Skip to main content

Systems biology approaches to cervical pre-cancer and cancer

Final Report Summary - SYSTEMCERV (Systems biology approaches to cervical pre-cancer and cancer)

Executive Summary:
SYSTEMCERV, Systems biology approaches to cervical pre-cancer and cancer, is a SME targeted collaborative research project funded through the European Union Seventh Framework Programme Health programme [Grant number 306037]. The project brings together a multidisciplinary team of four partners from three countries. The consortium together has the combined expertise in system biology, bioinformatics, protein chemistry, molecular biology, label free biosensor technology and clinical expertise to deliver a novel solution for use in cervical cancer screening and diagnostics

The overall aim of SYSTEMCERV is to develop and to validate, a system biology based proto-array for biomarker screening in cervical cancer and pre-cancer to improve clinical diagnosis, and management of disease. The project adopts a systems biology approach to cervical cancer screening through; developing new panels of biomarkers for use in cervical cancer screening; validation of these biomarker panels on clinical samples; development and testing of an integrated DNA protein copying technology with label free biosensor detection technology.

This project is divided into three scientific work packages, one management work package and one work package focusing on dissemination and exploitation. All work packages are integrated and involve contributions from all partners. Within the project using next generation sequencing technology and signal transduction pathway activity profiling technology, we have identified novel pathways and biomarkers implicated in cervical carcinogenesis. Combining these new datasets with existing published gene expression and protein expression data have identified selected panels of novel biomarkers which have been evaluated on over 100 cervical cancer and pre-cancer cases using standard immunohistochemistry approaches. Three of the biomarkers show promising results from clinical validation as markers of CIN progression.

In parallel, SYSTEMCERV has developed and tested an integrated demonstrator device which integrates ALU-FR DNA protein copying technology with Biametrics proprietary label free iRIfS technology and is capable of screening microarrays. The demonstrator device has been validated and characterised using several well established biological systems [avidin/biotin, DNA/Thrombin, and b2 Microglobin/anti-microglobin].

While the project achieved many of its original objectives as outline above, despite several attempts we were not able to produce reliable recombinant antibodies [using phage display technology] for our selected biomarker panels. This impacted on several aspects of the research programme, in that we were not able to produce the SYSTEMCERV proto-array for validation of cervical pre-cancer clinical specimens on the SYSTEMCERV demonstrator device.

Notwithstanding this the project has several successes; a novel panel of system biology derived biomarkers for cervical screening has been identified; clinical validation of the biomarker panels has been performed on over 100 cases of cervical pre-cancer; an integrated demonstrator device incorporating proto-arrays and label free detection technology has been developed and characterised.

Another major success of SYSTEMCERV has been its investment in people with the funding of three post-doctoral positions, four technicians, one engineer, 5 PhD students, and 3 MSc students. We have also been active in relation to dissemination activities with several representations at local and International conferences. Furthermore, through our collaborative and networking efforts, we are currently seeking to build upon and sustain aspects of our research programme through H2020 funding opportunities. Additionally, commercialisation via a spin out company, to commercialise of our aptamer generation approach [based on the results from the DNA/thrombin measurements] is being explored [ALU-FR].

Project Context and Objectives:
With 528,000 new cases every year, cervical cancer is the fourth most common cancer affecting women worldwide; it is most notable in the lower-resource countries of sub-Saharan Africa. It is also the fourth most common cause of cancer death (266 000 deaths in 2012) in women worldwide. Almost 70% of the global burden falls in areas with lower levels of development, and more than one fifth of all new cases are diagnosed in India [1].
Cervical cancer is usually preceded by a long phase of pre-invasive disease called cervical intraepithelial neoplasia [CIN]. This pre-invasive phase is characterised microscopically as a range of events progressing from cellular atypia to various grades of dysplasia, including cervical intraepithelial neoplasia (CIN), before progression to invasive carcinoma. This precursor phase is generally asymptomatic, and can occur over a long period of 10–20 years [2]. With the introduction of HPV vaccination, the landscape of cervical pre-cancer will change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (ie positive predictive (PPV) and negative predictive values (NPV)) of current diagnostic tests.
There are no specific clinical features or symptoms which indicate the presence of CIN. Initial diagnosis is usually made by cytological analysis of a Pap smear specimen. Currently histology is the gold standard procedure in terms of disease diagnosis. However, histological diagnosis of CIN is complicated by a variety of cellular changes associated with inflammation, pregnancy and/or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability [3]. This is reflected in poor inter-observer agreement between pathologists. In particular, the differential diagnosis between immature squamous metaplasia and CIN1/2, or between low-grade [CIN1] and high-grade [CIN2/3] lesions, tends to be difficult. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently [4, 5]. As the histological diagnoses determines the decision to treat, a low positive predictive value of the colposcopy-referring test may result in unnecessary treatments. The estimates of this burden has been assessed by several studies; one study in particular resulted in 20% of low-grade lesions being upgraded and 26% of high-grade lesions downgraded following review [6]. There is a clear need for better diagnostic methods to aid objective CIN lesion diagnosis and to identify true high grade cervical disease and to increase specificity and PPV [3].

Recent developments in the field of proteomics has opened new avenues for cancer-related biomarker discovery. With the advent of new and improved proteomic technologies such as the development of quantitative proteomic methods, high-resolution, high-speed, high-throughput, high-sensitivity mass spectrometry (MS) and protein chips, as well as advanced bioinformatics for data handling and interpretation, it is now possible to discover biomarkers that can reliably and accurately predict outcomes during cancer treatment and management. Onco-proteomics offers cutting-edge capabilities to accelerate the translation of basic discoveries into daily clinical practice. In the field of cervical cancer such efforts have already begun [7].

Concept
In SYSTEMCERV, we adopt a systems biology approach to address these requirements. The project builds on work performed within a previously funded FP7 collaborative research grant called AUTOCAST, where we have identified and validated a novel panel of RNA based biomarkers for detection of cervical pre-cancerous lesions. This panel of mRNA markers were developed using systems biology and data mining tools and have demonstrated high specificity [93%] and sensitivity [88%] for detecting CIN 2-3 lesions (in preparation). This compares to sensitivity and specificity metrics for cytology [specificity: 80-95%, sensitivity: 60-85%] and for HPV screening [specificity: 70-85%, sensitivity: 90-95%]. The combined biomarker panel has a specificity of 92% and sensitivity of 91%, for detecting CIN2+ disease in younger women under 30 years of age, where it outperforms HPV, whose specificity is unacceptably low. This enabling clinical validation work on the biomarker panel provided sufficient evidence that further exploration of existing data sets and biological pathways using a methodical systems biology approach was warranted. This approach combines de novo discovery and uses computational, and simulation approaches to extrapolate from existing data sets.
To complement the systems biology discovery component, high throughput biomarker analysis and validation was planned using the SYSTEMCERV approach. SYSTEMCERV proposed a novel strategy (Figure 1) for this which integrates several technologies to develop cervical pre-cancer and cancer specific protein arrays [ALU-FR DNA-to-protein copying technology] for subsequent down-stream generation of antibodies using phage display [CysDisplay] and antibody array technologies [Human Combinatorial Antibody Library (HuCAL)] antibody technology (CellCall). We then used a label free detection approach for the detection of proteins using imaging Reflectometric Interference Spectroscopy [iRIfS] technology (Biametrics). The ultimate goal of the project was to generate a panel of protein and specific detection antibodies for use in cervical pre-cancer screening and for stratification of patients with CIN disease (TCD and CellCall). This novel approach is not limited to cervical cancer biomarker discovery and validation; it can be translated to other diseases and biomarkers, and has personalised medicine applications.
SYSTEMCERV extends a systems biology approach on three levels
a). a main focus SYSTEMCERV was that the RNA based cervical pre-cancer and cancer markers derived in the FP7 funded project AUTOCAST were extended by systems biology means. As a first step corresponding antibodies would be produced using the proposed methodology as a pilot and to establish the value of the mRNA markers derived proteins. In the second step, parallel to the pilot phase of the technology, validated markers would be extended according to their proteome [iTRAQ proteomics and computational analysis] by translating existing RNA based cervical pre-cancer and cancer markers to proteins. In parallel,
b). SYSTEMCERV examined the EGFR [epidermal growth factor receptor] pathway. We had previously demonstrated that epiregulin is a marker that correlates with severity of cervical pre-cancer (unpublished data). Epiregulin is a member of the epidermal growth factor family and can function as a ligand for EGFR and other members of the ERBB/EGFR family of tyrosine kinase receptors. Epiregulin participates in a complex cellular network which leads to cell proliferation, epithelial mesenchymal transition [EMT] and migration, through activation of RAF, MEK, ERK, and JAKS. The EGFR pathway is the key regulator of proliferation in keratinocytes, so alteration of this pathway, as it is seen in non-melanoma skin cancers [8,9], is likely to be associated with the different proliferation potential of CIN lesions. We believe that the EGFR pathway is an important actuator of the cervical carcinogenesis process and the development of CIN lesions.
c). to complement these efforts targeted deep re-sequencing of key oncogenic drivers [including the EGFR pathway] was performed to investigate their role in cervical carcinogenesis, with particular focus on pre-cancerous lesions. The selection of genes and their target exons was based on current published data, the Sanger database, and our genome sequencing efforts.
To achieve these goals, within the given timeframe, we proposed the use and integration of a unique combination of novel technologies to develop cervical pre-cancer and cancer specific protein arrays [ALU-FR microfluidic DNA-to-protein copying technology] for subsequent down-stream generation of antibodies using phage display [Human Combinatorial Antibody Library (HuCAL) from Morphosys] in a microarrayed format. For readout we used a label free detection of proteins using imaging Reflectometric Interference Spectroscopy [iRIfS] technology. This will allow high throughput characterisation, regarding specificity and affinity, of all developed antibodies with ease and speed. This unique combination of technologies enables very fast turnaround times, featuring not only the protein array production and antibody-phage biopanning and selection in one day, but also the measurement of hundreds of samples for an array of proteins within the same timeframe. The ultimate goal of the project was to generate a panel of proteins and their respective highly specific antibodies [targeted against wild type and mutant protein] for use in cervical pre-cancer screening and for stratification of patients with CIN disease. The system was to be demonstrated and validated as proof of concept in clinical samples from patients with cervical pre-cancer and cancer. This approach has the potential to generate a highly specific panel of cervical pre-cancer and cancer protein markers which will have a high sensitivity, specificity, and PPV.

Objectives
The aim of SYSTEMCERV was to develop and to validate, a system biology based protoarray for biomarker screening in cervical cancer and pre-cancer to improve clinical diagnosis, and management of disease [outlined in Figure 1]. The complete system should:
•Ascertain the utility of a novel panel of pre-selected, experimentally and computationally systems biology derived proteomic biomarkers in cervical pre cancer and cancer disease, for more accurate grading and stratification of CIN.
•Establish a universal framework for validation of proteomic biomarkers exemplified by cervical disease stratification, which will have global applications across many other disease types and fields. This can be achieved by combining advanced technologies to establish the methodology in a robust and time-effective way, including;

o Cell free expression of protein array from DNA arrays (DNA-protein copying technology) to protein biomarkers in a microarray format
o Fast and reliable array-based protein marker search/validation and antibody-phage selection technologies
o Label-free detection via imaging Reflectometric Interference Spectroscopy (iRIfS) enabling the analysis of antibody characteristics (on- and off-rate, binding constant, cross-reactivity) used for validation and in clinical study setup
o High throughput deep targeted re-sequencing of major oncogenic driver genes in cervical pre cancer and cancer (deep sequencing)

•Assess the clinical feasibility of this approach through a validation “proof of principle” study to benchmark the performance of the biomarker

Project Results:
Overall the SYSTEMCERV project devolved into 4 main tasks; 1) Development of novel system biology derived biomarker panels [CellCall]; 2) Generation of phage antibodies [novel biomarker panels] for use in protoarray [CellCall]; 3) Development and validation of an integrated protein microarray and label free detection platform [ALU-FR, Biametrics]; 4) Clinical validation of novel biomarkers and the integrated technologies [TCD].

The project successfully delivered most aspects of the research programme; a novel panel of biomarkers were identified and clinically validated on cervical pre-cancer cases; a demonstrator device was developed and tested for on biological systems including biotin/streptavidin, DNA/thrombin, and b2 microglobin/anti–b2 microglobulin.

In this section, we describe the main achievements for each of the above tasks and discuss some of the technical difficulties we encountered.


1. Development of novel system biology derived biomarker panels [CellCall]

A key objective within SYSTEMCERV was to use systems biology approaches to derive novel biomarkers and gain more in depth understanding of the underlying pathways involved in the carcinogenic process. To achieve this we combined existing published datasets with new experimental data derived within the project.

-New datasets
To generate new experimental data we adopted several approaches; 1) Signal transduction pathway activity profiling to validate signal transduction pathway activities 2) two hybrid yeast screening to explore interactions between proteins in vivo 3) co-immunoprecipitation to analyse protein:protein interactions and 4) second generation sequencing for targeted re-sequencing of key oncogenic drivers in cervical cancer and pre-cancer.

Key signal transduction pathways linked with HPV and carcinogenesis such as p53 and SMAD pathway down regulation and STAT pathway up regulation were identified as well as novel pathways.

Targeted re-sequencing of 69 genes including EGFR, HRAS, KRAS, PIK3CA, BRAF, FGFR (I-III), ALK and MET was performed on 14 patient specimens [CIN1-CIN3 disease] on the Ion Torrent sequencing platform. Specific genes were identified with a higher mutation risk in cervical pre-cancer. These include; ALK, CCND2, MET, IGF2, TP63 CDC6, GMMN and TP53 among others.

-Systems biology
One of the major goals of the project was to discover and understand the regulatory pathways in a cervical cell and how human papillomavirus alters these pathways in cervical cancer. To achieve this we employed several bioinformatic and systems biology software packages including; Software Environment for BIological Network Inference (SEBINI) and Predictive Networks (PN), Supervised Inference of REgulatory NEtwork (SIRENE), Matlab Systems Biology Toolbox and Collective Analysis of Biological Interaction Networks (CABIN), to analyse datasets [published and new data], define and interrogate specific pathways to identify panels of novel markers.

We chose SIRENE as the most effective transcription network inference tool among several other software packages in the prediction of new transcription factor regulations. SIRENE is a software package for inference of gene regulatory networks from gene expression data and it retrieves in the order of 5 times more known regulations than other state-of-the-art inference methods [10]. The inference engine is based on support vector machines [SVM] for classification of input gene expression and regulatory data.
We focused our initial analysis on transcription factor-gene interactions using SIRENE initially and then further refined using GEPHI. Using Gephi, 21 different communities of genes representing different pathways were identified (Figure 2). In particular, community 16 representing the cell cycle regulatory pathway was notable as it contained 40% of all up-regulated genes in cervical cancer (Figure 3). Further exploration of Community 16 was performed using by PINA v2.0 based protein interaction mapping, which helped to elucidate more detailed functional annotation of this community. The map is used to construct new communities and these communities are also annotated functionally, to get a more detailed picture of their function (Figure 3). CXCL13, DSG3 and TP63 were major genes within this group and were taken forward along with Interferon regulatory pathway genes including IRF1 and several others as selected biomarkers for which phage antibodies would be produced for use on the SYSTEMCERV protoarray.

2. Generation of phage antibodies [novel biomarker panels] for use on the SYSTEMCERV protoarray [CellCall]

Our strategy for generating antibodies for use on the protoarray adopted CysDisplay technology for generating phage antibodies, which we licensed from Morphosys. The filamentous phage displaying the specific antibody fragment is generated by CysDisplay technology (Figure 4) which is an efficient display technology, based on the simultaneous periplasmic expression of engineered phage coat proteins and antibody fragments, both containing an unpaired cysteine residue. Disulphide bond formation between both partners results in heterodimerization and subsequent incorporation of these complexes into phage particles, leading to the display of functional antibody fragments on the phage surface. The cleavable disulphide bond allows efficient elution of the selected phage during the panning procedure.

The HuCAL PLATINUM Kit contains phage preparations for fourteen (seven VHVκ and seven VHVλ) sub-libraries providing the option to perform selections with separate sub-libraries or with a combined phage pool. Each sub-library comprises pools of one single VH master gene combined with either three Vκ or three Vλ light chain master genes.
Selection of antigen-specific antibody fragments from a phage library is performed by solid phase panning against several antigens including DSG3, TRKN3, PIK3AP1. During the preparation of antigens, we designed expression vectors and carried out large-scale protein expressions, despite significant efforts the process was suboptimal, required several rounds of redesign of the vector constructs and resulted in significant delays for the project.
Generally, phage selection is achieved using the following procedure; 1) coating of the antigen on MaxiSorp high protein-binding capacity ELISA plate wells; 2) blocking the remaining reactive sites to avoid the non-specific interaction; 3)addition of the HuCAL phage library, separated into two combined phage pools, one containing all VHVκ and the other all VHVλ libraries. These two sub-libraries are selected separately during the whole process; 4) phages bind to the immobilized protein by the specific Fab fragment displayed on their coat antigen gIIIp; 5) non-binding or non-specifically bound phages are removed by several washing steps; 6) the specifically bound phages are eluted from the antigen by addition of dithiothreitol (DTT) which breaks the disulphide bonds between the Fab and the phage; 7) the eluted phages are rescued and amplified by transfection into receptive E. coli bacterial cells; 8) to select phage particles for the next panning round, these phagemid carrying E. coli cells are then infected with helper phages which supply all the proteins required for the assembly of functional phages; 9) By adding IPTG this induces the expression of the HuCAL Fab as well as of gIIIp, both of which are encoded on the phagemid and linked via disulphide bond, for presentation of the Fab on the phage surface; 10) finally, usually three rounds of panning lead to an efficient enrichment of antigen-specific phage’s so this whole cycle needs to be repeated twice.

To date we have limited results (Table 1, Table 2). After the third panning round, for DSG3 protein, the results were promising when input and output phage titres were measured. The values matched expectations. After the third round of selection, the eluted phages were amplified by infecting E. coli bacterial cells, of which the polyclonal DNA was isolated and sub-cloned into expression vectors. These were transformed into E. coli suitable for protein expression, and then the single colonies were investigated. Six master plates with altogether 576 clones were screened by ELISA to identify reactive phage antibodies. Three clones were identified for DSG3. Under reducing condition, the disulphide linkage was cleaved between the VHCH1 and VLCL chains resulting two chains with roughly the same molecular weight of ~24 kDa. Consequently the lightest molecular weight corresponds to the reduced Fab fragments. Non-reduced Fab fragments are shown as the heaviest molecular weight band of ~47 kDa (Figure 5).

Despite these partial results and several selection attempts, no suitable, high-affinity phage antibodies were selected. We attribute this failure to the low proportion of phage bearing phages in the rescued pool, however as of yet no solution has been found to resolve this. This has impacted on other tasks and deliverables within the project.


3. Development and validation of an integrated protein microarray and detection platform [ALU-FR, Biametrics]
The original strategy was to integrate the protein copying technology [ALU-FR] with the iRIfS detection system [Biametrics] and to validate each prototype with cervical cytology specimens. Both technologies were successfully integrated for microarray screening and as the project evolved improved prototypes/demonstrators were developed.

The knowhow and components for iRIfs-detection were transferred from Biametrics to ALU-FR. This final setup (figure 6 and 7) allows the following features:

• A sample holder, holding 2 buffers (50 ml each) which can process up to 36 samples. The holder rack can be cooled to 8°C to preserve the samples in a cooled state.
• A flow cell holder to accommodate the flow cell and the detection slide. The holder has a heating/cooling unit enabling temperatures between 4 and 40 °C.
• A fluidic unit with two pumps, which enable flow conditions similar to a Biacore system.
• A detection unit yielding camera and optics to generate the iRIf signal.
• A steering unit to provide all electronic cards and power supplies for the heating/cooling system and the pumps.
• A pc with steering software to program the device and to calculate the iRIf signal.

The different functional units are depicted in figure 6, whilst figure 7 shows the final design and its realization.


In the absence of the specific biomarker antibodies, the SYSTEMCERV demonstrator device was characterised and benchmarked using alternative biological models including; Biotin/Streptavidin, DNA/thrombin, beta2-microglobulin/anti-beta2-microglobulin model systems and a novel system relevant for cervical pre-cancer screening the p16/anti-p16 model. We established a sensitivity of concentrations from 50 ng/ml to 280 ng/ml. In a model assay system, in which streptavidin [200 ng/ml] was spiked into a cell lysate [provided as “sample matrix” by TCD] a signal could be generated with 20 ng of the total amount of streptavidin (100 µl volume respectively). The binding constants (kD) were between 10 nM and 10 µM, in line with reference measurements from Biacore Label-free surface plasmon resonance (SPR) based technology.
Importantly, as the demonstrator device uses the label-free iRIfS detection technology, the resulting signal is dependent on the molecular mass of the analyte. p16ink4a being a relatively small analyte (16.5kDa) represents a challenging analyte for the system. It is anticipated that all other identified markers and analytes will generate higher signals and therefore be detected with a lower limit of detection. Nonetheless a limit of detection of 280ng/mL was detected for p16ink4A on the demonstrator device.
As a final proof of concept, we tested lysates from real cervical cytology samples which were spiked with streptavidin [0.2 µg/ml]. From this we established a limit of detection of 200ng/mL, with a dissociation constant kD of 1-10nM.
The sensitivity of the detection can be further improved by using an additional second antibody, increasing the mass on the surface, and thus increasing the iRIF-signal. Based on our thrombin antibody interaction analysis we conclude that such a measure will increase the signal by a factor of 4 to 10 depending on the initial mass of the detected target. For small targets larger increasing factors will be achieved. This is particularly beneficial for larger analytes.

4. Clinical validation of novel biomarkers and the integrated technologies [TCD]

Clinical validation was performed during the different phases of SYSTEMCERV. Initially our biomarker validation was focussed on assessment of biomarker tissue expression patterns across the different grades of CIN. Following this our assessment continued into cervical cytology specimens to progress towards demonstrating the “proof of principle” to benchmark the proposed biomarker panel and integrated protoarray and detection technology. This will be presented below in two parts; Part A: cervical tissue specific expression patterns and Part B: cervical cytology biomarker expression patterns.

This research was embedded within CERVIVA (The Irish Cervical Screening Research Consortium, www.cerviva.ie) based at the Coombe Women and Infants University Hospital. Specific research ethics approval has been obtained for SYSTEMCERV [Study no 15-2012] from the Coombe Women and Infants University Hospital Research Ethics Committee.

Part A: cervical tissue specific biomarker expression patterns
Within SYSTEMCERV we have identified a novel panel of biomarkers including DSG3, GAL 7, RTKN2, PIK3AP1, TP63, IL1RA, EREG, MUC5AC, CYP2C18 and CXCL13 for detection of cervical pre-cancerous lesions. This panel of markers has been developed using systems biology and data mining tools. Our primary assessment of these biomarkers was performed on cervical tissue specimens from patients with cervical pre-cancer [n=113 cases], using immunohistochemical and immunocytochemical methods. Over the course of the project and validation work our final biomarker panel evolved and consists of Desmoglein 3 (DSG3), Epiregulin (EREG) and TP63. This panel has been fully evaluated on 113 cases of cervical pre-cancer [CIN1, 2 and 3] and normal cervical epithelium and bench marked against the gold standard cervical pre-cancer marker p16ink4a (Figure 8).

Tumor protein p63 appears to be the most consistent marker in our panel. TP63 is a member of the p53 family of transcription factors, and plays a role in tumourigenesis and has recently been suggested as biomarker for cervical cancer. TP63 showed the strongest correlation to p16ink4a staining, indicating its potential use as a biomarker in CIN diagnosis. TP63 expression was both qualitatively and quantitatively different in CIN compared to normal cervical epithelium. This difference increased with disease progression, indicating the ability of TP63 to distinguish between CIN 1, 2 and 3. TP63 is strongly linked to differentiation and its over expression has been linked with many cancers including basal cell and squamous cell carcinomas of the head and neck, melanoma, and cervical [11]. It’s up regulation has also been previously demonstrated in CIN [12]. TP63 may also play a role in differentiating pure squamous or glandular from adenosquamous carcinomas of the cervix.

Desmoglein [DSG3] is a transmembrane glycoprotein that is exclusively expressed in the lower layers of stratified epithelium [12]. It is a calcium-binding transmembrane glycoprotein that has a key role in cell-to-cell adhesion [12]. Here we describe for the first time DSG3 expression in cervical pre-cancer. DSG3 was over expressed in CIN 2+ compared with histologically normal samples and there was a gradual increase in diffuse staining of this protein as disease grade progresses. However, results were variable with some CIN 3 cases showing a loss of DSG3. In these cases adjacent tissue demonstrated its normal staining pattern of basal and parabasal membrane staining. The precise role for the tissue specific expression patterns of desmogleins is not fully understood, but manipulation of desmosomal cadherin expression suggests that tight regulation of their expression pattern is critical to tissue homeostasis. Loss of expression may result in reduced intracellular adhesiveness. It also thought that overexpression of DSG3 promotes carcinogenesis through a plakoglobin-mediated signaling pathway involved in transcriptional activity and a further downstream effect on molecules, including c-myc, cyclin D1, and MMP-7, which may lead to malignant phenotypes[13, 14].

Epiregulin [EREG] functions as a ligand of epidermal growth factor receptor (EGFR). EREG demonstrated staining in the mid to upper layer of the epithelium in cervical tissue samples. This was most evident in epithelium showing viral changes and CIN1. EREG demonstrated positive cytoplasmic staining in CIN2 and CIN3, staining was predominately localized in the apical layer of the epithelium with mild staining in lower layers. Almost half CIN3 cases had minimal to negative staining for EREG. This diminishing expression from low grade to high grade lesions could be due to a HPV gene transition that occurs from progressive infections to transforming infections. During the life cycle of HPV, infected cells migrate upwards from the basal layer to the surface epithelium. As the cells migrate to the mid and upper layers HPV early proteins interacted with cellular proteins to create a replication proficient environment [15]. One early protein believed to contribute to this is E5. E5 is associated with enhanced ligand-dependent activation of the epidermal growth factor receptor (EGFR), this is achieved by inhibiting endosome fusion, preventing endosome maturation [16] allowing receptors to be recycled back to the cell surface [17]. Consequently their activity is extended resulting in enhanced cell cycle entry [18]. E5 expressing cells have been found to have an increased number of EGF receptors.


Part B: cervical cytology biomarker expression patterns.

The overall project concept was that as the project progressed and prototypes of the protoarrays and iRifS technology became available, these would be tested using cervical cytology specimens to develop the “proof of concept”. As there were delays in generating the phage display antibodies, no cervical cancer specific protoarray was available within the project to achieve this level of specific clinical validation.

Nonetheless, we did attempt alternative approaches to validate the biomarkers for use in cervical cytology samples using an ELISA based approach. Three different ELISA’s were developed for this purpose (GAPDH, p16INK4a, DSG3 and DSG3 recombinant). Protein extracts from cervical cytology samples were screened for these 3 biomarkers however despite several attempts we were not able to detect these biomarkers in cervical smear samples. We concluded that ELISA was not sensitive enough to detect the small number of biomarker positive cells in a background of normal cells.

Alternative approaches to examine expression patterns of the novel biomarker panels in cervical cytology specimens were taken. This include TaqMan PCR analysis to examine the mRNA expression patterns of TP63, DSG3 and EREG in high grade CIN compared to low grade disease. Total RNA was extracted from 55 ThinPrep samples; 4 Normal, 25 CIN 1, 12 CIN 2 and 14 CIN 3 samples.TP63 and DSG3 were significantly up-regulated in the majority of CIN 2 and CIN 3 cases and matched the protein expression patterns we observed by IHC in the same specimens. EREG was equally up-regulated and down-regulated in CIN 2 and 3 compared to Normal/CIN1. This data confirmed that biomarkers were detectable in cervical cytology samples and that ELISA was not sensitive enough, which is why the more sensitive label free iRifS technology was required.

References

1. Forouzanfar MH, Foreman KJ, Delossantos AM, Lozano R, Lopez AD, Murray CJ, Naghavi M. Breast and cervical cancer in 187 countries between 1980 and 2010: a systematic analysis. Lancet. 2011 Oct 22;378(9801):1461-84. doi: 10.1016/S0140-6736(11)61351-2. Epub 2011 Sep 14. Review. PubMed PMID: 21924486.
2. Colposcopy and treatment of cervical intraepithelial neoplasia. Colposcopy and treatment of cervical intraepithelial neoplasia: a beginners manual ed. S. R. 2003. 13–20.
3. Martin, C.M. and J.J. O'Leary, Histology of cervical intraepithelial neoplasia and the role of biomarkers. Best Pract Res Clin Obstet Gynaecol, 2011. 25(5): p. 605-15.
4. Joste, N.E. C.P. Crum, and E.S. Cibas, Cytologic/histologic correlation for quality control in cervicovaginal cytology. Experience with 1,582 paired cases. Am J Clin Pathol, 1995. 103(1): p. 32-4.
5. Tritz, D.M. et al., Etiologies for non-correlating cervical cytologies and biopsies. Am J Clin Pathol, 1995. 103(5): p. 594-7
6. Hopman, E.H. P. Kenemans, and T.J. Helmerhorst, Positive predictive rate of colposcopic examination of the cervix uteri: an overview of literature. Obstet Gynecol Surv, 1998. 53(2): p. 97- 106.
7. Higareda-Almaraz, J.C. et al., Proteomic patterns of cervical cancer cell lines, a network perspective. BMC Syst Biol, 2011. 5: p. 96.
8. Khan, M.H. M. Alam, and S. Yoo, Epidermal Growth Factor Receptor Inhibitors in the Treatment of Non-melanoma Skin Cancers. Dermatol Surg, 2011.
9. Lida, K., et al., EGFR gene amplification is related to adverse clinical outcomes in cervical squamous cell carcinoma, making the EGFR pathway a novel therapeutic target. Br J Cancer, 2011. 105(3): p.420-7
10. Madhamshettiwar PB, Maetschke SR, Davis MJ, Reverter A, Ragan MA Gene regulatory network inference: evaluation and application to ovarian cancer allows the prioritization of drug targets. Genome Med. 2012 May 1;4(5):41.
11. Matin R.N Chikh A, Chong S.L Mesher D, Graf M, Sanza' P, Senatore V, Scatolini M, Moretti F, Leigh IM, Proby CM, Costanzo A, Chiorino G, Cerio R, Harwood C.A Bergamaschi D (2013). " p63 is an alternative p53 repressor in melanoma that confers chemoresistance and a poor prognosis." J Exp Med. 210(3):581-603
12. Quade BJ, Yang A, Wang Y, et al. (2001). “Expression of the p53homologue p63 in early cervical neoplasia.” Gynecol Oncol; 80:24–29.
13. Delva E, (2009) The desmosome. Cold Spring Harb Perspect Biol. 1(2);1 – 21
14. Chen, Y.J Lee, L.Y Chao Y. K, Chang J. T, Lu Y. C, Li H.F Chiu C. C, Li Y. C, Li Y. L, Chiou J. F, Cheng A. J. (2013). "DSG3 Facilitates Cancer Cell Growth and Invasion through the DSG3-Plackoglobin-TCF/LEF-Myc/Cyclin D1/MMP Signaling Pathway." PLOS. 8(5); e4088
15. Doorbar, J. Quint, W. Bravo, I. G., Stoler, M. Broker, T. R., Stanley, M. A.. (2012). “The biology and life-cycle of human papillomaviruses.” Vaccine 30(5): 55-70.
16. Suprynowicz, F. A., et al. (2010). "The human papillomavirus type 16 E5 oncoprotein inhibits epidermal growth factor trafficking independently of endosome acidification." J Virol 84(20): 10619-10629.
17. Straight,S.W. P.M Hinkle, R.J Jewers, D.J McCance. (1993) “The E5 Oncoprotein of Human Papillomavirus Type 16 Transforms Fibroblasts and Effects the Downregulation of the Epidermal Growth Factor Receptor in Keratinocytes.” J Virol 67(8): 4521-32
18. Pedroza-Saavedra, A., et al. (2010). "The human papillomavirus type 16 E5 oncoprotein synergizes with EGF-receptor signaling to enhance cell cycle progression and the down-regulation of p27(Kip1)." Virology 400(1): 44-52.

Potential Impact:
Impact
Despite the introduction of cervical cancer screening and prevention strategies, cervical cancer remains the fourth most common cancer in women worldwide. With the introduction of HPV vaccination, the landscape of cervical pre-cancer [CIN] is expected to change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (ie positive predictive (PPV) and negative predictive values (NPV)) of current diagnostic tests.
Moreover, histology is the gold standard procedure in terms of disease diagnosis. However histological diagnosis of CIN is complicated by a variety of cellular changes associated with inflammation, pregnancy and/or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently.

Combined these two challenges highlight the need for new diagnostic biomarkers for cervical screening and to aid objective CIN lesion diagnosis and to identify true high grade cervical disease and to increase specificity and PPV

SYSTEMCERV has provided a novel panel of protein biomarkers which are likely progression markers for cervical pre-cancer but may also be of use in cervical screening for triage of HPV positive cases. This will potentially impact on clinical management of cervical disease, in that biomarkers can be used to stratify progressive lesions and will ultimately improve diagnosis and outcomes for women with cervical pre-cancer.

The general procedure for the production of diagnostic protein arrays in combination with label free read out developed within SYSTEMCERV will enable us to develop a vast variety of protein arrays for either IVD or R&D purpose. Finally within SYSTEMCERV we have expanded the application of microarray copying to aptamers.

Dissemination activities

The objectives of the SYSTEMCERV dissemination strategy are fourfold:
• to communicate the results of the cutting-edge biomarker validation approach developed in SYSTEMCERV to European stakeholders, potential lead users, and corresponding institutions beyond Europe through a structured dissemination activity,
• to provide background for development of successful individual exploitation plans by partners,
• to promote the overall acceptance of the approach and outcomes in the molecular diagnostic field to clinicians, and scientists internationally.
• to foster societal (non-profit) and industrial (profit) interest to translate the knowledge generated by the project into cost effective solutions to the advantage of the public, the national and international health systems and clinical organisations.

To date our dissemination and outreach activities have been primarily through websites, presentations at International and National Conferences and outreach events. These are detailed below.

Websites:
The main public web site for the project is located at http://www.cellcall.hu/systemcerv/. The individual project partners have also included links to the SYSTEMCERV website on their own websites:

http://www.zbsa.de/projects/ag-roth/projects/syscerv
http://biametrics.com/en/applications/
http://www.cerviva.ie/SystemCerv.html

Publications
There has been one peer reviewed publication emanating from the project to date in published in Engineering in Life Sciences. Several others are planned, in particular relating to the novel biomarker panels for cervical pre-cancer.

• Kilb, N., Burger, J. and Roth, G. (2014), Protein microarray generation by in situ protein expression from template DNA. Eng. Life Sci., 14: 352–364. doi: 10.1002/elsc.201300052

Presentations
There have been four poster presentations at International conferences, including the International Cancer Conference 2014, the International Federation of Cytology and Colposcopy 2014, and Eurogin 2013.

There were two invited presentations in which the SYSTEMCERV project concept and results were included.

Finally, the SYSTEMCERV project was included in a series of lectures delivered by Dr Cara Martin as part of the Irish National Cytopathology Training course for pathologists in training.

Demonstrations
As part of our outreach activities, SYSTEMCERV has demonstrated results and prototypes resulting from the project at public events:
• Cellcall presented the SYSTEMCERV project at their booth at EUROGIN 2013 Nov 3-6, 2013 Florence.
• Biametrics presented the SYSTEMCERV demonstrator prototype at the Analytica trade fair Munich 2014.
• ZBSA attended the "Innovation Summit 2014" in Washington to present device developed in SYSTEMCERV and distributed flyers specific for the SYSTEMCERV device.

Exploitation of Results

Each partner had a specific exploitation plan for the project. These are highlighted below.
CellCall (formerly Genoid) Ltd.
The technical solutions developed in SYSTEMCERV provide product leads for the evolution of Cellcall's product pipeline. During the project, there has been continuous knowledge transfer, including key individuals, from the research work performed within SYSTEMCERV ensuring that the results from SYSTEMCERV are anchored in the company product plans. SYSTEMCERV results have greatly influenced CellCall’s message on cervical cancer diagnostics. Arising from the project, a patented technology has emerged and its clinical validation will be completed and new product will be launched within a year.

CellCall will also carry forward its research activities in the cervical cancer field, focusing on the integration of earlier developments including HPV detection and genotyping assays. It is expected that the continued research on cervical cancer and pre-cancer will impact future product directions. This will ensure an increased maturity of CellCall’s diagnostic solutions and their applicability to actual market needs.

Biametrics GmbH
The SYSTEMCERV project enabled Biametrics to further develop its label-free biomolecular interaction devices. Combination with other innovative technologies from other project partners (like ZBSA) and technology transfer led to further improvement of Biametrics’ technologies and widened the spectrum of possible applications and therefore also the addressable markets and customers. The close collaboration with other partners (like Cellcall and TCD) enabled Biametrics to identify additional potential fields of application, on the one hand, and on the other hand, the SYSTEMCERV project raised the awareness and acceptance of those partners for new innovative technologies in general and Biametrics products in particular.

Overall, the SYSTEMCERV project enabled Biametrics to perform research and technological innovation, which would otherwise have been too risky to be financed. In the consequence Biametrics was able to employ additional personnel staff to permanent positions.
Trinity College Dublin
SYSTEMCERV has provided a novel panel of protein biomarkers which are likely progression markers for cervical pre-cancer. Our research group CERVIVA concentrate on health services research with a focus on improving diagnosis of cervical pre-cancer through HPV testing and second round biomarker triage. The biomarkers derived within SYSTEMCERV will be exploited for this purpose to define a set of progression markers for high grade CIN. This will have enormous potential for future cervical screening, where HPV testing will become the predominant test, replacing cytology and the need for second round triage and progression biomarkers will become very evident. Furthermore with vaccinated women coming through cervical screening programmes the performance of current screening and diagnostic approaches will change and the demand for new biomarkers will increase.
ALU-FR
SYSTEMCERV was an incubator for the microarray copying technology as well as the microfluidic detection part. It deepened the cooperation between all partners, in particular with Biametrics as they jointly improve now a setup derived from SYSTECERV within another EU-project (CE microarray) in which they will investigate antibodies against sepsis markers for marker panel selection. In addition several new sub-projects have evolved for which funding will be sought. In total four EU-proposals and four national proposals are under review for funding – all with the microarray copier as heart of it. In addition to the protein copying, the DNA copying of aptamers is now also a new product derived partially from SYSTEMCERV. Overall the SYSTEMCERV project enabled ALU-FR to establish a new group led by Günter Roth which is focused completely on the copying approach and the label-free detection system established within project. The group employs four PhDs and with further funding expects to increase its capacity.