Genetically Encoded, SH2 Domain-Based Fluorescent Reporters of Endogenous Receptor Tyrosine Kinase Activity in Living CellsJohn Fetter; Dmitry Malkov; Nathan Zenser; Keming Song

Sigma-Aldrich Research Biotech, Cell-Based Assays/Reporter Cell Lines, 2909 Laclede Avenue, Saint Louis, MO 63103 USA

Figure 3: Selectivity and specificity of the biosensor. Tyrphostin AG 1478, a selective inhibitor of EGFR (ErbB1), blocked the TagGFP:2X(SH2)Grb2 translocation to the plasma membrane and internalization. Most of the ligands to receptor tyrosine kinases were tested for activity with the TagGFP:2X(SH2)Grb2 sensor. However, heregulin-β1 was compared to EGF with a TurboGFP:2X(SH2)Grb2 sensor. Activity was seen with EGF and TGF-α, ligands specific for EGFR (ErbB-1). Heregulin-β1, a ligand for ErbB-3 and ErbB-4, did not show activity. HGF that binds HGFR also showed activity, but much less than EGF. GAS6, IGF-1, and heregulin- β1 were tested at 1 μg/ml. Insulin was tested at 2 μg/ml. The other ligands were tested at 100 ng/ml.

Ligand Activity

EGF +++

TGF-α +++

HGF +

Heregulin-β1 –

PDGF-AB –

Insulin –

IGF-1 –

NGF-Beta –

FGF-acidic –

Angiopoietin –

MSP –

GAS6 –

VEGF-D –

FLT-3 Ligand –

Selectivity and Specificity of the BiosensorAbstractReceptor tyrosine kinases (RTKs) are a subclass of cell-surface growth-factor receptors that dimerize and autophosphorylate upon ligand binding. They transduce signals across the cell membrane and thus regulate diverse cell functions. In contrast to biochemical assays or immunostaining, using a natural domain-based genetically encoded biosensor allows detection of RTK activation in live cells. The biosensors constitute a protein-binding domain, derived from natural domain-containing proteins, fused with a fluorescent protein that serves as a tracer.

A biosensor consisting of the SH2 domain of the adaptor protein Grb2 fused to GFP was developed and tested in cell lines with endogenous EGFR expression. The biosensor was transiently transfected or stably expressed using an exogenous promoter, such as CMV or PGK. The stable cell line was cloned from a single cell giving a homogeneous fluorescent population by microscopy imaging and flow cytometry. Within minutes after stimulation with EGF, the biosensor showed robust redistribution towards the cell membrane and subsequent internalization through endocytosis. The internalization kinetics of the sensor-receptor complex were quantified by granule counting. The reporter response was highly selective as it could be induced only by specific EGFR ligands and was abolished by a selective inhibitor of EGFR - Tyrphostin AG 1478. It was demonstrated that ratio imaging could increase the signal sensitivity by using a second differentially emitting fluorescent protein as a volume marker.

We also demonstrated that the biosensor could be integrated into the known location in the genome using zinc finger nuclease (ZFN) technology to be expressed under an endogenous promoter. The biosensor sequence was inserted directly in front of the first exon of the TUBA1B (alpha-tubulin isoform 1b) locus followed by the translation skip (2A) to preserve the host gene expression.

The successful creation of fluorescent indicators for imaging of endogenous EGFR activity suggests that this general strategy should be extensible to other RTKs. Optimized biosensors could be of great value in pharmaceutical screening for RTK modulators and studying their mechanisms of action.

Introduction We developed a novel platform that uses genetically encoded biosensors for live cell imaging. The biosensor constitutes a protein-binding domain, derived from natural domain-containing proteins, fused with a fluorescent protein. The protein-binding part of the biosensor is derived from screening of a protein domain library against the specific targets, such as protein kinases, whereas the fluorescent protein part serves as a tracer. Our first biosensor product monitors activation and kinetics of an RTK - EGFR in live cells. Upon RTK activation by a specific ligand, the binding domain of the biosensor will bind to the activated RTK at the tyrosine phosphorylation site through the domain-target interaction, whereas the fluorescent tag will track the binding and receptor internalization. Assay specificity is determined by three factors: binding affinity of the selected domain to the target RTK, expression level of the endogenous target RTK, and the specific ligand used to activate the target.

Materials and MethodsThe following reagents were obtained from Sigma: RPMI-1640 (R0883), L-glutamine (G7513), trypsin (T3924), puromycin (P9620), fetal bovine serum (F2442), hexadimethrine bromide (H9268), EGF (E9644). A549 were grown in RPMI-1640, 2 mM glutamine, 10% FBS in 5% CO2, 37 °C. For transient transfection experiments the biosensor was transfected into A549 cells using the recommended protocol on the Amaxa®. To make a stable cell line the biosensor was packaged in lentiviral particles and transduced into A549 at an MOI of 5 in media with 8 µg/ml hexadimethrine bromide. The cells were imaged with an automated Nikon TE2000 microscope. MetaMorph® was used for image analysis. GFP: ex 450–490 / em 500–550; tagFP635: ex 565–595 / em 610–650; DRAQ5: ex 618–673 / em 685–735; 40x/1.4 oil.

Nature Reviews Cancer 4, 361-370 (May 2004)

Figure 1: Structure of domain-based biosensors.

A: Generic structure of biosensors that consist of a target-specific binding domain and a fluorescent protein.

B: The structure of the EGFR biosensors, consisting of one or two SH2 domains from adapter protein Grb2 that specifically bind to activated EGFR and a green fluorescent protein tag (turboGFP or tagGFP).

A.

B.

N CBinder FP

N CFP Binder

N C

N CturboGFP or tagGFP

turboGFP or tagGFP

SH2Grb2

SH2Grb2 SH2Grb2

N CBinder FP

N CFP Binder

N C

N CturboGFP or tagGFP

turboGFP or tagGFP

SH2Grb2

SH2Grb2 SH2Grb2

ResultsStructure of Biosensors

A.Figure 5: Ratiometric imaging improves detection.

A: The EGFR-specific binder domain was fused with RFP and linked to GFP through the 2A sequence. The translational skip at the 2A site resulted in expression of two separate proteins: GFP as a volume marker and RFP-B as the sensor.

B: The fluorescence signal from TagFP635:2X(SH2)Grb2 was divided by the eGFP signal in MetaMorph. This removes the effect of changes in cell thickness on the florescence intensity of the sensor and improves the sensitivity at the thinner periphery of the cell.

TUBA1B locus

1 Kb

RFP GFP

B-2A

mRNA

N C

700 bpATG

ATG

ATG

coding region

untranslated region

intact splice signal

Two separate proteins

tagGFP SH2 SH2 2A

RFP B (binder) 2A GFP

Ratiometric Imaging

[EGF] = 100 ng/ml, bar = 10 µm

Rat

io:

TagF

P63

5 / e

GFP

- EGF (0 min) + EGF (3 min) + EGF (59 min)

eGFP

TagF

P63

5-2X

(SH

2 Grb

2)

B.

Summary• A tagGFP:2X(SH2)Grb2 sensor was tested for activity to receptor tyrosine kinases.

• Upon activation with EGF, the sensor showed translocation to the plasma membrane followed by internalization in mammalian cell lines.

• This sensor showed significant activity with EGF and much less activity with HGF.

• Activity could be inhibited with a small molecule specific for EGFR activity.

• The sensor was not activated by ligands to other receptor tyrosine kinases.

• Other SH2 sensors could possibly be developed to detect activity for other receptor tyrosine kinases.

• The domain-based biosensors provide a powerful tool to monitor the spatial and temporal change of endogenous RTKs in live cells.

Potential applications of domain-based biosensors:

– Screen for RTK inhibitors

– Study mechanism of inhibitor compounds

– Assess specificity of lead compounds (e.g. side effect)

– Assess selectivity of antibody-based RTK inhibitors

An A549 cell line stably expressing the biosensor was successfully validated in HCS mode by Dr. Hakim Djaballah’s screening group at Memorial Sloan-Kettering Cancer Center using known EGFR inhibitors and the LOPAC 1280 (LO1280) library.

Figure 6: Targeted integration of the biosensor to be expressed under endogenous promoter. Zinc finger nuclease (ZFN) technology was used to insert the biosensor sequence in front of the first exon of TUBA1B (NM_006082, α-tubulin isoform 1b) locus followed by the translation skip (2A) to preserve the host gene expression.

A: Schematics of TUBA1B locus showing the coding region (blue), untranslated region (gray), and the ZFN cut site (scissors). The donor (top) has the homology arms of indicated length and the insert (boxed) consisting of tagGFP (green), two SH2 domains from Grb2 (yel-low) and 2A (white) sequences. The first exon contains ATG only. To preserve its splice signal, the insert was integrated before the ATG. Another ATG was introduced in front of the insert to initiate translation.

B: The biosensor expressed under TUBA1B promoter shows activity similar to its transient expression. The donor construct was nucleo-fected into A549 cells along with ZFNs designed to cut near the genomic target site. After a few days the cells were sorted and imaged while applying 100 ng/ml EGF.

A.

B.

TUBA1B locus

1 Kb

RFP GFP

B-2A

mRNA

N C

700 bpATG

ATG

ATG

coding region

untranslated region

intact splice signal

Two separate proteins

tagGFP SH2 SH2 2A

RFP B (binder) 2A GFP

- EGF (0 min) + EGF (2 min) + EGF (63 min)

Targeted Genomic Integration of the Biosensor

- (0 min) + (3 min) + (50 min)

Her

egul

in-β

1H

GF

EG

FE

GF

+A

G 1

478

[EGF] = 100 ng/mL, [AG 1478] =1µM, [Heregulin-β1 (EGF Domain 176-246)] = 1µg/mL, [HGF] = 100 ng/ml, bar = 10µm

Development of a Stable Cell Line

Figure 4: Development of a stable cell line.

A: A lentiviral construct of tagGFP:2X(SH2)Grb2 was used to trans-fect A549. Single cell clones were selected and assayed for activity with EGF. This clone was homogenous for expression and EGF activity – 20x/0.75 objective, [EGF] = 100 ng/ml.

B: Flow cytometery analysis of the EGFR biosensor cell line showed a homogeneous population.

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

Wild Type

EGFR Biosensor

Sample

-EG

F (0

min

)+

EG

F (4

min

)

+ E

GF

(24

min

)+

EG

F (5

5 m

in)

50 µm

-EG

F (0

min

)+

EG

F (4

min

)

+ E

GF

(24

min

)+

EG

F (5

5 m

in)

50 µm

B.

A.A.

Figure 2A: EGF causes biosensor translocation to the plasma membrane followed by internalization. TagGFP:2X(SH2)Grb2 was trans-fected into A549 (Amaxa). The nucleus was labeled with 1 μM DRAQ5.

Figure 2B: EGF causes biosensor translocation to the plasma membrane followed by internalization. The redistribution of the biosensor over time was quantified using translocation and granularity analysis in MetaMorph.

DIC

- EGF (0 min) + EGF (3 min)

+ EGF (6 min) + EGF (17 min) + EGF (69 min)

Biosensor Translocation

- EGF (0 min) + EGF (3 min) + EGF (17 min) + EGF (69 min)

2

# 1

# 2

# 3# 4

# 5 5 74 107

Granule count for cell # 4

B.

761131031