Platelets prime hematopoietic–vascular niche to drive angiocrine-mediated liver regeneration

In mammals, the livers regenerate after chemical injury or resection of hepatic lobe by hepatectomy. How liver regeneration is initiated after mass loss remains to be defined. Here we report that following liver injury, activated platelets deploy SDF-1 and VEGF-A to stimulate CXCR7+ liver sinusoidal endothelial cell (LSEC) and VEGFR1+ myeloid cell, orchestrating hepatic regeneration. After carbon tetrachloride injection or hepatectomy, platelets and CD11b+VEGFR1+ myeloid cells were recruited to LSECs, and liver regeneration in both models was impaired in thrombopoietin-deficient (Thpo−/−) mice repressing production of circulating platelets. This impeded regeneration phenotype was recapitulated in mice with either conditional ablation of Cxcr7 in LSEC (Cxcr7iΔ/iΔ) or Vegfr1 in myeloid cell (Vegfr1lysM/lysM). Both Vegfr1lysM/lysM and Cxcr7iΔ/iΔ mice exhibited suppressed expression of hepatocyte growth factor and Wnt2, two crucial trophogenic angiocrine factors instigating hepatocyte propagation. Of note, administration of recombinant thrombopoietin restored the prohibited liver regeneration in the tested genetic models. As such, our data suggest that platelets and myeloid cells jointly activate the vascular niche to produce pro-regenerative endothelial paracrine/angiocrine factors. Modulating this ‘hematopoietic–vascular niche’ might help to develop regenerative therapy strategy for hepatic disorders.


MATERIALS AND METHODS Animals
C57BL/6J and LysM-Cre mice were obtained from Jackson laboratory. The Chd5(PAC)Cre ERT2 mice expressing tamoxifen-responsive Cre ERT2 driven by EC-specific VE-cadherin promoter 58,59 were provided by Dr Ralf Adams. Thrombopoietin (TPO)-deficient (Thpo − / − ) mice 60 were kindly offered by Dr Frederic J de Sauvage (Genentech, Inc., San Francisco, CA, USA). Mice harboring loxP site-flanked exon 3 of Cxcr7 (Cxcr7 LoxP/LoxP ) were kindly provided by ChemoCentryx, Inc. (Mountain View, CA, USA). Floxed Vegfr1 mice were kindly provided by Dr Guo-Hua Fong. Rosa-Cre ERT2 animals expressing tamoxifen-responsive inducible Cre were described previously. 15,39 The Chd5(PAC)Cre ERT2 mouse line was crossed with floxed Cxcr7 mice to generate Cxcr7 iΔEC/iΔEC mice and control Cxcr7 iΔEC/+ mice after treatment of tamoxifen at a dose of 250 mg kg − 1 for 6 days, and interrupted for 3 days after the third dose. Mice were rested for at least 20 days after the last injection. Cxcr7 LoxP/LoxP mice were also crossed with Rosa-Cre ERT2 mice to generate Cre + Cxcr7 loxP/loxP mice, resulting Cxcr7 deletion in adult mice (Cxcr7 iΔ/iΔ ). Floxed Vegfr1 mice were bred with LysMdriven Cre (Jackson, Bar Harbor, ME, USA) to generate mice lacking Vegfr1 in myeloid cells (Vegfr1 lyzM/lyzM ). Deletion of target genes was corroborated by quantitative PCR. Investigators who performed mouse experiments and who analyzed the pattern of cell distribution were randomly assigned with samples, and they were blinded to the genotype of the animals or samples from various groups. All animal experiments were carried out following the guidelines of Institutional Animal Care and Use Committee at Weill Cornell Medicine.

Mouse liver regeneration and repair models
In all, 70% PH model was used as previously described. 15 In brief, three most anterior lobes were resected without injuring the blood supply to the caudate and the right lobes after mice were anesthetized by 100 mg kg − 1 intraperitoneal (i.p.) ketamine and 10 mg kg − 1 xylazine followed by midline laparotomy. To induce liver injury, single injection of carbon tetrachloride (Sigma-Aldrich, St Louis, MO, USA) in oil at a concentration of 40% (0.64 mg ml − 1 ) was injected to mice at a dose of 1.6 mg kg − 1 to 1 induce acute liver injury. 35,61 Hepatic regeneration was assessed based on the following criteria: liver lobe weight, hepatocyte proliferation, alanine aminotransferase level and histology using hematoxylin and eosin staining. Six-to ten-week-old mice were utilized and compared.

Stimulation of thrombopoiesis
To stimulate thrombopoiesis, recombinant TPO, VEGF-A and/or SDF-1 (PeproTech, Rocky Hill, NJ, USA) was injected into Thpo − / − or WT mice i.p. at a dose of 25 μg kg − 1 on a daily basis 10 days before PH or CCL 4 injury and afterwards. Vehicle for individual cytokines was also injected as a control group. The degree of hepatic regeneration was evaluated with control group, including alteration in circulating platelets and parameters of hepatogenesis.
Cell proliferation in vivo was measured by 5-Bromo-2 0 -deoxyuridine (BrdU) uptake. Single dose of BrdU (Sigma, St Louis, MO, USA) at 50 mg kg − 1 was i.p. injected to mice 1 h before killing. Liver lobes were removed, weighed and slice of tissues were incubated with 1 M HCl at room temperature for 1 h, neutralized with 10 mM Tris (pH 8.5) for 15 min. After incubation with secondary antibody (Jackson ImmunoResearch), cells incorporated with BrdU were identified as proliferating hepatocytes.

Image acquisition and analysis
Histology analysis of liver sections was captured with Olympus BX51 microscope (Olympus America, Center Valley, PA, USA), and fluorescent images were recorded on AxioVert LSM710 (Carl-Zeiss, Thornwood, NY, USA) confocal microscope. Fluorescent signals in slide were independently evaluated by two investigators from randomly selected fields of view. Parameters from each individual animal were measured and averaged.

Flow cytometry analysis
Flow cytometry analysis of platelets and LSECs on isolated liver NPCs as previously described. 15,45,62 In brief, retrieved livers from killed animals were minced, digested in liver digestion medium (Invitrogen), and filtered through a 30-μm strainer. Single-cell suspensions were preincubated with Fc block (CD16/CD32; BD Biosciences) and then incubated with primary antibodies recognizing mouse LSECs and hematopoietic cells, as well as rat IgG2aκ and IgG2aβ isotype control. Primary antibodies were directly conjugated to different Alexa Fluor dyes or Quantum Dots (BD Biosciences) using antibody labeling kits (Invitrogen). Labeled cell populations were measured by a LSRII flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). Compensation for multivariate experiments was carried out with FACS Diva software (Becton Dickinson Immunocytometry Systems, Franklin Lakes, NJ, USA).

Gene expression analysis by real-time PCR
Total RNA was extracted using RNeasy kit (Qiagen, Germantown, MD, USA). After isolation, 500 ng of total RNA was transcribed into complementary DNA by using the superscript reverse transcriptase kit (Invitrogen). The detection of complementary DNA expression for the specific genes was performed by using the SYBR Green quantitative PCR (Applied Biosystems, Foster City, CA, USA). To selectively knockdown Cxcr7 in LSECs, shRNA lentiviruses were generated by cotransfecting 15 μg of shuttle lentiviral vector, 3 μg of pENV/VSV-G, 5 μg of pRRE and 2.5 μg of pRSV-REV in 293T cells. 35 Viral supernatants were concentrated by ultracentrifugation and used to transduce LSECs.

Statistical analysis
All data were presented as the mean ± s.e.m. Comparisons between different groups were made using one-way analysis of variance. Statistical significance was set at Po0.05. Each experiment was at least three times.

RESULTS
We first examined the localization of platelet-derived VEGF-A and SDF-1 after carbon tetrachloride (CCl 4 )-induced liver injury. VEGF-A and SDF-1 were co-stained with platelet surface marker CD41 in the liver 24 h after CCl 4 i.p. injection (Figures 1a and b). Compared to sham mice, CCl 4 injection caused significant deposition of CD41 + platelets on VEGFR3 + LSECs, with the majority of them stained for VEGF-A and SDF-1 (Figure 1c). Flow cytometric analysis of hepatic NPCs showed that CD41 + platelets constituted 24% of NPCs in CCl 4 -injured but not sham mice (Figure 1d), and platelet activation marker, P-selectin, was presented on the surface of 73% of CD41 + platelets in the damaged liver. Thus, CCl 4 injury caused recruitment and local activation of platelets secreting VEGF-A and SDF-1, which might activate hematopoietic and LSECs via VEGF-A and SDF-1 receptors.
To test the contribution of platelets in protecting against liver injury, we examined mice deficient of TPO (Thpo − / − ) after CCl 4 injection. Platelet number is decreased in Thpo − / − mice by 95% as compared to wild-type (WT) mice. 60,64 Hepatocyte proliferation after CCl 4 injection was significantly reduced in Thpo − / − mice than that of WT control group (Figures 2a-c). Meanwhile, hepatic injury was markedly increased, as indicated by elevation of plasma alanine aminotransferase activity (Figure 2d). Of note, the impaired hepatocyte proliferation and enhanced hepatic injury in Thpo − / − mice were rescued by injection of VEGF-A and/or SDF-1 (Figures 2a-d). Therefore, activated platelets recruited to the injured vascular bed supply SDF-1 and VEGF-A to stimulate hepatic repair.
We then further examined the contribution of platelets to liver regeneration after 70% PH. Two days after PH, CD41 + platelets were similarly recruited onto VEGFR3 + LSECs (Figure 3a). There was a co-localization of SDF-1 and VEGF-A with CD41 + platelets on the surface of VEGFR3 + LSECs (Figures 3a and b). Proliferation of hepatocytes and liver mass restoration after PH were diminished in Thpo − / − mice relative to WT group (Figures 3c and d). These data implicate that platelets produce VEGF-A and SDF-1 to prime LSECs, eliciting liver regeneration.
SDF-1 confers its pro-angiogenic activity 65 through the activation of two receptors, CXCR4 and CXCR7. CXCR7 expression is mainly enriched in ECs and subsets of lymphocytes, 55,57,66,67 and CXCR7 activation in EC is essential for the production of proregenerative angiocrine factor in organ repair. 35,39 Therefore, we examined the expression pattern and functional attributes of ECspecific SDF-1 receptor, CXCR7, after PH (Figure 4a). Immunostaining shows that the expression of CXCR7 was upregulated in LSEC 2 days post PH, compared to sham-operated mice (Figures 4a and b). Therefore, CXCR7 might serve as an inducible LSEC-specific SDF-1 receptor after acute liver injury.
To test functional contribution of CXCR7 in liver regeneration, we selectively deleted Cxcr7 in ECs of adult mice using an inducible tamoxifen-responsive Cre ERT2 that is specifically expressed in ECs 58 (Figure 4c). Mice expressing floxed Cxcr7 were bred with mouse line carrying EC-specific VE-Cadherin-Cre ERT2 / Cdh5(PAC)Cre ERT2 . 58 I.p. injection of tamoxifen to resulting offsprings induced 96% of Cxcr7 deletion in ECs of adult mice (Figure 4d). These mice lacking Cxcr7 in ECs (Cxcr7 iΔEC/iΔEC ) were subjected to PH, and liver regeneration was compared with control mice harboring endothelial haplodeficiency of Cxcr7 (Cxcr7 iΔEC/+ ). Hepatocyte proliferation and liver mass regeneration were markedly reduced in Cxcr7 iΔEC/iΔEC mice, as compared to those of control mice (Figures 4e-g). Thus, endothelial CXCR7 might be essential for promoting liver regeneration after PH.
After PH, LSECs produce hepatic-active paracrine/angiocrine growth factors such as HGF and Wnt2. 15 This angiocrine function of LSECs in liver regeneration depends on the activation of transcription factor inhibitor of DNA binding 1 (Id1) in LSEC. 15,37 We then assessed the effect of SDF-1 on cultured human LSECs. SDF-1 induced both upregulation and nuclear enrichment of Id1 protein in LSEC, which was abrogated by genetic silencing of Cxcr7 (Figures 5a and b). In addition, SDF-1-dependent Id1 upregulation in LSEC was recapitulated by Akt overexpression and suppressed by Wortmannin, an inhibitor of PI3 kinase-Akt pathway (Figure 5b). These data imply that SDF-1 stimulates CXCR7 in LSEC to trigger Akt-dependent activation of Id1 angiocrine pathway.
Platelets drive angiocrine-mediated liver regeneration K Shido et al crossed with Rosa-Cre ERT2 mice to generate Cre ERT2+ Cxcr7 loxP/loxP mice. Injection of tamoxifen induced deletion of Cxcr7 in adult mice (Cxcr7 iΔ/iΔ ; Figure 6b), and liver regeneration was compared between WT and Cxcr7 iΔ/iΔ mice after CCl 4 injection. There were significantly lower extent of hepatocyte proliferation (BrdU incorporation) and higher degree of hepatic injury in Cxcr7 iΔ/iΔ mice than those of WT mice (Figures 6c and d). Activation of Id1-HGF/Wnt2 angiocrine pathway was markedly prohibited in  Cxcr7 iΔ/iΔ mice, as compared to WT mice (Figure 6e). These results suggest that after liver injury, platelets supply SDF-1 to activate CXCR7 + LSECs, inducing the production of pro-regenerative angiocrine Wnt2 and HGF for hepatic repair (Figure 6f). In the CCl 4 -injured liver, platelets carry both SDF-1 and VEGF-A. Hence, we investigated the contribution of VEGF-A receptors in mediating liver repair. VEGFR1 activation mediates early protection after CCl 4 liver injury. 16,68 Thus, we tested the contribution of VEGFR1-expressing cells in the injured liver. 54 CCl 4 injection caused significant recruitment of VEGFR1 + CD11b + myeloid cells to the liver, adhering to VEGFR3 + LSECs (Figure 7a). To determine the contribution of VEGFR1 + myeloid cells to liver repair, Vegfr1 loxP/loxP mice were crossed with LysM-driven Cre to generate mice lacking Vegfr1 specifically in myeloid cells (Vegfr1 lysM/lysM ) (Figures 7b and c). Liver damage was substantially increased in Vegfr1 lysM/lysM , as evidenced by elevated plasma alanine aminotransferase activity (Figure 7d), and repeated injection of recombinant TPO in Vegfr1 lysM/lysM mice mitigated the injury. Of note, hepatogenic angiocrine Id1 pathway was suppressed in Vegfr1 lysM/lysM mice compared to control group, which was restored by TPO injection (Figure 7e). Thus, liver injury recruits platelets and VEGFR1 + myeloid cells to jointly activate LSEC niche, driving liver repair (Figure 7f).

DISCUSSION
LSECs lining hepatic sinusoids actively participate in liver repair and regeneration. Activation of transcription factor Id1 in LSECs leads to the elaboration of hepatic-active angiocrine factors. 15 In the mouse liver, we have identified a preferential distribution of Figure 6. After CCl 4 -induced liver injury, CXCR7 stimulates angiocrine-mediated hepatic regeneration. (a, b) Mice harboring loxP site-flanked Cxcr7 were crossed with Rosa-Cre ERT2 and treated six times with tamoxifen injection (250 mg kg − 1 ) to induce the deletion of Cxcr7 in adult mice (Cxcr7 iΔ/iΔ ). 15 Transcriptional level of Cxcr7 in the liver is shown in b; N = 6-9 mice per group. (c, d) Inhibition of SDF-1 signaling in LSEC abolished liver regeneration after CCl 4 injury. In Cxcr7 iΔ/iΔ mice, the decrease in cell proliferation was determined by staining for BrdU incorporation; N = 6 mice per group. Scale bar, 50 μm. (e) Increased hepatic damage in Cxcr7 iΔ/iΔ mice after CCl 4 injury was associated with abrogated induction of Id1-Wnt2/hepatocyte growth factor (HGF) angiocrine pathway; N = 5 mice per group. (f) Platelet-dependent activation of LSECs leads to the generation of hepatogenic angiocrine factors. Upon hepatotoxic injury, activated platelets generate SDF-1 to turn on Id1 pathway in LSECs, resulting in angiocrine production of HGF and Wnt2 that initiates hepatocyte propagation and liver repair.
SDF-1 receptor CXCR7 on LSECs, which was upregulated by PH. We have also revealed the functional role of endothelial CXCR7 in generating hepatic-active factors in both PH and CCl 4 models. As such, both hepatotoxic injury and loss of liver mass stimulate CXCR7 activation in LSECs, eliciting hepatic regeneration, and repair.
Recruited platelets in the injured liver initiates LSEC angiocrine signaling to trigger hepatic reconstitution. Post injury, platelets serve as circulating sentinel cells to promote tissue repair. 26,[42][43][44][45][46] Here our study has implied a paradigm in which platelet-supplied SDF-1 activates CXCR7 on LSECs and initiates subsequent angiocrine signaling. The beneficial effect of platelets in liver repair is in agreement with both preclinical and clinical findings that favorable prognosis of hepatic function correlates with higher circulating platelet count. 29 Whether recombinant TPO has the similar protective effect in infectious liver injury remains to be investigated. 47 The cytoprotective effect of VEGFR1 + myeloid cells after acute injury was evidenced by increased injury in mice with myeloidspecific Vegfr1 knockout (Vegfr1 lysM/lysM ). The rescue effect of TPO injection might depend on both platelets and myeloid cells. The effect of VEGFR1 + myeloid cells on angiocrine function was evidenced by the diminished Id1 pathway in LSECs of Vegfr1 lysM/lysM mice. Conceivably, increasing platelet number by TPO injection enhances platelet-dependent activation of myeloid cells, reinforcing endothelial activation and vascular niche-mediated liver regeneration.
The finding that, after liver injury, administration of TPO, SDF-1 and VEGF-A enhanced liver repair has clinical relevance. First, it is conceivable that in thrombocytopenic patients, liver repair is impaired. Thus, cautiously increasing the number of circulating platelet might offer tissue protection. Alternatively, after acute injury, administration of SDF-1 and VEGF-A might augment hepatocyte proliferation by stimulating angiocrine factor generation in LSECs. Furthermore, transfusion/transplantation of properly (e) Pro-regenerative Id1 angiocrine pathway was suppressed in Vegfr1 lyzM/lyzM mice, which was elevated by thrombopoietin injection (+TPO). Compared to WT control mice, transcriptional level of Id1 was lower in Vegfr1 lyzM/lyzM mice after CCl 4 injury. *P o0.05, compared to Vegfr1 lyzM/lyzM group. N = 5-7 mice per group. (f) Schema depicting the contribution of hematopoietic-vascular niche for liver regeneration. Upon liver injury, activated platelets are recruited to the liver and produce SDF-1 to activate CXCR7 + LSECs, initiating endothelial paracrine/ angiocrine-mediated liver repair. Activation of VEGFR1 + myeloid cells by platelet VEGF-A further stimulate Id1-Wnt2/HGF angiocrine pathway in LSEC, reinforcing liver regeneration.
primed platelets or myeloid cells can possibly offer optimal cell therapy approaches.
Taken together, we demonstrate a pro-regenerative interplay between platelets, myeloid cells and LSECs in liver repair. Platelets play an instrumental role in priming both angiocrine function of LSECs and myeloid cells post injury. Thus, platelet activation enables a hematopoietic-vascular 33 niche that orchestrates liver regeneration. Identifying critical pathways establishing this hepatogenic hematopoietic-vascular niche might aid in devising regenerative therapy for hepatic diseases.