Senescence and apoptosis block hematopoietic activation of quiescent hematopoietic stem cells with short telomeres

Telomere shortening limits the proliferative capacity of human cells and may contribute to aging-associated decline in hematopoietic stem cell (HSC) function.^1-5  Studies on DNA repair–deficient mice and telomerase knockout mice revealed that DNA damage accumulates in the HSC compartment and limits the functionality of HSCs by induction of DNA damage checkpoints.^6,7  Recent studies indicated that DNA damage is repaired when HSCs enter the cell cycle.⁸  However, telomere-free chromosome ends cannot easily be repaired because telomerase recruitment requires telomere repeats.^9,10  Moreover, the formation of chromosomal fusion represents an aberrant repair pathway that interferes with maintenance of chromosomal integrity in dividing cells.¹¹  Whether telomere shortening–induced DNA damage leads to accumulation of DNA damage and gene expression changes in HSCs, and whether these alterations are transmitted to hematopoietic progenitor cells, is currently unknown.

All mice used are C57BL/6 background. mTerc^+/− were crossed to generate G1 mTerc^−/−. mTerc^−/− mice were crossed until the third generation (G3mTerc^−/−). Mice were maintained and experiments were conducted according to protocols approved by the state government of Thuringia, Germany (Reg. No. 03-006/13).

Bone marrow cells were isolated by crushing bones from donor mice. Cells were stained and sorted by using the following surface markers combination: HSCs (CD34^loFlt3⁻ScaI⁺cKit⁺Lineage⁻), multipotent progenitor cells (MPPs; CD34⁺Flt3⁺ScaI⁺cKit⁺Lineage⁻), and myeloid cells (CD11b⁺). Quiescent HSCs and cycling HSCs were purified by using Pyronin Y (P9172; Sigma-Aldrich) and Hoechst33342 (B2261; Sigma-Aldrich). Bone marrow cells were stained with antibodies for HSCs first, then incubated with Hoechst33342 (DNA dye, 1 mg/mL) at 37°C for 30 minutes and Pyronin Y (RNA dye, 100 μg/mL) for a further 15 minutes under light-free conditions. Samples were analyzed by ARIA (BD Biosciences).

Comet assays were conducted by using the OxiSelect Comet Assay Kit according the manufacturer’s protocol. HSCs were sorted into ice-cold phosphate-buffered saline (PBS) with the concentration 1 × 10⁵ cells per mL and fixed onto the OxiSelect Comet together with Comet Agarose. Then, electrophoresis was applied, and Vista Green DNA Dye was used to develop tails.

To analyze gene expression changes in response to telomere shortening, HSCs (CD34⁻Flt3⁻Sca1⁺c-kit⁺Lineage⁻), MPPs (CD34⁺Flt3⁺Sca1⁺c-kit⁺Lineage⁻), and myeloid cells (CD11b⁺) were freshly isolated from 12-month-old G3mTerc^−/− mice and age-matched mTerc^+/+ mice (supplemental Table 1; see the Blood Web site). Gene expression profiling (original profiles were uploaded to Gene Expression Omnibus: GSE60164) revealed that 63 genes (fold change >2, P < .05) were differentially regulated in telomere dysfunctional HSCs compared with mTerc^+/+ HSCs (Figure 1A). In contrast, the comparison of gene expression from mTerc^+/+ vs G3mTerc^−/− revealed no differentially expressed genes at the level of MPPs and only 11 genes at the level of myeloid cells (supplemental Table 2; Figure 1A). Among the 63 genes differentially expressed in HSCs from G3mTerc^−/− compared with mTerc^+/+ mice, several DNA damage–related or apoptosis-related genes were present (highlighted in Figure 1A).

γH2AX staining (a marker of DNA breaks) revealed a significant increase of DNA damage in freshly isolated HSCs (Figure 1B), but not in MPPs and myeloid cells between G3mTerc^−/− and mTerc^+/+ mice (Figure 1C-D). Similar results were obtained for 53BP1 staining (data not shown). Furthermore, analysis of DNA breakage by the comet assay revealed that 47% of the HSCs from G3mTerc^−/− mice carried more than 30% DNA in comet tails compared with only 18% of HSCs from mTerc^+/+ mice (Figure 1E, P < .001), but no significant difference in MPPs and myeloid cells from G3mTerc^−/− compared with mTerc^+/+ mice (Figure 1F-G).

The previous results indicated that HSCs accumulate DNA damage and gene expression changes in response to telomere dysfunction, but these alterations are not transmitted to the progenitor cell level. To determine at what stage HSCs amass the alterations, 12 genes that are differentially expressed in HSCs between G3mTerc^−/− and mTerc^+/+ mice were selected (based on a reported high expression level in the HSC compartment¹² ) to be investigated in quiescent and cycling HSCs from G3mTerc^−/− compared with mTerc^+/+ mice (n = 11-13 mice per group pooled into 2 pools per group). The results revealed that most of the gene expression changes were present in quiescent HSCs but not in cycling HSCs (Figure 2A; n = 3 technical repeats, n = 2 pools per group). The higher magnitude of gene expression changes in quiescent compared with cycling HSCs correlated with a significantly elevated rate of DNA damage foci in quiescent vs cycling HSCs from G3mTerc^−/− compared with mTerc^+/+ mice (Figure 2B-C). Together, these results indicated that only the nondamaged HSCs contributed to the pool of cycling HSCs in aged telomere dysfunctional mice.

One possible explanation for the results obtained is that DNA damage checkpoints were less active in quiescent HSCs compared with activated HSCs; thus, damaged HSCs were eliminated or arrested at the transition from the quiescent to the activated stage. To test this interpretation, quiescent and activated HSCs were purified by Pyronin Y and Hoechst33342¹³  from 12-month-old G3mTerc^−/− mice that were treated with interferon-α to activate quiescent HSCs to enter the cell cycle.¹⁴  Cell cycle analysis showed significant cell cycle entry of HSCs in response to interferon-α treatment (supplemental Figure 1A-B). To monitor the activation of DNA damage checkpoints, the expression levels of p21,⁷  p16,¹⁵  BATF,¹⁶  and PUMA¹⁷  were analyzed by quantitative polymerase chain reaction in freshly isolated quiescent and activated HSCs. The experiments revealed that Puma was significantly increased in cycling compared with quiescent HSCs of G3mTerc^−/− mice (Figure 2D), but there was no difference for p21 and Batf expression (supplemental Figure 1G-H). Several other apoptosis-regulating genes were also upregulated in cycling HSCs compared with quiescent HSCs from G3mTerc^−/− mice (supplemental Figure 1C), but not in cycling HSCs from mTerc^+/+ stimulated with interferon-α (supplemental Figure 1D), indicating that apoptosis was induced in damaged HSCs of aged G3mTerc^−/− mice upon activation. Furthermore, annexin-V staining revealed a significant increase in apoptosis in activated HSCs compared with quiescent HSCs of G3mTerc^−/− mice (Figure 2E; supplemental Figure 1E), but not in mTerc^+/+ mice (supplemental Figure 1F).

The expression of p16 (a cell cycle inhibitor associated with senescence) was significantly elevated in quiescent HSCs from G3mTerc^−/− compared with mTerc^+/+ mice, and this increase was even higher in quiescent HSCs from interferon-α stimulated G3mTerc^−/− mice (Figure 2F). These data suggested that a subset of quiescent HSCs of telomere dysfunctional mice was arrested in senescence and could not exit quiescence to enter the cell cycle upon stimulation. In agreement with this interpretation, γH2AX staining revealed a significantly elevated amount of DNA damage in HSCs from G3mTerc^−/− mice that remained in the G0 gate after interferon-α stimulation compared with G0 HSCs from the control group (Figure 2G). To further test this hypothesis, a number of HSCs (CD34^loSca1⁺c-kit⁺Lineage⁻) from 12-month-old G3mTerc^−/− (332) and age-matched mTerc^+/+ (310) mice were purified and cultured individually. At 6 and 12 days after seeding, the cell numbers of each colony were assessed (Figure 2H-I). On day 6, 7.6% of HSCs from G3mTerc^−/− mice died compared with 0.32% of the HSCs from mTerc^+/+ mice (Figure 2H, P = .0083). Moreover, the percentage of HSCs that stayed alive as single cells and could not enter the cell cycle to form colonies was higher in HSCs purified from G3mTerc^−/− compared with mTerc^+/+ mice by day 6 (Figure 2H; 9.2% vs 2.9%, P = .043). Together with the in vivo data, these experiments indicate that both senescence cell cycle arrest and apoptosis represent checkpoint responses that block damaged HSCs from G3mTerc^−/− mice from entering the cell cycle upon stimulation and generating committed progenitor cells.

This study reveals that DNA damage and gene expression changes accumulate specifically in quiescent HSCs in the context of telomere shortening. Activation of senescence locks a subset of damaged HSCs in the G0 stage, which is consistent with a recent report showing a similar phenotype in muscle stem cells from geriatric mice.¹⁵  Some damaged HSCs from G3mTerc^−/− mice entered the cell cycle upon stimulation, which led to apoptosis induction. Taken together, both senescence and apoptosis checkpoints may cooperate to prevent damaged HSCs from generating hematopoietic progenitor cells. A similar mechanism has been disclosed in the intestinal stem cell system.¹⁷  In addition to the activation of checkpoints, it is also possible that DNA repair contributes to the decline of DNA damage and gene expression changes at the transition from quiescent HSCs to progenitor cells. Along these lines, a recent study showed that DNA repair signaling is compromised in quiescent HSCs but is activated when HSCs enter the cycle.⁸  Of note, many of the differentially expressed genes play a role in DNA damage repair and apoptosis, which may be a response and/or contributing factor to the accumulation of DNA damage. It is conceivable that some gene expression changes are induced by chromosomal, genetic, and epigenetic alterations that occur in response to telomere dysfunction.^17-19  Together, these findings improve our understanding of the evolution and dynamics of gene expression changes and DNA damage accumulation in aging hematopoietic stem and progenitor cells in the context of telomere dysfunction. The findings may influence the selection of aberrant HSC clones that are characteristic of human aging and leukemia development.^20,21 

The authors thank Isabell Blochberger for her excellent technical assistance.

This work was supported by grants from the Deutsche Forschungsgemeinschaft (Ru745-10 and RU-745-12), the European Union (ERC-2012-AdG 323136) (K.L.R.) (ERC-2013-AdG 322602) (C.A.K.), the Bundesministerium für Bildung und Forschung (GerontoSys – SyStaR 315894) (K.L.R.), and the German Cancer Aid (108246) (K.L.R. and C.A.K.).

Contribution: J.W. performed experiments, analyzed the data, and wrote the manuscript; X.L. and C.A.K. performed the gene array experiment and analyzed the data; K.L.R. designed research, analyzed the data, and wrote the manuscript; and V.S. conducted and analyzed mouse experiments.

Conflict-of-interest: The authors declare no competing financial interests.

Correspondence: Karl Lenhard Rudolph, Beutenbergstraße 11, 07745 Jena, Germany; e-mail: klrudolph@fli-leibniz.de.