiments with an off-target control. Here, the fusion protein is led to a locus that is not related to our test locus.

For repression experiments we performed experiments that alter the endogenous VEGF-A expression. For activation we tested the expression of a secreted reporter protein, the secreted embryonic alkalyine phosphatase, short SEAP.

Activation

Introduction

The aim of this subproject was to engineer a novel transcriptional activation component, based on an RNA guided synthetic dCas9-VP16 fusion protein. Therefore, we used a mutated form of Cas9 (dCas9). Impaired in its cleavage activity, dCas9 was fused to the 5' end of the trans-activation domain of VP16. To ensure nuclear localization of the whole construct, a nuclear localization signal (NLS) was fused to both the 5' and 3' ends of dCas9-VP16. A HA-epitope coding sequence was prefixed for efficient protein detection detections on western blots. Expression was set under control of SV40 or CMV promoters, and also a BGH terminator was suffixed. Figure 1 illustrates the above described design.

Figure 1: Design of the dCas9-VP16 fusion constuct.
dCas9 was fused to VP16 via a 3 amino acid linker. The resulting fusion protein was flanked by NLS sequences and tagged by a HA epitope. The SV40 or CMV promoter and the BGH terminator were chosen to control gene expression. BBa_K1150019 is set under the control of the SV40 promoter, BBa_K1150020 is under the control of the CMV promoter.

The Virus Protein 16 (VP16) is a transcription factor of Herpes simplex virus-1. Through complex formation with cellular host factors, VP16 can bind to a common regulatory element in the upstream promoter region of adjacent genes [1]. Through the trans-activating function of VP-16, expression rates of these genes can thus be enhanced. The protein consists of 490 amino acids - with two important domains: first, a core domain in its central region, which is necessary for the indirect DNA binding. Furthermore, a carboxy-terminal trans-activation domain [2][3]. The latter part of VP16 can be fused to DNA-binding domains of other protein - in order to efficiently upregulate expression of desired target genes [4].

Mechanism

By co-transfecting our RNA plasmid which includes the desired crRNA, dCas9-VP16 specifically binds to the targeted DNA sequence. With the help of the trans-activation domain of VP16, transcription factors are recruited and the pre-initiation complex can be built. By targeting this construct upstream of a promoter region any gene of interest can be activated.

Figure 2: Principle of trans-activation of mammalian gene expression by the fusion protein Cas9-VP16
The non-cleaving dCas9 fused to the VP16 domain serves as a crRNA-guided DNA-binding protein and trans-activates gene expression. Due to its helicase function, dCas9 unwinds the DNA-strand and the crRNA can bind sequence specifically.

Experimental setup

For testing this device we used HEK-293T cells, which were seeded at a densitiy of 65,000 cells/well in 24-well plates. After 24 hours RNA plasmids targeted against the indicated loci and the referring CMVmin:SEAP reporter plasmids were co-transfected to this device. The reporter plasmids contain the secreted alkaline phosphatase (SEAP) gene under the control of a CMV minimal promoter. 48 hours post transfection the activity of SEAP in the cell culture medium was measured. Additionally the luciferase Renilla was measured as an internal standard to eliminate variabilities of protein expression levels caused by the expression of the transfected plasmids. The dCas9-VP16 expression was assessed by Western blot analysis of cell lysates. Different crRNAs were tested and compared for their activation properties of the referring reporter plasmid.

Our controls were the following:
pRSet is a vector containing junk DNA and is transfected in the same DNA amount as the samples. As our negative control it reflects the basic SEAP level production in the absence of any activator protein. dCas9-VP16 driven by a SV40 or CMV promoter transfected without any crRNA is the off-target control. As dCas9-VP16 should only bind to the target locus in combination with the related crRNA no increase in the SEAP production should be visible in comparison to the pRSet control. This was observed so no off-target effects exist.

Results

With the dCas9-VP16 fusion protein different target loci have been tested by the usage of a SEAP reporter plasmid with a minimal CMV promoter (pKM602). The target sites can be determined by directing the crRNA against the desired sequence of interest. 4 Target sites with different distances to the promoter were tested (Figure 3 and Table 1).
We used two different promoters to drive the Cas9-VP16 expression: The SV40 promoter for a moderate level of Cas9-VP16 (BBa_K1150019) and the CMV promoter for a high level of our effector fusion protein (BBa_K1150020).

Figure 3: Position of the target loci on the SEAP reporter plasmid pKM602.
The tested target loci are located in front of the SEAP promoter. Several distances and combinations of the related crRNAs were tested.
Table 1: Overview of the tested crRNAs directed against different sites on the SEAP plasmid.
RNA plasmids containing the crRNAs EMXI, T2, T3 and T4 were transfected in combination with the activator dCas9-VP16. These RNAs target different loci on the SEAP plasmid. The crRNA sequences, their promoter distance and their binding site on the SEAP plasmid are displayed.
Name Distance to promoter Sequence GC content [%]
EMX1
-26 bp
GGAAGGGCCTGAGTCCGAGCAGAAGAAGAA
57
T2
-148 bp
AAGCATTTATCAGGGTTATTGTCTCATGAG
37
T3
-222 bp
AATGCCGCAAAAAAGGGAATAAGGGCGACA
47
T4
-443 bp
GACCGAGTTGCTCTTGCCCGGCGTCAATAC
60

By targeting EMXI and T2 with dCas9-VP16 driven by a SV40 promoter we gained 3 to 10 fold induction of SEAP expression (Figure 4). When we used these targets simultaneously a 17 fold activation could be achieved.

Figure 4: Results of the SEAP activation with Cas-VP16 under control of the SV40 promoter using different crRNAs.
HEK-293T cells have been transfected with different crRNA plasmids and Cas9 plasmids. The first line unter the blot shows the different targets, the second the transfected plasmid. pRSet (junk DNA) and the SV40:dCas9-VP16 construct without crRNA remain as negative controls.To quantify the activation properties of dCas9-VP16 the amount of SEAP expression was measured and divided through the expression level of the luciferase Renilla (internal standard). Each sample was measured in biological triplicates. The bright green columns reflect the negative controls while the dark green reflect the different samples. The fold induction above each column is related to the basic SEAP expression level of the pRSet (junk DNA) control.
Figure 4: Results of the SEAP activation with Cas-VP16 under control of the SV40 promoter using different crRNAs.

For the dCas9-VP16 construct under the CMV promoter the crRNA single targets EMXI, T2, T3 and T4 has been tested as well as the combination of two targets (T3 & T4 and EMXI & T2; Figure 5). All targets lead to an activation of the SEAP expression in comparison to the off target control. Compared to the samples without dCas9-VP16 only T4 showed no activation. The EMXI crRNA seems to be the most efficient locus for single target activation. We archievend an 11 fold induction. So T2 (BBa_K1150035) and EMXI (BBa_K1150040) proved as powerful activation sites.
By simultaniously using the EMXI and T2 loci the highest SEAP production could be achieved (29 fold induction; Figure 5).

Figure 5: Results of the SEAP activation with dCas9-VP16 under control of the CMV promoter using different crRNAs.
HEK-293T cells have been transfected with different crRNA plasmids and Cas9 plasmids. The first line unter the blot shows the different targets, the second the transfected plasmid. pRSet (junk DNA), the CMV:dCas9 and the CMV:dCas9-VP16 construct without crRNA remain as negative controls. To quantify the activation properties of dCas9-VP16 the amount of SEAP expression was measured and divided through the expression level of the luciferase Renilla (internal standard). Each sample was measured in biological triplicates. Bright green columns reflect the negative controls while dark green ones reflect the different samples. The fold induction above each column is related to the basic SEAP expression level of the pRSet (junk DNA) control. T3+4 crRNAs are transcribed from one RNA plasmid while T3 & T4 crRNAs are transcribed from two independent RNA plasmids.
Figure 5: Results of the SEAP activation with dCas9-VP16 under control of the CMV promoter using different crRNAs.

Discussion

We were able to efficiently activate the expression of a reporter protein from a transfected plasmid (up to 29 fold). Different levels of this activation were achieved by targeting different loci upstream of the promoter of the reporter protein. The best results were gained by targeting two loci simultaneously.
Very recently a 25 fold activation of transient GFP expression by dCas9-VP64 was shown by Gilbert et al. however in comparison to cells that were not transfected with a GFP plasmid [5]. So we were able to yield a higher effect of SEAP acitivation with dCas9-VP16.
With different promoters for dCas-VP16 the relations between the target sites were similar, but the activation of the SV40 driven dCas9-VP16 was slightly weaker. So SV40:dCas9-VP16 can be used to activate genes more moderately than with CMV:dCas9-VP16.

Click here for detailed information about the target sites.

References

(1) Weir, J. (2001). Regulation of herpes simplex virus gene expression. Gene 271: 117-130.
(2) Triezenberg, SJ. et al. (1988). Functional dissection of VP16, the trans-activator of herpes simplex virus immediate early gene expression. Genes Dev 2: 718-729.
(3) Greaves, R. and O’Hare, P. (1989). Separation of requirements for protein-DNA complex assembly from those for functional activity in the herpes simplex virus regulatory protein Vmw65. J Virol 63: 1641-1650.
(4) Hirai, H., et al. (2010). Structure and functions of powerful trans-activators: VP16, MyoD and FoxA. Int. J. Dev. Biol. 54: 1589-1596.
(5) Gilbert, LA., et al. (2013). CRISPR-Mediated Modular RNA-Guided Regulation of Transcription in Eukaryotes. Cell. 2013 Jul 18;154(2):442-51.

Epigenetics

Histone modification by dCas9-G9a

Introduction

For organisms it is crucial to have a tight control over their transcriptional machinery. As every cell has basically the same genetic information, different tissues have to be formed by alter the expression of this information. One of the most prominent mechanisms that give rise to this differential expression of genes are epigenetic modifications. Epigenetics are, by definition, inheritable changes in gene expression, that are not accompanied by changes in the nucleotide sequence [7].

There are several types of epigenetic modifications that may have severe impact on gene expression [8]. Basically there are two main types of modifications. The first type are the chemical modification of cytosine residues of nucleotides, better known as DNA methylation. At so-called CpG islands, that can be found clustered in front of promoters the nucleotides can be altered by methylation. This methylation is a hallmark of repressive transcriptional states. But not only the DNA may be altered, but also the protein-nucleotide complex, called the chromatin, that forms the highly variable system that has massive impact on the differential regulation of expression. The probably most prominent epigenetic modifications are histone modifications.

Histones are proteins that work as multi-histone complexes and form a backbone for the DNA to be wound around. The termini of these nucleosomes are protruding from the histone core complex and can be targets for a lot of different modifications, which influence the chromatin state. One very prominent modification is methylation at lysine 9 of histone 3 (H3K9me) which renders before-open chromatin inactive. Such methylations are a hallmark of gene repression. One interesting fact about histone modification is the capability to spread the activity state over the surrounding chromatin via reader proteins. So the information of e.g. "repressed state" can, once specifically introduced, be propagated over a whole locus.

Figure 6: CMV:dCas9-G9a
dCas9 was fused via a 3 amino acid linker to G9a. The resulting fusion protein was flanked by NLS sequences and tagged by a HA epitope. The CMV promoter and BGH terminator were chosen to control gene expression. BBa_K1150025 is set under the control of the SV40 promoter, BBa_K1150024 is under the control of the CMV promoter.

We extpand the uniCAS system by a histone methylase. By introducing specific histone modifications at several loci we should be able to regulate several genes at once. For this we fused a methyltransferase that is known to specifically perform certain histone tail methylations to the non-DNA-cleaving version of Cas9 (dCas9).

For our device we used a part of the murine EHMT2 gene, the G9a. It is described in literature that, when targeted to an open locus via zinc finger proteins, G9a is able to repress expression of this locus [9].

We assayed the function as an epigenetic repressor. As a test subject we chose the VEGF locus, as it is

  1. well characterized
  2. easy to measure by ELISA methods
  3. VEGF is known to be involved in tumorgenesis and therefore an interesting target for future medical approaches
  4. HEK293T cells have an open VEGF locus, so we do not have to artificially open the locus, before testing the system

Results

HEK293T cells were seeded into 24-well plates and transfected with our constructs, that were targeted to open regions [10] of the VEGF locus.

Figure 8: Target sites on the endogenous VEGF locus.
As the VEGF locus offers several accessible regions we chose four loci around the promoter. Arrows mark the regions. Numbers indicate position of the PAM sequence relative to the starting point of transcription.
By specific histone modification through G9a we should see repression of VEGF secretion into the medium. 12 hours after transfection the medium was change and 24 hours after medium change we harvested the supernatant and performed VEGF measurments by ELISA. Additionally we co-transfected a constitutive reporter (SEAP) as an internal standard to reduce mistake of non-transfected cells and cell number. A major problem of working with endogenous loci is the background of non-transfected cells which will display VEGF secretion, even though we have strong repression in other cells.

As a control that the repressive effect of our proteins is not based on the sterical block of the transcription, we tested against the catalytic inactive dCas9. So every detectable effect is due to the G9a targeted to this locus. [11] As we can see in the graph, we achieved a strong reduction in VEGF expression, especially when targeting the -8 and -573 loci. This is in line with literature results and plausible, when having in mind, that the promoter region is targeted and dense chromatin in promoters leads to repression.

Figure 9: Endogenous, stable repression by dCas9-G9a
HEK-293T cells have been trancfected with the dCas9-G9a device and the RNAimer plasmid containing the different crRNA target sites for the endogenous VEGF locus. 12 hours after transfection the medium was change and 24 hours after medium change we harvested the supernatant and performed VEGF measurments by ELISA. As a control that the repressive effect of our proteins is not based on the sterical block of the transcription, we tested against the catalytic inactive dCas9. So every detectable effect is due to the G9a targeted to this locus.(n=3, p<0.05 is marked by asterisks)

Discussion

Our results obviously show a strong repression of VEGF expression around 50%, depending on the locus targeted. We can conclude from our results and literature data that we were able to specifically methylate histones and change the transcriptional state of the VEGF locus. Off-target controls do not display differences in VEGF expression, which indicates the specificity of our protein.
As we always show the difference to the dCas9 protein without effector. This leads to the point that we only observe the catalytic activity of G9a. This results in a valuable repressor tool that is able to specifically change histone methylation patterns and can change transcriptional states. It leads to many possible applications such as cancer research, fundamental epigenetic science or even tissue engineering.

References

(7) Wolffe, A., et al. (1999). Epigenetics: Regulation Through Repression. Science 286169, 481.
(8) Jones P. and Baylin S. (2002). The Fundamental Role of Epigenetic Events in Cancer. Nature Reviews Genetics 3, 415-428.
(9) Snowden, A., et al. (2002). Gene-Specific Targeting of H3K9 Methylation Is Sufficient for Initiating Repression In Vivo. Current Biology 12, 2159-2166.
(10) Liu, PQ., et al. (2000). Regulation of an Endogenous Locus Using a Panel of Designed Zinc Finger Proteins Targeted to Accessible Chromatin Regions. The Journal of Biological Chemistry, 276, 11323-11334.
(11) Lee, D., et al. (2006). Histone 3 Lysine 9 Methyltransferase G9a Is a Transcriptional Coactivator for Nuclear Receptors. Journal of Biological Chemistry 281, 8476-8485.

Repression

Introduction

Krüppel-associated Box (KRAB) repressor domains are highly conserved polypeptide motifs and were first functionally characterized in 1991 [12]. Their occurence in about one third of all human zinc finger transcription factors suggests key regulatory features in higher eukaryotic transcriptomics [13]. In terms of tetrapod evolution, the role of their great abundance has been extensively discussed [14]. Even though KRAB minimal domains are usually no longer than ~ 50-75 amino acids, their mechanism of function remains complex. Common biochemical models suggest a key role in epigenetic silencing, by recruiting a scaffold of diverse proteins to the zinc fingers' binding site - amongst others histone deacetylases and histone methyltransferases [15]. To date, KRAB domains were attached to several DNA binding proteins such as lacR and tetR, thereby silencing gene expression downstream of designed reporter targets.

Figure 10: Principle of transcriptional repression by a dCas9-KRAB fusion
A double mutant Cas9 (D10A; H840A), fused to a Krüppel-Associated Box (KRAB) repression domain of Homo sapiens, serves as a crRNA-guided DNA-binding regulator protein.

In this work, KRAB was fused to enzymatically inoperable dCas9. Thus, a transcriptional repressor with the flexibility to target almost any DNA sequence of interest was yielded. Transient SEAP expression could thus be reduced by almost 60 %. In a second attempt, CMV-driven expression of the signaling scaffold protein CNK1 was targeted over 36h - being partially knocked down to background amounts [16]. Furthermore, GFP reporter expression was shown to be drastically reduced by dCas9-KRAB in Fluorescence Microscopy imaging. Endogenous levels of VEGF-A, a key factor in hypoxic tumor angiogenesis [17], were also successfully reduced and quantified through an Enzyme-Linked Immunosorbent Assay (ELISA).

Figure 11: Design of the dCas9-KRAB fusion construct
Through a seven amino acid-linker, dCas9 was fused to KRAB. The resulting fusion protein was flanked by two NLS sequences. An HA-tag was also suffixed for Western blot detections. To control gene expression, SV40 or CMV promoters and the BGH terminator were chosen to control gene expression.

Results

Repressional effects through our engineered dCas9-KRAB fusion protein were primarily evaluated by harnessing transient reporter assays.

First, we used Secreted Embryonic Alkaline Phosphatase (SEAP), an enzyme which is thermostable and secreted into the medium of transfected mammalian cells. By photometrically measuring conversion rates of an exogenously added substrate (pNPP) at 405 nm, we were able to precisely determine amounts of SEAP production. In our attempt, a tetO13-CMVmin-SEAP reporter plasmid was constitutively trans-activated by co-transfection of a second plasmid, coding for SV40-driven tetR-VP16. Thereby, CMVmin was efficiently upregulated and a high rate of SEAP production could be yielded. Thirdly, by transfecting dCas9-KRAB in combination with a crRNA that encodes a locus of ~ 40 basepairs upstream of the CMVmin promoter, a strongfold decreasement of the trans-activated SEAP production took place.

Figure 12: dCas9-KRAB repression effects on SEAP expression.
dCas9-KRAB was crRNA-directed towards the tetR-VP16-transactivated SEAP gene.

As a second prove-of-principle experiment, fluorescence microscopy was chosen. Therefore, we transfected HEK-293T cells with a self-built fluorescence-reporter plasmid, which contains a CMV promoter and thus constitutively drives high expressions of GFP. Co-transfection of dCas9-KRAB - and a crRNA corresponding to 30 basepairs of the CMV promoter - led to powerful disruption of fluorescence activity. Negative controls were conducted through co-transfections of similar amounts of non-coding pRSET and GFP-plasmid.

Figure 13: dCas9-KRAB repression effects on GFP fluorescence.
Upper images: As a negative control, non-coding pRSET was co-transfected with our GFP reporter.
Lower images: instead of pRSET, similar amounts of dCas9-KRAB were co-transfected.
Fluorescence images were taken with a light exposure of 250 ms and 80% saturation.

In order to counteract gene expression of a signaling pathway component, ectopically expressed Connector Enhancer of KSR 1 (CNK1) was utilized for detection. CNK1 acts as a scaffold protein for a vast variety of signaling processes in cells, including proliferation and apoptosis [16]. In a time-series experiment, repressive effects were to be shown on Western Blots. Therefore, CNK1 was co-transfected with similar quantities of either dCas9 or dCas9-KRAB, providing an insight into the functional effects of the KRAB domain. It could be shown that co-expression of dCas9-KRAB, in comparison with dCas9 alone, dramatically decreases CNK1 levels - at least on basis of a transient attempt.

Figure 14: dCas9-KRAB repression effects on CNK1 protein expression.
dCas9-KRAB was crRNA-directed towards a transiently expressed signaling protein. As a negative control, a similar amount of dCas9 was utilized.

Next, investigation of further effects was to be proven on endogenous stages in human cells, as well. Similar to our experiments with epigenetically acting dCas9-G9a, we decided to also direct dCas9-KRAB towards the promoter region of the Vascular Endothelial Growth Factor A (VEGF-A) on Chromosome no 6. VEGF-A protein amounts could be detected through a specifically optimized sandwich version of an Enzyme-linked Immunosorbent Assay (ELISA). Normalisation was performed by non-interacting SEAP. To assess dCas9-KRAB action, we co-transfected the fusion-protein plasmid with either crRNAs corresponding to the VEGF-A locus or to an EMX1 off-target on Chromosome no 2. The EMX1 protein is a prominent member of brain development stem cell regulators. On account of this, no effects on VEGF-A expression were expected for our off-target control of dCas9-KRAB actions.

Figure 15: dCas9-KRAB repression effects on CNK1 protein expression.
dCas9-KRAB was crRNA-directed towards a transiently expressed signaling protein. As a negative control, a similar amount of dCas9 was utilized.

dCas9-KRAB has shown a clear ability to downregulate VEGF levels. By contrast with the applied off-target control, protein amounts detected were almost halved - both on sites upstream and downstream of the corresponding promoter. VEGF is a key factor in angiogenesis and has thus an important impact on tumor growth under hypoxic conditions. Therefore, an efficient repression of VEGF-A may lead to the delivery of insights into both fundamental spheres of complex endogenous regulation and cancer therapy. Future attempts of dCas9-KRAB directed repression may include different and more sophisticated endogenous targets - namely those involved in intricate cancer development.

Figure 16: dCas9-KRAB repression effects on CNK1 protein expression.
dCas9-KRAB was crRNA-directed towards a transiently expressed signaling protein. As a negative control, a similar amount of dCas9 was utilized.

However, standardized versions of both SV40:dCas9-KRAB and CMV:dCas9-KRAB have yet failed to yield successful results. Hitherto, protein expression of these constructs could not be truly shown on Western-Blots. Effects on transient Secreted Embryonic Alkaline Phosphatase reporters, similar to the first experiment, were shown nevertheless, while undesired off-target effects have also lately emerged.

Discussion

Apparently, Figure 16 indicates an unfortunate off-target effect. Even though 30-nucleotide crRNAs against the SEAP promoter and EMX1 show non-similarity of more than 20 basepairs, repressive effects were shown to yield a certain similarity.

To date, we interpret these unforeseen data as follows. Transfection of SV40:dCas9-KRAB and CMV:dCas9-KRAB may be crucially inhibited for several reasons. First, one could suggest lower expression rates for both CMV and SV40 promoters which were used for our standardized constructs. This, in turn, does not fit with our findings that dCas9-VP16 and dCas9-G9a, our other transcriptional effectors, have been repeatedly well expressed. A more sophisticated hypothesis emerges from structural properties of the fusion protein. Both the enormous size of the transcript (4581 Bp), as well as a dozen structural coding deviances between standardized and non-standardized dCas9-KRAB, are very likely to having affected mRNA stability in a crucial manner. Expression values have been likewisely and repeatedly low for non-standardized constructs - Western Blot data not shown - but could be defected with certainty. Furthermore, following our assumptions, the associated fusion protein (1527 Aa) may have become impaired by unexpected post-translational modifications, especially referring to ubiquitylation and subsequent proteasomal degradation.

For these assumptions, we still believe in dCas9-KRAB effects and target specificities - which are still to be displayed after improving expression efficiencies. Since we consider it as a duty to point out the recently observed and unwanted off-target effects, we are keen to find out about the exact reasons. A very recent article of Gilbert et al. has already proven high functionality of this device - although strong repressive effects required stable lentiviral integration of dCas9-KRAB and HEK-cell pre-selected via Flow Cytometry had to be done [5].

References

(12) Rosati, M. et al. (1991). Members of the zinc finger protein gene family sharing a conserved N-terminal module. Nucleic acids research 19, 5661-5667.
(13) Witzgall, R. et al. (1994). The Krüppel-associated box-A domain of zinc finger proteins mediates transcriptional repression. Proc Nati Acad Sci 91, 4514-4518.
(14) Birtle, Z. and Ponting, C. (2006). Meisetz and the birth of the KRAB motif. Bioinformatics 22, 2841-2845.
(15) Urrutia, R. (2003). KRAB-containing zinc-finger repressor proteins. Genome Biology 4, 4:231.
(16) Fritz, R. and Radziwill, G. (2011). CNK1 and other scaffolds for Akt/FoxO signaling. Biochimica et biophysica acta 1813, 1971-1977.
(17) Bałan, B. and Słotwiński, R. (2008). VEGF and tumor angiogenesis. Centr Eur J Immunol 33, 232-236