Age-dependent effects of Igf2bp2 on gene regulation, function, and aging of hematopoietic stem cells in mice

:Increasing evidence links metabolism, protein synthesis, and growth signaling to impairments in the function of hematopoietic stem and progenitor cells (HSPC) during aging. The Lin28b/Hmga2 pathway controls tissue development and the postnatal downregulation of this pathway limits the self-renewal of adult vs. fetal hematopoietic stem cells (HSC). Igf2bp2 is an RNA binding protein downstream of Lin28b/Hmga2 , which regulates mRNA stability and translation. The role of Igf2bp2 in HSC aging is unknown. Here, we show in an analysis of wildtype and Igf2bp2 knockout mice that Igf2bp2 regulates oxidative metabolism in HSPC and the expression of metabolism, protein synthesis, and stemness-related genes in HSC of young mice. Interestingly, Igf2bp2


Introduction
Metabolic activity contributes to the maintenance of stem cell function by controlling cell proliferation/quiescence, self-renewal, and differentiation. [1][2][3] However, metabolism, cell growth controlling pathways and protein synthesis [4][5][6] can also lead to functional exhaustion of HSC. Pathways that control metabolism, growth, and protein synthesis may thus have an ambivalent role in controlling the full functionality of HSC but at the cost of driving HSC aging. One of the upstream regulators of metabolic activity is the Lin28/let-7 pathway controlling the expression of cell cycle regulators, mitochondrial protein encoding genes, and glucose metabolism. 7,8 The pathway controls the induction of metabolic pathways and growth signaling, such as IGF, PI3K, and mTOR. 9 Downstream effectors of the Lin28/let-7 pathway are Hmga2 (encoding for a chromatin modifier) and Igf2bp2 (also known as Imp2), which was originally identified as a fetal growth factor and a binding partner of Igf2 mRNA. 10,11 Additional targets of IGF2BP2 have been identified, including major regulators of cellular metabolism and mitochondrial function, such as Igf1r, c-myc, Sp1 12 , Lamb2 13 , Ucp1 14 , as well as various non-coding RNA 15,16 .
In the hematopoietic system, Lin28 is expressed during development but declines in adult hematopoiesis. 17 The postnatal suppression of Lin28 limits selfrenewal and changes differentiation of HSC at the transition from fetal to adult hematopoiesis. Overexpression of Lin28a/b or its downstream target Hmga2 enhances the self-renewal capacity of adult HSC to similar levels as possessed by fetal HSC, whereas knockout of Hmga2 abrogates the elevated self-renewal of fetal HSC. 18,19 In addition, Lin28 controls fetal B-lymphopoiesis 17 and the differentiation of adult HSC. 20 However, the possible role of this pathway in HSC aging has not been explored. Here, we investigated aging of HSC in homozygous germline Igf2bp2 knockout mice (Igf2bp2 -/-) and wildtype mice (Igf2bp2 +/+ ). The study reveals a new role of Igf2bp2 in regulating the expression of metabolism, protein synthesis, and stemness related genes in HSC of young mice, which is required for full capacity of young HSC to form colonies in culture or to repopulate hematopoiesis in transplanted mice. Igf2bp2 expression and its gene regulatory function is almost completely lost during aging. The analysis of Igf2bp2 -/mice indicates that both (i) Igf2bp2-dependent gene regulation in young HSC and (ii) the decline of Igf2bp2-dependent function in aged HSC contribute to the development of distinct phenotypes of HSC aging.

Mice
Igf2bp2 knockout (Igf2bp2 -/-) mice carry an inverted exon-3, which leads to a frameshift, a premature stop codon, and a complete knockout of the gene encoded protein expression (Supplemental Figure 1).

Flow cytometry
Details of the procedures for cell isolation and staining combination are provided in supplemental methods (Supplemental Figure 2).

In vivo transplantation assay
Freshly purified HSC (with indicated marker combination; 100 HSC from young donor or 1,000 HSC from old donor, CD45.2 + ) were transplanted into lethally -irradiated (12 Gy) recipients (CD45.2, 4 months old) by intravenous (i.v.) injection along with 1x10 6 total BM cells from age-matched competitor mice (CD45.1). After transplantation, all recipients were treated with antibiotic water (0.01%; Baytril) for one week. Chimerism and lineage composition in PB and bone marrow of recipient mice were analyzed by FACS at indicated timepoints.

HSC colony forming assay
Freshly isolated HSC were seeded in methylcellulose medium (Stem Cell Technologies; M3434; 1.2 ml/well in 6-well plate) at 500 cells/duplicate. For serial rounds of plating, cells harvested from the previous plating were seeded at 5,000 cells/duplicate. Colony numbers were scored after 10 days.

Homing assay
Previous protocol 21 was used to examine the homing potential of myeloidbiased HSC (see supplementary methods for details).

Respirometry analysis
Freshly purified HSPC were used for Agilent Seahorse Cell Mito Stress assay and Real-time ATP production assay according to the manufacturer's protocols (see supplement).

Statistical analysis
Numbers of biological replicates and experimental repetitions are stated in the figure legends. Statistical tests and corrections for multiple testing were performed as indicated in the figure legends. Normality of the data, whenever required by the test method, was determined by the Shapiro-Wilk test. The significance level was set at p≤0.05 for tests unless otherwise specified. All statistical analyses were performed with GraphPad Prism 7.01 software, except for RNA-seq and proteomic analysis (see supplementary information) or where specifically indicated.

Lentivirus infection of HSC, Cell culture and inhibitor treatment, MitoRed measurement, Proteomics analysis of Igf2bp2-overexpressing stem cells
See supplemental methods for details.

Data sharing statement
Bulk and single cell RNA-seq are deposited in the National Center for Biotechnology Information's Gene Expression Omnibus (GSE166176). The mass spectrometry proteomics data are deposited to the ProteomeXchange Consortium via the PRIDE 22 partner repository (dataset identifier PXD018535).

Lin28b/Hmga2/Igf2bp2 expression and downstream pathways decline in
HSPC during aging. mRNA expression was analyzed in freshly isolated CD150positive (high and low) HSC (CD150 + CD34 − LSK) and MPP (CD34 + LSK) of young (3-6 month) and old (22-28 month) C57Bl/6JRj mice. Lin28b, Hmga2, and Igf2bp2 mRNA were expressed in HSC and MPP of young mice, but strongly declined during aging ( Figure 1A,B). Igf2bp2 is a mediator of Lin28/Hmga2-regulated metabolic activity and growth. 7 In line with the Igf2bp2 expression data, immunofluorescence staining of HSPC revealed a significant reduction in p-AKT in HSC and MPP during aging ( Figure 1C-F). In addition, FACS-based analysis of p-mTOR expression showed a significant decrease in mTOR activity in aging HSC but not in MPP ( Figure   1G,H).
Igf2bp2 controls metabolism, protein synthesis, and stemness-related genes in HSC of young mice but its function is lost in HSC of aged mice. In mice, the total population of CD150-positive HSC (CD150 + CD34 -LSK) can be further separated, based on the level of CD150-expression, into HSC with myeloid-biased differentiation (CD150 high CD34 -LSK) or balanced (lymphoid/myeloid) differentiation (CD150 low CD34 -LSK). Previous studies revealed that myeloid-biased (CD150 high ) HSC exhibit a more latent repopulation but enhanced self-renewal compared to balanced (CD150 low ) HSC. 23 To identify Igf2bp2-controlled genes in HSC, mRNA sequencing was conducted on myeloid-biased HSC from young (3 months) and aged (22-26 months) Igf2bp2 -/vs. Igf2bp2 +/+ mice. This sub-population of HSC was chosen, because it showed the highest expression of Igf2bp2 (Supplemental Figure   3A). In line with previous studies 24 , mRNA profiles showed clustering separation of myeloid-biased HSC from old vs. young mice ( Figure 2A). Interestingly, Igf2bp2 gene status separated the clustering of myeloid-biased HSC from young mice but not from the old mice ( Figure 2A). Concordantly, a much larger number of differentially expressed genes (DEG) was identified in myeloid-biased HSC of Igf2bp2 -/vs.
Igf2bp2 +/+ mice at young age (1421 DEG) compared to old age (26 DEG) ( Figure   2B). These data indicated that Igf2bp2 regulates mRNA expression in myeloid-biased HSC of young mice, but its gene regulatory function is lost during aging.
Since IGF2BP2 is an RNA binding protein to regulate gene expression, the DEG in myeloid-biased HSC from Igf2bp2 -/vs. Igf2bp2 +/+ from young mice were compared with mRNA that are directly bound to IGF2BP2 in mouse brown fat 14 . mRNA that are bound by IGF2BP2 in brown fat 14 exclusively overlapped with downregulated but not with upregulated DEG in myeloid-biased HSC of young Igf2bp2 -/vs. Igf2bp2 +/+ mice ( Figure 2B). This suggested that upregulated genes in response to Igf2bp2 deletion may represent secondary responses and not direct IGF2BP2 targets. Further analysis focused on mRNA that showed decreased expression in Igf2bp2 -/vs. Igf2bp2 +/+ myeloid-biased HSC. Gene Ontology (GO) analysis revealed a significant reduction in the expression of genes related to GOterms of mitochondrial metabolism & protein synthesis in Igf2bp2 -/vs. Igf2bp2 +/+ myeloid-biased HSC of young mice but not in myeloid-biased HSC of old mice ( Figure 2C,D). ANOVA analysis on this DEG subset showed that the expression of these genes significantly declined in myeloid-biased HSC from aged vs. young Igf2bp2 +/+ mice ( Figure 2D, p=0.048). However, there was no significant difference in the expression of these genes in myeloid-biased HSC from aged Igf2bp2 +/+ vs.
Igf2bp2 -/mice (p=0.49). These data indicated that Igf2bp2 controls the expression of genes related to metabolism and protein synthesis in myeloid-biased HSC of young mice but this gene regulatory function of Igf2bp2 is lost during aging.
Igf2bp2 expression is enriched in a subcluster of HSC from young mice cosegregating with expression of Lin28, Igf/Pi3k, and stemness-related genes.
Single cell (sc) RNA sequencing was conducted on freshly isolated myeloid-biased HSC from a pool of 6-week old, male, wildtype mice (n=5). This age and gender was chosen, because it showed the highest expression of Igf2bp2 (Supplemental Figure   3B and C). Through cell clustering (via the Louvain algorithm in Seurat), scRNA-seq identified a sub-cluster (Cluster-3), of myeloid-biased HSC, which showed an increased percentage of Igf2bp2 expressing cells compared to the other 8 subclusters that were identified by specific markers of each cluster ( Figure 3A-C, Supplemental Figure 4A). In agreement with this observation, Cluster-3 was also enriched for an increased percentage of cells expressing other components of the Lin28 pathway ( Figure 3D,E, Supplemental Figure 4B,C) as well as downstream targets that are known to be bound and regulated by Igf2bp2 ( Figure 3F-I).
Compared to the unprimed HSC clusters (0, 4, 5, and 8), the expression signature of Cluster-3 showed an upregulation of IGF/PI3K signaling (Igf1, Pik3r1) and a downregulation of inhibitors of AKT signaling (Igfbp4 and Cmtm7) ( Figure 3J). IGF/PI3K/AKT represent major regulators of cellular metabolism and growth that are known to be activated by Igf2bp2 mediated RNA expression. 25-28 Interestingly, Cluster-3 was also enriched for the expression of genes related to quiescence and stemness of HSC including Mllt3 29 , Cdkn1c/p57 30 , Procr 31 , Xbp1 32 and Slfn2 33 ( Figure 3J). Moreover, Cluster-3 showed enrichment for genes that are known to be regulated by imprinting and highly expressed in quiescent-enriched, long-term Igf2bp2 +/+ ( Figure 2). This analysis revealed that the marker genes of Cluster-3 that overlapped with DEG in the RNA-seq analysis of pooled myeloid-biased HSC from Igf2bp2 -/vs. Igf2bp2 +/+ mice were all regulated in the expected direction ( Figure 3K).
Igf2bp2 deletion impairs the functional capacity of HSC from young mice more than HSC from aged mice. The colony forming unit (CFU) capacity of HSC was determined using freshly isolated myeloid-biased HSC and balanced HSC from Igf2bp2 -/and Igf2bp2 +/+ male mice at young (3-6 months) and old age (22-26 months). Igf2bp2 deletion inhibited the CFU capacity of myeloid-biased HSC and the effect size of Igf2bp2-deletion on inhibition of CFU capacity was higher for myeloidbiased HSC from young compared to aged mice ( Figure 4A). Similar results were obtained for balanced HSC, albeit the effect size of Igf2bp2 genotype was overall reduced (Supplemental Figure 6).
The in vivo repopulation capacity of HSC was assessed by transplantation of freshly isolated myeloid-biased HSC from young mice. For old mice, CD150 + (high and low) HSC were used. Igf2bp2 deletion significantly reduced the repopulation capacity of young myeloid-biased HSC during long-term engraftment in primary recipients ( Figure 4B, left). In contrast, Igf2bp2 gene status did not affect the longterm repopulating capacity of CD150 + HSC of aged mice ( Figure 4B, right). Based on the difference in purification of test donor HSC, we cannot exclude that the transplantation of purified myeloid-biased (CD150 high ) HSC could have revealed an inhibitory effect of Igf2bp2 deletion on the repopulation function of myeloid-biased HSC from old mice. However, the analysis of the colony forming capacity demonstrated that Igf2bp2 deletion had a stronger inhibitory effect on the function of myeloid-biased HSC from young mice compared to myeloid-biased HSC from old mice. Igf2bp2 gene status had no effect on the homing capacity of myeloid-biased HSC ( Figure 4C) as measured by previously established protocols. 21 However, Igf2bp2 deletion led to an increase in the fraction of young donor-derived HSC (CD150 + ) in quiescence (G 0 ) and a decrease in G 1 ( Figure 4D,E). In aged donor HSC (CD150 + ), Igf2bp2 gene status had no effect on these cell cycle parameters ( Figure   4D,E). Together, these data indicate that Igf2bp2 is required for full repopulation function and colony forming capacity of young, myeloid-biased HSC but the deletion of Igf2bp2 has less effects on the function of aged HSC.  Figure 7C). Igf2bp2 deletion also led to a reduction of ATP-production in LSK cells of mice, which was not seen at a significant level in aged HSC ( Figure 5D). Igf2bp2-dependent lowering of the OCR of young LSK cells did not lead to changes in ROS levels (Supplemental Figure 7D).
However, GO-terms that were significantly enriched and overall down-regulated in the bulk RNA sequencing of myeloid-biased HSC of young, Igf2bp2 -/vs. Igf2bp2 +/+ mice ( Figure 2) included twenty GO terms that were related to mitochondrial stress (Supplemental Figure 7E) out of the top 200 significant GO terms (see Supplemental Table 1, for GO-terms that were significantly enriched and overall up-regulated in the same experiment see Supplemental Table 2). Together, it appears that mitochondrial stress was enhanced in young Igf2bp2 +/+ vs. Igf2bp2 -/myeloid-biased HSC despite the lack of detectable, genotype-related differences in ROS, possibly due to the lack of sensitivity of currently used, dye-based methodology. Interestingly, intracellular ATP concentration did also not show a reduction in CD150 + (high and low) HSC and MPP of young Igf2bp2 -/mice vs. wildtype mice (Supplemental Figure 7F) despite the observed genotype-dependent differences in ATP-production rates ( Figure 5D). It is possible that the inhibitory effect of Igf2bp2 deletion on protein synthesis ( Figure 2) reduces ATP consumption thus outweighing reduction in ATP-synthesis in Igf2bp2 depleted cells.
The Lin28/let-7 pathway is upstream of Hmga2/Igf2bp2 in regulating cell growth, protein synthesis, and metabolism by activation of mTOR and PI3K pathways. 9 To test whether this pathway contributes to Igf2bp2-mediated differentiation of HSC in culture, Igf2bp2-cDNA vs. control vector infected CD150 + (high and low) HSC were treated with inhibitors of mTOR (by rapamycin) or PI3K signaling (by LY294002). In culture, freshly isolated HSC differentiate mainly into the myeloid lineage (CD11b + ) thus reflecting the HSC capacity to differentiate into the myeloid lineage. Both inhibitors abrogated Igf2bp2-overexpression induced increases in the mitochondrial membrane potential ( Figure 5E) and myeloid differentiation of HSC ( Figure 5F). Together, these experiments show that iatrogenic induction of Igf2bp2 activates mitochondria, protein synthesis pathways and differentiation of HSC in an mTOR/PI3K-dependent manner.
Enhanced mitochondrial and mTOR activity can lead to loss of HSC maintenance. 38,39 To determine consequences of Igf2bp2 overexpression, freshly isolated CD150 + (high and low) HSC from young and aged mice were transplanted along with competitor cells into lethally irradiated recipients. 12 weeks after transplantation, peripheral blood analysis revealed a significant loss of repopulation capacity (Supplemental Figure 8A-B) and aggravated myeloid skewing (Supplemental Figure 8C) in recipients of HSC overexpressing Igf2bp2 compared to controls, especially when HSC from aged mice were transplanted. These results indicated that overexpression of Igf2bp2 cannot rescue the function of HSC from aged mice but leads to HSC exhaustion possibly due to the supra-physiological induction of mitochondria and PI3K/mTOR activity.
Igf2bp2 deletion ameliorates aging-associated HSC expansion and myeloid skewing. Increases in the number of phenotypic HSC represent a hallmark of the aging hematopoietic system in mice and men. [40][41][42][43] Here, the number of phenotypic-defined HSC was investigated in Igf2bp2 +/+ and Igf2bp2 -/mice at young (3-6 months), middle (9-15 months) and old age (18-27 months). In agreement with previous results 14 , there was a reduction in body weight and an elongated lifespan in Igf2bp2 -/mice vs. Igf2bp2 +/+ littermate controls (Supplemental Figure 9).
Analyzing the number of CD150 + (high and low) HSC, revealed the expected increase in the HSC number during aging ( Figure 6A,B). Igf2bp2 deletion significantly reduced this age-related increase in HSC numbers in male and female mice ( Figure   6A,B). In addition, the slope of HSC number increases from young adult to old age was also significantly lower in Igf2bp2 -/vs. Igf2bp2 +/+ mice ( Figure 6C), implying that  Figure 10A,B). Igf2bp2 deletion also led to a significant reduction in the ratio of myeloid-biased HSC vs. balanced HSC in aged Igf2bp2 -/vs. Igf2bp2 +/+ mice of both genders ( Figure 6H,I). Igf2bp2 gene status had no effect on progenitor cell populations, such as MPP, MPP subpopulations, CMP, or total bone marrow cells (Supplemental Figure 10C-F). Only, the number of common lymphoid progenitors (CLP) was reduced in young (both gender) and middle-aged (male) Igf2bp2 -/vs. Igf2bp2 +/+ mice (Supplemental Figure 10G). Igf2bp2 -/vs. Igf2bp2 +/+ mice did not show a reduction in MPP4 (Supplemental Figure 10D). Since MPP4 represent a subpopulation of MPP primed to undergo lymphoid differentiation 48,49 , it is conceivable that Igf2bp2 is required for the capacity of MPP4 to generate CLP.
Igf2bp2 +/+ mice were analyzed. Igf2bp2 deletion partially rescued aging-associated myeloid skewing in blood cell production resulting in an improved balance of myeloid to lymphoid cells in peripheral blood of aged mice of both genders ( Figure 6J and K).
To determine whether HSC-intrinsic mechanisms contributed to this rescue, an HSC transplantation experiment was conducted using CD150 + (high and low) HSC for transplantation. As expected from previous studies 44,46 the ratio of myeloid cells vs.  Figure 10H). When purified, myeloid-biased (CD150 high ) HSC from young donors were used for transplantation ( Figure 6L, right graph), the myeloid skewing in transplant recipients was similar to what was seen in recipients of CD150 + HSC from aged mice ( Figure 6L, left graph) and it was not significantly altered by Igf2bp2 deletion (Figure 6L, right graph). Together, these results indicate that the ameliorated increase in the ratio of myeloid-biased to balanced HSC in aged Igf2bp2 -/vs. Igf2bp2 +/+ mice ( Figure 6H) contributes to the reduction in myeloid skewing in peripheral blood of Igf2bp2 -/vs. Igf2bp2 +/+ mice.

Discussion
This study reveals experimental evidence that in young HSC the endogenous, physiological expression level of Igf2bp2 regulates mitochondrial metabolism and the