Diagnostic assessment and therapeutic management of IBD spectrum disorders currently derives from a combination of clinical examination, laboratory indices and invasive procedures 
. Given the emerging association of circulating miRNA species with a spectrum of physiological and disease conditions 
we have systematically explored the utility of secretory miRNAs isolated from different hematological fractions as non-invasive predictors of Ulcerative Colitis.
To address this question, we have investigated broadly across distinct whole blood fractions comprising both cellular (platelets, PBMC) and acellular compartments (exosomal/micro-vesicular) that are known to express and shuttle miRNA cargo through circulation. A set of distinct, differentially expressed maps of annotated miRNAs have been constructed from a cohort of 20 diseased versus 20 normal individuals through hybridization to miRNA arrays. Fraction-specific differentially expressed miRNA signatures have been interrogated through SAM 
at a 1% FDR 
that identify primarily up-regulated miRNA candidates in the full dataset with only a minority of miRNAs displaying down-regulation of expression. Through molecular classifiers derived from a Support Vector Machine model and a multiple random sub-sampling strategy, our results () have isolated a distinct diagnostic signature of 31 miRNAs with increased expression that derive specifically from the platelet fraction and can stratify UC patients from normal individuals with 92.8% accuracy, 96.2% specificity and 89.5% sensitivity. Given the low number of differentially expressed miRNAs from the PBMC and micro-vesicular fractions, no predictive miRNA species could be identified from the PBMC isolates and only a poorly performing minority 6-marker panel derived from independent analysis of the micro-vesicular fraction. Additionally, no added benefit is obtained from combination of the platelet and micro-vesicular strata which generate a 29 marker panel consisting 27 platelet-derived and 2 micro-vesicular markers with comparative predictive performance in distinguishing case from control (92.3% accuracy, 96.8% specificity and 87.8% sensitivity). Given the significant overlap observed between the platelet-derived and platelet+micro-vesicle-derived profiles () and the poor performance metrics of the non-platelet miRNAs we conclude that the origin of the majority of disease-specific predictors in UC map back to the anucleate platelet fraction. Additionally, since classifier analysis did not identify any down- regulated miRNA biomarkers, this data also suggests that the primary physiological response in Ulcerative Colitis is to enhance expression of miRNA subsets. The divergence of platelet vs. micro-vesicular signatures is interesting and indicates distinct sub-cellular addresses for these miRNA species. Taken together, this supports the hypotheses that the isolated micro-vesicle fraction mapped in this study either do not originate from platelet progenitors or very selectively sort platelet derived miRNA cargo into classes destined for trafficking through the hematopoietic circuit. Given the significant overlaps observed in previous studies between platelet and micro-vesicular signatures 
, this disparity may be indicative of disease specific characteristics of Ulcerative Colitis. The validity of the identified platelet-derived class can be additionally verified with 88% success through qPCR assays run on pooled patient and control samples () – thus providing independent confirmation of the differential regulation of this panel. Furthermore, these markers segregate into 4 highly correlated intensity clusters, often containing specific pairs of miRNAs that share proximity of sub-chromosomal location (Table S5
). Cumulatively these results suggest distinct sets of biomarkers that may be coordinately regulated or share functional regulatory sub-networks closely synchronized under the given physiological conditions.
The biology of IBD spectrum disorder is known to encompass a complex interplay of innate and adaptive immunity that ultimately regulates a regional intestinal inflammation response 
. Circulating blood platelets representing non-immune cells and engaged classically in the hemostatic function of surveying vessel integrity are emergent mediators of active inflammation modulating local intestinal tissue injury 
. Enriched in a small but diverse transcriptome 
, platelets contain a large proportion of mature miRNA that can potentially engage in the regulation of gene expression. In order to understand the diversity of our identified miRNA biomarker panel, we have assessed the overlap between our maps and miRNA profiles from normal or disease-specific signatures surveyed across different studies. These analyses identify between 6% (2/31) to 29% (9/31) overlap between the 31-platelet-derived biomarker panel with miRNAs derived from platelet populations of normal individuals 
and no overlap with UC specific miRNAs profiled from peripheral blood 
. Additionally comparison to miRNA maps of non-platelet origin involved in a broad spectrum of inflammatory disorders 
reveal only two overlapping miRNAs: hsa-miR-140 and hsa-miR-146b with the majority of the 31 marker panel thus being exclusive to our study. Taken together, these comparative analyses clearly delineate that members of the platelet-derived 31 biomarkers are (a) enriched in miRNA species that are typically not detected in the normal platelet transcriptome, (b) are highly compartment specific and detectable only on enrichment of the platelet fraction and (c) display specificity of expression restricted to Ulcerative Colitis compared to other inflammatory disorders.
Given that miRNAs are known to be collective regulators of approximately 30% of human genes 
understanding of the mechanism of gene regulation mediated through the 31-miRNA biomarker panel is of importance. The prevalence of biological functional processes was evaluated through analysis of GO terms on 3304 non-redundant predicted mRNA targets representing approximately 13% of the estimated number of protein coding genes in the human genome 
. Stratification of the top canonical groups and pathways reveal enrichment of genes engaged in actin-cytoskeleton assembly, vesicular secretary pathway, biosynthesis and metabolism of diverse lipids such as glycerol phospholipds, pterin and tri-acylglycerol, implicated in a variety of cell-regulatory and immune stimulation roles (). Specific examples include (a) regulation of the members of the Rho-ROCK pathway involved in cytoskeletal reorganization/cell adhesion response to wound healing (
and references therein), (b) involvement of sphingolipids as modulators of growth factor receptors and second messengers of known agonists of the inflammatory response such as tumor necrosis factor-α, interleukin-1β 
, (c) regulation of Pterin metabolism- in response to IFN-γ stimulation 
, (d) genes involved in ionic transport and regulation of calcium homeostasis that may influence membrane permeability and (e) modulators of classic inflammatory regulators such as TNF-α. The predicted mRNA target spectrum is also dominated by transcription factors/mediators such as PI3K, KLF5, etc. that we hypothesize play a role in modulation of key genes involved in the regulation of the other diverse functions of metabolism, biosynthesis, transport, etc. 
. Additionally, we also find that a subset of the biomarkers (i.e., hsa-miR-1271, hsa-miR-874, hsa-let-7i-star and hsa-miR-20b-star) can target ATGs16L, a gene necessary for autophagy and implicated in IBD susceptibility 
. Therefore, comparison of these targets to published studies 
reveal a high degree of concordance in functional groups but often limited overlaps in terms of specific gene targets – suggesting that the landscape of physiologically relevant mRNAs in IBD is far from being saturated. Therefore, this study complements our current understanding of the UC transcriptome and provides interesting insight into the expression divergence of this disorder.
In principal, the platelet derived miRNAs could contribute to the pathophysiology of UC through regulation of the platelet transcriptome, consequently influencing downstream interaction with the intestinal mucosa. Adhesion of platelets to the mucosal endothelium is known to occur though a CD40/CD40L contact-dependent process resulting in platelet activation 
and a chemokine mediated immune response. The ultimate outcome of this trigger is the recruitment of leucocytes to the endothelium which results in a focal inflammatory response. These mechanisms are consistent with the functional classes of mRNA target genes identified in this study and implicated in the modulation of cellular architecture and inflammation. Taken together, these results support a hypothesis which offers a direct role for miRNAs in influencing platelet activation and cell-cell mediated immune response in gut inflammation.
Given that micro-vesicles are known mediators of cell-cell communication 
it is intriguing to speculate if the miRNA content of platelets can potentially be transferred to endothelial cells through vesicular fusion. In this model, the epithelial transcriptome would be directly targeted by the identified platelet miRNA regulators. Analysis of differentially expressed miRNA profiles from colonic epithelial tissues from UC patients can isolate 34% (30 out of 89) significantly down-regulated mRNAs 
corresponding to representative targets of 57% (17 out of 30) miRNAs from our biomarker panel. Comparison of fold change densities of these anti-correlated target mRNAs against all down-regulated transcripts on the array, reveal a statistically significant difference (p-value of 0.011 by Binomial test and 0.007 by Fisher Exact Test) between the two groups (Fig S7
). This suggests that the platelet miRNAs could possibly be a significant factor in the down-regulation of endothelial gene expression in the context of IBD.
Given the prevalence and increasing disease-specific role of miRNAs, it is of interest to catalogue the association of these sequences with naturally occurring polymorphisms in polygenic disorders 
. Mechanistically, single nucleotide changes can have significant functional impact through either impeding miRNA biogenesis, or through inhibition of binding to target miRNA sequences. Examples of such functional polymorphisms are numerous 
) and has been demonstrated to impact a variety of disease states 
. We can map polymorphisms in the mature miRNA sequence of 35% of the miRNAs (11 of 31) in our biomarker signature and 20% (673 out of 3304) of their computationally predicted mRNA targets (http://compbio.uthsc.edu/miRSNP/
). It is intriguing to speculate whether the patient who is misclassified by the platelet-derived miRNA signature is impacted by this type of variation. A link between polymorphisms and miRNAs were also explored by examining possible associations of the 31-biomarker prognostic panel with IBD susceptibility loci identified through genome-wide association studies. We can identify a close physical link (220 Kb) between hsa-miR-941 and rs2297441, a SNP mapping to the 20q telomere that is associated with high confidence to UC 
. As more evidence demonstrating the significance of polymorphisms affecting miRNA synthesis, binding, stability or functional activity accumulates, it will become important to investigate these changes in the context of disease associations and diagnostic accuracy of miRNA biomarkers.
The abundance and stability of miRNAs in circulation holds the promise of easy and rapidly accessible biomarkers. Given the prognostic and diagnostic utility of this class in cancer and the inherently vast regulatory potential of miRNA molecules, we have assessed the ability of fraction-specific miRNAs in delineating incidences of Ulcerative Colitis. To our knowledge the current results outline for the first time platelet-specific miRNA signatures for an inflammatory disorder, with exceedingly high predictive measures. The potential of these biomarkers to influence gene expression in context of modulating intestinal homeostasis is explored and the spectrum of putative physiological targets evaluated for their regulatory potential. In summary, results obtained from this study support a model where expression and coordinate control of various polygenic components of UC pathogenesis may be mediated by the interplay of circulating miRNA species and their physiological targets. As universally applicable tests are developed a combinatorial use of these indicators would perhaps provide the most rational strategy for disease determination and clinical therapy.