High-accuracy determination of internal circadian time from a single blood sample

Nicole Wittenbrink, Bharath Ananthasubramaniam, Mirjam Münch, Barbara Koller, Bert Maier, Charlotte Weschke, Frederik Bes, Jan de Zeeuw, Claudia Nowozin, Amely Wahnschaffe, Sophia Wisniewski, Mandy Zaleska, Osnat Bartok, Reut Ashwal-Fluss, Hedwig Lammert, Hanspeter Herzel, Michael Hummel, Sebastian Kadener, Dieter Kunz, Achim Kramer, Nicole Wittenbrink, Bharath Ananthasubramaniam, Mirjam Münch, Barbara Koller, Bert Maier, Charlotte Weschke, Frederik Bes, Jan de Zeeuw, Claudia Nowozin, Amely Wahnschaffe, Sophia Wisniewski, Mandy Zaleska, Osnat Bartok, Reut Ashwal-Fluss, Hedwig Lammert, Hanspeter Herzel, Michael Hummel, Sebastian Kadener, Dieter Kunz, Achim Kramer

Abstract

Background: The circadian clock is a fundamental and pervasive biological program that coordinates 24-hour rhythms in physiology, metabolism, and behavior, and it is essential to health. Whereas therapy adapted to time of day is increasingly reported to be highly successful, it needs to be personalized, since internal circadian time is different for each individual. In addition, internal time is not a stable trait, but is influenced by many factors, including genetic predisposition, age, sex, environmental light levels, and season. An easy and convenient diagnostic tool is currently missing.

Methods: To establish a validated test, we followed a 3-stage biomarker development strategy: (a) using circadian transcriptomics of blood monocytes from 12 individuals in a constant routine protocol combined with machine learning approaches, we identified biomarkers for internal time; and these biomarkers (b) were migrated to a clinically relevant gene expression profiling platform (NanoString) and (c) were externally validated using an independent study with 28 early or late chronotypes.

Results: We developed a highly accurate and simple assay (BodyTime) to estimate the internal circadian time in humans from a single blood sample. Our assay needs only a small set of blood-based transcript biomarkers and is as accurate as the current gold standard method, dim-light melatonin onset, at smaller monetary, time, and sample-number cost.

Conclusion: The BodyTime assay provides a new diagnostic tool for personalization of health care according to the patient's circadian clock.

Funding: This study was supported by the Bundesministerium für Bildung und Forschung, Germany (FKZ: 13N13160 and 13N13162) and Intellux GmbH, Germany.

Keywords: Bioinformatics; Diagnostics; Genetics; Molecular diagnosis.

Conflict of interest statement

Conflict of interest: The study was supported by Intellux Berlin GmbH. JDZ and MZ are employees of Intellux GmbH. DK is shareholder of Intellux GmbH. Charité Universitätsmedizin Berlin has filed a patent application (EP17177735) relating to the BodyTime assay, in which AK, BA, and NW are listed as inventors; decision is pending.

Figures

Figure 1. Biomarker discovery strategy, sampling schemes,…
Figure 1. Biomarker discovery strategy, sampling schemes, and study cohorts.
(A) Biomarker discovery pipeline. (B) Sampling scheme and composition of the BOTI study cohort (n = 12 subjects) by sex, age, and DLMO. Blood samples were drawn at regular 3-hour intervals over a period of 40 hours (M = 14 samples per subject). Each sample was assigned an external time (Central European Time) and an internal time (hours past DLMO, derived from saliva melatonin profiles). The displayed sampling scheme is representative of the subject highlighted by a circle in the study cohort plot (green lines indicate sampling times on the second day). (C) Sampling scheme and composition of the VALI study cohort (n = 28 subjects) by sex, age, and DLMO. The spread of the BOTI study cohort (B) in the same coordinate system is shaded in gray. In contrast to the BOTI study, the VALI study includes extreme and moderately extreme chronotypes. For each subject 2 blood samples were obtained, drawn 6 hours apart (M1, morning sample; M2, afternoon sample). Each sample was assigned an external time (Central European Time) and an internal time (hours past DLMO derived from saliva melatonin secretion profiles).
Figure 2. Extraction of candidate biomarkers and…
Figure 2. Extraction of candidate biomarkers and migration to the NanoString platform.
(A) Cumulative frequency distributions of the absolute prediction errors for 4 types of ZeitZeiger internal cross-validation predictors of either internal or external time by 1- or 2-sample mRNA abundance profiles (n = 136). Each type of predictor was built for 9 combinations of the ZeitZeiger parameters sumabsv = {1, 2, 3} and nSPC = {1, 2, 3}. Insets show the average number of genes ± SD in the internal cross-validation predictors as a function of sumabsv and nSPC. (B) Global gene sets of the best-performing internal cross-validation predictors shown in A. Each column depicts a predictor defined by the type of the predictor variable (internal or external time), the format of the data input (1-sample or 2-sample), and the ZeitZeiger parameters (sumabsv, nSPC). Each predictor includes 10 leave-one-subject-out cross-validation runs, i.e., 10 gene sets. The ordering (from top to bottom) and the colors indicate how often a gene was identified as time-telling and assigned to SPC1, SPC2, or both among those 10 gene sets. Thirty-four genes that showed a high frequency of identification among cross-validation runs and were consistently identified across the best-performing predictors were chosen as a candidate biomarker set for internal time and migrated to the NanoString platform (highlighted in bold font). (C) Impact of platform migration on the performance of the candidate biomarkers for internal time. Given are cumulative frequency distributions of the absolute prediction errors of ZeitZeiger internal cross-validation models built on either RNA-Seq (blue) or NanoString data (red) obtained from the same RNA preparations. Platform comparison was performed for 4 types of predictors of either internal or external time by 1- or 2-sample mRNA abundance profiles (n = 136).
Figure 3. Composition and properties of the…
Figure 3. Composition and properties of the final NanoString BodyTime predictors.
(A) One-sample and 2-sample predictors trained on the NanoString data of the BOTI study (n = 154 samples) for sumabsv = {1, 2} and nSPC = 2. Genes assigned to SPC1 or SPC2 as well as their loadings are shown. (B) NanoString expression profiles of the BOTI study’s samples (n = 154) in the SPC space of the 1-sample 12-gene predictor. Colors indicate bins of the internal time. (C) Time course of expression of the genes building the 1-sample 12-gene predictor. Colors indicate the individual subjects of the BOTI study. Each time course starts with the internal time of the first sample of a subject (M1, day 1) and ends with its last (M14, day 2).
Figure 4. External validation and performance of…
Figure 4. External validation and performance of the NanoString BodyTime predictors.
(A) Cumulative frequency distributions of the absolute prediction errors of the 1-sample and 2-sample NanoString BodyTime predictors when they were applied to the VALI study data set. In the case of the 1-sample assay, the internal time stamps of all morning (M1) or afternoon (M2) samples were predicted; in the case of the 2-sample assay, the time stamp of the sample ratio was predicted (M1/M2). Proportion refers to the number of predictions with an absolute error that is less than or equal to the specified value divided by the total number of predictions (1-sample, M1: n = 28; 1-sample, M2: n = 28; 2-sample, M1/M2: n = 28). (B) Correlation of DLMO estimated by the BodyTime predictors and DLMO derived from saliva melatonin concentrations (gold standard) assayed by RIA; circular Pearson correlation coefficients (r) and P values are indicated. (C) Bland-Altman analysis of the bias between DLMO derived from saliva melatonin profiles and BodyTime estimations. The dashed horizontal line indicates the mean of the differences (bias); dotted lines represent the upper and lower limits (mean of the differences ± 2 SDs) with their 95% confidence intervals shaded light gray. The morning sample of 1 subject was excluded from A–C because its 12-gene predictor maximum likelihood curve was ambiguous.

References

    1. Yang G, Paschos G, Curtis AM, Musiek ES, McLoughlin SC, FitzGerald GA. Knitting up the raveled sleave of care. Sci Transl Med. 2013;5(212):212rv3. doi: 10.1126/scitranslmed.3007225.
    1. Zhang R, Lahens NF, Ballance HI, Hughes ME, Hogenesch JB. A circadian gene expression atlas in mammals: implications for biology and medicine. Proc Natl Acad Sci U S A. 2014;111(45):16219–16224. doi: 10.1073/pnas.1408886111.
    1. Anafi RC, Francey LJ, Hogenesch JB, Kim J. CYCLOPS reveals human transcriptional rhythms in health and disease. Proc Natl Acad Sci U S A. 2017;114(20):5312–5317. doi: 10.1073/pnas.1619320114.
    1. Montaigne D, et al. Daytime variation of perioperative myocardial injury in cardiac surgery and its prevention by Rev-Erbα antagonism: a single-centre propensity-matched cohort study and a randomised study. Lancet. 2018;391(10115):59–69. doi: 10.1016/S0140-6736(17)32132-3.
    1. Dallmann R, Okyar A, Lévi F. Dosing-time makes the poison: circadian regulation and pharmacotherapy. Trends Mol Med. 2016;22(5):430–445. doi: 10.1016/j.molmed.2016.03.004.
    1. Hoyle NP, et al. Circadian actin dynamics drive rhythmic fibroblast mobilization during wound healing. Sci Transl Med. 2017;9(415):eaal2774. doi: 10.1126/scitranslmed.aal2774.
    1. Long JE, Drayson MT, Taylor AE, Toellner KM, Lord JM, Phillips AC. Morning vaccination enhances antibody response over afternoon vaccination: a cluster-randomised trial. Vaccine. 2016;34(24):2679–2685. doi: 10.1016/j.vaccine.2016.04.032.
    1. Bonten TN, et al. Effect of aspirin intake at bedtime versus on awakening on circadian rhythm of platelet reactivity. A randomised cross-over trial. Thromb Haemost. 2014;112(6):1209–1218.
    1. Ballesta A, Innominato PF, Dallmann R, Rand DA, Lévi FA. Systems chronotherapeutics. Pharmacol Rev. 2017;69(2):161–199. doi: 10.1124/pr.116.013441.
    1. Roenneberg T, et al. Epidemiology of the human circadian clock. Sleep Med Rev. 2007;11(6):429–438. doi: 10.1016/j.smrv.2007.07.005.
    1. Hsu PK, Ptáček LJ, Fu YH. Genetics of human sleep behavioral phenotypes. Meth Enzymol. 2015;552:309–324.
    1. Stothard ER, et al. Circadian entrainment to the natural light-dark cycle across seasons and the weekend. Curr Biol. 2017;27(4):508–513. doi: 10.1016/j.cub.2016.12.041.
    1. Wright KP, McHill AW, Birks BR, Griffin BR, Rusterholz T, Chinoy ED. Entrainment of the human circadian clock to the natural light-dark cycle. Curr Biol. 2013;23(16):1554–1558. doi: 10.1016/j.cub.2013.06.039.
    1. Allebrandt KV, et al. Chronotype and sleep duration: the influence of season of assessment. Chronobiol Int. 2014;31(5):731–740. doi: 10.3109/07420528.2014.901347.
    1. Horne JA, Ostberg O. A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms. Int J Chronobiol. 1976;4(2):97–110.
    1. Danilenko KV, Verevkin EG, Antyufeev VS, Wirz-Justice A, Cajochen C. The hockey-stick method to estimate evening dim light melatonin onset (DLMO) in humans. Chronobiol Int. 2014;31(3):349–355. doi: 10.3109/07420528.2013.855226.
    1. Klerman EB, Gershengorn HB, Duffy JF, Kronauer RE. Comparisons of the variability of three markers of the human circadian pacemaker. J Biol Rhythms. 2002;17(2):181–193. doi: 10.1177/074873002129002474.
    1. American Academy of Sleep Medicine, eds. The International Classification of Sleep Disorders – Third Edition (ICSD-3). Darien, Illinois, USA: American Academy of Sleep Medicine; 2014.
    1. Ueda HR, et al. Molecular-timetable methods for detection of body time and rhythm disorders from single-time-point genome-wide expression profiles. Proc Natl Acad Sci U S A. 2004;101(31):11227–11232. doi: 10.1073/pnas.0401882101.
    1. Kasukawa T, et al. Human blood metabolite timetable indicates internal body time. Proc Natl Acad Sci U S A. 2012;109(37):15036–15041. doi: 10.1073/pnas.1207768109.
    1. Hughey JJ. Machine learning identifies a compact gene set for monitoring the circadian clock in human blood. Genome Med. 2017;9(1):19. doi: 10.1186/s13073-017-0406-4.
    1. Laing EE, Möller-Levet CS, Poh N, Santhi N, Archer SN, Dijk DJ. Blood transcriptome based biomarkers for human circadian phase. Elife. 2017;6:e20214.
    1. Goossens N, Nakagawa S, Sun X, Hoshida Y. Cancer biomarker discovery and validation. Transl Cancer Res. 2015;4(3):256–269.
    1. Keller M, et al. A circadian clock in macrophages controls inflammatory immune responses. Proc Natl Acad Sci U S A. 2009;106(50):21407–21412. doi: 10.1073/pnas.0906361106.
    1. Spies CM, et al. Circadian rhythms of cellular immunity in rheumatoid arthritis: a hypothesis-generating study. Clin Exp Rheumatol. 2015;33(1):34–43.
    1. Veldman-Jones MH, et al. Evaluating robustness and sensitivity of the NanoString technologies nCounter platform to enable multiplexed gene expression analysis of clinical samples. Cancer Res. 2015;75(13):2587–2593. doi: 10.1158/0008-5472.CAN-15-0262.
    1. Wallden B, et al. Development and verification of the PAM50-based Prosigna breast cancer gene signature assay. BMC Med Genomics. 2015;8:54.
    1. Minors DS, Waterhouse JM. The use of constant routines in unmasking the endogenous component of human circadian rhythms. Chronobiol Int. 1984;1(3):205–216. doi: 10.3109/07420528409063897.
    1. Hughey JJ, Hastie T, Butte AJ. ZeitZeiger: supervised learning for high-dimensional data from an oscillatory system. Nucleic Acids Res. 2016;44(8):e80. doi: 10.1093/nar/gkw030.
    1. Bleeker SE, et al. External validation is necessary in prediction research: a clinical example. J Clin Epidemiol. 2003;56(9):826–832. doi: 10.1016/S0895-4356(03)00207-5.
    1. Archer SN, et al. Mistimed sleep disrupts circadian regulation of the human transcriptome. Proc Natl Acad Sci U S A. 2014;111(6):E682–E691. doi: 10.1073/pnas.1316335111.
    1. Scheiermann C, Kunisaki Y, Frenette PS. Circadian control of the immune system. Nat Rev Immunol. 2013;13(3):190–198. doi: 10.1038/nri3386.
    1. Benloucif S, et al. Measuring melatonin in humans. J Clin Sleep Med. 2008;4(1):66–69.
    1. Shishkin AA, et al. Simultaneous generation of many RNA-seq libraries in a single reaction. Nat Methods. 2015;12(4):323–325. doi: 10.1038/nmeth.3313.
    1. Afik S, et al. Defining the 5′ and 3′ landscape of the Drosophila transcriptome with Exo-seq and RNaseH-seq. Nucleic Acids Res. 2017;45(11):e95. doi: 10.1093/nar/gkx133.
    1. Wang L, Wang S, Li W. RSeQC: quality control of RNA-seq experiments. Bioinformatics. 2012;28(16):2184–2185. doi: 10.1093/bioinformatics/bts356.
    1. Dobin A, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. doi: 10.1093/bioinformatics/bts635.
    1. Derr A, et al. End Sequence Analysis Toolkit (ESAT) expands the extractable information from single-cell RNA-seq data. Genome Res. 2016;26(10):1397–1410. doi: 10.1101/gr.207902.116.
    1. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26(1):139–140. doi: 10.1093/bioinformatics/btp616.
    1. R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Accessed June 29, 2018.
    1. Hastie T, Tibshirani R, Friedman JH, eds. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. 2nd ed. New York, New York, USA: Springer; 2001.
    1. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol. 1995;57(1):289–300.

Source: PubMed

Sponsorer og samarbejdspartnere

Medicinske tilstande

Narkotikainterventioner

3
Abonner