TIMP-1 deficiency subverts cell-cycle dynamics in murine long-term HSCs

The capability of HSCs to maintain the homeostasis of the hematopoietic system is the result of a finely tuned balance between self-renewal and differentiation. The mechanisms responsible for this balance comprise both intrinsic and extrinsic factors, whose crosstalk eventually dictates the fate of stem cells in the framework of the BM niche.^1-3  Beside the well-established structural function, the dynamic network of interacting macromolecules that constitutes the extracellular matrix (ECM) represents one of the most powerful sources of extrinsic factors generated by the BM microenvironment.⁴  The intricate architecture created by these molecules not only guarantees protection and mechanical support to the stem cell pool but also plays an active role in regulating their behavior. By binding growth factors, regulating their bio-availability, and enabling the interaction with cell-surface receptors, ECM components have been shown to modulate a variety of cellular functions, such as proliferation, survival, and differentiation.⁵ 

ECM dynamic remodeling is controlled by metalloproteinases (MMPs), a class of Zn⁺⁺-dependent proteinases, such as collagenases, gelatinases, and stromelysins, that participate in the digestion of many ECM components, under both physiologic and pathologic conditions.⁶  The enzymatic activity of MMPs is counterbalanced by several natural inhibitors, including the tissue inhibitors of metalloproteinases (TIMPs).⁷  Both MMPs and TIMPs are expressed by hematopoietic and stromal cells⁸  and are decisive regulators of the crosstalk between these 2 cellular entities. The mammalian TIMP family comprises 4 highly conserved members that reversibly block MMP-dependent proteolysis by forming noncovalent 1:1 stoichiometric complexes. Alterations in the balance between the enzymatic activities of MMPs and TIMPs have been linked to developmental defects and are associated with specific tumor microenvironments.⁹ 

Although TIMPs were initially described as mere inhibitors of MMPs, recent findings have offered a different perspective on their biologic role, unveiling their multifaceted nature.^10,11  In addition to inhibiting MMPs, TIMP-1 has been proven to play MMP-independent, cytokine-like activities and to be involved in cell growth, angiogenesis, apoptosis, and migration.^12,13  For instance, Nakajima et al¹⁴  recently found that TIMP-3 plays a role in recruiting HSCs into the cell cycle.

Despite intense investigation, the coexistence of MMP-dependent and -independent functions has hindered the thorough dissection of the signaling pathways activated by TIMP-1, leaving the interpretation of its different biologic effects controversial and difficult to reconcile. Liu et al¹⁵  described the ability of TIMP-1 to protect human breast epithelial cells from apoptosis through the focal adhesion kinase/PI3K and MAPK signaling pathway. A similar activity has been described in the erythroleukemic cell line UT-7, with activation of the JAK2/PI3K/Akt cascade.¹⁶  The mechanisms underlying the activation of the molecular pathways downstream of TIMP-1 are also a matter of debate. The tetraspanin receptor CD63 protein has been identified as putative cell-surface receptor for TIMP-1 in human breast epithelial cells.¹⁷  In this model, TIMP-1 promotes cell survival through the activation of a CD63/integrin complex on the membrane of MCF10A cells. However, according to Lambert et al,¹⁸  TIMP-1 would form at the cell surface a ternary complex with pro-MMP-9 and CD44, which would in turn activate the signaling cascade, preventing apoptotic death in UT-7 cells.

Here, we investigate the functional behavior of HSCs in a model of TIMP-1^−/− mice, showing that the obliteration of TIMP-1 alters the cell-cycle dynamics of long-term HSCs (LT-HSCs), thus affecting their capability of repopulating recipients after transplantation. According to our model, TIMP-1 deficiency increases p53 levels, thus blocking the transition of HSCs from G₁ to S, through the p53-dependent down-regulation of CD44. Our study highlights a novel biologic role of TIMP-1 in the regulation of the HSC compartment and provides a new interpretative tool for the molecular dissection of the crosstalk between HSCs and the environment within they reside.

Wild-type (WT) C57Bl/6, CD45.1 mice and homozygous C57Bl/6, CD45.2 TIMP-1^−/− mice were obtained from The Jackson Laboratory and used at 6-12 weeks of age for all experiments. Animals were maintained in pathogen-free conditions at Baylor College of Medicine. The hematopoietic progenitor cell line 32D was obtained from American Type Culture Collection and maintained in RPMI 1640 medium, supplemented with 10% FBS and rIL-3. All animal studies were approved by the Baylor College of Medicine Animal Care and Use Committee.

The enforced expression of TIMP-1 was induced in WT Sca-1⁺ hematopoietic progenitors by retroviral transduction with murine stem cell virus–internal ribosome entry site–enhanced green fluorescent protein (MSCV-IRES-eGFP) or MSCV-TIMP1-IRES-eGFP as previously described.¹⁹  After transduction, cells were transplanted into lethally irradiated C57BL/6, CD45.1 recipient mice (50 000 cells/mouse).

Side population⁺c-Kit⁺Sca-1⁺Lineage⁻ (SP^KLS) HSCs or KSL cells were isolated from the BM of WT and TIMP-1^−/− mice and sorted on a MoFlo cytometer into 96-well plates (1 cell/well), containing M3434 MethoCult medium (StemCell Technologies), and incubated at 37°C, 5% CO₂. After 12-14 days of culture, colonies were counted. For in vitro CFU (CFU-C) of cells overexpressing TIMP-1, 32D cells were transduced with retroviral vectors (MSCV-IRES-eGFP or MSCV-TIMP1-IRES-eGFP) as described and cultured in transduction medium at 37°C, 5% CO₂ for 48 hours. For primary cells, Sca-1⁺ cells were transduced as described.¹⁹  GFP⁺Sca-1⁺ cells were then sorted into 96-well plates and cultured as above.

BM cells (200 000) from C57Bl/6, CD45.2 WT or TIMP-1^−/− mice were transplanted into C57Bl/6, CD45.1 lethally irradiated recipients. Thirteen days after transplantation, spleens were harvested, fixed in Bouin solution for 30 minutes, and spleen CFUs (CFU-Ss) were counted under a dissecting microscope. Similarly, Sca-1⁺ cells, obtained from C57Bl/6, CD45.2 mice treated with 5-fluorouracil, were retrovirally transduced as described with MSCV constructs, and 35 000 Sca-1⁺–transduced cells were transplanted into C57Bl/6, CD45.1 recipient mice. Spleens were fixed and CFU-Ss were scored as above.

BM cells from 3 mice (sex- and age-matched) were harvested from tibias and femurs of TIMP-1^−/− and WT C57Bl/6, CD45.2 mice. Whole BM (WBM) cells (200 000) from each group were transplanted into primary C57Bl/6, CD45.1 recipients, along with an equal number of CD45.1 WBM competitor cells (1:1). Peripheral blood (PB) engraftment and lineage distribution of CD45.2 cells was analyzed periodically at monthly intervals. Repopulation units were calculated as per Harrison.²⁰  A separate set of transplantations was performed to test the effects of a TIMP-1^−/− microenvironment on engraftment. CD45.1 WT BM cells (400 000) were transplanted into C57Bl/6, CD45.2 WT, or TIMP-1^−/− recipients. PB chimerism and lineage distribution of CD45.1 cells was analyzed periodically at 3, 6, and 9 weeks after transplantation.

PB from untreated or mice that received a transplant was collected from the retro-orbital sinus. The samples were pretreated with red blood cell lysis buffer (Tris Cl pH 7.8 and ammonium chloride, 1:9) and then stained for 20 minutes on ice with the following Ab cocktail: anti-CD45.2 (or CD45.1) APC-conjugated, anti-CD4 Pacific Blue, anti-CD8 Pacific Blue, anti-B220 Pacific Blue, anti-B220 PE-cyanine 7, anti-Mac1 PE-cyanine 7, and anti-Gr1 PE. Cells were resuspended in Hanks buffer containing propidium iodide (PI; 1 μg/mL). The analysis of common lymphoid progenitors (CLPs) and common myeloid progenitors (CMPs), as well as MMPs, short-term HSCs, and LT-HSCs, was performed as previously described.²⁰  Samples were analyzed on an LSRII flow cytometer (Becton Dickinson).

Cell cycle analysis was performed as previously described.²¹  Briefly, SP^KLS HSCs²²  were sorted on top of 200 000 B220⁺ splenocytes. Subsequently, cells were stained for 45 minutes at 37°C with Hoechst 33342 (20 μg/mL) and verapamil (50 μg/mL; Sigma-Aldrich) in PBS plus 3% FBS. Pyronin Y (1 μg/mL; Sigma-Aldrich) was then added to the staining solution, and cells were incubated for 15 minutes at 37°C. Ki-67 staining was performed as previously described.²⁰  Samples were analyzed on an LSRII flow cytometer (Becton Dickinson).

B220⁺ splenocytes (200 000) were sorted into collection tubes. Subsequently, SP^KLS HSCs were sorted into the same tubes, spun down, and resuspended in Annexin Binding Buffer (10mM HEPES, 140mM NaCl, 2.5mM CaCl₂, pH 7.4) and stained for 15 minutes at room temperature with anti-Annexin V antibody and PI. Samples were immediately analyzed on an LSRII flow cytometer (Becton Dickinson).

Transmigration assays were performed in 24-well transwell plates (diameter, 6.5 mm; pore size, 5 μm; Corning International). Sca-1⁺ cells (250 000) were loaded into the upper chamber in 100 μL of StemPro Medium, supplemented with nutrient supplement, L-glutamine, and penicillin/streptomycin. Lower chambers were loaded with 600 μL of the same media, supplemented with 125 ng/mL CXCL12 (R&D Systems). Plates were incubated at 37°C, 5% CO₂ for 4 hours. Migration rate was calculated as follows: (migrated cells/loaded cells) × 100. CD150⁺48⁻KSL migration rate was calculated on the basis of their frequency in the original population and in the migrated cells.

SP^KLS cells from WT and TIMP-1^−/− mice (n = 3 per group) were sorted into lysis buffer for RNA, and real-time PCR was performed as previously described. The fold change was calculated from the formula 2^ΔΔCt. For each gene the WT value was set to 1. The error bars represent the corresponding SEM.

Statistical significance was determined by 2-tailed Student t test. P values are indicated as *P < .05, **P < .01, and ***P < .001.

To begin to investigate the role of TIMP1 in regulation of HSCs, we first examined the hematopoietic system of TIMP-1^−/− mice. We initially investigated whether the constitutive lack of TIMP-1 would affect the production and release of mature cells in the PB. TIMP-1^−/− mice showed no major alteration in the respective percentages of circulating myeloid cells (Mac-1⁺), as well as B lymphocytes (B220⁺) and T lymphocytes (CD4⁺CD8⁺), when assessed on a flow cytometer (Figure 1A). Likewise, complete blood count analyses performed on knockout (KO) mice did not show any significant deviation from the PB profile of the WT counterpart (L.R., unpublished data). Similarly, the analysis of BM and spleens of KO mice did not show any relevant difference in the linage distribution of hematopoietic cells (Figure 1A; supplemental Figure 1A, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

We next analyzed the BM compartment and surprisingly observed that TIMP-1^−/− mice consistently presented with a decreased BM cellularity (average decrease of 24.8%; P < .001), compared with WT mice (Figure 1B). The decreased BM cellularity is not ascribable to a smaller size of the KO mice, because their weight in grams is comparable to that of WT mice (supplemental Figure 1B). Interestingly, we found that TIMP-1^−/− mice also present a slightly decreased percentage of HSCs (Figure 2A), when quantified on the basis of the SP^KSL phenotype. SP^KSL absolute numbers also presented a 43% decrease, compared with controls (4.9 × 10³ ± 0.7 × 10³ and 8.6 × 10³ ± 1.1 × 10³, respectively; P < .05; Figure 2B). Similar results were obtained when HSC percentages and absolute numbers were quantified on the basis of the CD150⁺48⁻KSL phenotype (Figure 2C-D; supplemental Figure 2). The CD34/Flk2 markers were also used to analyze HSC frequencies. With this particular staining scheme the percentage of LT-HSCs in TIMP-1^−/− mice appeared to be significantly decreased compared with WT animals (P = .05; Figure 2E; supplemental Figure 1C-D).

Conversely, the KSL compartment showed a slight increase compared with WT mice (Figure 2F), although the absolute number of KSLs still remains decreased. To determine whether the KSL compartment would present any specific skewing, we also analyzed the relative percentages of CMPs and CLPs. As assessed by flow cytometry, none of the myeloid progenitors (megakaryocyte-erythroid precursors, CMPs, and granulocyte macrophage progenitors), as well as the CLPs, seemed to be affected by TIMP-1 deficiency, suggesting that the slight expansion KSL cells is equally distributed among the different classes of progenitors (Figure 2G).

These data suggest that TIMP-1 deficiency affects the hematopoietic system, especially in terms of BM cellularity. Nonetheless, the production of mature PB components seems not to be significantly affected. We therefore hypothesized the existence of compensatory mechanisms promoted by extramedullary hematopoietic sites, such as the spleen. We found that the percentage of KSL progenitors in TIMP-1^−/− spleens was significantly higher than the controls (1.5-fold increase; P < .05; Figure 2H), suggesting a pivotal role of splenic hematopoiesis in rescuing the PB phenotype.

In an attempt to investigate how the described phenotype reflects on BM functionality, we first assessed the clonogenic potential of BM-derived SP^KLS cells in vitro. The number of colonies generated in semisolid cultures after 10 and 14 days was significantly increased (1.4-fold at day 10, P < .001; 1.2-fold at day 14; P < .05; Figure 3A) compared with the controls. However, at day 18, although the number of WT colonies kept increasing, the number of CFU-Cs generated from TIMP-1–deficient SP^KLS cells remained the same. Interestingly, we noticed that the size of the colonies generated from TIMP-1^−/−SP^KLS cells was also enlarged compared with the WT (data not shown). Conversely, KSL cells did not display any significant difference in terms of clonogenic potential (supplemental Figure 3), when cultured in the same conditions.

In addition, BM cells from TIMP-1^−/− mice were transplanted into lethally irradiated mice, and the appearance of spleen colonies (CFU-S) was quantified. Thirteen days after transplantation, the total number of CFU-Ss generated by TIMP-1^−/− BM cells was considerably expanded with respect to the WT (14.0 and 9.2 CFU-Ss, respectively; P < .05; Figure 3B).

To further investigate the role of TIMP-1 in cell proliferation, we transduced 32D cells with retroviral vectors driving the coordinated expression of TIMP-1 and eGFP (Figure 3C). We found that the enforced expression of TIMP-1 reduced cell proliferation of 32D cells in liquid culture (42.3% decrease; P < .05; Figure 3D) and significantly suppressed the generation of CFU-Cs in vivo (P < .05; Figure 3E). With the use of the same retroviral vectors, we also transduced primary WT Sca-1⁺ hematopoietic progenitor cells/HSCs and tested how the overexpression of TIMP-1 would affect CFU-S generation after transplantation. We found the CFU-S output to be significantly decreased after the induction of TIMP-1–enforced expression (1.8 fold-decrease; P < .001; Figure 3F). These findings reinforce the supposition that TIMP-1 plays a role in controlling proliferative dynamics.

We subsequently evaluated the capability of TIMP-1^−/− HSCs to replenish the hematopoiesis of lethally irradiated recipients, by inducing long-term, multilineage reconstitution on transplantation. WBM cells were transplanted along with WT competitors (1:1), and their engraftment was periodically evaluated as contribution to PB chimerism (Figure 4A). Under these conditions, we observed that TIMP-1^−/− BM cells do not reconstitute recipients as efficiently as WT cells (21.9% vs 53.3%; P < .05; 12 weeks after transplantation; Figure 4B). Of note, a profound decrease of 5.9-fold in repopulation units was observed in primary recipients of TIMP-1^−/− WBM transplants (Figure 4C). We also analyzed the KSL compartment of mice that received a transplant for contribution from donor-derived cells, as measured by expression of CD45.2 Ag. We found that, 12 weeks after transplantation, mice that received a transplant with WT BM cells present a 12.2% contribution of donor-derived BM KSLs, whereas in mice receiving TIMP-1 KO WBM cells, the contribution to the KSL compartment drastically drops to 4.8% (Figure 4D; P = .0005). Distribution among the different hematopoietic lineages was also evaluated, although no significant differences were observed compared with the control (L.R., unpublished data). These findings indicate that, despite the ability of TIMP-1^−/− BM cells to sustain multilineage reconstitution, their long-term reconstitution is significantly affected when in direct competition with their WT counterparts.

HSC function and survival highly rely on interactions with the hematopoietic niche. Because of the role that TIMP-1 plays in tuning MMP function, we considered that the impaired repopulation activity of TIMP-1^−/− BM cells might be secondary to alterations at the interface between HSCs and the niche. To address this possibility, we transplanted WT WBM cells into WT or TIMP-1^−/− recipients and periodically evaluated their engraftment and lineage distribution (Figure 5A). Over a period of 9 weeks, we found that the percentage of donor-derived chimerism in the PB was essentially the same in the 2 groups (88.1% and 83.9% for TIMP-1 KO mice and WT mice, respectively; Figure 5B), suggesting that the capability of WT BM cells of repopulating lethally irradiated mice is not significantly affected by the constitutive lack of TIMP-1 in the BM niche. The lineage distribution of PB cells also did not appear to significantly diverge from the control (Figure 5C), although B-cell production appeared to be slightly decreased at the latest time point. Because of the role played by MMPs/TIMPs in regulating HSC trafficking, we hypothesized that TIMP-1 deficiency may determine a decreased BM cellularity by increasing HSC mobilization into the PB. However, the percentage of circulating KSLs in KO mice did not appear to be significantly different from the controls (Figure 5D). Furthermore, compared with WT controls, TIMP-1^−/− cells showed the same homing rate toward the BM or the spleen (L.R., unpublished data), suggesting that the hematopoietic impairment observed in TIMP-1^−/− mice most probably depends on a defect intrinsic to the HSCs, rather than their interaction with the BM environment.

The decreased BM cellularity and the impaired BM repopulation after transplantation led us to hypothesize that TIMP-1 deficiency may be responsible for cell-cycle deregulation in HSCs. We used Pyronin Y staining in conjunction with Hoechst, that distinguishes cells in the G₀ (PyroninY⁻/Hoechst⁻), G₁ (PyroninY⁺/Hoechst⁻), and S-G₂M (Hoechst⁺) phases (Figure 6A).^21,23  In agreement with our hypothesis, we found that the cell cycle distribution of TIMP-1^−/− SP^KLS cells was significantly deregulated (Figure 6B). As shown in Figure 6B, the percentages of SP^KLS cells in the G₀ phase of the cell cycle were decreased in TIMP-1^−/− mice compared with the control (79.5% ± 3.5% vs 87.6% ± 2.3%, respectively; P < .01), whereas a 2.6-fold increase in the percentage of HSCs in the G₁ phase was accordingly recorded (17.4% ± 3.2% for TIMP-1^−/− SP^KLS cells vs 6.8% ± 1.7% for WT SP^KLS cells; P < .01). The percentage of TIMP-1^−/− HSCs found to be in the S/G₂/M phase also appeared decreased compared with the WT counterpart. Such a cell-cycle distribution suggests that TIMP-1^−/− mice are depleted in long-term, quiescent HSCs (which are assumed to be mainly in G₀). In addition, in the constitutive absence of TIMP-1, the physiologic cell-cycle completion appears to be impaired, because TIMP-1^−/− HSCs present a dysregulation at the level of the G₁-S transition. The role of TIMP-1 deficiency in HSC cell cycle was also investigated by measuring Ki-67, a nuclear marker of proliferation. TIMP-1^−/− SP^KLS cells express significantly higher levels of Ki-67 than the WT counterpart, thus confirming the finding that TIMP-1^−/− mice have a lower percentage of G₀ HSCs.

Because TIMP-1 has previously been suggested to play a role in cell survival,^16,18  we also considered the possibility that the impaired reconstitution ability of TIMP-1^−/− BM cells could be secondary to a higher rate of cell death in the hematopoietic compartment. We addressed this possibility by comparing the percentage of CD150⁺48⁻ KSL cells undergoing spontaneous apoptosis in TIMP-1^−/− and WT mice, as measured by annexing V/PI staining. We observed that the percentage of annexin V⁺ SP^KLS HSCs was essentially the same in the 2 groups, (Figure 6D; supplemental Figure 4). These data suggest that, at least under the conditions we investigated, HSC survival is not significantly affected by TIMP-1 deficiency.

To further investigate the role of TIMP-1 in HSC cell cycle, we analyzed by real-time PCR the expression level of some of the prominent regulators of cycling activity^24,25  in SP^KLS cells (Figure 6E). We found that p21 mRNA levels display a 3-fold increase in TIMP-1^−/− SP^KLS cells. Likewise, the expression levels of p57, found to play a pivotal role in HSC quiescence,^26,27  showed a 4.7-fold increase in TIMP-1^−/− SP^KLS cells, whereas p27 levels in KO HSCs were only slightly decreased compared with WT HSCs (Figure 6E).

Efficient homing and subsequent engraftment of HSCs into the BM microenvironment relies on 2-way molecular interactions between donor-derived hematopoietic cells and the recipient niche.²⁸  Alterations in the ability of each of these counterparts to efficiently crosstalk could compromise the entrance of HSCs in the niche and their subsequent survival. Although we showed that a TIMP-1–deficient niche does not hinder the engraftment of WT BM cells (Figure 5), the capability of TIMP-1 HSCs per se to properly migrate toward the BM microenvironment still remained unclear. We therefore considered the possibility that the decreased repopulation observed for TIMP-1^−/− BM cells could be ascribed to an impaired migratory pattern, in addition to the aforementioned proliferation defects. To address this possibility, we analyzed the capability of TIMP-1^−/− progenitors of migrating toward a gradient of the chemokine CXCL12 (stromal cell-derived factor 1), which represents the most powerful regulator of HSC directional migration. The percentage of Sca-1⁺ hematopoietic progenitor cells migrating in vitro toward CXCL12 was significantly lower for TIMP-1^−/− mice than for controls, with a 39% decrease in the migration rate (P < .05; Figure 7A). To define more precisely how TIMP-1 deficiency affects HSC migration, we calculated the percentage of CD150⁺48⁻KSLs that underwent migration in both groups. Consistent with previous results, we observed that CD150⁺48⁻KSLs from KO mice are less responsive to CXCL12 gradients and migrate at a significantly lower rate (P = .05; Figure 7B). Of note, we observed that the membrane expression of CXCR4, the conjugate receptor for CXCL12, is not significantly affected by TIMP-1 deficiency. When analyzed in a pool of BM CD150⁺48⁻KSL HSCs, no difference in the MFI of CXCR4 was observed (Figure 7C; supplemental Figure 5), indicating that CXCR4 expression cannot account for the impaired migration of TIMP-1^−/− HSCs.

Because of the correlation between migratory behavior and expression of adhesion molecules, we also considered whether their expression in TIMP-1^−/− HSCs would be affected. Interestingly, we found that the expression of CD44, a glycoprotein highly expressed on LT-HSCs,^28-34  was significantly influenced by the constitutive lack of TIMP-1. When assessed by flow cytometry, CD44 MFI emerged to be decreased in TIMP-1^−/− SP^KLS HSCs, with a 14.5% reduction compared with the WT controls (P < .05; Figure 7D-E; supplemental Figure 6). Because of the role that CD44 has been suggested to play in both migration and proliferation,²⁹  the finding that this glycoprotein is down-regulated in TIMP-1^−/− HSCs might provide a possible explanation for their deficient responsiveness to CXCL12 gradients, as well as their aberrant cell cycle, which could ultimately be responsible for the decreased BM cellularity in TIMP-1–deficient mice. Interestingly, Godar et al³⁵  showed that CD44 expression is directly inhibited by the tumor suppressor p53. By real-time PCR we measured p53 expression in SP^KLS cells and found that p53 levels are 3.6 times higher in TIMP-1^−/− HSCs compared than in controls (Figure 7F).

Although the role of TIMP-1 in ECM turnover has been extensively investigated, a thorough analysis of its MMP-independent, cytokine-like functions in hematopoiesis is yet to be provided. In this study we demonstrate that TIMP-1 subtly affects the HSC compartment, by subverting cell-cycle dynamics of long-term repopulating HSCs. BM cellularity in TIMP-1^−/− mice is significantly decreased, with a concomitant reduction in the absolute number of LT-HSCs. When tested in a competitive setting, TIMP-1^−/− BM cells displayed a defective engraftment, showing that TIMP-1 deletion functionally hinders HSC capability of sustaining regular hematopoiesis. Haviernik et al³⁶  have previously performed BM competitive transplantation assays with the use of TIMP-1^−/− WBM cells, concluding that TIMP-1 deficiency does not affect HSC engraftment. In our study, we injected 200 000 WBM cells per mouse, whereas Haviernik et al³⁶  transplanted 5 times as many cells. Furthermore, we found that the engraftment rate of TIMP-1^−/− BM cells was significantly lower than the WT counterpart at 1 month after transplantation, but it increased over the following 4 weeks. Similarly to Haviernik et al,³⁶  we found that, at 2 months after transplantation, the engraftment of TIMP-1^−/− cells did not significantly diverge from the control group. In contrast, we found that, at 12 weeks, the rate of PB chimerism of TIMP-1^−/− cells has dropped significantly (P = .014), suggesting that the specific time point at which the engraftment is measured is critical for interpreting TIMP-1 role in hematopoiesis.

In parallel, we found that progenitor-driven, short-term hematopoiesis is actually enhanced in TIMP-1^−/− mice, with an increased output of both CFU-Cs and CFU-Ss. This apparent paradox is probably the result of a compensatory mechanism, activated to limit the detrimental effects of an impaired hematopoietic system. Indeed, we found the percentage of KSL progenitors in the spleen to be significantly higher than the control, suggesting that the extramedullary splenic hematopoiesis contributes to rescue the PB phenotype.

Because TIMP-1 is a soluble protein, it still remained unclear whether the impaired function of TIMP-1^−/− HSCs was intrinsic to the stem cell compartment or could rather be attributed to the lack of TIMP-1 in the BM niche. To answer this question, we transplanted WT hematopoietic cells into WT and TIMP-1^−/− recipients and found no significant differences in the engraftment rates. This suggests that, despite that TIMP-1 is a secreted protein, its influence on HSCs is cell autonomous and intrinsically linked to the genotype of the transplanted cells rather than to the surrounding microenvironment. Similarly, Soloway et al³⁷  demonstrated that TIMP-1^−/− mice display the same behavior of the WT counterpart when tested as hosts in metastasis assays, suggesting that TIMP-1 influence on tumor depends exclusively on the TIMP-1 genotype of neoplastic cells rather than the one of the host. Nonetheless, alternative scenarios are still possible. The chronic, rather than acute, lack of TIMP-1 in the BM niche could play a role in the regulation of HSCs, and the general absence of TIMP-1 may be responsible for the hematopoietic phenotype, regardless of the tissue or cell type producing TIMP-1. In addition, the constitutive lack of TIMP-1 could be critical, especially during the early phases of hematopoietic development, rather than during adult life. Thus, we cannot completely rule out an environmental role for TIMP that would be consistent with some of our data.

MMP-2 and MMP-9 have been extensively investigated as key players in HSC trafficking. The release of MMPs from neutrophil granules, secondary to G-CSF mobilization regimens, leads to CXCL12 cleavage, thus favoring the mobilization of HSCs from the BM to the PB.³⁸  Because of the capability of binding pro-MMP-9 and inhibiting its proteolytic activity, we expected TIMP-1 deficiency to enhance HSC mobilization. Nevertheless, the percentage of KSL cells circulating in the PB, in steady-state conditions, did not appear to be affected in TIMP-1^−/− mice. Similarly, Haviernik et al³⁶  demonstrated that TIMP-1 deficiency does not influence HSC mobilization in mice undergoing G-CSF treatment, suggesting that other proteases might play a compensatory role.

Our findings indicate that the functional impairment of the stem cell pool in TIMP-1^−/− mice is linked to subversion of cell-cycle dynamics. HSCs in TIMP-1 KO mice present an abnormal cell cycle distribution, with a significant decrease in the percentage of quiescent cells in G₀. However, despite that an increased percentage of HSCs leaves quiescence and enters G₁, a lower percentage of HSCs is in S/G₂M, suggesting a defect in the G₁/S transition. Failure to efficiently achieve cell-cycle completion may help explain the decline in BM cellularity in KO mice and the impaired engraftment of TIMP-1^−/− BM cells, when tested in a competitive setting. These findings suggest that TIMP-1 may play a role in the preservation of HSC quiescence, which is essential to ensure lifelong maintenance of hematopoiesis and to prevent premature exhaustion under conditions of hematopoietic stress, such as transplantations, infections, or chemotherapeutic insults. On the contrary, despite the antiapoptotic role in cancer cell lines that TIMP-1 has been shown to play, HSCs in TIMP-1–deficient mice do not seem to present any evident survival defect. Although we cannot exclude the possibility that a survival defect might arise after a more intense stress, it seems probable that inhibition of apoptosis in cellular pools critical for survival, such as stem cells, would be protected by redundant molecular pathways, capable of compensating for TIMP-1 absence.

Despite intense focus on the signaling pathways activated by TIMP-1, many of the controversies concerning TIMP-1 molecular mechanisms are yet to be reconciled. Among several hypotheses, one of the most intriguing is based on the existence of a specific membrane receptor, the CD63 protein,³⁹  whose binding to TIMP-1 is assumed to lead to proliferation and survival. CD63 was the first-identified tetraspanin, a member of a large family of receptors involved in cell motility, adhesion, and survival.³⁹  Tetraspanins have been shown to form complex networks with other membrane proteins, generating tetraspanin-enriched microdomains. Interestingly, Jung et al¹⁷  demonstrated that TIMP-1 contributes to the formation of the ternary complex TIMP-1/CD63/β1-integrin, which inhibits apoptosis in MCF10A breast epithelial cells. Of note, β1-integrin has been previously shown to play a pivotal role in HSCs, by controlling homing and survival.^40,41  For instance, Yoshihara et al⁴²  demonstrated the role of THPO/MPL signaling in preserving HSC quiescence through the activation of β1-integrin. Therefore, it seems conceivable that the capability of TIMP-1 in stabilizing the TIMP-1/CD63/β1-integrin complex and keeping β1-integrin in an activated conformation could play a role in preserving HSC quiescence.⁴³  Interestingly, β1-integrin blocking can cause an increase in p53 nuclear levels, thus interfering with proliferation,⁴⁴  and drug-resistance in B-cell chronic lymphocytic leukemia cells has been suggested to involve p53 inhibition on activation of α4β1-integrin.⁴³  Consistent with these findings, we found that disrupting TIMP-1/CD63 interaction by knocking down TIMP-1 expression compromises HSC quiescence and destabilizes cell cycle dynamics.

The glycoprotein CD44 is an adhesion molecule widely expressed in many cell types, including HSCs. Although CD44 was first described as a key regulator of cell adhesion, it is now widely accepted that its engagement activates signaling cascades involved in cell proliferation, survival, and migration.^45,46  Godar et al³⁵  effectively showed that CD44 is fundamental for the proliferation of tumorigenic mammary epithelial cells and demonstrated that its expression is directly inhibited by the tumor suppressor p53, a master regulator of cell proliferation. The investigators showed that, by binding to a noncanonical p53-binding sequence in the CD44 promoter, p53 inhibits CD44 expression and down-regulates the growth-promoting cascade downstream of CD44.³⁵  Interestingly, we found that TIMP-1^−/− SP^KLS cells not only present cell- cycle dysregulation at the level of the G₁/S transition, but they also express significantly lower levels of CD44. We also showed that p53 levels are increased in TIMP-1^−/− HSCs. These findings suggest the involvement of the p53/CD44 axis in cell-cycle dysregulation.

On the basis of the aforementioned findings, we propose a hypothetical mechanistic model for TIMP-1 control of proliferation in HSCs. As shown in supplemental Figure 7, the ternary complex TIMP-1/CD63/β1-integrin contributes to maintain β1-integrin activated. When stimulated, β1-integrin keeps p53 levels low, thus promoting cell proliferation through CD44. In contrast, in a TIMP-1^−/− model, TIMP-1/CD63/β1-integrin assembling is compromised, resulting in the down-regulation of β1-integrin cascade, with subsequent increase of p53 levels. By binding to the CD44 promoter, p53 represses CD44 expression and causes cell-cycle alteration. Incidentally, because of the role that CD44 plays in CXCL12-driven HSC migration, the decreased expression of CD44 could explain, concomitantly with cell-cycle dysregulation, the decreased responsiveness of TIMP-1^−/− HSCs to CXCL12 chemotactic gradients.

TIMP-1 has been proposed as an alternative prognostic marker for the stratification of patients with metastatic breast cancer. Contrary to what the role of TIMP-1 in inhibiting the pro-metastatic factor MMP-9 might suggest, high TIMP-1 levels actually correlate with a worse prognosis and a faster progression of the disease.^47,48  This phenomenon seems to be related to the capability of TIMP-1 of promoting cell survival and proliferation of cancer cells, making it an appealing target for novel therapeutic strategies for the treatment of breast cancer, as well as fibrosis secondary to hypertension⁴⁹  or liver diseases.^50,51  Although several compounds are currently under investigation as TIMP-1 inhibitors, the potential side effects of such therapeutic approaches are still controversial and poorly investigated. On the basis of our observations, drugs targeting TIMP-1 could potentially induce complications in terms of hematologic toxicity, advising a cautious use of these compounds. In conclusion, our study shows that TIMP-1 is involved in regulating the LT-HSC cell cycle, thus providing a novel tool for understanding HSC behavior.

We thank Chris Threeton and Xuejun (Tony) Zhang for help with flow cytometric sorting. We also thank all the members of the Goodell Lab (Baylor College of Medicine), as well as Roberto M. Lemoli, Michele Baccarani, and Sante Tura (University of Bologna) for helpful discussion and support.

This work was supported by a grant from the Italian Association for Leukemias and Lymphomas (BolognAIL; L.R.) and by the National Institutes of Health (grants DK58192, EB005173, and HL081007) and the American Heart Association (grant 0740020N).

National Institutes of Health

Contribution: L.R. designed the research, performed experiments, analyzed the data, and wrote the paper; A.V.E. performed the real-time evaluation of TIMP-1 expression; and M.A.G. designed the research and wrote the paper.

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

Correspondence: Margaret A. Goodell, One Baylor Plaza, N1030, Houston, TX 77030; e-mail: goodell@bcm.edu; or Lara Rossi, Via Massarenti, 9, 40138 Bologna, Italy; e-mail: lararossi79@yahoo.it.