A FLI1-KLF6 Axis Regulates Aging in Human Hematopoietic Stem and Progenitor Cells and Normalization of KLF6 Levels in Aged Cells Leads to Their Rejuvenation

Aging causes a gradual decline in hematopoietic stem cell (HSC) function, which increases the risk for hematological malignancies. While much has been done in murine models, human HSC aging impairment is less understood. We recently showed that Krüppel-like transcription factor 6 (KLF6) is among the top downregulated genes during human HSC aging, which correlates with H3K27ac loss at several upstream putative enhancers. Moreover, loss of KLF6 in human CD34 ⁺ cells resulted in impaired in vitro differentiation, increased colony-forming potential and a transcriptional profile similar to that of aged CD34 ⁺ CD38 ^- cells. We hypothesized that age-acquired deregulation of KLF6 may be a key player in age-related HSC dysfunction and sought to fully characterize this. Thus, we isolated CD34 ⁺ cells from young (<32 y.o) and aged (>65 y.o.) healthy donors and performed CRISPR-Cas9 genome editing and transcriptional activation of KLF6, respectively, followed by epigenetic and transcriptional reprogramming, in vivo hematopoietic reconstitution, and analysis of DNA damage, apoptosis, and reactive oxygen species (ROS) levels.

KLF6 knock-out (KO) and non-targeting control (NTC) cells from young healthy donors were engrafted into immunodeficient NSGS mice. Hematopoietic reconstitution analysis showed that KLF6 KO cells led to increased myeloid and reduced lymphoid reconstitution in peripheral blood (PB; p<1.62 ^-7) and an increase in immunophenotypically defined HSC and CD34 ⁺ CD38 ^- progenitor fractions in the bone marrow (BM; p=0.02, and p=0.04, respectively). H3K27ac analysis of KLF6 KO cells revealed a loss of 3,390 ChIP-seq peaks (FDR < 0.05) and 285 peaks gained. Functional annotation using ChIP-Enrich showed that H3K27ac loss associates with myeloid homeostasis, erythroid differentiation and oxidative stress (FDR < 0.05). Three putative enhancer (E) regions upstream of the KLF6 locus showed loss of H3K27ac with aging. Depletion of the E1 but not E2 or E3 regions phenocopied in vitro and in vivo findings of KLF6 KO. Transcription factor (TF) ChIP-seq data analysis revealed FLI1, ERG, and RUNX1 binding overlapping the E1 region. Knockdown of FLI1 but not ERG or RUNX1 led to an increase in KLF6. Notably, FLI1 mRNA levels, but not ERG or RUNX1, are increased during normal aging.

We next performed in vitro KLF6 activation in aged CD34 ⁺ (KLF6a) cells using a dCas9-VP64 system to test if we could rejuvenate these cells. KLF6a cells exhibited a decrease in their in vitro myeloid differentiation potential, compared to aged NTC CD34 ⁺ cells (p<0.0041), and behaved instead similar to young controls. ChIP-seq analysis of KLF6a showed marked decrease of H3K4me1 (n=3,273 peaks) with relatively few regions with increased H3K4me1 (n=602) (FDR < 0.05). In contrast, we observed an increase in H3K27ac (n=3,361 peaks) with only 71 peaks lost compared to aged NTC (FDR < 0.05). Regions that gained H3K27ac in KLF6a were associated with platelet activation, cell junction and adhesion. In vivo analysis of KLF6a cells injected into NSGS mice revealed a significant reduction in the PB myeloid fraction compared to NTC (p<1.2-8), with a concomitant expansion in the lymphoid compartment (p<4.4 ^-11). BM composition analysis at week 16 showed a decrease in the HSC fraction in KLF6a cells (p=0.0029) as well as a reduction in CD34 ⁺CD38 ^-, CD34 ⁺CD38 ⁺ and MEPs (p=0.036, p<0.0001 and p=0.041, respectively). We next examined the impact of KLF6 modulation on DNA damage and observed that young human KLF6 KO cells had a significant increase in gH2AX and 53BP1 (p<0.0001, for both) whereas KLF6a in aged CD34 ⁺ cells exhibited reduced gH2AX and 53BP1 foci in comparison to aged NTC (p<0.0001, for both). In addition, apoptotic levels in KLF6 KO cells were higher than in NTC cells (p=0.006) whereas aged KLF6a cells showed a reduction in the incidence of apoptotic cells compared to NTC (p=0.019). Finally, ROS analysis in young KLF6 KO showed increased levels of total and mitochondrial ROS compared to NTC (p=0.0008 and p<0.0001, respectively) whereas both ROS fractions were reduced in KLF6a cells (p=0.0002 and p<0.0001, respectively).

In summary, these results show that the FLI1-KLF6 axis plays a key role in regulating HSPC aging and that KLF6 is required for normal HSPC function and differentiation. In addition, normalization of KLF6 levels in aged HSPCs resulted in reprogramming and rejuvenation HSPCs, confirming the central role of this TF in aging HSPC biology.