|Home | About | Journals | Submit | Contact Us | Français|
The immune system develops in waves during ontogeny, being initially populated by cells generated from fetal hematopoietic stem cells (HSCs) and later by cells derived from adult HSCs. Remarkably, the genetic programs that control these two distinct stem cell fates remain poorly understood. We report that Lin28b is specifically expressed in mouse and human fetal liver and thymus, but not in adult bone marrow or thymus. We demonstrate that ectopic expression of Lin28 reprograms hematopoietic stem/progenitor cells (HSPCs) from adult bone marrow, endowing them with the ability to mediate multi-lineage reconstitution that resembles fetal lymphopoiesis, including increased development of B-1a, marginal zone B, gamma/delta (γδ) T cells, and natural killer T (NKT) cells.
Hematopoietic stem cells give rise to all the erythroid, myeloid and lymphoid lineages and are used clinically in transplants to treat patients with a wide variety of blood and immune system disorders that include immunodeficiencies and malignancies. In mice, the major site of hematopoiesis transitions from the fetal liver (FL) to the fetal spleen (FS) and bone marrow (BM) during late embryogenesis. FL HSCs differ phenotypically and functionally from adult BM HSCs and a developmental switch is thought to occur within the first weeks of post-natal life (1–3). Adoptive transfer and fetal thymic organ culture studies established that FL HSPCs, unlike adult BM HSPCs, preferentially give rise to lymphocyte subsets that can be collectively referred to as innate-like lymphocytes. For example, CD5+ B cells, referred to as the B-1a subset, and certain γδ T cell subsets develop almost exclusively in the fetus and/or neonate (4–7). More recently, a fetal source of HSPCs was found to preferentially give rise to the B-1a and marginal zone (MZ) B cell lineages (8). In contrast, adult BM HSPCs are less efficient at reconstituting the B-1a, and fetal γδ-T cell subsets compared to conventional B and T lymphocyte pools. Although the development of these innate-like lymphocyte subsets requires recombination activating gene (RAG)-mediated V(D)J recombination at the T or B cell antigen receptor loci, their repertoire has been described as semi-invariant or oligo-clonal (9). Similar to cells of the innate immune system, they are often strategically localized in the body to respond rapidly to a limited set of conserved antigens. Despite being potent mediators of host defense, the molecular basis for their preference to develop in the fetus or neonate largely remains a mystery. The ability to generate B-1a B cells and Vγ3+ γδ-T cells (nomenclature used) is intrinsically programmed in the FL HSCs (5, 10, 11). Sox17 was identified as a transcription factor that is specifically expressed in and required for the maintenance of fetal but not adult HSCs (12). Thus, transcriptional control can distinguish the fetal from adult HSC fate. In this study, we show that a layer of post-transcriptional control involving Lin28b and let-7 microRNAs (miRNAs) contributes to the distinct differentiation potential of these two stem cell types.
To identify the molecular differences between fetal and adult lymphoid progenitors, we performed global miRNA-expression profiling of FL and adult BM progenitor B (pro-B) cells (B220+CD43+IgM−veCD19+CD24+). This population was chosen because we could readily sort sufficient numbers of these progenitors to high purity and led to the observation that expression of the let-7 family of miRNAs was largely absent in FL pro-B cells. In the mouse genome, there are twelve let-7 paralogs that result in the expression of nine distinguishable mature miRNA isoforms (isomiRs) (Fig. S1). The let-7 miRNAs are broadly and highly expressed in the hematopoietic system (13). Surprisingly, the eight let-7 isomiRs that were detectable in adult BM pro-B cells were either significantly reduced in expression or undetectable in FL pro-B cells (Fig. 1A). From the same FACS sort, we had isolated the B220+CD43+IgM−ve CD19−veCD24−ve precursors to the pro-B cells (pre-pro-B cells). We determined that mature let-7g and let-7c isomiRs were abundantly expressed in pre-pro-B cells from adult BM but not from FL using quantitative RT-PCR (qRT-PCR) (Fig. 1B). Next, we examined if the observed global repression of let-7 miRNAs in fetal B-cell progenitors was mediated by endogenous Lin28 or Lin28b, encoded by two evolutionarily conserved paralogs, known to specifically block the biogenesis of let-7 miRNAs post-transcriptionally by binding to the terminal loop region of the let-7 primary or precursor miRNAs (pri- or pre-miRNAs) (14–17). We determined in FACS-sorted populations that Lin28b mRNA was abundant in all stages of FL B cell development (Fig. 1C) and also in FL lineage (Lin)−veSca-1+c-Kit+ (LSK) HSCs and Lin−ve c-Kitint Sca-1int IL7Rα+ common lymphoid progenitors (CLPs) (Fig. 1D) but absent the corresponding populations in adult BM. In contrast, Lin28 mRNA was not detectable in fetal HSPCs (Fig. S2A). These findings indicate that Lin28b expression is limited to HSPCs of fetal origin and could account for the observed reduction in mature let-7 expression.
Classic fetal thymus (FT) graft studies suggest that fetal HSPCs seed the thymus between 10 and 13 days of gestation, whereas cells derived from adult HSPCs begin to substantially dilute the first generation thymocytes at day 7 after birth, and have largely replaced them by two weeks of age (18). To test if Lin28b mRNA expression in the thymus correlates with the presence of fetal HSC-derived cells we performed qRT-PCR on various FACS-sorted thymocyte populations ranging from the double negative (DN)2/3 (CD4−veCD8−veCD25+CD44int) stage to the CD4+CD8−ve single positive (CD4SP) stage from 1 day, 1 week and 4 week-old mice. We observed abundant Lin28b transcripts in all the tested thymocyte subsets from 1 day-old thymi with the greatest amount observed in the γδ-TCR+CD3+ thymocytes (Fig. 1E). The abundance of endogenous Lin28b mRNA expression was dramatically decreased in thymocytes by 7 days after birth and was undetectable by 4 weeks of age. Declining Lin28b expression coincided with rising levels of mature let-7g (Fig. S2B). These data demonstrate that endogenous Lin28b mRNA expression declines in concordance with the decrease in the proportion of thymocyte progenitors of fetal origin in the postnatal mouse. Interestingly, Lin28b expression was not maintained in the thymus or peripheral lymphoid populations in sub-lethally irradiated adult Rag1−/− mice reconstituted with FL HSPCs (Fig. S2C). These findings suggest that Lin28b expression is turned on during fetal hematopoiesis but not sustained during ontogeny.
To determine if these observed fetal signatures in mouse HSPCs are conserved in humans, we compared Lin28b and let-7g expression in human FL, FT, FS and CD34+ umbilical cord blood (CB) cells, with that in adult BM, lymph node and thymus. As in mice, Lin28b mRNA expression was exclusively detected in fetal human hematopoietic tissues (Fig. 1F), whereas mature let-7g was more abundant in the adult (Fig. S2D). Taken together, these expression analyses reveal a previously unknown molecular difference between HSPCs of fetal and adult origin that is conserved between human and mice and prompted us to identify a heterochronic role of the Lin28b/let-7 axis in regulating hematopoiesis during mammalian ontogeny.
To test if ectopic expression of Lin28 can confer characteristics of fetal lymphopoiesis in adult mice, we used a retrogenic BM chimera system. Lin28 was chosen because an excellent antibody was available to confirm its expression. HSPCs from lineage-depleted adult BM of wild-type C57BL/6 mice were transduced with a Lin28-encoding retrovirus and transplanted into sub-lethally irradiated Rag1−/−recipients. Full-length mouse Lin28 cDNA was cloned into a mouse stem cell virus-based vector upstream of an internal ribosomal entry site followed by GFP (Lin28-RV) (19). Lineage-depleted C57BL/6 bone marrow cells from adult mice were transduced with Lin28-RV or control GFP-RV ex vivo and injected intravenously into sub-lethally irradiated Rag1−/− recipients (Fig. S3A). The initial transduction efficiency varied between 20–60% as determined by frequency of GFP+ cells 24 hours post-infection (Fig. 2A). HSPCs transduced with Lin28-RV mediated multi-lineage reconstitution with grossly normal proportions of T, B and myeloid cells in the peripheral lymphoid organs 6–8 weeks following transplantation in recipient mice hereafter referred to as Lin28-RV BM chimeras (Fig. 2B, S3). Ectopic Lin28 protein expression was confirmed by western blot in thymocytes of Lin28-RV BM chimeras (Fig. 2C). To assess the effect of ectopic Lin28 protein on miRNA expression, we performed global miRNA profiling of sorted GFP+ and GFP−ve pre-selection double positive (DP) thymocytes (CD3loCD4+CD8+) (Fig. S4A). Our results demonstrate that the impact of Lin28-RV on miRNA expression is dramatic and highly specific to the let-7 family members (Fig. 2D). Diminished expression of mature let-7a and let-7g was further validated by qRT-PCR (Fig. 2E). To assess the impact on the transcriptome, we performed deep sequencing of cDNA generated from poly-A+ RNA (RNA-seq) comparing sorted GFP+ and GFP−ve pre-selection DP thymocytes in Lin28-RV BM chimeras (Fig. S4). Statistical analysis using the Sylamer software demonstrated enrichment only of the complementary let-7 family heptameric seed sequence TACCTCA in 3′ UTRs of genes up-regulated in GFP+ DP thymocytes (Fig. S5) and not the 969 other heptameric seed sequences that were also tested. Out of the 175 up-regulated genes containing this seed motif and passing Bonferroni-corrected E-value threshold of 0.01 identified using Sylamer, we observed a higher than expected overlap of 47 and 67 genes with let-7 target genes phylogenetically predicted by PicTar and TargetScan respectively (Tables S1 and S2) (20, 21). Thus, Lin28-RV mediated a dramatic decrease in steady-state levels of mature let-7 miRNAs in our retrogenic system with minimal effects on other miRNAs, resulting in the specific global de-repression of putative let-7 target mRNAs. These data reveal that despite abundant let-7 expression throughout the adult hematopoietic system, we can readily knock-down their expression by Lin28-RV transduction without inhibiting multi-lineage reconstitution.
It has been reported that FL HSPCs preferentially differentiate into B-1a cells compared to adult HSPCs (6, 7). To test for the effect of Lin28-RV transduction on the development of B-1a cells in adult mice, the peritoneal cavities of Lin28 retrogenic BM chimeras were analyzed 6–8 weeks post-transplant for donor-derived B cells. Remarkably, Lin28-RV transduced adult HSPCs efficiently generated the B-1a subset, whereas untransduced and GFP-RV transduced adult HSPC controls were biased towards the generation of conventional B-2 cells (Fig. 3A, 3B and S6). In addition to the B-1a subset, growing evidence suggests that a significant component of the MZ B-cell compartment is also of fetal origin (8, 22, 23). Consistent with this notion, Lin28-RV transduced HSPCs reproducibly gave rise to an increased percentage of B220+CD1d+CD23−ve splenic MZ B cells compared to their untransduced (GFP−ve) or GFP-RV transduced controls (Fig. 3C and 3D). Similar results were obtained from hLin28b-RV BM chimeras (Fig. S7A and S7B). These data indicate that the Lin28b/let-7 axis may account for the long-known differences between fetal and adult fate bias in B cell development and that ectopic Lin28 or Lin28b expression in HSPCs is sufficient to confer fetal-like B-lymphopoieisis.
Signaling through the IL-7 receptor α-chain (IL7Rα) has been reported to be required for adult B cell development but dispensable for fetal B-lymphopoiesis (1). To further examine the notion that ectopic Lin28 expression reprograms adult BM HSPCs to acquire fetal-like B-cell development we investigated whether Lin28-RV could restore B cell potential in adult Il7rα−/−BM HSPCs. Recipients reconstituted with Il7rα−/−HSPCs transduced with Lin28-RV were analyzed 7 weeks post-transplant for the presence of B cells in the peritoneal cavity and the spleen. In accordance with previous reports, control GFP−ve IL7Rα-deficient donor cells failed to give rise to any detectable population of mature CD19+B220+ B cells. In contrast, sizable populations of GFP+ Lin28-induced B cells were detected in peritoneal cavities and spleens of Lin28-RV BM chimeras (Fig. 3E). Reminiscent of fetal lymphopoiesis, we observed an over-representation of the B-1a and MZ B cell lineages. These results demonstrate that Lin28 acts as a potent reprogramming factor in adult HSPCs to confer fetal-like aspects of B cell development that include independence from IL7Rα-mediated signaling.
Adult thymocytes mainly differentiate into αβ-T cells but also Vγ2, Vγ1.1 and Vγ5 expressing γδ-T cell subsets while fetal thymocytes preferentially rearrange the Vγ3 and Vγ4 TCR gene segments (24). In addition, one sub-population of Vγ1.1+ T cells is the CD4SP+ Vγ1.1+Vδ6.3+ innate-like T cell subset and has been described to develop perinatally originating primarily from late embryonic precursors (25). Strikingly, we observed a over 15-fold increase in the percentage of total γδ-TCR+ CD3+ cells among GFP+ thymocytes in Lin28-RV BM chimeras compared to either the GFP−ve thymocytes in the same mouse or GFP+ thymocytes in the GFP-RV BM chimeras (Fig. 4A and 4B). The Lin28-induced γδ T cells are primarily CD4SP (Fig. 4C), contributing to an expansion of the overall CD4SP thymocyte compartment (Fig. S8A), and do not express the adult-specific Vγ2-TCR chain or the fetal-specific the Vγ3-TCR chain. The latter is expected since the development of Vγ3+ T cells is known to require both fetal HSCs and a fetal thymus environment (5). Remarkably, we identified Lin28-induced γδ-T cells to be almost exclusively carrying the Vγ1.1+Vδ6.3+ innate-like γδ-TCR that normally arise perinatally (Fig. 4D). Both the innate-like CD4+Vγ1.1+Vδ6.3+ and invariant Vα14+Jα18+ NKT (iNKT) cells are known to express the PLZF transcription factor, and the development of PLZF+ CD4 T-cells is a physiological process in humans during fetal and perinatal stages of ontogeny (26). Indeed, we were able to demonstrate an increased representation of PLZF+γδ-TCR+CD3+ thymocytes in Lin28-RV BM chimeras (Fig. S8B). The ability of Lin28 to induce γδ T cell development correlates with the finding that γδ-TCR+ thymocytes naturally express the highest levels of Lin28b in neonatal mice (Fig. 1E). In line with our findings regarding B cell development, these data support the ability of Lin28 to confer fetal-like lymphopoiesis in the γδ T cell lineage.
Although the invariant NKT (iNKT) cell lineage has not been directly demonstrated to preferentially arise from fetal precursors, several groups have described a transient burst of rapidly proliferating iNKT cell precursors around day 12 after birth (27, 28), consistent with a biased ontogenic window of development. Our discovery of an increased PLZF+ thymocyte population in thymi of Lin28-RV BM chimeras prompted us to further evaluate the effect of Lin28 on iNKT cell development. We observed a 100-fold increase in the percentage of iNKT (CD1dPBS57+ CD3+) cells among GFP+ thymocytes in the Lin28-RV BM chimeras compared to the GFP−ve thymocytes within the same mouse or a 30-fold increase compared to the GFP+ thymocytes in the GFP-RV BM chimeras (Fig. 5A, Fig. 5B). This dramatic increase in thymic iNKT cells contributes to a Lin28-induced expansion of the overall CD4SP and DN thymocyte compartments (Fig. S8A) as well as the PLZF+ thymocyte pool (Fig. S8B). A dramatic increase in the frequency of GFP+ iNKT cells was also observed in the spleen, lymph node and liver of Lin28-RV BM chimeras (Fig. 5B), translating to higher than expected iNKT cell absolute numbers in the spleen (Fig. S8C). Positive selection of iNKT cells relies on thymocyte-thymocyte interactions mediated by CD1d (29). To determine whether CD1d is required for the development of Lin28-induced iNKT cells, we reconstituted Rag1−/− recipients with Cd1d−/− HSPCs transduced with Lin28-RV. No iNKT cells were detectable in these recipients consistent with the notion that their positive selection requires CD1d (Fig. S9).
Since it is not clear whether iNKT cells develop early during ontogeny, we performed flow cytometric analysis to resolve the proposed developmental stages of the iNKT cell lineage subsequent to the DP thymocyte stage: 1 (CD44−veNK1.1−ve), 2 (CD44+NK1.1−ve) and 3 (CD44+NK1.1+) (27). The composition of the Lin28-induced thymic CD3+CD1dPBS57+ population resembled that of a 2-week-old thymus, characterized by a marked increase in the percentage of immature stage 1 iNKT cells (Fig. 5C and 5D). Since DP thymocytes are precursors of stage 1 iNKT cells, we quantified the efficiency with which DP thymocytes in wild-type C57BL/6 mice of varying ages give rise to stage 1 iNKT by calculating the ratio of stage 1 iNKT cells to DP thymocytes. Consistent with previous reports, our results indicate a peak in iNKT cell potential between 2–4 weeks of age followed by a sharp decline (Fig. 5D and 5E). Thus, our data point towards a higher iNKT cell potential in DP thymocytes of embryonic or neonatal HSPC origin.
Increased numbers of thymic PLZF+ innate-like T cells can produce large amounts of IL-4 and thereby promote thymic development of Eomesodermin (Eomes)+ memory-like CD8 T cells, characterized by high surface expression of CD44 and low surface expression of CD24 (30–32). This phenomenon is not normally detectable in adult wild-type C57BL/6 mice due to the infrequency of PLZF+ thymocytes in this strain (31, 33). However, we observed a significant and global increase in memory-like CD8 T cells among both GFP+ and GFP−ve thymocytes in Lin28-RV BM chimeras but not in the GFP-RV control chimeras (Fig. S10). This result demonstrates the ability of Lin28-induced PLZF+ innate-like T cells to support the differentiation of memory-like CD8 T cells within the thymus and is consistent with the well-characterized IL4-mediated bystander effect that is not cell intrinsic (34).
It has been reported that the potential to develop into innate-like lymphocytes is intrinsically programmed in HSCs (5, 11). Therefore, we performed cell cycle and gene expression analyses to look for Lin28-induced fetal characteristics in the BM HSCs of Lin28-RV chimeras. Cell cycle analysis of Lin28-RV transduced BM HSPCs in culture compared to controls did not reveal any significant differences (Fig. S11B). However, this is an artificial system that is highly mitogenic for HSPCs. Furthermore, a number of surface markers including CD34, CD93 (AA4.1), and CD11b are preferentially expressed on FL HSCs, likely due to their heightened proliferative state (35–40). We did not detect any Lin28-induced changes in the expression of these markers, nor that of ESAM, CD48, and Tie-2 within the LSK HSC compartment in Lin28-RV chimeras (Fig. S12A). Together, our results suggest that ectopic Lin28 does not grossly affect the cell intrinsic proliferative capacity or the overall composition of the adult LSK HSC compartment and that increased proliferation alone cannot account for fetal-like developmental potential of HSPCs.
To characterize the effect of ectopic Lin28 on the long-term competence of HSCs, we performed FACS and qRT-PCR analysis of known long-term reconstituting (LT) HSC markers on GFP+ and GFP−ve LSK HSCs from aged Lin28-RV and GFP-RV BM chimeras 10 months following transplantation. CD150 surface expression marks long-term (LT) HSCs (41) and Mecom (Evi1) and Tal1 encode for transcription factors critical in regulating HSC quiescence and long-term competence (42, 43). The presence of GFP+ Lin28-RV transduced LSK HSCs 10 months post-adoptive transfer along with comparable expression of surface CD150 protein, Evi1 and Tal1 mRNA between Lin28-RV and control LSK HSCs suggest that ectopic Lin28 does not interfere with LT-HSCs (Fig. S12).
Hmga2 and Igf2 have been reported to be targets of the Lin28/Let-7 axis (44–46). Consistent with Lin28b being critical in conferring fetal HSPC identity, both Hmga2 and Igf2 are highly expressed in fetal but not adult HSCs (47). To assess if ectopic Lin28 confers fetal-like gene expression profile in adult BM HSCs we performed qRT-PCR of FACS-sorted LSK HSCs and detected marked enhancement in the expression of both genes in the GFP+ fraction from Lin28-RV BM chimeras (Fig. S12B). Collectively, these results demonstrate the ability of ectopic Lin28 expression to confer fetal-like gene expression signatures in LSK HSCs.
In summary, these results identify Lin28b expression as a key molecular feature distinguishing fetal from adult BM HSPCs in mice and humans, shedding new understanding on a longstanding question in the field of lymphocyte development (Fig. S13). We demonstrate that ectopic expression of Lin28 in adult murine BM HSPCs dramatically skews the lymphocyte repertoire towards the production of innate-like lymphocyte subsets. Thus, we propose that Lin28b is a master regulator gene for fetal HSPC identity that is capable of reprograming adult BM HSPCs to acquire fetal-like characteristics in a cell autonomous manner. As a result, Lin28b is a switch that turns on the development of major subsets of innate-like lymphocytes. Together, these findings highlight how post-transcriptional regulation of gene expression can be pivotal in determining cell fate decisions.
Our findings have potential clinical implications for improving the reconstitution of innate-like lymphocyte populations upon adult BM transplantation. Most HSC transplants utilize adult BM and it is not known whether B-1 and MZ B cells develop following hematopoietic reconstitution. Due to the scarcity and ethical limitations of human fetal HSCs, they have not been seriously considered for clinical use. Umbilical cord blood is used clinically, but typically contains too few HSCs to reconstitute an adult (48). In the mouse, innate-like B-1 cell subsets play a role in T-independent host defense against pathogens such as Streptococcus pneumoniae (49); GM-CSF producing, protective innate response activator B cells differentiate from B-1a cells during sepsis (50); Vδ6.3+γδ-T cells provide critical protection against Listeria monocytogenes (51). Furthermore, the ability of iNKT cells to produce large amounts of cytokines including IFNγ, IL-4, and TNF allows them to modulate a broad spectrum of diseases including cancer, graft rejection, autoimmunity and infectious diseases (52). Thus, given the growing awareness of the ability of innate-like lymphocytes to influence immunity, it is of clinical interest to determine if fetal-like HSCs can more effectively achieve reconstitution of the full continuum of the innate and adaptive immune system.
We thank W.E. Paul for his generous support and advice throughout; T. Bender, J. Daniel, B.J. Fowlkes, R. Germain, M. Lenardo, W.E. Paul, and M. Schlissel for critical reading of this manuscript and constructive suggestions; K. Laky and B.J. Fowlkes for valuable advice and antibodies; NIH Tetramer Core Facility for reagents; M. Holt for NKT cell discussions; P. Burr for sequencing; J. Edwards and C. Eigsti for cell sorting; M. Foster for animal care; R. Zahr for technical support; N. Bartonicek (EMBL-EBI, Hinxton, UK) for providing the RefSeq 3′ UTR library for Sylamer analysis. We gratefully acknowledge the high performance computational capabilities of the Biowulf Linux cluster at the National Institutes of Health (biowulf.nih.gov) and NIAID Office of Cyber Infrastructure and Computational Biology HPC cluster (hpcweb.niaid.nih.gov) required for massively parallel sequencing analyses. The data reported in this paper are tabulated in the main text and in the Supporting Online Materials. The RNA-seq and NanoString data are available in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo) under the accession numbers GSE34854, GSE35081 and GSE35107. We apologize that we could not cite all the relevant references due to space limitations. The Integrative Immunobiology Unit is supported by the Intramural Research Program of the NIAID, NIH.
The authors declare no competing financial interests.