Innovative molecular testing strategies for adjunctive investigations in hemostasis and thrombosis

Abstract

Clinicians and scientists in the fields of hemostasis and thrombosis have been among those first to integrate new molecular strategies for the purpose of enhancing disease diagnosis and treatment. The molecular diagnosis and introduction of gene therapy approaches for hemophilia are obvious examples of this tendency. In this chapter, we have summarized information concerning three molecular technologies that have reached various stages of translational potential for their incorporation into the clinical management of disorders of hemostasis. Chromatin conformation assays are now being used to capture structural knowledge of long-range genomic interactions that can alter patterns of gene expression and contribute to quantitative trait pathogenesis. Liquid biopsies in various forms are providing opportunities for early cancer detection, and in the context of tumor-educated platelets, as described here, can also characterize tumor type and the extent of tumor progression. This technology is already being trialed in patients with unprovoked venous thrombosis to assess the potential for occult malignancies. Lastly, advances in single cell transcriptome analysis, provide opportunities to definitively determine molecular events in rare cells, such as antigen-specific regulatory T cells, within the context of heterogeneous cell populations.

Introduction:

Scientists interested in the biology and pathobiology of hemostasis can now turn to a range of innovative molecular technologies that have the potential to provide an enhanced pathophysiologic understanding of hemostatic events. In some instances, the integration of these new testing strategies may also offer opportunities to extend the sensitivity of current diagnostic approaches for hemostatic pathologies.

In this review, we will summarize advances in three areas of molecular investigation that are showing significant promise in furthering our appreciation of mechanisms responsible of disease processes that often involve the hemostatic system. We will initially discuss how improvements in chromatin conformation technologies might provide insights into quantitative traits involving hemostatic proteins. Next, we will provide an overview of the practical considerations and potential application of liquid biopsies as adjunctive testing in the investigation of patients with unprovoked thrombosis. Lastly, we will summarize the current status of single cell transcriptome analysis and suggest possible applications of this technological innovation in the field of hemostasis (Table 1).

Table 1:

Summary of the innovative molecular methodologies described in this review. The source material for molecular analysis and possible basic/clinical scenarios involving hemostasis and thrombosis in which these testing methodologies might be applied.

Molecular Testing Methodology	Test Material	Hemostasis-Thrombosis Applications
Chromatin Conformation Capture (3C-related analysis)	Native cells expressing the coagulation protein of interest.	Evidence of chromatin structure changes with regulatory sequence (enhancer/promoter) variants – contributing to quantitative traits – eg type 1 VWD, fibrinogen deficiency.
Liquid Biopsies	Circulating tumor cells, cell-free DNA, exosomes, Tumor-educated platelets.	Search for occult cancer in unprovoked cases of VTE.
Single Cell RNA-seq	Progenitor/stem cells, immune cells, rare cells in heterogeneous cell populations.	Antigen-specific T cells in anti-FVIII immune responses. Assessment of coagulation factor expression in heterogeneous endothelial cell populations.

Chromatin Conformation Analysis:

The regulation of gene expression is recognized to involve a complex interaction between genetic sequence structure and DNA binding proteins that either activate or repress RNA transcription.¹ In the molecular pathology of inherited disorders of hemostasis, there are well documented examples of transcriptional variants that disrupt these DNA-protein interactions, resulting in subsequent protein deficiency states.^2–4 In most of these cases, the DNA variants are found in proximal regulatory sequences, within a relatively short distance of the transcriptional start site of the cognate gene (<1 kb). However, there is a growing appreciation that DNA sequence elements located at far greater distances from the transcriptional start site (multiple kilobases), derive their activating function through long-range chromatin interactions with the proximal promoter regulatory elements.⁵ These enhancer sequences possess site and orientation independent activating properties and can sometimes be found in clusters that may be referred to as locus control regions or super-enhancer elements.^6,7 To date, the pathobiological significance of enhancer sequences is unclear, but there are quantitative traits affecting hemostasis, such as type 1 von Willebrand disease, where disruption of enhancer sequence function may well be playing a role.⁸

As indicated above, the function of enhancer elements requires the generation of long-range chromatin interactions with proximal regulatory sequences, and over the past few years major advances have been made in developing laboratory tests to evaluate these events. In all of these test procedures the aim is to evaluate native chromatin conformation in the presence of wild type and variant enhancer sequences. In the investigation of a quantitative trait, evidence that a sequence variant has been identified in a critical enhancer element would be substantially strengthened through determination of a disrupted chromatin conformation.

The conserved regulation of gene expression involves the coordinated involvement of topologically associated chromatin domains (TADs) that comprise 100kb-1Mb regions of interactive DNA sequences. These chromatin structures likely represent key functional transcriptional regulatory units.

There are two methodologic strategies for evaluating chromatin spatial organization: microscopy and a range of molecular assays. Light microscopy has a resolution of 50–100 nm in single cells, and electron microscopy provides unprecedented resolution but fails to deliver specific structural information relating to chromatin interactions. In contrast, a variety of recently developed molecular techniques are well suited to determine the relative spatial organization and interactive relationship of chromatin. All of these molecular tests provide a population-based average of interaction frequencies of two (or more) genomic loci based upon their spatial proximity in a three-dimensional nuclear environment.⁹ The range of assays available to assess these chromatin interactions are all based upon the foundational chromatin conformation capture (3C) assay (Figure 1).¹⁰

Fig1:

Schematic illustration of 3C methodology: Cells are treated with formaldehyde to cross-link DNA-Protein interactions. Chromatin is digested with a restriction enzyme. Digested fragments are ligated under a condition that favors intra-molecular ligation. Cross-links are reversed and the quantity of ligated fragments is assessed by qPCR.

In each of these chromatin configuration experiments the same initial biochemical procedures are performed: chromatin is first cross-linked in the nucleus by exposure to formaldehyde and the cross-linked chromatin is fragmented with a restriction endonuclease. The resulting chromatin fragment ends are subsequently re-ligated in a dilute solution to favor intra- versus inter-molecular interactions (ie. to bias production of chimeric interactive chromatin products). Finally, in a step that can be modified in the various iterations of the 3C strategy, the ligated products are characterized by a quantitative DNA sequence analysis, in 2019, most often either qPCR or next generation sequencing. The relative population averaged prevalence of chimeric sequences compared to appropriately selected control sequences will provide an indication of the interactive frequency of chromatin elements.

In standard 3C studies, the genomic locations of the proposed interactive elements need to be known to determine a quantitative assessment of their frequency of spatial co-localization. This is the situation where a hypothesis has been proposed in which a distant regulatory element is spatially interactive with a proximal element and provides functional enhancement of transcriptional activation.¹¹ If pathogenic sequence variants are present in either the distant or proximal elements, the frequency of this interaction may be reduced with a concomitant reduction in transcription of the cognate coding sequence. This scenario could pertain to the putative VWF super-enhancer element, ~45 kb 5’ of the VWF gene, and contribute to the quantitative trait, type 1 VWD.¹²

Details of the 3C protocol can be revised depending upon the question being addressed. As one example, choice of the restriction enzyme employed to fragment the ligated chromatin regions can significantly alter the complexity of the end-product library. Six base pair cutter enzymes (as opposed to more frequent 4bp cutters), that digest DNA approximately every 4kb and give rise to >16 million genomic fragments, are more appropriately used for studying long range enhancer-promoter interactions and will result in reduced complexity of the 3C library for subsequent quantitative sequence analysis.

Additional recent modifications of the 3C methodology have addressed multiple genome-wide interactions of a single chromatin fragment (4C assay),¹³ and the Hi-C assay that determines the genome-wide interactive frequency of chromatin regions in the nucleus under defined (patho) physiological conditions. This latter assay provides a global interactive chromatin map and is able to assess both cis and trans-interactions over long distances. The challenges of these latter iterations of the 3C methodology relate to the increasing complexity of the quantitative sequence analysis required to interpret the interactive frequency of fragments.

While molecular technologies like the 3C-derived assays can now provide precise population averaged frequencies of relative spatial organization of genomic elements, these structural studies have important limitations. They do not evaluate where in the nucleus the chromatin interactions occur (interactions in the nuclear periphery are less likely to be productive than interactions in the interior of the nucleus), and the functional consequences of the chromatin interactions cannot be predicted. This latter deficit must be addressed by complementary assays to assign functional significance to the chromatin interactions.¹⁴

In conclusion, while still in a relatively early phase of development, evaluation of the frequency of chromatin interactions in regulating gene expression has the potential to contribute to our understanding of the phenotypic influence of many intergenic sequence variants associated with quantitative traits such as those causing disorders of hemostasis.

Liquid Biopsies:

The development of a venous thromboembolic (VTE) occurrence without a clear provoking event should always prompt the consideration of an occult malignancy. There is substantial pathobiological evidence linking the presence of cancer to an enhanced likelihood for VTE development, and a large systematic review has documented that the prevalence of occult cancer was up to 10% at one year after an unprovoked VTE event, compared to a prevalence of 2.6% in cases of provoked VTE.¹⁵ Nevertheless, despite this association, screening strategies using various imaging modalities to detect occult cancers in patients with unprovoked VTE have not proved to be effective for earlier cancer identification. Thus, currently, aside from ensuring that patients are up to date with age- and gender-specific cancer screening (colon, breast, cervix and prostate) no additional investigations are recommended in follow up of these patients.¹⁶

The potential for using blood-derived biomarkers as a means to diagnose solid organ cancers is clearly attractive, and recent studies using these liquid biopsies suggests that this objective is feasible.^17,18 It has been known for a long time that blood is a rich source of tumor associated biomarkers. There is an extensive literature describing the detection of circulating tumor cells,¹⁹ extracellular tumor-derived vesicles, cell-free tumor DNA²⁰ and RNA,²¹ and platelets containing a range of tumor associated molecules.²² The presence of these tumor-derived blood biomarkers can be explained by two principal mechanisms: the release of cellular contents during tumor cell apoptosis and the result of a systemic activity of the tumor.

The integration of liquid biopsy-based technologies to clinical practice would result in several potential benefits: a) earlier detection of cancer, b) prognostic information concerning the stage and spread of cancer, c) prediction of therapeutic response, d) real time monitoring of treatment and e) early detection of disease recurrence. For the remainder of this section of the review, we will focus on the possible use of tumor-educated platelets as a source of liquid biopsy material.

Human platelets are the second most abundant cell type in blood. They circulate for 7–10 days after their release from megakaryocytes in the bone marrow and lung. While platelets are anucleate cells, they have been demonstrated to possess a complex transcriptome derived from both their cell of origin, but also from other tissues including tumors, through processes including microvesicle-mediated exchange.²³ Platelets possess a range of RNA molecules, including mRNA transcripts representative of ~5,000 genes.²⁴ Pre-mRNA molecules can be processed by platelet resident spliceosome machinery²⁵ and there are also complex and dynamic populations of regulatory miRNAs, long non-coding RNAs and circular RNAs.²⁶

While there has been longstanding historic evidence linking cancer diagnostics to platelet number and quality, the analysis of platelet RNA profiles provides a level of interrogation that offers major advances in both sensitivity and specificity of discovery. There is now a growing body of evidence demonstrating that the RNA content of tumor-educated platelets can be qualitatively distinct and offer insights into the type, location and stage of a malignant tumor.^22,27,28 As just a few specific examples: platelets sequester mutant RNAs such as the EGFRvIII variant in glioblastoma, and KRAS and EGFR mutations in non-small cell lung cancer.²⁹ In addition, alternatively spliced RNA signatures have been demonstrated in association with several cancers including non-small cell lung cancer, glioblastoma and breast cancer.²⁹

A recent proof-of-principle study involving 283 platelet samples from 55 healthy controls and 228 cancer patients (6 different tumor types) employed RNA profiling of platelets combined with a machine learning classification algorithm. This study was able to diagnose the presence of cancer with >95% accuracy, and in addition, identified the type of malignancy in >70% of cases, suggesting the existence of tumor type-specific RNA profiles.³⁰

While, the clinical application of liquid biopsies for cancer in the form of tumor-educated platelets appears to have potential, much work needs to be done to validate the recent proof-of-principle studies. Additional investigations in patients with inflammatory diseases and other pathologies that might be expected to alter platelet RNA profiles³¹ need to be conducted and larger cancer study populations need to be evaluated.

Of critical diagnostic importance, all blood-based molecular testing requires careful attention to detail in the various stages of the analytical process. For platelet-based RNA studies, particular care will need to be exercised in the following areas of the diagnostic procedure. The collection of blood samples and isolation of platelets must be standardized with special attention to avoiding contamination with other cellular elements and platelet activation. Throughout the diagnostic process, quality control assessment should be undertaken to ensure the quality of the platelet RNA. Lastly, considerable work needs to be completed to improve the bioinformatic algorithms used to analyze the RNA profile results. These analyzes should incorporate measures of RNA sequence quality in addition to the application of machine learning strategies to provide the optimal approach for identifying tumor-associated and tumor-type specific RNA signatures.

Collectively, the promise of liquid biopsy strategies for non-invasive cancer diagnosis may provide an important advance for the screening of patients with unprovoked thrombosis. A prospective clinical trial is currently in progress (the PLATO study - ClinicalTrials.gov Identifier: NCT02739867) to evaluate the role of tumor-educated platelets for the diagnosis of cancer in a population of >450 adult patients with a first episode of an unprovoked venous thrombosis or pulmonary embolism. The results of this study are eagerly awaited.

Single Cell Transcriptome Analysis:

Recent advances in nucleic acid sequencing technologies have now progressed to the point where informative results can be derived from single cells. Indeed, with the current state of technology it is feasible to perform “multi-omic” analyzes on single cells with combined genomic, epigenomic, and transcriptomic determinations to correlate with the cell’s structural and functional phenotype. Evaluation of the detailed single cell protein content remains a challenge, although with the development of mass cytometry strategies in which multiple antibodies can be labelled with heavy metal ions and quantified by time-of-flight detectors it is now feasible to begin to determine a partial cellular protein content.³² In lieu of a strategy for single cell proteomic characterization, the utility of single cell transcriptome analysis (scRNA-seq) has now proven to be a feasible and informative experimental approach.^33,34

Single cell RNA-seq has now been applied to a range of biological and pathobiological scenarios in which knowledge of single cell structure and function is critical for understanding critical cellular roles. Examples of scRNA-seq applications include the definition of developmental processes in various tissues,^35,36 the characterization of rare cell populations in large heterogeneous pools of cells, and in immunology, the identification of immune modulatory T cell subsets in an immune response.³⁷ scRNA-seq will also enable the characterization and quantification of variant gene expression patterns involving processes such as alternative splicing events³⁸ and monoallelic gene expression.³⁹

Since the original description of scRNA-seq in 2009,⁴⁰ the technical aspects of the process have undergone significant advances. Most important is the isolation of single cells of a defined phenotype for subsequent RNA-seq analysis. This process has many variables that need to be taken into consideration when embarking on a scRNA-seq study. There are a number of alternative strategies that can be employed to isolate single cells, that in the past have most often employed flow cytometry and laser microdissection technologies. Now however, the integration of droplet-based methodologies with the potential to encapsulate thousands of single cells in individual partitions enables subsequent single cell lysis, reverse transcription reactions and molecular tagging to proceed in the context of the droplet platform.⁴¹

Single cell isolations also need to take into account the variable success with different cell populations (better yields from blood and lymphoid tissues versus other solid organs) and the influence of other pre-analytical events. Thus, removal of single cells from their native environment may effect rapid changes to their transcriptome, and a delay in single cell capture and processing may amplify any artifactually induced changes to their transcript profile. These variances can be mitigated by interventions such as cell cryopreservation that has been documented to provide good fidelity of scRNA-seq profiles compared to cells that have undergone immediate RNA isolation.⁴²

Following single cell isolation, polyadenylated transcripts are captured with poly(T) primers and reverse transcribed into their corresponding cDNAs. The capture of non-polyadenylated RNAs is more challenging and requires additional specialized techniques.

cDNAs derived from the captured polyadenylated RNAs are subsequently amplified, by PCR, and molecular tags are integrated into the PCR primers to provide unique molecular identifiers for the amplified pooled cDNA products. These tagged cDNA pools are then subjected to next generation sequencing (NGS). Finally, following bioinformatic evaluation of the NGS quality control and technical variability, the data is evaluated by computational and bioinformatic strategies to report biologically robust outcomes.

In the design of scRNA-seq experiments a number of key variables need to be considered. These include the establishment of sequencing approaches to detect either full-length or 3’ end biased transcripts, and the integration of strategies to influence the number of cells and depth of sequence acquired from the test material. The decision of how to proceed for each of these variables will depend largely on the aims of the scRNA-seq experiment. Thus, as one example, cell-specific variability in splicing patterns will require sequencing strategies that provide comprehensive information on full-length transcript patterns and abundance.

Questions of how many cells should be analyzed and at what depth of sequence also require reference to the aims of the study. Increasing cell numbers, but with a relatively low read depth will provide sufficient power to detect transcript patterns for cells comprising <1% of a total cell population, but for genes that are expressed at low levels, the depth of sequencing will also need to be increased.

scRNA-seq analysis is more prone to variability than bulk cell evaluations due to technical noise in the process and the inherent biological variability of transcript profiles in pooled cell populations. For this reason, the major challenges facing investigators conducting scRNA-seq are the bioinformatic and computational assessments of sequence quality control and variability, and the correct assignment of biological significance.⁴³ Suffice it to say that this is an area of rapidly advancing methodology with the incorporation of machine learning strategies to facilitate the analysis of the multidimensional data.^33,44 The comparative expression of 1000s of genes in two different cell populations can be most easily interpreted by reduction in the dimensionality of the data to display results in formats such as principal component analysis.⁴⁵

The technical feasibility of scRNA-seq has advanced rapidly in the past decade and obtaining single cell transcriptome data is now relatively straightforward, even for investigators inexperienced in the art. However, the post-analytical evaluation and interpretation of this data remains complex and challenging and awaits further advances in computational and bioinformatic resources to make this technology more widely accessible.