High-resolution, virome-wide infection analysis in type 1 diabetes birth cohorts

PROJECT ABSTRACT/SUMMARY
Type 1 Diabetes (T1D) affects >20 million people globally, causing life-long dependency on exogenous insulin
and reduced life expectancy. Although the disease has a strong genetic basis, a substantial fraction of risk is not
explained by genetic factors alone, and neither is the marked increase in T1D incidence over recent decades.
The identification of environmental determinants of T1D is important because it can reveal underlying
mechanisms of pathogenesis and highlight intervention strategies that may be developed and/or deployed for
high-risk subjects. While infection by particular viruses has long been postulated as a risk factor for T1D, prior
studies in this area have been limited in their power and/or breadth, with many investigating relatively small
cross-sectional cohorts. Large, prospective, longitudinal studies – including TEDDY (in the US / Europe) and
ENDIA (in Australia) – offer opportunities to overcome these limitations, however existing viral analyses of these
cohorts have so far been limited in their resolution (focused on infections defined symptomatically) and/or
sensitivity (relying on the direct detection of viruses at the time/site of infection). In this project, we will overcome
these limitations by using an innovative technology (“PepSeq”) to perform a first-in-class, longitudinal, virome-
wide serology-based analysis of the TEDDY and ENDIA cohorts. PepSeq uses programmable, highly-
multiplexed libraries of DNA-barcoded peptides to enable virome-wide serology at epitope-level resolution. We
recently showed that PepSeq can be combined with longitudinal sampling and an innovative analysis algorithm
(PSEA) to enable the detection of temporally-resolved, virome-wide infection events at subspecies resolution.
Building directly on this work, we will use PepSeq+PSEA to study ~14,000 longitudinal plasma samples from
TEDDY+ENDIA case/control subjects and comprehensively catalog the timing and identity of viral infection
events. We will begin by extending our existing PepSeq human virome library (which is already capable of
subspecies resolution) to enable highly-controlled comparisons across ~600-700 virus subtypes (Aim 1). Next,
we will test whether and when these viruses are associated with autoimmunity in cases vs. matched controls,
using selected longitudinal samples from (i) the first year of life, (ii) the year preceding the onset of T1D/IA, and
(iii) mothers during pregnancy. Overall, our approach combines the power of longitudinal sampling (to
establish a temporal link with autoimmunity), the sensitivity of serology (to detect infections across all body
sites) and the breadth and resolution of PepSeq (using 10,000s of peptides to address the virome
comprehensively and at subspecies resolution). Together, these innovations have the potential to reveal viral
associations with autoimmunity that could not have been detected with previous, less powered approaches.
Moreover, by generating the largest known longitudinal dataset of virome-wide, subtype-resolved infection
events in two highly-characterized birth cohorts, this project will enable future studies focused on the natural
history and consequences of viral infections during ontogeny.

Grant Number: 5R01DK140795-02
NIH Institute/Center: NIH
Principal Investigator: John Altin