Page 94 - 2021_10-Haematologica-web

L. Grassi et al.
functional roles in the hematopoietic system. We estimat- ed the transcriptome complexity of each cell type, as it had previously been reported that whole blood is one of the least transcriptionally complex tissues.24 We found that, out of a mean of ~10,000 expressed protein-coding genes, the number accounting for 75% of each transcrip- tome ranged from 168 in PLT to 2,422 in resting HUVEC. These genes displayed an enrichment for GO terms relat- ing to basic cellular functions, rather than for terms relat- ing to the different functional phenotypes or identities, the only exception being PLT (Figure 2). These findings indi- cate that the genes allowing each cell type to perform its functions have a wide range of expression values and they form a unique, although partially overlapping, transcrip- tional signature. They also suggest that basic cellular func- tions are maintained even in those cell types with an extremely short half-life, such as neutrophils (Online Supplementary Table S5).
To identify the unique gene expression signatures of each cell type, we classified genes according to their cell type specificity after grouping the most similar cell types into functional categories because otherwise they would mutually erase their signals (Figure 3, Online Supplementary Table S2). Perhaps not surprisingly given the uniqueness of their function in the coagulation process, the largest signa- ture belongs to the MK/PLT category with 3,502 genes. The smallest signature (186 genes) belongs to the CD8TC group largely due to the overlap with the CD4TC group (Online Supplementary Table S7). As expected, the identi- fied signatures showed GO-term enrichment correspon- ding to the core functions of each cell type, with the exception of BAS, M0 and MONO. This is likely due to the considerable overlaps between the gene expression programs in many of these cell types, which causes the genes to which the primary functions of these cells are ascribed to, not to be selected. Overall, we found that almost 60% of known genes are differentially expressed in the hematopoietic system. Half of these have a high mean expression (log expression >0), whilst only a minority (3.5%) of the non-differentially expressed genes have high mean expression.
The annotation-agnostic nature of RNA-sequencing led us to identify novel genes using guided transcriptome assembly. This approach allowed us to identify 645 multi- exonic novel transcripts from 400 novel genes. The prop- erties of these novel genes, such as their overlap with transposons and repeat elements, low conservation, low expression levels, and high cell type specificity (Figure 4), are in agreement with observations in known lncRNA, as previously shown by Schwarzer and colleagues.53 The high cell type specificity, in particular, most likely explains why these transcripts have not been identified previously using more coarsely fractionated samples. Moreover, the nature of the library preparation (ribo-depletion, inde- pendent of poly-A tail) allowed us to expand the catalog of circRNA transcribed in blood and show that these ncRNA display high levels of cell type specificity (Figure
5). Our findings support the notion that some lncRNA and circRNA may have roles in determining cell fate and func- tions in hematopoiesis,53,54 in line with findings in other tissues and organs.55 However, further work is needed to understand the underlying mechanisms. Finally, to allow a wider access to these data, we created a web-based appli- cation (https://blueprint.haem.cam.ac.uk/bloodatlas/). Here, expression values at gene and transcript levels, as well as, expression values for microRNA, novel genes and circRNA can be visualized. Moreover, publication-ready graphical representations, together with expression values can also be downloaded.
Disclosures
PF is a member of the scientific advisory boards of Fabric Genomics, Inc., and Eagle Genomics, Ltd. None of the other authors has any conflicts of interest to declare.
Contributions
LG, OGI, NANJ, DS, RP, MK FP and ET performed analy- ses; MB, FB, SF, NF, JJL, SR, EMR and KD collected samples and generated data; FJM, AF, JMM, LC and PF provided data infrastructure; XE, PF, JHAM, MLY, HGS, WHO, PF and MF provided funding and infrastructure; FJM, AF, PF, FP, ET and MF supervised analyses; LG, OGI, PF, ET and MF conceptu- alised the analyses.
Acknowledgments
The authors would like to acknowledge the participation of National Institute of Health Research (NIHR) Cambridge BioResource volunteers and thank the NIHR Cambridge BioResource staff for their support.
Funding
The work was funded by a grant from the European Commission 7th Framework Program (FP7/2007–2013, grant 282510, BLUEPRINT) to XE, PF, JHAM, MY, HGS and WHO. WHO is an NIHR senior investigator and receives fund- ing from Bristol-Myers Squibb, the British Heart Foundation, the Medical Research Council and the NIHR. OGI, FJM, AF, JMM, LC and PF are funded by the Wellcome Trust (WT108749/Z/15/Z) with additional funding for specific proj- ect components such as GENCODE from the National Human Genome Research Institute of the National Institutes of Health (2U41HG007234). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. KD is a HSST trainee supported by NHS Health Education England. NF is funded by the NIHR Cambridge Biomedical Research Centre. FP is supported by the Fundação Carlos Chagas Filho de Amparo à Pesquisado Estado do Rio de Janeiro (FAPERJ; E-26/203.229/2016). NANJ is a recipient of a scholarship from the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES; Finance Code 001). The work by DS was supported in part by an Isaac Newton fellowship to MF. MF is supported by the British Heart Foundation (FS/18/53/33863).
References
1.Bagger FO, Sasivarevic D, Sohi SH, et al. BloodSpot: a database of gene expression profiles and transcriptional programs for healthy and malignant haematopoiesis.
Nucleic Acids Res. 2016;44(D1):D917-24. 2.Novershtern N, Subramanian A, Lawton LN, et al. Densely interconnected transcrip- tional circuits control cell states in human
hematopoiesis. Cell. 2011;144(2):296-309. 3. Laurenti E, Doulatov S, Zandi S, et al. The transcriptional architecture of early human
hematopoiesis identifies multilevel control of lymphoid commitment. Nat Immunol. 2013;14(7):756-763.
4. Caron H, van Schaik B, van der Mee M, et al. The human transcriptome map: clustering of highly expressed genes in chromosomal domains. Science. 2001;291(5507):1289-1292.
2622
haematologica | 2021; 106(10)