MARCO METODOLÓGICO Y TEÓRICO
B. MARCO TEÓRICO SOBRE LAS VARIABLES DE ESTUDIO a El microcrédito
Reverse phase protein arrays (RPPAs), first introduced in 2001 [Paweletz et al.,
2001], are a high-throughput technology capable of quantitative measurement of (phosphorylated) protein levels in thousands of biological samples simultaneously. The experimental procedure uses antibodies to detect proteins, and in this respect it is similar to Western blots and ELISA. The essence of the procedure is illustrated in Figure 2.4 and outlined below.
Cells are lysed and the solution is robotically spotted onto a nitrocellulose coated slide. The microarray slide contains many spots in a grid allowing for many samples to be immobilised and tested simultaneously, usually in replicate and using
PROTEIN MICROARRAY FORMATS
Protein microarray formats can be divided into two major classes: forward phase arrays and reverse phase arrays (RPAs).1In the forward phase array format, the analyte(s) of
interest is captured from the solution phase by a capture molecule, usually an antibody, that is immobilized on a sub- stratum and acts as bait molecule (1, 2) (Fig. 1). In a forward phase array, each spot contains one type of immobilized antibody or bait protein. Each array is incubated with one test sample such as a cellular lysate or serum sample representing a specific treatment condition, and multiple analytes from that sample are measured simultaneously. In contrast, the RPA format immobilizes an individual complex test sample in each array spot such that an array is comprised of hundreds of different patient samples or cellular lysates. In the RPA format, each array is incubated with one detection protein (i.e.anti- body), and a single analyte end point is measured and directly compared across multiple samples (17, 24, 26 –29) (Fig. 1). Probing multiple arrays spotted with the same lysate concom- itantly with different phosphospecific antibodies provides the
effect of generating a multiplex readout. Efforts are ongoing in our laboratory to multiplex the arrays even further through the use of dual color infrared dye-labeled antibodies as well as quantum dots. Using these technologies, it is hoped that multiple analytes can be measured on the same spot on the same array (30, 31). The utility of reverse phase protein mi- croarrays lies in their ability to provide a map of known cell signaling proteins. Identification of critical nodes, or interac- tions, within the network is a potential starting point for drug development and/or the design of individual therapy regimens (21, 22). The array format is also amenable to extremely sensitive analyte detection (Fig. 2) with detection levels ap- proaching attogram amounts of a given protein and variances of less than 10% (1, 32). Detection ranges could be substan- tially lower in a complex mixture such as a cellular lysate; however, the sensitivity of the RPAs is such that low abun- dance phosphorylated isoforms can still be measured from a spotted lysate amount of less than 10 cell equivalents. This level of sensitivity combined with analytical robustness is critical if the starting input material is only a few hundred cells from a biopsy specimen.
The reverse phase protein array has demonstrated a unique ability to analyze signaling pathways using small numbers of cultured cells or cells isolated by laser capture microdissec- tion from human tissue procured during clinical trials (17, 24, 26, 27). Using this approach, microdissected pure cell popu- lations are taken from human biopsy specimens, and a protein lysate is arrayed onto nitrocellulose-coated slides (Fig. 3). Key technological components of this method offer unique advan- tages over tissue arrays (33) or antibody arrays (34, 35). First the RPA can use denatured lysates so that antigen retrieval, which is a large limitation for tissue arrays, is not problematic. Protein microarrays can also consist of non-denatured lysates derived directly from microdissected tissue cells so that pro- tein-protein, protein-DNA, and/or protein-RNA complexes can be detected and characterized. Each patient sample is printed on the array in serial dilutions, providing an internal standard. When an internal reference standard of known and fixed amounts of the analyte are applied to the same array, a direct and quantitative measurement of the phosphorylated end point can be attained within the linear dynamic range of the assay. Finally RPAs do not require direct labeling of the patient sample as a readout for the assay, which provides a marked improvement in reproducibility, sensitivity, and ro- bustness of the assay over other techniques (36).
The RPA platform has been used to explore a variety of signaling pathways involved in malignant progression and tumor biology (17, 26 –29, 37). For example, in a study of prostate tissue, pathway profiling of microdissected cells from normal, stroma, and prostate tumors revealed the preliminary finding that activation of protein kinase C␣is down-modu- lated in prostate cancer progression (26). If validated, this finding could have profound effects on the rationale behind some current therapies (38) and illustrates the importance of
1
The abbreviation used is: RPA, reverse phase array.
FIG. 1.Classes of protein microarray technology.Forward phase
arrays (top) immobilize a bait molecule such as an antibody designed to capture specific analytes with a mixture of test sample proteins. The bound analytes are detected by a second sandwich antibody or by labeling the analyte directly (upper right). Reverse phase arrays immobilize the test sample analytes (e.g.lysate from laser capture microdissected cells) on the solid phase. An analyte-specific ligand (e.g.antibody;lower left) is applied in solution phase. Bound antibod- ies are detected by secondary tagging and signal amplification (lower
right).
Reverse Phase Arrays for Molecular Network Analysis
Figure 2.4: Protein microarrays. (a) Forward phase protein array. Protein
antibodies are immobilised onto the slide which capture specific (phospho)proteins of interest (analytes) from cell lysate (a single sample). The analytes are detected using a sandwich antibody (as in ELISA), together with a labelled secondary antibody, or the analyte can be labelled directly to allow detection. (b) Reverse phase protein array. Cell lysates (multiple samples) are spotted onto the microarray slide, which is then probed with a single (phospho)protein specific primary antibody. A labelled secondary antibody is used to detect the primary antibody. Multiple slides can be used, each probed with different antibodies, to detect different proteins of interest.
Figure reproduced from Sheehanet al.[2005].
dilution series (we explain dilution series further below). The array is then incubated with a primary antibody that binds specifically to the phosphoprotein of interest. This antibody is then detected using a labelled secondary antibody (as in ELISA and Western blot) and signal amplification. The emitted signal is quantified using a
software package. Full details of RPPA protocol can be found in Tibeset al. [2006]
and Hennessyet al.[2010]. If multiple microarray slides are spotted simultaneously
with the same samples, each slide can be probed with a different antibody, thereby providing readouts for multiple samples and multiple proteins [see e.g. Sheehan et al., 2005].
The samples are spotted onto the array in dilution series. Protein and phos- phoprotein concentrations can vary greatly, so accurate measurements over a wide dynamic range are required. The dynamic range of measurements is extended by diluting each sample several times and spotting onto the array at each dilution step. Hence, if the protein concentration in the original undiluted sample is near saturation, it can still be detected in the diluted samples. Dilution series also aid the accurate quantification of protein concentrations. Quantification is usually car- ried out using response curves, that relate the observed signal intensities to the (phospho)protein concentrations. The fact that a single antibody is used for the whole slide motivates the use of a single response curve for all samples on the slide. For the RPPA data used in this thesis, a logistic model was used for the response
curve (R package ‘SuperCurve’ developed by the Department of Bioinfomatics and
Computational Biology in MD Anderson Cancer Center [Huet al., 2007]).
There are two main types of protein microarrays, of which RPPAs are one. ‘Reverse phase’ refers to the fact that the cell lysate is immobilised on the slide and the array is probed with an antibody, which is a reversal of the procedure for antibody arrays, the other type of protein microarray. Antibody arrays are also referred to as forward phase protein arrays. For the antibody array, antibodies are immobilised onto the slide which capture proteins of interest from cell lysate. Each array spot contains a single type of antibody and the array is incubated with one sample only, providing readouts for multiple proteins from one sample (see Figure 2.4).
RPPAs are an emerging technology that enable measurements for a single (phospho)protein of interest to be obtained for hundreds to thousands of samples in a fast, automated, quantitative and economical manner. Multiple arrays can be used to probe for multiple (phospho)proteins; tens of proteins are often measured in the same experiment, providing an advantage over flow cytometry. The technique is also highly sensitive, requiring very small amounts of sample to enable detection of
analytes; only 103cells are required for an RPPA experiment, compared with 108for
mass spectrometry and 105 for Western blotting. Therefore, while mass spectrome-
try is a promising approach, RPPA is currently more sensitive and cost-effective. An advantage of RPPAs over antibody arrays is that either fewer antibodies are required or samples do not need to be directly labelled to allow detection of the analyte of in- terest. This can improve robustness and reproducibility of results. Also, RPPAs can use denatured lysates (proteins in the lysate have lost their three-dimensional con- formation) which can allow antibodies to bind that previously would not have been able to do so, providing an advantage over tissue microarrays (another reverse-phase assay).
The main limitation of RPPAs is specificity of primary and secondary an- tibodies. The signal from a microarray spot could be due to cross-reactivity from unspecific binding and it is not possible to determine if this is the case from the RPPA results themselves. Therefore antibodies have to be carefully validated by
Western blotting prior to their use in RPPA assays. Hennessy et al. [2010] is an
example of such a validation study. The number of available validated antibodies is continuously growing.
RPPAs have been used in many studies to investigate cancer cell signalling,
both in cancer cell lines [Tibeset al., 2006] and in primary tumour samples [Sheehan
et al., 2005]. These studies include the profiling and comparison of active signalling pathways in different contexts; for example, between primary and metastatic tu-
mours [Sheehan et al., 2005] or between cancer subtypes [Boyd et al., 2008], the
identification of signalling biomarkers that are predictive of response to certain anti-
cancer agents [Boyd et al., 2008], the identification of optimal drug combinations
[Iadevaiaet al., 2010] and structure learning of signalling networks [Bender et al.,
2010]. For further studies see, for example, Spurrier et al. [2008]; Hu et al. [2007]
and references therein. RPPAs have promising utility in the development of per- sonalised therapies; using RPPAs to investigate and compare signalling profiles in patient tumour cells and normal cells, and to monitor changes in phosphorylation through time, both pre- and post-treatment, could provide information that guides the discovery and application of targeted therapies. Indeed, RPPAs are currently
involved in several clinical trials [Muelleret al., 2010].
In Chapter 4, RPPA data for 20 phosphoproteins from a breast cancer cell line are used for structure learning of a signalling network, and in Chapter 5 RPPA data for 39 phosphoproteins across 43 breast cancer cell lines covering two breast cancer subtypes are used for clustering and network structure learning.