healthmedicinet

SH2-PLA assay scheme

We chose the epidermoid carcinoma cell line A431, which overexpresses wild type epidermal
growth factor receptor (EGFR), as a developmental platform. The premise of the SH2-PLA
assay is that 1) EGF stimulation induces tyrosine phosphorylation of the intracellular
domain of EGFR, which creates specific binding sites for SH2 domains such as Grb2,
Src, PLC?1, Vav2, etc.; 2) GST-SH2 domain coupled to anti-GST 5? Prox-Oligo binds to these phosphotyrosines
in cell lysate; 3) This interaction brings the anti-GST 5? Prox-Oligo and anti-EGFR
3? Prox-Oligo probes together in close proximity, thus allowing for the proximity
ligation reaction to be subsequently quantified by real-time PCR (Fig. 1). In other words, detection of a specific assay signal requires the creation of the
quaternary complex: anti-EGFR 3?Prox-Oligo probe:phosphorylated EGFR:GST-SH2 protein:anti-GST
5? Prox-Oligo probe.

Fig. 1. In-solution SH2 domain binding assay using proximity ligation and real-time PCR. a, Schematic Illustration of SH2-PLA. A pair of PLA probes is used to detect the interaction
of tyrosine phosphorylated EGFR and a GST-SH2 domain. The 3? SH2-PLA probe consists
of an anti-EGFR antibody conjugated with the 3? proximity oligonucleotide (3? Prox-Oligo).
The 5? SH2-PLA probes consists of an anti-GST antibody conjugated with the 5? Prox-Oligo
and a GST-SH2 domain. When the GST-SH2 domain binds to tyrosine phosphorylation sites
of EGFR, the 5? and 3? PLA probes are brought in close proximity, allowing ligation
of the two Prox-Oligos which is detectable by real-time PCR. b, Experimental workflow of SH2-PLA Method 1. Lysates are prepared with or without EGF
stimulation. Biotinylated anti-GST and anti-EGFR antibodies are conjugated with the
5? and 3? Prox-Oligos, respectively, and stored at ?20 °C. The 5? SH2-PLA probe is
mixed with purified GST-SH2, and the 3? SH2-PLA probe is mixed with cell lysates allowing
the antibodies to bind their respective epitopes. Subsequently, the 5? and 3? PLA
probe solutions are combined to induce interaction between the SH2 and pEGFR. Then,
the amount of the complex is quantified by proximity ligation and real-time PCR. An
alternative method is also possible (Additional file 1: Figure S2). Estimated assay runtime including sample-handling steps for each procedure
is noted on the right

Performance of TaqMan protein expression assay

Prior to customizing the TaqMan Protein Expression Assay for SH2-PLA (i.e., modular domain binding assay), we tested the performance of the original proximity
ligation assay using the kit-supplied anti-ICAM1 assay probes and Raji B-cell lymphoma
lysate. The 3? and 5? oligonucleotide-conjugated anti-ICAM1 assay probes were incubated
with the Raji lysate for 60 min followed by the ligation reaction. After heat inactivation,
the ligation product was quantified using the TaqMan real-time PCR system. The experiment
was performed in a 96 well plate with approximately three hours of assay time. We
observed, with high-precision, approximately three orders of magnitude of linear dynamic
range for a cell input of 0.24 – 250 cells per ?l (intra-assay %CV??1.1 %) (Additional
file 1: Figure S1A). When two independent experiments were compared, they showed very strong
correlation (Pearson correlation r?=?0.99) (Additional file 1: Figure S1B). We concluded that the performance of the TaqMan Protein Expression
Assay was sufficient for development of an SH2 domain binding assay.

Development of the SH2-PLA assay

Biotinylated anti-GST and anti-EGFR polyclonal antibodies were conjugated with 5?
and 3? oligonucleotides following the probe development protocol (see Methods). To
evaluate the anti-GST probes (5? Prox-Oligo- and 3? Prox-Oligo-conjugated anti-GST
antibodies) and anti-EGFR probes (5? Prox-Oligo- and 3? Prox-Oligo-conjugated anti-EGFR
antibodies), purified GST protein and A431 cell lysate were serially diluted and respective
TaqMan protein expression assays were performed as in the ICAM1 expression assay.
In these separate detection experiments for GST and EGFR, we observed that the GST
probe had a linear increase in Ct values between 0.13 and 4.17 nM of GST input, while
the EGFR probe for A431 lysate was linear from 1.9 to 30 ?g/ml (Additional file 1: Figure S1C and D).

Next we tested the ability of the 5? Prox-Oligo-conjugated anti-GST antibody and 3?
Prox-Oligo-conjugated anti-EGFR antibody pair to detect an interaction between the
GST-SH2 and EGFR in A431 lysate. Among multiple attempts performed to optimize the
binding conditions, we found two distinct methods that provided a favorable signal-to-noise
profile (Fig. 1b Additional file 1: Figure S2). In Method 1, antibodies are premixed with target proteins prior to SH2
binding, while in Method 2, GST-SH2 domains and EGFR (lysate) are incubated prior
to antibody binding (See Methods for more details). As the results from both methods
were nearly equivalent, Method 1 (Fig. 1b) was used for all experiments described in this report.

Specificity of the SH2-PLA

A431 cell lysates were prepared in the presence or absence of EGF stimulation and
the SH2-PLA assay was performed as outlined in Fig. 1b. We employed several SH2 domain containing proteins for validation that are known
to be physiological ligands of EGFR such as Grb2, Vav2, and PLC?1 26]–28]. Figure 2a shows a representative real-time PCR amplification plot of the SH2-PLA assay using
the Grb2 SH2 domain probe and A431 cell samples. PCR product in the EGF-stimulated
A431 sample was amplified more rapidly than in the unstimulated sample resulting in
a lower threshold cycle (Ct) value. The difference in Ct values between the two samples
(?Ct) is an indicator of enhanced binding by the Grb2 SH2 domain probe to tyrosine
phosphorylated EGFR (pEGFR). To validate the specificity of the assay, we compared
signal from a GST-SH2 domain probe and GST control. The Ct value for the GST control
was unchanged with EGF stimulation (lanes 1 vs. 2, Fig. 2b upper panel). On the other hand, the SH2 domains of Grb2 and PLC?1 showed a marked
reduction in their Ct values upon stimulation (lanes 5 vs. 6 and 9 vs. 10). Using
the same set of samples and SH2 domains, far-Western blotting was performed as a reference
(Fig. 2b middle panel). In far-Western, proteins were separated on polyacrylamide and transferred
to a nitrocellulose membrane which was then probed with HRP-labeled GST-SH2 domains
23]. The identity of a major band at approximately 180 KDa in the SH2-far-Western blotting
has previously been confirmed to be EGFR by anti-EGFR immunodepletion (data not shown).
As shown in the middle panel of Fig. 2b, the signal profiles of the SH2-PLA and far-Western are similar despite their use
of distinctive assay readouts (Ct values vs. bands). SH2 domains are known to have
both unique and overlapping ligand binding characteristics 7], 29]–32]. To determine if the SH2 binding is tyrosine site dependent, a synthesized phosphopeptide
corresponding to EGFR tyrosine 1068, containing the Grb2 SH2 consensus binding site,
was added as a blocker. In both assays, Grb2 SH2 binding was significantly reduced
in the presence of the blocker, while the blocking effect on PLC? SH2 domain binding
was relatively modest (lanes 6 vs. 8 and 10 vs. 12). Taken together, these results
indicate that, like far-Western, the SH2-PLA assay performed with the anti-GST 5?
Prox-Oligo antibody and anti-EGFR 3? Prox-Oligo antibody probe pair is specific enough
to distinguish between EGF-stimulated and control cell samples.

Fig. 2. Validation of the SH2-PLA assay. a, Representative PCR amplification plot for SH2-PLA experiments. Increased binding
between SH2 and pEGFR upon EGF stimulation is expressed as a reduced threshold cycle
value (Ct). Here, ?Ct is defined as [Ct
_control
– Ct
_{EGF stimulated}
]
_.b, Specificity of SH2-PLA. SH2-PLA (top panel) and far-Western (middle panel) results
for EGF-stimulated and control A431 cell samples are shown. Results for GST control
probe are shown in lanes 1–4; Grb2 SH2 probe in lanes 5–8; and PLC?1 tandem SH2 probe
in lanes 9–12. Lanes 3, 4, 7, 8, 11, and 12 show the assay result in the presence
of pY1068 blocking peptide, which contains the Grb2 SH2 consensus binding site of
EGFR. c, SH2-PLA assay performance. The SH2-PLA assay was performed three times using a two
fold dilution series of EGF-stimulated and control A431 cell lysates. Average Ct values,
normalized to non protein control (NPC), are shown in the upper panel. The intra-assay
variation for Ct values was 0.07-2.36 (mean 0.60) and the inter-assay %CV was 0.32-3.08
(mean 1.28). Since EGF-stimulated samples always showed a greater signal (lower Ct)
than the unstimulated control throughout the dilution series, the range of assay detection
is estimated to be at least 1.1–1100 ?g/ml of lysate concentration, and the lower
limit of detection is approximately 2 ng of protein per assay. The lower panel shows
the approximately linear region of the mean Ct plot against log input lysate concentrations,
and the ?Ct (unstimulated – stimulated) of about three cycles. The log
₂
fold change between EGF-stimulated and control samples was estimated to be 6.0 – 6.4
using the ProteinAssist software tool (Additional file 1: Figure S3). d, Adoption of other phosphotyrosine recognizing domains. The SH2-PLA methodology was
applied to protein tyrosine phosphatase (PTP) and phosphotyrosine binding (PTB) domains.
ShcA PTB domain and the substrate-trapping mutant of PTP1B PTP domain displayed activity
comparable to Grb2 SH2 (lanes 1–4 and 7–8). Signal was undetectable for the wild type
(wt) PTP1B PTP domain, likely due to the intrinsic phosphatase activity (lanes 5–6)

Performance of the SH2-PLA assay

To evaluate assay performance, including the limit of detection, linearity, and precision,
we performed the SH2-PLA assay using a serial dilution of lysate. In a 96 well plate,
EGF-stimulated and control A431 cell lysates at concentrations between 1.1 and 1100 ?g/ml
were incubated with the Grb2 SH2 probe. Surprisingly, at all 11 lysate concentrations
tested, EGF-stimulated samples showed a greater signal (lower Ct value) than unstimulated
samples, demonstrating the high sensitivity of the system (limit of detection: ~1 ?g/ml
or 2 ng protein per assay). However, at higher lysate concentrations (300 ?g/ml),
suppressed signal was observed (Fig. 2c upper panel), consistent with other reports using homogeneous proximity ligation
assays 12], 19], 22]. This is often referred to as the “high-dose hook effect” and has been observed in
other antibody based assays 33]–35]. As a result, the assay had a linear signal response range of a 1–2 order of magnitude
(Fig. 2c lower panel). In addition, the slopes of stimulated and unstimulated samples, an
indicator of PCR amplification efficiency, were slightly different suggesting that
conventional ??Ct or standard curve methods are not suitable for relative quantification
of SH2 binding. Therefore, we utilized ProteinAssist, a software tool designed for
fold change estimation in TaqMan Protein Expression Assays based the ?Ct squared method
12]. With this method, the log
₂
fold change between EGF-stimulated and control samples was estimated to be 6.0 – 6.4
(Additional file 1: Figure S3). Based on three independent assays, the intra-assay variation of ?Ct
values was 0.07-2.36 (mean 0.60) and the inter-assay %CV was 0.32-3.08 % (mean 1.28 %).
Taken together, these results suggest that the limit of detection and precision are
favorable, but the linear signal response range is modest, likely due to the binding
characteristics of antibodies and probes.

Application to other pTyr recognition domains

In addition to SH2 domains, members of the phosphotyrosine binding (PTB) and tyrosine
phosphatase (PTP) domain families are also known to recognize phosphotyrosine residues
and play regulatory roles in tyrosine kinase pathways 1], 36], 37]. Considering their potential applications in phosphoproteomics research, we tested
if the same SH2-PLA methodology is applicable to these domains. GST-fusion proteins
of the ShcA PTB domain, wild type PTP1B PTP domain, and catalytically inactive (substrate-trapping)
mutant PTP1B PTP domain were purified and subjected to the assay using the same protocol
as for SH2 domains. As shown in Fig. 2d, ShcA PTB and the catalytically inactive PTP domain of PTP1B showed binding activity
to pEGFR comparable to the Grb2 SH2 domain, while the wild type PTP domain, having
intrinsic PTP activity, showed no binding.

Estimation of EGFR phosphotyrosines at the limit of detection

The serial dilution experiment indicated that the limit of detection is about 2 ng
protein per assay in the case of A431 cells, although this threshold could change
for other cell lines with different EGFR expression levels. A titration experiment
using a “spike-in” pEGFR protein control would address the ambiguity, but preparation
of such a reagent is challenging. Therefore, we took a retrospective approach in which
a series of quantification methods were combined to measure the absolute amount of
EGFR phosphotyrosine in the minimum amount of cell lysate necessary for SH2-PLA.

First, using a baculovirus expression system, we generated recombinant GST fused c-Abl
protein to serve as the standard for phosphotyrosine. Then we treated the Abl protein
with the tyrosine specific phosphatases PTP1B and TC-PTP. Anti-phosphotyrosine blots
of treated and untreated Abl proteins indicated that phosphorylated Abl protein was
mostly dephosphorylated by phosphatase treatment. Following the treatment, we quantified
the amount of free phosphate, which is the hydrolyzed product of phosphotyrosine,
using a phosphate standard curve generated with a malachite green phosphatase assay
(?Abl PO
₄^3?
?=?5.7 pmol/?g, Fig. 3a).

Fig. 3. Estimation of EGFR phosphotyrosines at the limit of detection. To define an absolute
lower limit of detection (LOD) for SH2-PLA, the total amount of EGFR phosphotyrosines
in sample cell lysate was estimated using a phosphotyrosine standard sample and quantitative
dot blotting analyses. a, Recombinant c-Abl protein, the pTyr-standard sample, was treated with tyrosine specific
phosphatases PTP1B and TC-PTP. The amount of hydrolyzed phosphotyrosine was quantified
by malachite green phosphatase assay (?Abl PO
₄^3?
). Left panel shows anti-Abl and anti-phosphotyrosine blots for phosphatase-treated
(Abl PTP+) and -untreated (Abl PTP-) samples. After the PTP treatment, the level of
c-Abl tyrosine phosphorylation was greatly reduced but weak phosphorylation was still
detectable with longer exposure time. The right panel shows a plot of the phosphate
standard used for the quantification. Red circle, untreated c-Abl; blue circle, PTP-treated
c-Abl; yellow circles, the kit supplied phosphate standard. From this analysis, ?Abl
PO
₄³
was estimated to be 5.7 pmol per ?g of the c-Abl protein. b, Quantitative dot blotting. The total pTyr in the EGF-stimulated Cos1 cell lysate
was estimated from a pTyr standard curve generated from anti-phosphotyrosine dot blotting.
Upper panel shows raw anti-Abl and anti-pTyr blots. Serially diluted c-Abl pTyr standard
(left to right 3.1–0.02 ng per spot) and 0.01 ?g EGF-stimulated Cos1 samples were
spotted on nitrocellulose membrane (performed in triplicate). The middle panel shows
the resulting pTyr standard plot with the quantified signal intensities. The pTyr
amount in the EGF-stimulated Cos1 lysate was estimated to be 0.08 pmol per ?g lysate.
Subsequently, an anti-pTyr Western analysis for A431 and Cos1 samples was performed,
relative intensities of the EGFR bands were calculated, and the amount of EGFR pTyr
in the EGF-stimulated A431 sample was estimated to be 0.122 pmol/?g. Thus 2 ng of
EGF-stimulated A431 sample, which is the lower limit for SH2-PLA detection, would
contain 0.243 femtomole EGFR pTyr. See Methods and Additional file 1: Figure S4 for more information

Next, anti-phosphotyrosine dot blotting was performed and the amount of total phosphotyrosine
in EGF-stimulated Cos 1 lysate was estimated using the Abl standard (Fig. 3b). Finally, the tyrosine phosphorylation of EGFR was estimated by comparing the EGFR
band intensity with the whole band intensity on an anti-phosphotyrosine Western blot
(Additional file 1: Figure S4). According to these analyses, we estimated that 0.122 pmol phosphotyrosine
is present on EGFR in 1 ?g of EGF-stimulated A431 cell lysate. Thus in 2 ng lysate,
which is the quantification limit of SH2-PLA (Fig. 2b), there is 0.243 fmol of phosphotyrosine on EGFR (Fig. 3b). In addition, assuming that A431 has 2.5 million EGFR per cell 38]–40] and 150 pg of total protein per cell 41], we estimated that on average 4–5 tyrosines out of 22 putative tyrosine phosphorylated
sites 42] per EGFR molecule are phosphorylated in EGF-stimulated A431 cells (Additional file
1: Figure S4).

Practical limit of detection from cell culture

Since the lysate dilution experiment indicated that the lysate requirement for SH2-PLA
is very low, we next determined the lower limit of the assay by cell numbers. Serially
diluted A431 cells were seeded in a 96-well plate, starved 16 h, and stimulated with
EGF. Cells were lysed in the same volume of buffer and interaction between EGFR and
Vav2 SH2 was analyzed by SH2-PLA. As shown in Fig. 4a, the assay detected EGF dependent SH2 interaction in the lysate equivalent of 16
cells (780 cells per well lysed in 50 ?l) or approximately 2.5 ng which is close to
the detection limit calculated above (Fig. 2c). In addition to EGFR-overexpressing A431 cells, we performed a similar lysate dilution
experiment using EGF stimulated Cos1 cells and found that a lysate concentration approximately
four times higher is required for detection, consistent with modest EGFR phosphorylation
in Cos1 cells (Additional file 1: Figure S4 and data not shown). These results demonstrate that SH2-PLA is capable
of detecting interaction between SH2 domains and pEGFR using a very small number of
cultured cells.

Fig. 4. Applications of SH2-PLA assay. a, Practical limit of detection from cell culture. Two-fold serial dilutions of A431
cells were seeded to wells in a 96-well plate. Image series shows various 10x magnifications
of diluted A431 cells with cell number indicated. Vav2 SH2:pEGFR interaction of starved
or EGF-stimulated cell lysates was quantified by the SH2-PLA assay to resolve the
assay detection limit. The Ct values from real-time PCR are shown with approximate
numbers of cells per culture well or per assay (in brackets) underneath the chart.
b, Time course and dose response of EGF stimulation. A431 and Cos1 cells were starved
and stimulated with EGF at various times and concentrations as indicated. Upper panel
shows far-Western blotting with Grb2 SH2 (25 ?g lysate loaded per lane) and control
blotting with anti-actin. Bottom panel shows Ct values of comparable SH2-PLA experiments
loading 0.4 ?g lysate per assay well. c, Correlation between far-Western and SH2-PLA assay. Using experimental results shown
in B, EGFR band intensities of far-Western blot (X-axis) and average Ct values of SH2-PLA
(Y-axis) in panel B were plotted and showed strong correlation. a.u., arbitrary unit;
r, Pearson correlation coefficient. d, Application of SH2-PLA for cancer tissue analysis. The SH2-PLA/Western/far-Western
analyses were performed using 10 lung cancer tissue samples. Upper panel shows Western
and far-Western results with antibody/probe names indicated on the left. Only one
sample (#3) shows an EGFR size band which also overlapped with bands detected by anti-pTyr
and Grb2 SH2 (far-Western image represents 60-min exposure). The tyrosine phosphorylation
level of the band is similar to the weak phosphorylation of EGFR in unstimulated A431
cells (right panel). The PLA-SH2 results for the same set of lung cancer samples are
shown on the bottom. Consistent with the Grb2 far-Western result, only sample #3 had
significant signal beyond the no protein control (NPC). The BG line indicates the
background Ct value

Correlation between SH2-PLA and far-Western in kinetic analyses

Defining phosphorylation kinetics in growth factor stimulated cells is important in
cell signaling studies. We applied the SH2-PLA approach to determine time-dependent
and dose-dependent changes in SH2 binding to pEGFR using A431 and Cos1 cells (Fig. 4b). For cross-validation, we performed far-Western analysis and compared the results
side by side. As shown in Fig. 4b, the dose- and time-dependent increases in Grb2 SH2 binding to EGFR were obvious
both by far-Western (~180 KDa band on upper panel) and SH2-PLA (Ct value in lower
panel). When corresponding signal values (band intensity and Ct values) of both assays
were compared, there was a high correlation (Pearson coefficient r?=?0.97) (Fig. 4c). These results suggest that SH2-PLA is capable of analyzing cells with various
levels of EGFR expression and is able to produce quantitative results similar to those
obtained from established protein-protein interaction assays (e.g., pull-down and immunoprecipitation) in a significantly shorter assay time while using
10–100 times less sample.

Application of SH2-PLA for cancer tissue analysis

To explore the translational application of SH2-PLA, we analyzed binding of Grb2 SH2
to EGFR in lung cancer patient tissues. Empirically, the tyrosine phosphorylation
analysis of patient-derived solid tumors is challenging due to intrinsically high
protease and phosphatase activities requiring phospho-enrichment and larger starting
material 43]. We asked if SH2-PLA is capable of detecting a specific signal within human lung
cancer tissues without pTyr enrichment. In collaboration with the Haura group at the
Moffit Cancer Center, we obtained 10 non-small cell lung cancer samples in a single-blind
manner (neither molecular nor pathological characteristics were provided). We lysed
the frozen tissues and performed the SH2-PLA assay (with the Grb2 SH2-anti EGFR probe
pair), Western blotting (with anti-phosphotyrosine antibody), and far-Western blotting
(with Grb2 SH2 probe) for comparison (Fig. 4d). In Western and far-Western, only one sample showed tyrosine phosphorylation at
a band corresponding to EGFR size (#3, upper and middle panels). In the SH2-PLA assay
using 0.5 ?g of lysate per assay (equivalent to ~3,000 cells), we also observed a
modest but significant signal (lower Ct value) in the same sample confirming the presence
of tyrosine phosphorylated EGFR (lower panel), even though its level of phosphorylation
was significantly lower than that of A431 cell samples (upper right panel). This result
indicates that SH2-PLA is capable of detecting tyrosine phosphorylated SH2 domain
recognition sites in weakly phosphorylated tumor tissues without requiring phosphopeptide
enrichment.