OUP user menu

Krüppel-like factor 2 improves neovascularization capacity of aged proangiogenic cells

Reinier A. Boon, Carmen Urbich, Ariane Fischer, Ruud D. Fontijn, Florian H. Seeger, Masamichi Koyanagi, Anton J.G. Horrevoets, Stefanie Dimmeler
DOI: http://dx.doi.org/10.1093/eurheartj/ehq137 371-377 First published online: 21 May 2010


Aims Coronary artery disease (CAD) patients have less circulating proangiogenic cells (PACs), formerly known as endothelial progenitor cells, which exhibit impaired neovascularization properties. Inverse correlations were also found between PAC function and risk factors like age. Krüppel-like factor 2 (KLF2) is expressed by mature endothelial cells (ECs), is induced by both shear stress and statins, and provokes endothelial functional differentiation. The aim of this study is to identify whether KLF2 can reverse negative effects of ageing on PAC function.

Methods and results We describe that progenitor cells in the bone marrow and PACs also express KLF2 at a comparable level to mature ECs and that senescence decreases KLF2 levels. To study the effects of ageing on KLF2 levels, we compared progenitor cells of 4 weeks and 16- to 18-month-old C57BL/6 mice. In addition to the three-fold reduction of circulating Sca1+/c-Kit+/Lin progenitor cells and the 15% reduction of Sca1+/Flk1+ endothelial-committed progenitor cells, the spleen-derived PACs and bone marrow-derived progenitor cells isolated from aged mice showed a lower level of KLF2 when compared with young mice. Lentiviral overexpression of KLF2 increased human PAC numbers and endothelial nitric oxide synthase expression by 60% during in vitro culture. Endothelial lineage-specific KLF2 overexpression in aged bone marrow-derived mononuclear cells strongly augments neovascularization in vivo in a murine hind-limb ischaemia model.

Conclusion These results imply that KLF2 is an attractive novel target to rejuvenate PACs before autologous administration to CAD patients.

  • Angiogenesis
  • KLF2
  • Ischaemia
  • Cell therapy
  • Ageing


The coordinated growth of new blood vessels is essential for tissue repair after ischaemia. This can be mediated by the migration and proliferation of neighbouring endothelial cells (ECs), but also through incorporation of circulating proangiogenic cells (PACs), previously known as endothelial progenitor cells (EPCs).1 Proangiogenic cells are circulating bone marrow-derived cells, which can home to sites of neovascularization and differentiate into ECs or promote angiogenesis by secreting paracrine factors, and have been successfully used for therapeutic angiogenesis.2,3 However, patients in most need of neovascularization are often burdened with co-morbidity factors, known to be detrimental for PAC numbers and quality.4,5

Shear stress, the frictional force generated by blood flow, is one of the factors that play a role in the differentiation of PACs to ECs.6 Shear stress was described to induce Krüppel-like factor 2 (KLF2) expression,7 where it induces functional quiescence.8,9 Krüppel-like factor 2 was also found to be highly expressed in pluripotent stem cells10 and can be used as one of the factors described for the generation of induced pluripotent stem cells.11 Therefore, we hypothesized that KLF2 is important for the stem/progenitor cell-like function of PACs.

To test this hypothesis, we analysed the expression of KLF2 and we conclude that KLF2 levels in aged and senescent PACs, ECs, and progenitor cells are lower compared with control cells. Finally, we describe that KLF2 overexpression increases the number of human peripheral blood-derived PACs during ex vivo culture and that endothelial lineage-restricted overexpression of KLF2 improves the age-induced impairment in neovascularization capacity of PACs.


Animal models

C57BL/6 mice were obtained from Charles River (Sulzfeld, Germany). The animal experiments were approved by the Regional Board of the State of Hessen, Germany. The hind-limb ischaemia mouse model and subsequent laser-Doppler perfusion measurement were performed as described.12 Briefly, lentivirally transduced PACs cultured from bone marrow of 18-month-old C57BL/6 mice were counted and equal cell numbers were injected in the tail vein of 4-week-old C57BL/6 mice (2.5 × 105 cells per mouse), 1 day after the hind-limb ischaemia operation was performed.

Cell isolation and culture

Ex vivo PAC differentiation from circulating mononuclear cells (MNCs) was assayed as described previously to isolate early EPC.4 In brief, MNCs were isolated by density gradient centrifugation with Biocoll separating solution (density 1.077; Biochrom AG, Berlin, Germany) from human peripheral blood buffy coats or full blood. For the isolation of spleen-derived PACs, murine MNCs were isolated from homogenized splenic tissue by density gradient. Immediately after isolation, 8 × 106 MNCs/mL of medium were plated on culture dishes coated with human fibronectin (Sigma) and maintained in endothelial basal medium (EBM; Cambrex, East Rutherford, NJ, USA) supplemented with EGM SingleQuots (Cambrex) and 20% foetal calf serum (FCS; GIBCO BRL, Carlsbad, CA, USA). After 3 days in culture, non-adherent cells were removed by thorough washing with phosphate-buffered saline. Adherent cells were stained with 2.4 µg/mL DiI-Ac-LDL (Harbor Bio-Products, Norwood, MA, USA) at 37°C for 1 h. Human umbilical vein endothelial cells (HUVECs) were purchased from CellSystems and cultured in EBM (Cambrex) supplemented with EGM SingleQuots (Cambrex) and 10% FCS (GIBCO BRL). Proangiogenic cells are characterized as KDR+/CD105+/CD34/CD133/CD45+ cells. Mouse bone marrow cells were sorted in lineage-positive and -negative cell populations using a magnetic-activated cell sorting kit (mouse lineage depletion kit, Miltenyi Biotec, Bergisch Gladbach, Germany). Proangiogenic cells and HUVECs were exposed to shear stress (15 dynes/cm2) using a cone-and-plate device, as described previously,13 for 18–24 h.

Endothelial differentiation of embryonic stem cells

ES D3 cells, a 129/Sv-derived embryonic stem (ES) cell line, were cultured as described previously.14 To initiate ES cell differentiation and embryoid body formation, ES cells were trypsinized and suspended in IMDM (GIBCO BRL) with 15% foetal bovine serum, 10 µg/mL insulin (Sigma), 100 U/mL penicillin, 100 µg/mL streptomycin, 450 µmol/L monothioglycerol, and endothelial growth factors including 50 ng/mL recombinant human vascular endothelial growth factor (VEGF) (PeproTech, Rocky Hill, NJ, USA), 2 U/mL recombinant human erythropoietin (Cilag AG), 100 ng/mL human basic fibroblast growth factor (Genzyme, Cambridge, MA, USA), and 10 ng/mL murine interleukin 6 (Genzyme).

Flow cytometry analysis

Murine bone marrow (24 h after lentiviral transduction) or peripheral blood-derived cells were incubated for 30 min at 4°C with FITC-, APC- or PE-labelled antibodies against Flk1 (KDR, VEGFR2), Sca-1, c-kit, and/or an anti-lineage marker antibody cocktail (BD Biosciences, Franklin Lakes, NJ, USA). Cells were then washed and surface expression was quantified using a FACS, CantoII (BD Biosciences).

Real-time reverse transcription–polymerase chain reaction and western blotting

Total RNA was isolated using the Qiagen RNA extraction kit. cDNA was synthesized using Oligo(dT) as primer and M-MLV reverse transcriptase, according to the manufacturers protocol (Invitrogen, Carlsbad, CA, USA). The reverse transcriptase–polymerase chain reaction (RT–PCR) was composed of LightCycler–RNA MasterPLUS SYBR Green I (Roche, Basel, Switzerland), using the LightCycler (Roche) real-time thermocycler according to the instructions of the manufacturer (Roche). Amplification was performed with 40 cycles and an annealing temperature of 60°C. The specificity of the amplification reaction was determined by a melting curve analysis. Western blotting and the KLF2 antibody were described previously.15

Lentiviral constructs

Polymerase chain reaction amplification was used to add terminal NheI (5′) and SpeI (3′) restriction sites to a 787 bp fragment containing TIE2 promoter/enhancer sequences (kindly provided by Dr J.A. Kuivenhoven).16 This fragment was ligated in the pRRL-cPPT-CMV-X2-PRE-SIN vector, according to our recently described procedure.17 The original CMV promoter and multiple cloning site were removed by combined digestion with SpeI and ClaI and a new multiple cloning site was introduced by ligation of a synthetic adapter containing restriction sites SpeI–AgeI–BclI–SalI–XbaI–ClaI. Subsequently, either KLF2 or eGFP cDNA were inserted as SalI–XbaI fragments. Finally, the SV40 late poly(A) signal was introduced as XbaI–ClaI fragment. This vector was also used for insertion of the 566 bp proximal KLF2 promoter as NheI–SpeI PCR fragment to yield the lentivector used in flow cytometry experiments. The lentivector with KLF2 under control of the PGK promoter and the production of lentivirus containing supernatants were described previously.8

Statistical analysis

Data were analysed with Graphpad Prism 5 using unpaired Student's t-tests when comparing two conditions, or a one-way ANOVA with Bonferroni's correction for multiple comparisons. Probability values of <0.05 were considered significant and tests were performed two-sided. Data are presented as mean and error bars depict the standard error of the mean.


Krüppel-like factor 2 is expressed in embryonic stem cells, mature endothelial cells, and proangiogenic cells, but is reduced by senescence in vitro

Since the transcription factor KLF2 was identified as a marker for undifferentiated stem cells10 as well as a functional endothelial marker gene,8,9 we analysed its expression in an established model of endothelial differentiation of mouse ES cells (ESCs).14 Reverse transcriptase–polymerase chain reaction analysis revealed that KLF2 is indeed expressed by ESCs (Figure 1A). Krüppel-like factor 2 expression levels then drop during initial differentiation stages, but rise again during commitment to endothelial differentiation, as marked by the induction of VEGF receptor 2 (Flk1), CD31, endothelial nitric oxide synthase (eNOS) and the angiopoietin receptor Tie2 within 4–10 days (Figure 1A). Interestingly, direct transcriptional targets of KLF2 in ECs, thrombomodulin, and eNOS, which are considered endothelial markers, are also expressed in ESCs.

Figure 1

Krüppel-like factor 2 (KLF2) is expressed bi-phasicly in differentiating embryonic stem cells and proangiogenic cells (PACs) and is repressed by replicative and cell culture-induced senescence. (A) mRNA levels of murine KLF2, thrombomodulin (THBD), eNOS, Tie2, CD31, Flk1, and GAPDH were analysed by reverse transcriptase–polymerase chain reaction and gel-electrophoresis. The first lane represents undifferentiated embryonic stem cells and the second, third, and forth lanes represent embryonic stem cells under endothelial differentiation conditions for 4, 7, and 11 days, respectively. (B) In proangiogenic cells (static), proangiogenic cells exposed to shear stress, human umbilical vein endothelial cells (static) and human umbilical vein endothelial cells exposed to shear stress, mRNA levels of human KLF2 were measured by real-time reverse transcriptase–polymerase chain reaction. Expression levels were corrected for GAPDH mRNA levels (n = 3–9). (C) Reverse transcriptase–polymerase chain reaction analysis of KLF2 and GAPDH mRNA levels in human umbilical vein endothelial cells (HUVECs) cultured for 5, 8, 14, or 16 passages. (D) Krüppel-like factor 2 and GAPDH mRNA levels in proangiogenic cells that were cultured for 3, 4, 6, 7, 10, or 12 days. *P < 0.05 compared with static controls. Representative images of three experiments are shown.

Next, we analysed the expression levels of KLF2 in human PACs and compared them with mature ECs, using real-time RT–PCR. Human PACs express KLF2 at a comparable level to mature ECs and KLF2 can be induced by shear stress in PACs to almost the same extent as is observed for HUVECs (Figure 1B).7 On the basis of reports that expression of eNOS, a direct transcriptional target of KLF2, decreased in senescent ECs,18 we assessed the expression of KLF2 in replicative senescent HUVECs, compared with relatively young cells (Figure 1C). Young HUVECs (passages 5 and 8) express higher levels of KLF2 mRNA when compared with senescent HUVECs (passages 14 and 16). Krüppel-like factor 2 levels in PACs also decrease during cell culture-induced senescence (Figure 1D). This in vitro model for senescence has been shown to reduce telomere length without affecting basal apoptosis,18,19 but de-differentiation cannot be excluded and therefore we next studied ageing in vivo.

Age reduces progenitor cell and proangiogenic cell numbers and represses Krüppel-like factor 2 expression

To study KLF2 during in vivo ageing, we compared progenitor cells of young (4 weeks old) and aged (16–18 months old) C57BL/6 mice. Results indicate that in aged mice, numbers of acetylated LDL-positive PACs are reduced by ∼40%, when compared with young mice (Figure 2A). Immunofluorescent flow cytometry was used to analyse cellular surface expression levels of Sca1, c-Kit (both stem cell markers), and Flk1, an endothelial lineage marker, in combination with a negative selection for lineage markers (Lin). Aged mice have three-fold less Sca1+/c-Kit+/Lin progenitor cells as well as ∼15% less Sca1+/Flk1+ endothelial-committed progenitor cells, when compared with young control mice (Figure 2B). The spleen-derived PACs isolated from aged mice also have ∼60% lower levels of KLF2 compared with young mice (Figure 2C). To identify which progenitor cell populations express KLF2, we used a lentiviral GFP reporter construct containing the KLF2 promoter on bone marrow cells isolated from young and aged mice in combination with flow cytometry analysis for Sca1, c-Kit, and Flk1. This revealed that KLF2 is mainly expressed by Sca1/Flk1 double-positive and c-Kit/Flk1 double-positive progenitor cells (see Supplementary material online, Figure S1A), also confirmed by real-time RT–PCR analysis of sorted lineage-positive and -negative cells (see Supplementary material online, Figure S1B). The number of Sca1/c-Kit/KLF2 triple-positive cells shows a trend towards a reduction by age (Figure 2D, P = 0.06). Moreover, the levels of KLF2 in either Sca1/Flk1 or c-Kit/Flk1 double-positive cells are robustly decreased by age (Figure 2E). Together, these results show that in addition to reduced numbers of circulating PACs in aged mice, the expression level of KLF2 is also markedly reduced in the remaining cells.

Figure 2

Aged mice have less progenitor cells and proangiogenic cells, which also express lower KLF2 levels than young mice. (A) Proangiogenic cells were isolated from spleens from young (4 weeks) and aged (16–18 months) mice and acetylated LDL uptake was measured by fluorescence microscopy (n = 5). (B) Mononuclear cells were isolated from peripheral blood from young and aged mice. Flow cytometry was used to analyse Sca1+/c-Kit+/Lin, Sca1+/c-Kit+ and Sca1+/Flk1+ cell populations (n = 5). (C) Proangiogenic cells were isolated from the spleens of young and aged mice and KLF2 levels were measured by real-time reverse transcriptase–polymerase chain reaction. GAPDH was used as a control (n = 5). (D and E) Bone marrow was isolated from young (4 weeks) and aged (16–18 months) mice and transduced with a lentiviral KLF2 promoter-GFP reporter construct. Sca1+/c-Kit+/GFP+ cell number (D) and mean fluorescence intensity (MFI) of the GFP signal in Sca1+/Flk1+ or c-Kit+/Flk1+ cells (E) were measured by flow cytometry. **P < 0.01, *P < 0.05, and #P = 0.06.

Krüppel-like factor 2 overexpression increases proangiogenic cell number ex vivo and augments the impaired neovascularization capacity of aged proangiogenic cells in vivo

To identify a potential role for KLF2 in PACs, KLF2 was overexpressed using lentivirus.15 Three days after lentiviral transduction with mock or KLF2 overexpression virus, KLF2 mRNA and protein levels were measured using RT–PCR and western blot analysis (Figure 3A), which confirmed overexpression of KLF2. As a measure of KLF2 activity, we analysed an established transcriptional target of KLF2, eNOS, by real-time RT–PCR (Figure 3B), which showed that KLF2 induces eNOS in PACs. Next, we assessed PAC numbers (Figure 3C), which revealed that lentiviral overexpression of KLF2 increases the number of PACs by 60% in vitro. Additionally, KLF2 overexpression in bone marrow cells increases Flk+ cell numbers (see Supplementary material online, Figure S2).

Figure 3

Lentiviral overexpression of KLF2 improves age-impaired neovascularization. (A) Three days after lentiviral transduction of human proangiogenic cells with mock- or KLF2-lentivirus, protein and RNA were isolated. Krüppel-like factor 2 and GAPDH mRNA levels were measured using reverse transcriptase–polymerase chain reaction. Krüppel-like factor 2 and α-tubulin protein levels were analysed by western blot and (B) eNOS mRNA levels were measured by real-time reverse transcriptase–polymerase chain reaction. (C) Three days after mock- and KLF2-transduction, the number of proangiogenic cells was counted using DiI-labelled acetylated LDL (n = 5). (D and E) Mouse bone marrow-derived proangiogenic cells from young (4 weeks old) or old (18 months old) mice were transduced with lentivirus containing either KLF2 or GFP under the control of the Tie2 promoter. These transduced proangiogenic cells or phosphate-buffered saline (control) were injected in the tail vein, 1 day after the hind-limb ischaemia operation was performed on 4-week-old C57BL/6 mice. Perfusion was measured using laser-Doppler imaging (D) and calculating capillary density (E) was determined on histological sections (number of CD31+/α-smooth muscle actin vessels per laminin+ myocyte, three random sections per animal) (n = 3–4). *P < 0.05, **P < 0.01.

To identify whether KLF2 induces neovascularization in vivo, we generated lentivirus encoding either KLF2 or GFP under the control of the endothelial lineage-specific Tie2 promoter. Bone marrow mononuclear cells from 18-month-old C57BL/6 mice were transduced with Tie2-GFP or Tie2-KLF2 lentivirus and maintained for 3 days under standard PAC conditions. The hind-limb ischaemia operation was performed and equal numbers of transduced PACs were injected intravenously 1 day post-operation. Laser-Doppler perfusion measurement was performed 14 days post-operation and lower limb perfusion was calculated as the ratio of the ischaemic (right) hind limb vs. the control (left) hind limb (Figure 3D). Since perfusion measurements can also be influenced by vasodilation, neovascularization was further quantified by measuring capillary density in the ischaemic limbs (Figure 3E). This showed that aged PACs transduced with the GFP-control lentivirus indeed fail to induce neovascularization in the ischaemic hind limb and that endothelial lineage-specific KLF2 expression potently induces the neovascularization capacity of aged PACs, albeit still not to the level observed when using young PACs (perfusion ratio ∼0.8).20 Although we cannot exclude additional effects on vasodilation, these data demonstrate that KLF2 improves the neovascularization capacity of PACs in an endothelial lineage-specific manner.


The present study demonstrates that KLF2 is expressed by PACs, but is reduced by ageing and senescence, indicating a positive correlation between KLF2 levels and PAC function. Moreover, overexpression of KLF2 increases the number of human PACs when cultured ex vivo and endothelial lineage-specific KLF2 overexpression restores the neovascularization capacities of aged PACs in vivo.

The observation that PAC quality of patients most in need of neovascularization is limiting, raised the concept that improving PAC functionality in vitro, prior to re-administering these cells to the patient, might serve as a better therapeutic tool for improvement of impaired neovascularization in ageing. A number of factors have been proposed for this so-called in vitro rejuvenation of PACs, including statins. We show that restoring KLF2 levels in aged PACs might be a useful strategy for the ‘in vitro rejuvenation’ of these PACs. HMG-CoA reductase inhibitors (statins) have been shown to induce PAC numbers and function, partly through the inhibition of senescence,21,22 and are used as a standard component of the in vitro differentiation protocol used in clinical trials.23 Statins are also known to augment one of the best-known inducers of PAC function, eNOS, which we show to be up-regulated in PACs by KLF2 (Figure 3B). Interestingly, the KLF2-induced increase in PAC numbers (Figure 3C) is identical to the previously reported statin-mediated increase in PAC numbers (∼1.6-fold).22 These published findings, combined with the results reported here, suggest that the superior PAC function induced by statins could be mediated through KLF2.

In mature ECs, KLF2 has been shown to potently inhibit angiogenesis,24 which stands in apparent contrast to the findings reported here. Proangiogenic cells are however very distinct from mature ECs. Although mature ECs respond to a gradient of VEGF with loss of polarity, migration towards the source of VEGF, and proliferation, which is inhibited by KLF2 via multiple mechanisms,8,25,26 PACs are mobilized from the bone marrow, home to the site of neovascularization guided by an SDF1 gradient, and either incorporate into the endothelium or transiently stimulate angiogenesis of the pre-existing endothelium by paracrine factors.27 In our experiments, where we inject PACs that overexpress KLF2, the pre-existing vascular bed does not express altered KLF2 levels and is in principle still capable of angiogenesis, which is augmented by KLF2 overexpressing PACs.

KLF2 seems to be able to positively affect several features of PACs. First, KLF2 is expressed in pluripotent ESCs as well as in ESCs differentiated towards the endothelial lineage (Figure 1A). Therefore, KLF2 could induce a PAC phenotype more reminiscent of stem cells. Secondly, KLF2 could skew PACs more towards an endothelial phenotype, because KLF2 is highly expressed in endothelial-committed cells, induces Flk+ cell numbers, and induces eNOS expression (Figure 3B and Supplementary material online, Figure S2). Interestingly, KLF2 induces apoptosis in total bone marrow cells under ex vivo culture conditions (see Supplementary material online, Figure S2), likely due to its inhibition of NFκB,28 a known anti-apoptotic factor.29 However, KLF2 does induce PAC numbers (Figure 3C), correlating to an increase in AnnexinV/Flk+ cells (see Supplementary material online, Figure S2), indicating that KLF2 favours the survival of the endothelial-committed population. Thirdly, monocytic features are also known to be essential for PAC function12 and KLF2 overexpression was shown to prevent monocytes from differentiating to macrophages.30 Together, the synergistic effects of KLF2 on these three functions could thus maintain proper differentiation of PACs and thereby augment their functionality. Collectively, these results indicate KLF2 to be an attractive novel target for the improvement of PAC function prior to autologous transfer.


This study was supported by the European Vascular Genomics Network (EVGN). R.A.B. is supported by the Netherlands Organization for Scientific Research (NWO).

Conflict on interest: none declared.


View Abstract