Ann. N.Y. Acad. Sci. ISSN 0077-8923
A N N A L S O F T H E N E W Y O R K A C A D E M Y O F SC I E N C E S Issue: Companion Diagnostics
Functional proteomic biomarkers in cancer Banafshe Larijani,1 Michela Perani,2 Kholoud Alburai’si,2 and Peter J. Parker2,3 1
Cell Biophysics Laboratory, Ikerbasque, Basque Foundation for Science and Unidad de Biof´ısica (CSIC-UPV/EHU), Leioa, Spain. 2 King’s College London, Guy’s Campus, London, United Kingdom. 3 London Research Institute Cancer Research UK, Lincoln’s Inn Fields, London, United Kingdom Address for correspondence: Prof. Peter J. Parker, New Hunt’s House, Guy’s Campus, London SE1 1UL, UK. [email protected]
Beyond penetrant germline and somatic mutations, there are substantial challenges in extrapolating phenotypes from linear DNA sequences and transcriptomics. This brings a molecular pathology emphasis to the properties of the main players responsible for executing actions, proteins. The proteomic attribute most frequently determined in pathology is (relative) content, but for many candidate biomarkers this is not the most important feature to understand. In keeping pace with the depth of knowledge of the mechanisms underlying pathologies, we need to ask more sophisticated questions about the state of proteins, for example, their oligomerization status, modification status, and location. This demands hitherto nonroutine approaches to proteomics, which we will discuss in this brief perspective. Keywords: cancer proteomic biomarkers; protein states; coincidence detection; FRET-FLIM; a-FRET
Introduction The era of genomic medicine has witnessed the accumulation of a wealth of genomic and transcriptomic patient data. These data have accrued on the back of advances in our genomics technology capabilities that has seen the time and costs for DNA sequencing reduced from years and billions to days and approaching $1000.1 The data accumulated in this genomics tsunami have provided extraordinary insights and advances in our understanding of the genetics of disorders (familial diseases) and a greatly enriched, if complex, panorama and predicted etiology for somatic diseases, that is, cancers (reviewed in Ref. 2). In cancer genomics, the identification of some highly penetrant mutations has provided a powerful validation of targets for which novel therapeutics are already in the clinic (e.g., see recent reviews on Raf inhibitors3 and Alk inhibitors4 ). More broadly, the genomic and transcriptomic profiles in particular tumor types have given rise to prognostic markers informing on likely outcomes, e.g., Onco type DX , MammaPrint , and ProsignaTM in breast cancer (recently reviewed, see Ref. 5). Such R
R
prognostic signatures are exactly that, signatures, with no known (as yet) linkage to mechanistic drivers that determine pathological behavior. It is pertinent to emphasize the distinction between signatures (correlative markers) and biomarkers (which can be defined as having a traceable mechanistic impact on the pathology observed). Thus, for example, phosphatase and tensin homolog (PTEN), once identified as a tumor suppressor gene on 10q23 displaying loss of expression in association with advanced cancers (see Ref. 6), might once have been a useful signature, but has since become a biomarker informed by a traceable mechanistic pathway involving increased output of the phosphoinositide 3 (PI3)-kinase pathway (through loss of PTEN’s lipid phosphatase function7 ). In distinguishing signatures and biomarkers, it is evident that there is a need to define the activities of the protein players that are proximal to the pathology, and so establish proteomic readouts that are directly informative on pathology. Typically, routine proteomics applied to tumor samples rely on immunohistochemistry (IHC) to determine relative protein content, or on extraction and processing (e.g., enzyme-linked immunosorbant
doi: 10.1111/nyas.12749 C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
1
Functional cancer proteomes
Larijani et al.
assay (ELISA)). Although conventional IHC minimizes manipulation and retains tissue organization, it is limited in specificity (one-site assay); by contrast ELISA-type approaches can bring specificity (twosite assays) but require processing (challenging on sample size) and destroy spatial information. More experimental, hardware-demanding, approaches, such as imaging mass spectrometry (see Ref. 8), provide a distinct proteomic view, although in this specific case coverage is severely limited by the abundance of the proteins observable. These constraints on proteomics are intensified for questions more demanding than simply concentration; but we would argue that it is these demanding questions we need to address. Functional proteomics Functional proteomics, as intended here, refers to the state of proteins beyond their simple levels of expression. This includes the localization (and tissue/tumor heterogeneity) of the protein, where this is relevant to action. For example, the presence of cytochrome C in the cytosol, as opposed to the mitochondrion, is associated with the assembly of the apoptosome, consequent pro-caspase9 activation, and apoptosis.9 Defining the cytoplasmic location of cytochrome C, therefore, is informative of its action in driving apoptosis. The oligomerization state of a protein is another informative property that can define a functional state. This is evident in many cell surface receptors where ligand binding leads to the assembly of complexes. In the case of the epidermal growth factor receptor (EGFR) family, receptor homo- and heterodimers, and the consequent intracellular assembly of multiple effector complexes (in different compartments), reflect ligand induced activation events (recently reviewed, see Ref. 10). Posttranslational modifications are also functional attributes critically informative of protein function. This is well understood for protein phosphorylation, which is frequently exploited to switch on or off cell functions. Addressing EGFR family functions again, the routine analysis of activation is afforded by monitoring autophosphorylation with site-specific reagents to defined receptor tyrosine phosphorylation sites.11 These functional proteomic properties are analyzed with varying degrees of facility in experimental systems, where manipulation of conditions permits 2
1 Ligation
2
Ag1 Ag2
Figure 1. The principles of the proximity ligation assay are illustrated. The two antigens in a complex (Ag1/2) are detected by species-different primary antisera (purple and yellow). These in turn are bound by species-specific secondary Fab fragment ligated to oligonucleotides (cyan and green). Hybridization of these to a pair of complementary oligonucleotides (1 and 2; black) provides the templates that enable ligation to take place to generate a circular DNA fragment. This can be amplified by rolling circle amplification and then detected with a complementary fluorescent oligonucleotide probe. Components illustrated are not to scale.
exquisite specificity to be brought to bear on the analyses. However, to specify these functional attributes in biopsies is many-fold more difficult. So, what options do we have for such analyses? Proximity ligation assay Proximity ligation assay (PLA) refers to a coincidence recognition technology that enables the detection of two antibodies in close proximity (reviewed in Ref. 12). The marketed system from Olink Bioscience employs the covalent modification of secondary antibody reagents with oligonucleotides that when in physical proximity (e.g., antibody bound on a tissue section; distance limits 40 nm) will permit the coincident association of a pair of circle forming oligonucleotides that can then be ligated and used as a template for rolling circle amplification and fluorescence detection (Fig. 1; also see http://www. olink.com/products/duolink/). In a commercial development of the technology the conjugation of the PLA oligonucleotide arms directly to primary antibodies of choice reduces the theoretical maximal distance to approximately 28 nm (see www. olink.com/sites/default/files/files/FAQ˙Duolink˙ Probemaker.pdf). This technology has been employed extensively in discovery science (hundreds of publications) across many areas of research from cell survival,13 through cell migration,14 to cell death.15 It has been employed also in the context of biomarker validation (see Ref. 16), and has been used for functional proteomics in paraffin embedded breast
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
Larijani et al.
Functional cancer proteomes
Figure 2. Amplified-FRET–based proximity assay. The scheme illustrates the principles of dual recognition of the kinase Akt (Site1) and its activation loop phosphorylation site (Site2; T308) using amplified FRET. The two primary antisera are detected with species-specific Fab fragments that are labeled with oregon488 (ORG488) or horseradish peroxidase (HRP), as indicated. HRP is exposed to the substrate tyramide-alexa594 (TSA-ALX594) for a limited time to create an enrichment of acceptor fluorophore in proximity to the HRP. The ensuing FRET can be detected by changes in the lifetime of the donor (ORG488) excited state.
cancer tissues.17,18 In this last set of studies, the assembly of human epidermal growth factor receptor 2 (HER2) homotypic and heterotypic receptor complexes and effector complexes is described. It is reassuring to note that the extent of HER2– HER2 dimer formation correlated very strongly with HER2 amplification18 consistent with the view that upregulation of HER2 is a cancer driver. This PLA technology has gained a number of proponents as the technical process has become more user friendly. However, there are limitations relating to the distances for which standard PLA, or directly conjugated antibody probes, can function as a proximity marker (28–42 assuming an antibody is 7 nm19 ). The relatively long upper distance limit (compared to the 24 nm constraint on FRET processes making the same antibody assumption; see below) limits the degree to which specific conclusions might be drawn. For example, a positive proximity readout by PLA could be reporting on membrane domain clustering of receptors as much as it does on complex formation. With respect to quantitation, the counting of dots quantified independently of the intensity can pose problems for measurements in crowded fields where there are significant, albeit not insurmountable, problems (see Ref. 20).
FRET-based imaging A photophysical event ideal for the detection of two coincident events is F¨orster resonance energy transfer (FRET). This refers to a fluorescence property of fluorophores manifest when a fluorophore (donor) is in proximity with a second (acceptor) fluorophore––ideal for a two-site assay (10 nm proximity for the fluorophores; accounting for antibodies, this amounts to an operating distance limit of 24 nm). The covalent modification of primary antisera with suitably chosen fluorophores provides a direct route to coincidence detection of appropriately paired antisera through FRET approaches. Studies have shown that the functional status of EGFR determined by FRET using high throughput fluorescent lifetime imaging (FLIM; see below) in head and neck tumor arrays, provides more useful prognostic information than measuring EGFR expression levels using conventional IHC alone.21 Two main types of determinations of FRET are possible. These are steady state (intensity-based FRET) and time-resolved (detection of chromophore lifetime; FLIM) measurements; timeresolved measurements yield additional information about FRET. One of the major advantages of measuring fluorescence lifetime is that it is independent of fluorophore concentration, unlike
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
3
Functional cancer proteomes
Larijani et al.
Signal readout
DNA consensus sequence DM1
DM2
L2 L1
RM2
RM1
P Figure 3. The principle of the coincidence biodetector. The biodetector has the ability to generate a specific signal only when the detection modules (DM1 and DM2 ) simultaneously bind their DNA consensus sequence to form a heterotrimeric complex with their cognate DNA used as the signal readout. The signal can be generated only when the two recognition modules (RM1 and RM2 ) bind their specific epitope targets in close proximity. The linkers (L1 and L2 ) can modulate the distance reached between the two target epitopes.
steady-state measurements, which are intensity based.22,23 In contrast to steady-state measurements, where a change in intensity of the recorded emission would be observed and thus becomes prone to artefacts that lead to overestimation of FRET efficiency, FRET detected by FLIM is a highly effective method of quantifying events. The phosphorylation status of intracellular molecules in archived formalin-fixed paraffin-embedded tumor tissue has been monitored by FLIM, exploiting the coincidence detection to drive the specificity of the readout, a particular problem with phosphorylation site epitopes.21,24,25 In a recent development on this theme, a two-site amplified FRET (a-FRET) platform has been implemented. Importantly, this enables the use of unmodified antibodies, while allowing greater signal generation with minimal noise (Fig. 2). Detection in this context derives from the use of a repertoire of species-specific fluorescently-labeled secondary anti-IgG Fab fragments. This development provides
4
an application that could be deployed routinely on fixed tissues by users applying their “favorite” unmodified antisera. In a recently published work using a-FRET, it was shown that dysregulation of Akt/PKB, is associated with poor prognosis in several human tumors.26 This technology provides perhaps the best quantitation and also the most rigorous assessment of proximity, with the latter limited to 24 nanometers. However, it is the case that the hardware required to generate data using FRET–FLIM is more demanding than other approaches. Alternative coincidence detection methods There are a number of additional coincidence methods that can be referred to as complementation approaches. This is the case for biomolecular fluorescence complementation (BiFC; recently reviewed in Ref. 27) and enzyme fragment complementation, such as applied to luciferase (reviewed in
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
Larijani et al.
Ref. 28). However, these approaches have not been applied to the analysis of biopsies and are more typically and indeed effectively used in imaging modalities. A further technology affording coincidence biodetection, one that provides an additional method applied to bring specificity to the analysis of tissue sections, relies upon oligonucleotide capture. The technique encompasses entirely recombinant protein biodetectors produced in bacterial expression systems that either recognize and visualize posttranslational modifications on specific proteins or can identify two proteins as part of a complex within their normal cell/tissue context. The technology is based on the properties of nuclear receptors29 (detection modules; DM) and their ability to form heterotrimeric complexes with their target double stranded DNA sequences (DNA consensus sequence). The requirements for coincident binding are two distinct affinity probes (recognition modules, RM), which bind two different sites/epitopes in the same target molecule, or in two target proteins in close proximity. Thus, the overall purpose is achieved by two recombinant proteins, each comprising a DM and a RM joined by a linker (L) (see Fig. 3). The linkers can be modulated in size spanning theoretically between just a few nanometers (
A N N A L S O F T H E N E W Y O R K A C A D E M Y O F SC I E N C E S Issue: Companion Diagnostics
Functional proteomic biomarkers in cancer Banafshe Larijani,1 Michela Perani,2 Kholoud Alburai’si,2 and Peter J. Parker2,3 1
Cell Biophysics Laboratory, Ikerbasque, Basque Foundation for Science and Unidad de Biof´ısica (CSIC-UPV/EHU), Leioa, Spain. 2 King’s College London, Guy’s Campus, London, United Kingdom. 3 London Research Institute Cancer Research UK, Lincoln’s Inn Fields, London, United Kingdom Address for correspondence: Prof. Peter J. Parker, New Hunt’s House, Guy’s Campus, London SE1 1UL, UK. [email protected]
Beyond penetrant germline and somatic mutations, there are substantial challenges in extrapolating phenotypes from linear DNA sequences and transcriptomics. This brings a molecular pathology emphasis to the properties of the main players responsible for executing actions, proteins. The proteomic attribute most frequently determined in pathology is (relative) content, but for many candidate biomarkers this is not the most important feature to understand. In keeping pace with the depth of knowledge of the mechanisms underlying pathologies, we need to ask more sophisticated questions about the state of proteins, for example, their oligomerization status, modification status, and location. This demands hitherto nonroutine approaches to proteomics, which we will discuss in this brief perspective. Keywords: cancer proteomic biomarkers; protein states; coincidence detection; FRET-FLIM; a-FRET
Introduction The era of genomic medicine has witnessed the accumulation of a wealth of genomic and transcriptomic patient data. These data have accrued on the back of advances in our genomics technology capabilities that has seen the time and costs for DNA sequencing reduced from years and billions to days and approaching $1000.1 The data accumulated in this genomics tsunami have provided extraordinary insights and advances in our understanding of the genetics of disorders (familial diseases) and a greatly enriched, if complex, panorama and predicted etiology for somatic diseases, that is, cancers (reviewed in Ref. 2). In cancer genomics, the identification of some highly penetrant mutations has provided a powerful validation of targets for which novel therapeutics are already in the clinic (e.g., see recent reviews on Raf inhibitors3 and Alk inhibitors4 ). More broadly, the genomic and transcriptomic profiles in particular tumor types have given rise to prognostic markers informing on likely outcomes, e.g., Onco type DX , MammaPrint , and ProsignaTM in breast cancer (recently reviewed, see Ref. 5). Such R
R
prognostic signatures are exactly that, signatures, with no known (as yet) linkage to mechanistic drivers that determine pathological behavior. It is pertinent to emphasize the distinction between signatures (correlative markers) and biomarkers (which can be defined as having a traceable mechanistic impact on the pathology observed). Thus, for example, phosphatase and tensin homolog (PTEN), once identified as a tumor suppressor gene on 10q23 displaying loss of expression in association with advanced cancers (see Ref. 6), might once have been a useful signature, but has since become a biomarker informed by a traceable mechanistic pathway involving increased output of the phosphoinositide 3 (PI3)-kinase pathway (through loss of PTEN’s lipid phosphatase function7 ). In distinguishing signatures and biomarkers, it is evident that there is a need to define the activities of the protein players that are proximal to the pathology, and so establish proteomic readouts that are directly informative on pathology. Typically, routine proteomics applied to tumor samples rely on immunohistochemistry (IHC) to determine relative protein content, or on extraction and processing (e.g., enzyme-linked immunosorbant
doi: 10.1111/nyas.12749 C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
1
Functional cancer proteomes
Larijani et al.
assay (ELISA)). Although conventional IHC minimizes manipulation and retains tissue organization, it is limited in specificity (one-site assay); by contrast ELISA-type approaches can bring specificity (twosite assays) but require processing (challenging on sample size) and destroy spatial information. More experimental, hardware-demanding, approaches, such as imaging mass spectrometry (see Ref. 8), provide a distinct proteomic view, although in this specific case coverage is severely limited by the abundance of the proteins observable. These constraints on proteomics are intensified for questions more demanding than simply concentration; but we would argue that it is these demanding questions we need to address. Functional proteomics Functional proteomics, as intended here, refers to the state of proteins beyond their simple levels of expression. This includes the localization (and tissue/tumor heterogeneity) of the protein, where this is relevant to action. For example, the presence of cytochrome C in the cytosol, as opposed to the mitochondrion, is associated with the assembly of the apoptosome, consequent pro-caspase9 activation, and apoptosis.9 Defining the cytoplasmic location of cytochrome C, therefore, is informative of its action in driving apoptosis. The oligomerization state of a protein is another informative property that can define a functional state. This is evident in many cell surface receptors where ligand binding leads to the assembly of complexes. In the case of the epidermal growth factor receptor (EGFR) family, receptor homo- and heterodimers, and the consequent intracellular assembly of multiple effector complexes (in different compartments), reflect ligand induced activation events (recently reviewed, see Ref. 10). Posttranslational modifications are also functional attributes critically informative of protein function. This is well understood for protein phosphorylation, which is frequently exploited to switch on or off cell functions. Addressing EGFR family functions again, the routine analysis of activation is afforded by monitoring autophosphorylation with site-specific reagents to defined receptor tyrosine phosphorylation sites.11 These functional proteomic properties are analyzed with varying degrees of facility in experimental systems, where manipulation of conditions permits 2
1 Ligation
2
Ag1 Ag2
Figure 1. The principles of the proximity ligation assay are illustrated. The two antigens in a complex (Ag1/2) are detected by species-different primary antisera (purple and yellow). These in turn are bound by species-specific secondary Fab fragment ligated to oligonucleotides (cyan and green). Hybridization of these to a pair of complementary oligonucleotides (1 and 2; black) provides the templates that enable ligation to take place to generate a circular DNA fragment. This can be amplified by rolling circle amplification and then detected with a complementary fluorescent oligonucleotide probe. Components illustrated are not to scale.
exquisite specificity to be brought to bear on the analyses. However, to specify these functional attributes in biopsies is many-fold more difficult. So, what options do we have for such analyses? Proximity ligation assay Proximity ligation assay (PLA) refers to a coincidence recognition technology that enables the detection of two antibodies in close proximity (reviewed in Ref. 12). The marketed system from Olink Bioscience employs the covalent modification of secondary antibody reagents with oligonucleotides that when in physical proximity (e.g., antibody bound on a tissue section; distance limits 40 nm) will permit the coincident association of a pair of circle forming oligonucleotides that can then be ligated and used as a template for rolling circle amplification and fluorescence detection (Fig. 1; also see http://www. olink.com/products/duolink/). In a commercial development of the technology the conjugation of the PLA oligonucleotide arms directly to primary antibodies of choice reduces the theoretical maximal distance to approximately 28 nm (see www. olink.com/sites/default/files/files/FAQ˙Duolink˙ Probemaker.pdf). This technology has been employed extensively in discovery science (hundreds of publications) across many areas of research from cell survival,13 through cell migration,14 to cell death.15 It has been employed also in the context of biomarker validation (see Ref. 16), and has been used for functional proteomics in paraffin embedded breast
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
Larijani et al.
Functional cancer proteomes
Figure 2. Amplified-FRET–based proximity assay. The scheme illustrates the principles of dual recognition of the kinase Akt (Site1) and its activation loop phosphorylation site (Site2; T308) using amplified FRET. The two primary antisera are detected with species-specific Fab fragments that are labeled with oregon488 (ORG488) or horseradish peroxidase (HRP), as indicated. HRP is exposed to the substrate tyramide-alexa594 (TSA-ALX594) for a limited time to create an enrichment of acceptor fluorophore in proximity to the HRP. The ensuing FRET can be detected by changes in the lifetime of the donor (ORG488) excited state.
cancer tissues.17,18 In this last set of studies, the assembly of human epidermal growth factor receptor 2 (HER2) homotypic and heterotypic receptor complexes and effector complexes is described. It is reassuring to note that the extent of HER2– HER2 dimer formation correlated very strongly with HER2 amplification18 consistent with the view that upregulation of HER2 is a cancer driver. This PLA technology has gained a number of proponents as the technical process has become more user friendly. However, there are limitations relating to the distances for which standard PLA, or directly conjugated antibody probes, can function as a proximity marker (28–42 assuming an antibody is 7 nm19 ). The relatively long upper distance limit (compared to the 24 nm constraint on FRET processes making the same antibody assumption; see below) limits the degree to which specific conclusions might be drawn. For example, a positive proximity readout by PLA could be reporting on membrane domain clustering of receptors as much as it does on complex formation. With respect to quantitation, the counting of dots quantified independently of the intensity can pose problems for measurements in crowded fields where there are significant, albeit not insurmountable, problems (see Ref. 20).
FRET-based imaging A photophysical event ideal for the detection of two coincident events is F¨orster resonance energy transfer (FRET). This refers to a fluorescence property of fluorophores manifest when a fluorophore (donor) is in proximity with a second (acceptor) fluorophore––ideal for a two-site assay (10 nm proximity for the fluorophores; accounting for antibodies, this amounts to an operating distance limit of 24 nm). The covalent modification of primary antisera with suitably chosen fluorophores provides a direct route to coincidence detection of appropriately paired antisera through FRET approaches. Studies have shown that the functional status of EGFR determined by FRET using high throughput fluorescent lifetime imaging (FLIM; see below) in head and neck tumor arrays, provides more useful prognostic information than measuring EGFR expression levels using conventional IHC alone.21 Two main types of determinations of FRET are possible. These are steady state (intensity-based FRET) and time-resolved (detection of chromophore lifetime; FLIM) measurements; timeresolved measurements yield additional information about FRET. One of the major advantages of measuring fluorescence lifetime is that it is independent of fluorophore concentration, unlike
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
3
Functional cancer proteomes
Larijani et al.
Signal readout
DNA consensus sequence DM1
DM2
L2 L1
RM2
RM1
P Figure 3. The principle of the coincidence biodetector. The biodetector has the ability to generate a specific signal only when the detection modules (DM1 and DM2 ) simultaneously bind their DNA consensus sequence to form a heterotrimeric complex with their cognate DNA used as the signal readout. The signal can be generated only when the two recognition modules (RM1 and RM2 ) bind their specific epitope targets in close proximity. The linkers (L1 and L2 ) can modulate the distance reached between the two target epitopes.
steady-state measurements, which are intensity based.22,23 In contrast to steady-state measurements, where a change in intensity of the recorded emission would be observed and thus becomes prone to artefacts that lead to overestimation of FRET efficiency, FRET detected by FLIM is a highly effective method of quantifying events. The phosphorylation status of intracellular molecules in archived formalin-fixed paraffin-embedded tumor tissue has been monitored by FLIM, exploiting the coincidence detection to drive the specificity of the readout, a particular problem with phosphorylation site epitopes.21,24,25 In a recent development on this theme, a two-site amplified FRET (a-FRET) platform has been implemented. Importantly, this enables the use of unmodified antibodies, while allowing greater signal generation with minimal noise (Fig. 2). Detection in this context derives from the use of a repertoire of species-specific fluorescently-labeled secondary anti-IgG Fab fragments. This development provides
4
an application that could be deployed routinely on fixed tissues by users applying their “favorite” unmodified antisera. In a recently published work using a-FRET, it was shown that dysregulation of Akt/PKB, is associated with poor prognosis in several human tumors.26 This technology provides perhaps the best quantitation and also the most rigorous assessment of proximity, with the latter limited to 24 nanometers. However, it is the case that the hardware required to generate data using FRET–FLIM is more demanding than other approaches. Alternative coincidence detection methods There are a number of additional coincidence methods that can be referred to as complementation approaches. This is the case for biomolecular fluorescence complementation (BiFC; recently reviewed in Ref. 27) and enzyme fragment complementation, such as applied to luciferase (reviewed in
C 2015 New York Academy of Sciences. Ann. N.Y. Acad. Sci. xxxx (2015) 1–6
Larijani et al.
Ref. 28). However, these approaches have not been applied to the analysis of biopsies and are more typically and indeed effectively used in imaging modalities. A further technology affording coincidence biodetection, one that provides an additional method applied to bring specificity to the analysis of tissue sections, relies upon oligonucleotide capture. The technique encompasses entirely recombinant protein biodetectors produced in bacterial expression systems that either recognize and visualize posttranslational modifications on specific proteins or can identify two proteins as part of a complex within their normal cell/tissue context. The technology is based on the properties of nuclear receptors29 (detection modules; DM) and their ability to form heterotrimeric complexes with their target double stranded DNA sequences (DNA consensus sequence). The requirements for coincident binding are two distinct affinity probes (recognition modules, RM), which bind two different sites/epitopes in the same target molecule, or in two target proteins in close proximity. Thus, the overall purpose is achieved by two recombinant proteins, each comprising a DM and a RM joined by a linker (L) (see Fig. 3). The linkers can be modulated in size spanning theoretically between just a few nanometers (
