Generation of an isoform-level transcriptome atlas of macrophage activation

Abstract

RNA-seq is routinely used to measure gene expression changes in response to cell perturbation. Genes upregulated or downregulated following some perturbation are designated as genes of interest, and their most expressed isoform(s) would then be selected for follow-up experimentation. However, because of its need to fragment RNA molecules, RNA-seq is limited in its ability to capture gene isoforms and their expression patterns. This lack of isoform-specific data means that isoforms would be selected based on annotation databases that are incomplete, not tissue specific, or do not provide key information on expression levels. As a result, minority or nonexistent isoforms might be selected for follow-up, leading to loss in valuable resources and time. There is therefore a great need to comprehensively identify gene isoforms along with their corresponding levels of expression. Using the long-read nanopore-based R2C2 method, which does not fragment RNA molecules, we generated an Isoform-level transcriptome Atlas of Macrophage Activation that identifies full-length isoforms in primary human monocyte-derived macrophages. Macrophages are critical innate immune cells important for recognizing pathogens through binding of pathogen-associated molecular patterns to toll-like receptors, culminating in the initiation of host defense pathways. We characterized isoforms for most moderately-to-highly expressed genes in resting and toll-like receptor–activated monocyte-derived macrophages, identified isoforms differentially expressed between conditions, and validated these isoforms by RT-qPCR. We compiled these data into a user-friendly data portal within the UCSC Genome Browser (https://genome.ucsc.edu/s/vollmers/IAMA). Our atlas represents a valuable resource for innate immune research, providing unprecedented isoform information for primary human macrophages. RNA-seq is routinely used to measure gene expression changes in response to cell perturbation. Genes upregulated or downregulated following some perturbation are designated as genes of interest, and their most expressed isoform(s) would then be selected for follow-up experimentation. However, because of its need to fragment RNA molecules, RNA-seq is limited in its ability to capture gene isoforms and their expression patterns. This lack of isoform-specific data means that isoforms would be selected based on annotation databases that are incomplete, not tissue specific, or do not provide key information on expression levels. As a result, minority or nonexistent isoforms might be selected for follow-up, leading to loss in valuable resources and time. There is therefore a great need to comprehensively identify gene isoforms along with their corresponding levels of expression. Using the long-read nanopore-based R2C2 method, which does not fragment RNA molecules, we generated an Isoform-level transcriptome Atlas of Macrophage Activation that identifies full-length isoforms in primary human monocyte-derived macrophages. Macrophages are critical innate immune cells important for recognizing pathogens through binding of pathogen-associated molecular patterns to toll-like receptors, culminating in the initiation of host defense pathways. We characterized isoforms for most moderately-to-highly expressed genes in resting and toll-like receptor–activated monocyte-derived macrophages, identified isoforms differentially expressed between conditions, and validated these isoforms by RT-qPCR. We compiled these data into a user-friendly data portal within the UCSC Genome Browser (https://genome.ucsc.edu/s/vollmers/IAMA). Our atlas represents a valuable resource for innate immune research, providing unprecedented isoform information for primary human macrophages. The use of RNA-seq is a primary strategy in biomedical research to identify genes involved in biological processes of interest and how gene expression is impacted upon gene editing or use of chemical or biological agonists. Notably, short-read sequencing technology has reliably been used to quantify changes in gene expression levels or the inclusion level of individual exons and splice junctions. However, because short-read RNA-seq relies on fragmenting RNA molecules before sequencing, even advanced computational tools fail at leveraging this ubiquitous data type into isoform-level information (1Pertea M. Pertea G.M. Antonescu C.M. Chang T.-C. Mendell J.T. Salzberg S.L. StringTie enables improved reconstruction of a transcriptome from RNA-seq reads.Nat. Biotechnol. 2015; 33: 290-295Crossref PubMed Scopus (4506) Google Scholar, 2Bankevich A. Nurk S. Antipov D. Gurevich A.A. Dvorkin M. Kulikov A.S. Lesin V.M. Nikolenko S.I. Pham S. Prjibelski A.D. Pyshkin A.V. Sirotkin A.V. Vyahhi N. Tesler G. Alekseyev M.A. et al.SPAdes: A new genome assembly algorithm and its applications to single-cell sequencing.J. Comput. Biol. 2012; 19: 455-477Crossref PubMed Scopus (13451) Google Scholar, 3Grabherr M.G. Haas B.J. Yassour M. Levin J.Z. Thompson D.A. Amit I. Adiconis X. Fan L. Raychowdhury R. Zeng Q. Chen Z. Mauceli E. Hacohen N. Gnirke A. Rhind N. et al.Full-length transcriptome assembly from RNA-seq data without a reference genome.Nat. Biotechnol. 2011; 29: 644-652Crossref PubMed Scopus (12578) Google Scholar). Short-read RNA-seq ultimately falls short in providing comprehensive and accurate full-length isoform structures as well as the level of expression of each isoform under specific conditions. More recently, long-read technologies from Pacific Biosciences and Oxford Nanopore Technologies (ONT) have been used for sequencing and analyzing full-length cDNA molecules at the transcriptome scale (4Gupta I. Collier P.G. Haase B. Mahfouz A. Joglekar A. Floyd T. Koopmans F. Barres B. Smit A.B. Sloan S.A. Luo W. Fedrigo O. Ross M.E. Tilgner H.U. Single-cell isoform RNA sequencing characterizes isoforms in thousands of cerebellar cells.Nat. Biotechnol. 2018; https://doi.org/10.1038/nbt.4259Crossref Scopus (134) Google Scholar, 5Workman R.E. Tang A.D. Tang P.S. Jain M. Tyson J.R. Razaghi R. Zuzarte P.C. Gilpatrick T. Payne A. Quick J. Sadowski N. Holmes N. de Jesus J.G. Jones K.L. Soulette C.M. et al.Nanopore native RNA sequencing of a human poly(A) transcriptome.Nat. Methods. 2019; 16: 1297-1305Crossref PubMed Scopus (200) Google Scholar, 6Lebrigand K. Magnone V. Barbry P. Waldmann R. High throughput error corrected nanopore single cell transcriptome sequencing.Nat. Commun. 2020; 11: 4025Crossref PubMed Scopus (54) Google Scholar, 7Wyman D. Balderrama-Gutierrez G. Reese F. Jiang S. Rahmanian S. Zeng W. Williams B. Trout D. England W. Chu S. Spitale R.C. Tenner A. Wold B. Mortazavi A. A technology-agnostic long-read analysis pipeline for transcriptome discovery and quantification.bioRxiv. 2019; ([preprint])https://doi.org/10.1101/672931Crossref Scopus (0) Google Scholar). In contrast to RNA-seq, this technology can determine which isoforms, down to the exact transcription start and poly(A) sites, are expressed at what level by each gene. The comprehensive transcriptome scale isoform information these technologies provide has the potential to remove the need for targeted and work intensive methods like RT-PCR and 5’/3’ RACE to identify and characterize transcript and/or protein isoforms expressed by a gene. Therefore, comprehensive transcriptome scale isoform information is bound to simplify and improve the outcome of single gene focused follow-up studies which include knock-down and knock-out experiments, overexpression assays, Western Blots, ELISAs, pull-downs, and many more. This is because of the fact that these assays rely on prior knowledge of what isoform(s) the gene of interest actually expresses in the condition and experimental system being investigated. Finally, detailed knowledge of transcription start sites (TSSs) for each expressed gene in a cell type will also improve the use of CRISPR interference technology to knock down genes because guide RNAs can be targeted to TSSs with greater accuracy. To build on our previous work (8Byrne A. Beaudin A.E. Olsen H.E. Jain M. Cole C. Palmer T. DuBois R.M. Forsberg E.C. Akeson M. Vollmers C. Nanopore long-read RNAseq reveals widespread transcriptional variation among the surface receptors of individual B cells.Nat. Commun. 2017; 8: 16027Crossref PubMed Scopus (196) Google Scholar, 9Robinson E.K. Jagannatha P. Covarrubias S. Cattle M. Safavi R. Song R. Viswanathan K. Shapleigh B. Abu-Shumays R. Jain M. Cloonan S.M. Wakeland E. Akeson M. Brooks A.N. Carpenter S. Inflammation drives alternative first exon usage to regulate immune genes including a novel iron regulated isoform of Aim2.bioRxiv. 2020; ([preprint])https://doi.org/10.1101/2020.07.06.190330Crossref Scopus (0) Google Scholar) and further push the limits of long-read technology to provide a resource for the innate immune research community, we set out to generate an isoform-level transcriptome atlas of macrophage activation by determining (1) what isoform of a given gene is expressed, (2) at what level, and (3) how isoform and gene expression change following toll-like receptor (TLR) activation. Macrophages are a key cellular component of the innate immune system which represents the first line of host defense against infection and is critical for the development of adaptive immunity (10Medzhitov R. Horng T. Transcriptional control of the inflammatory response.Nat. Rev. Immunol. 2009; 9: 692-703Crossref PubMed Scopus (766) Google Scholar, 11Carpenter S. Aiello D. Atianand M.K. Ricci E.P. Gandhi P. Hall L.L. Byron M. Monks B. Henry-Bezy M. Lawrence J.B. O’Neill L.A.J. Moore M.J. Caffrey D.R. Fitzgerald K.A. A long noncoding RNA mediates both activation and repression of immune response genes.Science. 2013; 341: 789-792Crossref PubMed Scopus (727) Google Scholar). Macrophages recognize conserved structures of microbial-derived molecules or pathogen associated molecular patterns using TLRs. The regulation of this TLR repertoire fundamentally alters the response to infection (12Kawai T. Akira S. Toll-like receptor and RIG-I-like receptor signaling.Ann. N. Y. Acad. Sci. 2008; 1143: 1-20Crossref PubMed Scopus (752) Google Scholar). TLR activation induces the expression of hundreds of genes that encode inflammatory response genes including cytokines, type I interferons, antimicrobial proteins, and regulators of metabolism and regeneration; these molecules in turn mediate inflammation, antimicrobial immunity, and tissue regeneration. Here, we investigated transcriptional responses of human macrophages treated with lipopolysaccharide (LPS), Pam3CSK4 (PAM), R848, and poly(I:C) which activate TLR4, TLR1/2, TLR7/8, and TLR3, respectively. Using our ONT-based R2C2 method, we then generated a total of ∼15 million full-length cDNA reads at a median accuracy >99% (Q20) and processed this data into isoforms which we characterized in depth and provide alongside deep Smart-seq2 short-read RNA-seq data as a UCSC Genome Browser session for easy exploration. To generate a comprehensive isoform-level transcriptome atlas of TLR-dependent macrophage activation, we collected peripheral blood mononuclear cells from two individuals (Rep1 and Rep2) from which we isolated monocytes. From these monocytes, we generated monocyte-derived macrophages (MDMs). We treated these MDMs with TLR ligands LPS, PAM, R848, or poly(I:C) and included a no treatment (NoStim) control. After 6 h, we collected the stimulated and nonstimulated MDMs and proceeded to extract RNA from each sample. We reverse transcribed the poly(A) fraction of this RNA using a modified oligo(dT) primer and a template switch oligo to generate full-length cDNA with known sequences on both ends. We then amplified this cDNA using PCR and used the resulting double-stranded full-length cDNA as input for both Illumina-based Smart-seq2 (13Picelli S. Faridani O.R. Björklund A.K. Winberg G. Sagasser S. Sandberg R. Full-length RNA-seq from single cells using Smart-seq2.Nat. Protoc. 2014; 9: 171-181Crossref PubMed Scopus (1952) Google Scholar) and ONT-based R2C2 (14Volden R. Palmer T. Byrne A. Cole C. Schmitz R.J. Green R.E. Vollmers C. Improving nanopore read accuracy with the R2C2 method enables the sequencing of highly multiplexed full-length single-cell cDNA.Proc. Natl. Acad. Sci. U. S. A. 2018; 115: 9726-9731Crossref PubMed Scopus (90) Google Scholar) sequencing protocols (Fig. 1). To identify genes differentially expressed upon TLR activation following treatment with LPS, PAM, R848, or poly(I:C), we performed Illumina-based Smart-seq2 (15Picelli S. Björklund A.K. Reinius B. Sagasser S. Winberg G. Sandberg R. Tn5 transposase and tagmentation procedures for massively scaled sequencing projects.Genome Res. 2014; 24: 2033-2040Crossref PubMed Scopus (378) Google Scholar) sequencing as previously described (16Cole C. Byrne A. Adams M. Volden R. Vollmers C. Complete characterization of the human immune cell transcriptome using accurate full-length cDNA sequencing.Genome Res. 2020; 30: 589-601Crossref PubMed Scopus (17) Google Scholar, 17Byrne A. Supple M.A. Volden R. Laidre K.L. Shapiro B. Vollmers C. Depletion of hemoglobin transcripts and long-read sequencing improves the transcriptome annotation of the polar bear (Ursus maritimus).Front. Genet. 2019; 10: 643Crossref PubMed Scopus (11) Google Scholar) (Fig. S1, see Experimental procedures). We generated approximately 15 to 30 million reads per sample (Table S1) and processed the resulting data using a standard workflow which includes STAR (18Dobin A. Davis C.A. Schlesinger F. Drenkow J. Zaleski C. Jha S. Batut P. Chaisson M. RNA-seq 2013; 29: PubMed Scopus Google Y. W. for reads to 2014; 30: PubMed Scopus Google and W. S. of change and for RNA-seq data with Biol. 2014; PubMed Scopus Google Scholar). of our biological we the LPS, PAM, R848, and poly(I:C) to the control. the expression of to genes and conditions, genes a between and two and some to (Fig. Using analysis M.J. A. B. R. K. A. A. A of protein and by Res. PubMed Scopus Google Scholar, Q. A. P. S. and with the Res. PubMed Scopus Google we that genes between for biological processes including to and inflammatory (Table Notably, the genes to for the to biological because is an To this we the full-length double-stranded cDNA using our ONT-based R2C2 (14Volden R. Palmer T. Byrne A. Cole C. Schmitz R.J. Green R.E. Vollmers C. Improving nanopore read accuracy with the R2C2 method enables the sequencing of highly multiplexed full-length single-cell cDNA.Proc. Natl. Acad. Sci. U. S. A. 2018; 115: 9726-9731Crossref PubMed Scopus (90) Google Scholar, 17Byrne A. Supple M.A. Volden R. Laidre K.L. Shapiro B. Vollmers C. Depletion of hemoglobin transcripts and long-read sequencing improves the transcriptome annotation of the polar bear (Ursus maritimus).Front. Genet. 2019; 10: 643Crossref PubMed Scopus (11) Google Scholar, R. Vollmers C. multiplexed single-cell full-length cDNA sequencing of human immune cells with and 2020; Scopus (0) Google Scholar) (Fig. 1). R2C2 cDNA and then the resulting to generate long molecules of the The resulting long is then and into We then these to generate accurate reads of the cDNA To the of this macrophage isoform atlas we to the R2C2 we of cDNA by sample into the oligo(dT) as well as using highly for cDNA (Fig. 1). the we used these two to and biological (Table to in on the sequencing between and that and cDNA of and we generated R2C2 reads at a median of cells (Fig. the experiments, we improved R2C2 per read accuracy by the read of R2C2 We this by a that the read of our R2C2 from to with a new this the median per accuracy of our R2C2 this accuracy (16Cole C. Byrne A. Adams M. Volden R. Vollmers C. Complete characterization of the human immune cell transcriptome using accurate full-length cDNA sequencing.Genome Res. 2020; 30: 589-601Crossref PubMed Scopus (17) Google Scholar). Here, our most sequencing an accuracy of (Fig. and including accurate experiments, the ∼15 million reads generated for this a median accuracy of to of this improved accuracy and the of isoforms from the R2C2 reads we we a new of our pipeline among includes improved using the and improved of isoform ends. this pipeline on the ∼15 million read data we identified isoforms with a median of and a per accuracy of which the for accuracy K. T. R. M. Olsen H.E. C. J. K. N. S. T. S. V. et al.Nanopore sequencing and the de assembly of human Biotechnol. 2020; PubMed Scopus (134) Google Scholar) (Fig. we which genes these isoforms transcribed from and to what isoform on gene expression levels. Our analysis identified expressed genes reads per million and Our analysis identified at isoform for of these Our ability to identify at isoform for a gene the gene expressed at a (Fig. We identified at isoform for of genes between and This with to and differentially expressed genes to have expression (Fig. we identified at isoform for of the genes expressed in and of genes expressed in conditions. The of isoforms identified per gene also with (Fig. However, the median of isoforms per gene not even in highly expressed This is because of the we used which isoforms with of the R2C2 reads at a and minority we a pipeline to genes isoforms differentially expressed between (Fig. like be to this would isoforms from differentially expressed To genes isoform usage between A isoform A isoform following we a each we first the isoforms transcribed from that gene. we the usage for each isoform in each condition by the of R2C2 reads associated with that isoform by the of R2C2 reads associated with the isoforms in that We then the to the resulting usage To we the genes with at R2C2 reads in at two experimental conditions. After this genes with isoform usage with (Table of these genes not identified as differentially expressed by the Smart-seq2 short-read The genes are to isoforms usage between conditions. To we the standard of this isoform usage between for each isoform in each gene (Fig. the genes we by the standard among their isoforms, we that this identify genes with isoforms that have highly usage between conditions. To further the of this we validated the changes in the usage of alternative individual exons for the genes with isoform usage following treatment using macrophages from an Using we that in we exon usage in the in the or with many change to that by R2C2 in the two (Fig. The of the genes alternative TSSs which in in and first exons (Fig. To how the use of exons is isoforms and how this their we first the isoforms we To this we used the M. de L. C. C. M. M. M. K. M. I. J. M. Mortazavi A. et characterization of long-read transcript sequences for control in full-length transcriptome and Res. 2018; PubMed Scopus Google Scholar) algorithm which isoforms with genes and as novel in novel not in and (Fig. and isoforms are as or the of an are as isoforms that use splice sites in are as isoforms that use at splice (Fig. isoforms isoforms isoforms and isoforms of the isoforms we identified (Table associated with a isoforms of to a of the gene associated of of of and of isoforms, which a of the gene transcribed from (Fig. isoforms which not a to from the transcript in their and the and of isoforms in their to the and sites of the transcript associated with (Fig. into that and of a isoform be to the or of transcript in their we that of the isoforms we and a from or The of with in isoforms at in be by exons being on first exons (Fig. isoforms, which not a of the gene associated a new splice between and which would a isoforms which not a of the gene associated with might from the by a to encode a or new exons (Fig. we focused on isoforms to new We a new exon as a of a transcript does not with a known exon at (Fig. In the isoforms we new exons of which first and new exons among the exons of the isoforms we we would new first and that first and exons are in this set of new exons (Fig. these new exons exons in the annotation or exons identified by (Fig. the of new and exons the of exons with exons being to first and Finally, the of new exons be validated with short read Smart-seq2 leading into these exons for two for in Smart-seq2 reads generated from the cDNA in of This that these new exons are highly to be in the cDNA we In to new isoforms in long noncoding RNA of isoforms associated with genes as with for both and This is because of the fact that are expressed at levels in a tissue specific genes which their comprehensive annotation T. R. G. A. S. Tilgner G. D. A. J. L. X. Y. T. et of human long noncoding of their gene and Res. 2012; PubMed Scopus Google Scholar). Our data set enables the of these and their in macrophage activation. In to isoforms, isoforms into the are also to noncoding transcripts genes have been in the human In isoforms as as not in the annotation (Table these isoforms, isoforms in turn into of these with from deep short-read data C. L. M. B. A. annotation of human noncoding RNAs reveals and specific 2011; PubMed Scopus Google Scholar). This that specific cell under has the potential to identify new previously like are for are to follow-up potential from a or an RNA-seq Our data set this by providing information on which of the isoforms in the annotation is actually expressed by a gene of interest and at what level is expressed with isoforms of that gene. To this type of of the data set is as a UCSC genome session (https://genome.ucsc.edu/s/vollmers/IAMA). This session gene expression isoform and as well as R2C2 and Smart-seq2 read To the of the Isoform-level transcriptome Atlas of Macrophage Activation we an of and both to be differentially upregulated by TLR ligands based on short-read RNA-seq these in the genome that a of R2C2 reads to and that short RNA-seq reads to are therefore from the isoforms based on these R2C2 reads also no for for (Fig. the isoforms in the some the the isoforms that is expressed the in the of R2C2 reads in that The of can then be for analysis by its in the analysis of a expression and in a specific under experimental conditions, to and quantify This is because of two of these (1) the use of short-read RNA-seq methods which the and of isoforms and (2) using long-read methods are limited to the analysis of a limited of at which will isoforms specific to a and under experimental conditions. The data set and we will be of use to human macrophages at both and TLR activation and provide a for studies short and long-read transcriptome The analysis of the transcriptome we generated that isoform expression between is limited and most associated with the usage of TSSs which is to we have previously in and human macrophages E.K. Jagannatha P. Covarrubias S. Cattle M. Safavi R. Song R. Viswanathan K. Shapleigh B. Abu-Shumays R. Jain M. Cloonan S.M. Wakeland E. Akeson M. Brooks A.N. Carpenter S. Inflammation drives alternative first exon usage to regulate immune genes including a novel iron regulated isoform of Aim2.bioRxiv. 2020; ([preprint])https://doi.org/10.1101/2020.07.06.190330Crossref Scopus (0) Google Scholar). This that the of genes is between the we investigated. We further that most isoforms we identify the of an isoform not its and We also hundreds of new for first and The of these exons from the annotation be by or biological of previous studies and annotation Short-read RNA-seq, which is most used as the for is with transcript ends. these new exons might be expressed in or macrophages and even then at levels. we the data set a into macrophage the in to to their gene or genes of We that for might be in a gene expressed in macrophages activation will the isoform with the expression in that condition from our data set to its exact for follow-up studies or for its for CRISPR interference

Topics & Concepts