Hematopoietic chimerism in liver transplantation patients and hematopoietic stem/progenitor cells in adult human liver


  • Potential conflict of interest: Nothing to report.


Liver transplantation (LT) is a cure for many liver diseases. Blood chimerism of donor origin can develop after LT, which raises the possibility of the existence of hematopoietic stem/progenitor cells (HSPCs) in the liver. We characterized the blood chimerism in a large cohort of 249 LT patients and analyzed putative HSPCs in adult human livers. The overall incidence of chimerism was 6.43%, of which 11.11% was among short-term (1 day to 6 months) and 3.77% was among long-term (6 months to 8 years) LT patients. Hematopoietic LinCD34+CD38CD90+ populations have been demonstrated to generate long-term lymphomyeloid grafts in transplantations. In human adult livers, we detected LinCD34+CD38CD90+ populations accounting for 0.03% ± 0.017% of the total single liver cells and for 0.05% ± 0.012% of CD45+ liver cells. Both LinCD34+ and LinCD45+ liver cells, from extensively perfused human liver grafts, were capable of forming hematopoietic myeloid-lineage and erythroid-lineage methylcellulose colonies. More importantly, LinCD45+ or CD45+ liver cells could be engrafted into hematopoietic cells in an immunodeficient mouse model. These results are the first evidence of the presence of putative HSPC populations in the adult human liver, where the liver is a good ectopic niche. The discovery of the existence of HSPCs in the adult liver have implications for the understanding of extramarrow hematopoiesis, liver regeneration, mechanisms of tolerance in organ transplantation, and de novo cancer recurrence in LT patients. Conclusion: The human adult liver contains a small population of HSPCs. In LT patients, there are two types of chimerisms: transient chimerism, resulting from mature leucocytes, and long-term chimerism, derived from putative HSPCs in the liver graft. (HEPATOLOGY 2012)

Liver transplantation (LT) has saved the lives of many patients with end-stage liver diseases and has become a curative procedure for certain liver diseases.1 Similar to other types of organ transplantation, chimerism can develop after LT with the chimeric cells either circulating or integrated into the parenchyma.2 Several types of reciprocal chimerisms after LT have been reported, including (1) recipient-derived cells in the donor organ3, 4; (2) hematopoietic chimerism of donor origin in the recipient blood5-7; and (3) donor origin cells in the skin and lymph nodes.8 Donor lymphocyte chimerism is common after LT, but it usually decreases and often disappears within 3 weeks.6 However, it has been shown that blood chimerism can last for years.8 Complete donor hematopoietic chimerism, in which the whole lineage of blood cells are of donor origin, has been detected in a LT recipient 3 years after LT.7

Clinically, the effect of chimerism in the recipients of solid-organ transplants is uncertain. Some researchers consider that developing a hematopoietic chimerism could be a desirable situation after LT, because the chimerism is often associated with allograft tolerance and therefore immunosuppression-related side effects could be reduced by obviating the need for immunosuppression therapy.9 Thus, attempts have been made to enhance chimerism by the intravenous infusion of donor bone marrow (BM) cells on the same day as transplantation, which have shown significant augmentation of chimerism and graft survival.10 However, other observations have implied that chimerism is not necessarily associated with allograft tolerance.11

During embryonic development, hematopoiesis occurs in the fetal liver before transition to the BM in adult life.12 Research has also indicated that the adult liver remains a compatible environment for hematopoiesis. However, it remains uncertain whether hematopoietic stem cells (HSCs) or hematopoietic progenitor cells (HPCs) are present in the adult liver. Even if HSCs are present in the adult liver, the origin of these cells is still unknown, because they could be mobilized from the BM. There have been reports of the isolation of HSCs from mouse adult livers,13, 14 but the cell-surface markers used for purification are not consistent and are even contradictory. One study identified a c-kit+ Sca-1+ Lin lo/ population, representing HSCs in mouse adult livers,13 whereas another study found that the purified CD45+ side population is more phenotypically similar to HSCs from adult mouse BM, but this population is c-kit negative.14 Thus, the markers used to isolate putative HSCs from mouse adult liver are not consistent. Moreover, to date, there has been no report on the identification of HSCs in human adult livers.

In the present study, we investigated the incidence of blood cell chimerism of donor origin in 249 LT survival patients; the shortest time after LT was 1 day, and the longest time after LT was 8 years. We also analyzed the putative hematopoietic stem/progenitor cells (HSPCs) in adult human livers. The overall incidence of blood chimerism was 6.43%. The incidence was 11.11% among patients tested shortly after LT (1 day to <6 months), whereas the incidence was 3.77% in long-term LT survival patients (6 months to 8 years). In human adult livers, we detected a LinCD34+CD38CD90+ population representing 0.03% ± 0.017% of the total single liver cells and 0.05% ± 0.012% of CD45+ liver cells. Both LinCD34+ and LinCD45+ liver cells were capable of forming myeloid-lineage and erythroid-lineage methylcellulose colonies; more importantly, LinCD45+ or CD45+ liver cells could be engrafted into hematopoietic cells in immunodeficient mice. Thus, we provide the first evidence of a putative HSPC population in the adult human liver, with the liver acting as a good ectopic niche.


APC, allophycocyanin; BM, bone marrow; CFU, colony-forming unit; Cy7, cyanin-7; DMEM, Dulbecco's modified Eagle's medium; FACS, fluorescence-activated cell sorting; FITC, fluorescein isothiocyanate; gDNA, genomic DNA; HCC, hepatocellular carcinoma; HPCs, hematopoietic progenitor cells; HSCs, hematopoietic stem cells; HSPCs, hematopoietic stem/progenitor cells; LT, liver transplantation; NOD-SCID, nonobese diabetic/severe combined immunodeficiency; PCR, polymerase chain reaction; PE, phycoerythrin; SD, standard deviation; STR, short tandem repeat.

Materials and Methods

Human Specimens.

This was a retrospective study of 249 LT patients who received orthotopic LT at Queen Mary Hospital (Pok Fu Lam, Hong Kong) between 2000 and 2011. Peripheral blood was collected from recipients at various times after LT and from matched donors. Patients who received liver allografts from close relatives were excluded. For liver specimens, before transplantation, a small wedge of liver tissue from human cadaveric or living donor graft was collected after extensive perfusion with the University of Wisconsin solution for cadaveric donor grafts and histidine/tryptophan/ketoglutarate solution for live donor grafts to remove peripheral blood. The processed tissues were then kept in Dulbecco's modified Eagle's medium (DMEM) medium at 4°C until further study. The study was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority of Hong Kong.

DNA Extraction.

Genomic DNA (gDNA) was isolated from peripheral blood mononuclear cells using a DNA mini or midi kit (QIAGEN GmbH, Hilden, Germany). To avoid cross-mixing samples during the DNA-extraction procedure, recipient and donor DNA samples were extracted by different groups of researchers.

Polymorphic Microsatellite Marker Polymerase Chain Reaction and Analysis.

Short tandem repeat (STR) DNA loci were amplified with an AmpFlSTR Profiler PCR Kit, following the manufacturer's instructions, which coamplifies nine STR loci and the gene for sex identification (Applied Biosystems, Foster City, CA). Briefly, 1.5-2.5 ng of gDNA was used for polymerase chain reaction (PCR), and paired PCR products of the recipients and donors were then run on an ABI Prism 310 Genetic Analyzer on the same day (Applied Biosystems). For the putative positive samples, the PCR was repeated two times, independently. The STR genotype and the precision of the data between LT recipients and donors were determined, following instructions in the User's Manual,15 including the exclusion of stutters and other extrapeak confusion.

Liver Cell Preparation.

Liver tissue specimens suspended in DMEM were minced and digested with collagenase V (100 units/mL; Sigma-Aldrich, St. Louis, MO) at 37°C for 15 minutes, followed by filtering through a 40-μm nylon mesh to remove debris. Cells were then collected by centrifugation (440×g for 20 minutes) at 4°C and either resuspended in fluorescence-activated cell-sorting (FACS) buffer (phosphate-buffered saline, 0.5% bovine serum albumin, and 0.1% NaN3) for cell-surface marker study or resuspended in R medium (Dulbecco's phosphate-buffered saline, 2% fetal bovine serum, and 1 mM of ethylene diamine tetraacetic acid) for cell sorting.

Flow Cytometry.

To determine the LinCD34+CD38CD90+ population in liver tissue, 5-10 × 105 single liver cells from the unsorted sample, or from a sample previously sorted for CD45+, were incubated with antihuman lineage cocktail 1/fluorescein isothiocyanate (FITC) (BD Immunocytometry Systems, San Jose, CA), anti-CD34-APC (allophycocyanin) anti-CD90-PE (phycoerythrin) (BD Pharmingen, San Diego, CA), and anti-CD38-PE/Cy7 (cyanin-7) (BioLegend, San Diego, CA) antibodies for 30 minutes at room temperature, followed by two washes with FACS buffer. As a control, cells were also labeled with FITC, APC, PE, and PE/Cy7 isotype control antibodies (BioLegend). Cell-surface markers were then analyzed by FACSCalibur (BD Immunocytometry Systems).

Cell Sorting.

LinCD34+ cells were isolated by magnetic cell sorting. Lin liver cells were obtained by incubation with negative selection (two times) of human progenitor enrichment cocktail antibodies (anti-CD2, anti-CD3, anti-CD11b, anti-CD14, anti-CD16, anti-CD19, anti-CD24, anti-CD56, anti-CD66b, and anti–glycophorin A; StemCell Technologies, Vancouver, Canada), followed by magnetic separation. Sorted Lin cells were then enriched for either CD34+ or CD45+ cells two times using a human CD34 or CD45 selection kit (StemCell Technologies).

Methycellulose Culture for Hematopoietic Colony Forming Units.

Cells (2,000-5,000) from Lin-depleted and CD34- or CD45-enriched cell populations from liver cell suspensions were seeded in complete methylcellulose (MethoCult GF+ H4435 or GF H4034; StemCell Technologies) in pretreated 35-mm dishes, following the manufacturer's instructions, and incubated at 37°C, 5% CO2, and 95% humidity for 14-16 days. Hematopoietic colonies were then scored based on size, color, and morphology, with each colony containing at least 40 cells.


Nonobese diabetic/severe combined immunodeficiency (NOD-SCID) mice (NOD.CB17-Prkdcscid/J; The Jackson Laboratory, Bar Harbor, ME), at 6-10 weeks of age, were irradiated with 2-3 Gy (Cs137, MDS Gammacell; MDS Nordion, Freiburg, Germany) 4 hours before transplantation. Magnetic sorted LinCD45+ or CD45+ liver cells (2 × 105 to 2 × 106) (for the case with low cell count only, CD45+ cells were selected) in 20-49 μL were transplanted by femur injection using a 31-gauge insulin syringe. Transplanted mice were kept in individual ventilation cages and supplemented with 0.001% enrofloxacin (Bayer HealthCare, Berlin, Germany) in sterile drinking water. Engraftment was evaluated at 6-9 weeks by determining the presence of human CD45+ populations in mouse blood and BM. The percentage of human CD45+ cells was calculated as the proportion of labeled human CD45+ over isotype antibody control. For multilineage engraftment, human CD45+CD33+ and CD45+CD71+ cells were measured in mouse BM; human CD45+CD19+ and CD45+CD4+ cells were measured in mouse peripheral blood by flow cytometry. Anti-CD45-FITC and anti-CD4-APC antibodies were from BD Pharmingen; anti-CD19-PE, anti-CD33-APC, and anti-CD71-APC antibodies were from BioLegend.

Statistical Analysis.

Data are presented as the percentage and the mean ± standard deviation (SD). Fisher's exact test and the Student t test were performed using SPSS software (v. 16.0; SPSS, Inc., Chicago, IL). P < 0.05 was regarded as statistically significant.


Hematopoietic Chimerism Development in LT Patients.

Although blood chimerism development is not uncommon in LT patients, the related reports have presented only a single or a few cases. There has not been any study on hematopoietic chimerism in a large cohort and in long-term LT patients. We investigated hematopoietic chimerism of donor origin in 249 LT survival patients; the shortest time after LT was 1 day, and the longest time after LT was 8 years. The overall incidence of blood chimerism was 6.43% (16 of 249; Table 1). The incidence of chimerism was 11.11% (10 of 90) among patients evaluated a short time after LT (1 day to <6 months), whereas the incidence was 3.77% (6 of 159) among long-term LT survival patients (6 months to 8 years; Table 2). There were 6 patients with chimerism lasting more than 7 months, with the longest lasting 4.0-4.5 years (case 351; Fig. 1A). Thus, the short time after LT group had a significantly higher blood chimerism (P = 0.03; Table 2). Blood chimerism of donor origin could result from resident leukocytes/lymphocytes in the liver graft6, 16, 17; it could also result from HSPCs present in the liver. If the chimerism results from donor HSPCs, then the type of donor liver, the sex of the donor, and the age of the donor may have an effect on the development of blood chimerism. We found that there were no statistically significant associations between donor liver type (i.e., cadaveric and living), donor sex (male and female), or donor age (<50 and ≥50 years old), and chimerism formation (Table 2). Thus, liver graft type and sex and age of the donor had no significant effects on the development of chimerism.

Figure 1.

Identification of blood chimerism of donor origin in LT patients by comparing STR loci between recipient and donor. Extracted blood-cell DNA from LT patients and liver graft donors were analyzed by PCR for 9 STR loci. GeneScan electropherogram (Applied Biosystems, Foster City, CA) of the informative alleles of LT recipient and donor are shown in parallel. (A) Allele THO1-08, highlighted with a red asterisk in LT patient 351 (upper column), was from the donor (lower column). (B) Allele D3S1358-16, labeled with a red asterisk in LT patient 823 (upper column), was derived from the donor (lower column). (C) Two alleles of CSF1PO-10 and CSF1PO-12, labeled with red asterisks in LT patient 887 (upper column), were derived from the donor (lower column).

Table 1. Incidence of Hematopoietic Chimerism in LT Patients
LT PatientsGender (M/F)AgeHematopoietic Chimerism (Donor Origin) (%)
  1. Abbreviations: M, male; F, female.

n = 249205/4419-726.43 (16/249)
Table 2. Comparison of Hematopoietic Chimerism
Comparison Blood Chimerism (%)P Value
  • Abbreviations: d, day; m, month; yr, year.

  • *

    P < 0.05; statistical significance.

LT time1 d to less than 6 m post-LT11.11 (10/90)0.03*
6 m to 8 yr post-LT3.77 (6/159)
Donor typeCadaveric5.59 (9/161)0.33
Living7.95 (7/88)
Donor genderMale8.33 (9/108)0.23
Female4.96 (7/141)
Donor ageYounger than 50 yr8.13 (13/160)0.13
Older than 50 yr3.37 (3/89)

Interestingly, chimerism-positive cases were 7.57% (14 of 185) in non–hepatocellular carcinoma (non-HCC) LT patients. These non-HCC LT patients included those with cirrhosis or cirrhosis with acute complication, chronic or acute hepatitis; and congenital or heritable diseases. By comparison, there were 3.13% (2 of 64) positive cases in HCC patients. This distribution suggests that blood chimerism can develop in patients with all types of liver disease that require LT.

Kinetics and Characteristics of Hematopoietic Chimerism in LT Patients.

Given that patients evaluated a shorter time after LT had a higher incidence of chimerism than those patients evaluated a longer time after LT, the observed blood chimerism may be derived from residual lymphocytes in the liver graft. We therefore assessed blood chimerism over time after LT. LT patients 723, 739, and 860 displayed STR loci of donor origin in the blood on day 2 after LT, but these loci disappeared 1 week or longer after LT (Table 3). One female LT recipient (case 823) was positive for the amelogenin Y locus (from a male donor) on 1 day after LT; the presence of this locus became undetectable 1 month after LT, although another locus persisted 3 months after LT (Fig. 1B; Table 3). For case 887, although STR could not be measured shortly after LT, 3 loci of donor origin were detectable 7 months after LT (Fig. 1C; Table 3). These were unlikely to be derived from residual leucocytes/lymphocytes from the donor liver graft. The data suggest that there could be two types of blood cells present in liver grafts: residual mature leucocytes/lymphocytes responsible for short-term chimerism and putative HSPCs resulting in long-term chimerism of donor origin. These two types of chimerism might occur simultaneously, as demonstrated by the fact that partial chimerism patients showed multiple loci of donor origin shortly after LT, but were positive for only a single locus of donor origin at later time points after LT (Table 3).

Table 3. Chimerism Kinetics After LT
CaseSex/AgeLiver Graft TypeTime Post-LTNo. of informative Alleles
  1. Abbreviations: M, male; F, female; NT, not tested.

723M/45Cadaveric2 days4
7 days0
739M/23Cadaveric1 day1
8 days0
780M/46Cadaveric2 days1
16 days1
823F/48Cadaveric1 day2
1 month1
3 months1
832F/45Cadaveric1 day5
2 months1
860F/63Living1 day1
1 year0
887M/50Living1 dayNT
7 months3

HSPC LinCD34+CD38CD90+ Population in Liver Grafts.

The blood chimerism phenomenon raises the question of whether HSPCs exist in the adult liver or that residual leukocytes/lymphocytes in liver grafts could be the source of the chimerism. Attempts have been made to isolate hematopoietic stem cells from mouse adult livers using disparate panels of different cell-surface markers.13, 14 There has not been any report regarding HSPCs in human adult livers. A LinCD34+CD38CD90+ population purified from human umbilical cord blood has been demonstrated to have the ability to give rise to long-term multipotent grafts in serial transplantations.18, 19 We therefore attempted to determine whether LinCD34+CD38CD90+ HSCs were present in the human adult liver. Single-cell suspensions isolated from healthy donor livers were analyzed using either the total cell population (n = 9) or cells sorted for CD45+ (n = 7). Average sizes of the LinCD34+CD38CD90+ populations were 0.03% ± 0.017% in total liver cells and 0.05% ± 0.012% in CD45+ liver cells (Fig. 2A). The LinCD34+CD38CD90+ population was significantly higher in CD45+ liver cells than in total liver cells (Fig. 2A; P = 0.043), indicating that CD45+ selection enriched for potential HSPCs. Representative flow-cytometry results of the population are shown in Fig. 2B,C. These results suggest the presence of a LinCD34+CD38CD90+ HSPC population in human adult livers. It is important to point out that the LinCD34+CD38CD90+ population is limited to its ability to generate lymphomyeloid engraftment with no T-cell engraftment;18 therefore, it does not represent multipotent HSCs, which can give rise to most (if not all) of the differentiated cell populations in the blood.

Figure 2.

Flow-cytometry analysis of LinCD34+CD38CD90+ populations in adult liver graft. Single cells were isolated from donor liver grafts after extensive perfusion. The liver cells in total, or in magnet sorted for CD45+ populations, were analyzed for expression of lineage marker, CD34, CD38, and CD90 by flow cytometry. (A) Average percentages of LinCD34+CD38CD90+ cells from total liver cells (n = 9) and from CD45+ liver cells (n = 7). (B) FACS plots of the LinCD34+CD38CD90+ population in liver cells. Cells were gated on Lin (left panel), followed by gating on LinCD34+CD38 (middle panel), and then LinCD34+CD38CD90+ (right panel). Data shown are representative analyses of total liver cells. (C) Data shown are representative analyses of CD45+ liver cells.

HSPCs Methylcellulose Colony Formation From Human Adult Liver Cells.

We further determined the presence of HSPCs in human adult livers by a methylcellulose-based colony-forming unit (CFU) assay. Because of the limited availability of healthy liver grafts, both in terms of number and size, we performed the CFU assay using magnet-bead isolated LinCD34+ or LinCD45+ liver cells, instead of FACS-sorted LinCD34+CD38CD90+ cells. The exclusion of Lin cells excluded most mature cells with lineage markers; CD34 is a marker of hematopoietic progenitor cells,20, 21 and CD45 is expressed on all nucleated hematopoietic cells.22 Liver grafts were subjected to either standard extensive perfusion or without perfusion. The overall CFU colony formation was 0.2% ± 0.15% in LinCD34+ or LinCD45+ liver cells from perfused liver grafts (n = 6) and 0.08% ± 0.06% in LinCD34+ or LinCD45+ liver cells from nonperfused liver grafts (n = 12) (Fig. 3A; P = 0.096). Both LinCD34+ (n = 11) and LinCD45+ (n = 9) liver cells were equally capable of forming CFUs (Fig. 3B; P = 0.224). Given that HSCs are known to circulate,23 it is possible that the CFUs from liver grafts preceding perfusion could be derived from HSCs in the blood. However, CFUs indeed formed in 6 of 6 liver grafts that went through extensive perfusion, thus demonstrating that it was likely they were generated from HSPCs preexisting in the liver graft and not from blood cells (Fig. 3A, column 1). There were 4 liver samples (18%; 4 of 22) that did not result in any colony growth in methylcellulose culture.

Figure 3.

Hematopoietic colony formation of LinCD34+ and LinCD45+ liver cells. For single liver cell isolation, a wedge of each liver specimen was collected after standard extensive perfusion or unperfused. Isolated single liver cells were magnetically sorted into LinCD34+ or LinCD45+ populations and plated into complete methylcellulose medium and incubated for 12-14 days. (A) Average percentage (mean ± SD) of CFU formation from single liver cells obtained from perfused liver grafts (n = 6, column 1) or from liver grafts without perfusion (n = 12, column 2). The percentage of CFUs was calculated by number of CFUs formed versus number of cells seeded. An independent t test was used for statistical comparison. (B) Statistical comparison of CFU formation (mean ± SD) between LinCD34+ (n = 11) and LinCD45+ (n = 9) liver cells from all liver grafts, regardless of perfusion status. (C) CFUs were scored based on morphology. The percentage of each type of CFU of the total pool of CFUs from LinCD34+ liver cells is indicated. (D) The percentage of each type of CFU of total CFUs from LinCD45+ is indicated. (E) Representative colony pictures of CFU-E, BFU-E, CFU-G, CFU-M, and CFU-GM grown from either LinCD34+ or LinCD45+ liver cells. Original magnification: ×200 for CFU-E; ×40 for CFU-G and CFU-GM; and ×100 for BFU-E and CFU-M. (F) Wright-Giemsa–stained cytospin preparations of cells from dissociated CFU colonies (i-iii; magnification ×1,000); (i) erythroblast; (ii) myeloblast; and (iii) maturing myeloid cells.

The pool of all colonies formed was classified into different lineages according to colony size, color, and morphology. Colonies formed by both LinCD34+ and LinCD45+ liver cells consisted of all lineages of hematopoietic cells: CFU-E; BFU-E; CFU-G; CFU-M; and CFU-GM (Fig. 3C,D). A high proportion of colonies appeared to be CFU-E, representing more mature erythroid progenitors (Fig. 3C-E). There were still BFU-E colonies, representing primitive erythroid progenitors (Fig. 3C-E). CFU-G, CFU-M, and CFU-GM were formed, although the total number of these types of colonies was not high (Fig. 3C-E). We detected only one CFU-GEMM colony (from perfused liver graft) in all experiments, which are derived from multilineage progenitor cells. Representative CFU types are shown in Fig. 3E. Dissociated single cells from colonies were stained with Wright-Giemsa, and both mature and progenitor hematopoietic cells were observed (Fig. 3F). All of these results provide evidence that HSPCs were present in the adult human liver. It needs to be noted that the presence of CFUs does not distinguish multipotent HSCs versus HPCs. This is because, from the CFU-distribution pattern (Fig. 3C,D), they more closely fit the profile of derivation from HPCs (CFU-E, BFU-E, CFU-G, CFU-M, and CFU-GM), rather than multipotential progenitor cells (CFU-GEMM). However, apart from long-term propagation and in vivo reconstitution experiments, there is no efficient culture assay to distinguish multipotent HSCs.

Engraftment of Putative HSPCs From Adult Human Liver in NOD-SCID Mouse.

To confirm that LinCD34+ or LinCD45+ liver cells contained putative HSPCs, LinCD45+ or CD45+ liver cells (2 × 105 to 2 × 106) sorted from extensively perfused liver grafts were transplanted into ionizing radiation–treated NOD-SCID mice to evaluate engraftment. Six to nine weeks after transplantation, human CD45+ hematopoietic cells were observed in the peripheral blood (Fig. 4A) and BM (Fig. 4B) of immunodeficient mice. Overall engraftment rate was 88.9% (8 of 9 transplantations), although the repopulation number was not high (0.25% ± 0.25% in blood and 0.3% ± 0.12% in BM). Furthermore, for multilineage engraftment, human CD33, CD71, CD19, and CD4 markers were measured from human CD45+ cells in BM and blood cells of engraftment mice. We found 0.04% of hCD45+CD33+ myeloid progenitor cells and 0.06% of hCD45+CD71+ erythroid precursor cells in the BM and 0.09% of hCD45+CD19+ B-lymphoid cells and 0.32% of hCD45+CD4+ T-lymphoid cells in the peripheral blood (Fig. 4C). Thus, multilineage profiles of myeloid, erythroid, and both B- and T-lymphoid cells could be detected in engrafted mice. We noticed that relatively large numbers of cells had to be used for transplantation, and that engraftment capacity is low (Fig. 4).

Figure 4.

Hematopoietic repopulation of human liver LinCD45+ or CD45+ cells in NOD-SCID mice. Magnetically sorted LinCD45+ or CD45+ cells (2 × 105 to 2 × 106) from extensively perfused human liver grafts were transplanted into NOD-SCID mice that had been subjected to 2.7 Gy of irradiation 2 hours before transplantation. Nine weeks after transplantation, the human CD45+ population was determined in the peripheral blood and BM of mice. (A) Mouse peripheral blood flow-cytometry analysis of human CD45+ cells (right panel), compared to isotype immunoglobulin G (IgG) control (left panel). Dot plots are representative of the gated positive population (CD45-PE) in the FSC/SSC setting, where the negative population is excluded. Percentages of dot plots were calculated as positive cells versus negative cells. (B) BM flow-cytometry analysis of human CD45+ cells (right panel), compared to isotype controls (left panel), from the same mouse as (A). (C) Multilineage engraftment by flow-cytometry analysis is indicated by the presence of hCD45+CD33+ myeloid progenitor cells and hCD45+CD71+ erythroid precursor cells in the BM and hCD45+CD19+ B-lymphoid and hCD45+CD4+ T-lymphoid cells in the peripheral blood. Cells were gated on CD45+ (FITC), followed by gating on CD33+ (APC), CD71+ (APC), CD19+ (PE), and CD4+ (APC), respectively, where the negative population is excluded. Percentages of dot plots are presented as double-positive populations in all cells tested after subtraction of isotype controls.


Chimerism Was Derived From Both Mature Leukocytes and Putative HSPCs.

Blood chimerism of donor origin in LT patients has been considered to be donor leukocyte or lymphocyte chimerism, because a substantial number of residual leukocytes and lymphocytes are observed in donor liver grafts after extensive perfusion.6, 16 Of our large cohort study of 249 LT patients from 1 day to 8 years after LT, 16 patients with detectable donor STR loci were identified, of which 6 cases were long-term LT survival patients (7 months to 4.5 years). The results suggest that there must be two types of hematopoietic cells in donor liver grafts capable of causing blood chimerism: (1) residual leukocytes and lymphocytes, which contribute to transient chimerism that usually disappears within 3 weeks after LT (Table 3), and (2) HSPCs, which can self-renew and differentiate, contributing to long-term chimerism (Table 3; Figs. 3 and 4). The present study revealed that the overall incidence of chimerism was not high in LT patients. Based on the results of ours and others' studies, we hypothesize that the low chimerism could result from the following: (1) mature leukocytes/lymphocytes derived chimerism would disappear in 3 weeks6 (Table 3) and (2) putative HSPCs, which represent a very small population in the liver graft (Figs. 2 and 3), leading to a lower degree of chimerism in long-term LT patients.

Putative HSPCs in Human Adult Liver.

Long-term donor-origin blood chimerism must be derived from HSPCs in liver grafts. There has been no report on the identification of HSPCs in human adult livers. Recently, it has been demonstrated that a multipotent progenitor population with the LinCD34+CD38CD90CD45RA immunophenotype19 and an HSC population with the LinCD34+CD38CD90+ immunophenotype,18 both from human umbilical cord blood, are able to give rise to long-term lymphomyeloid grafts in immunodeficient mice.18, 19 Therefore, we analyzed LinCD34+CD38CD90+ cells in human adult livers and determined that this population represented 0.03%-0.05% of isolated single liver cells or CD45+ liver cells (Fig. 2). It is important to point out that the LinCD34+CD38CD90+ population is limited to its ability to generate lymphomyeloid engraftment with no T-cell engrafment;18 therefore, it does not represent multipotent HSCs. Because of the limitation of the size of adult donor livers (typically 2 × 106 total cells were isolated), it was not possible to purify LinCD34+CD38CD90+ cells by FACS for biological study. Instead, we determined the methycellulose colony-forming ability of magnet-sorted LinCD34+ or LinCD45+ liver cell populations. Indeed, 82% (18 of 22) of donor liver cell samples sorted into LinCD34+ and LinCD45+ populations were able to form colonies, including myeloid-lineage colonies (CFU-GM, CFU-G, and CFU-GM; Fig. 3) and erythroid-lineage colonies (BFU-E and CFU-E; Fig. 3). More convincingly, LinCD45+ or CD45+ liver cells from perfused liver graft were able to repopulate in NOD-SCID mice by detection of human CD45+ cells in the BM and blood of mice (Fig. 4A,B). These human CD45+ hematopoietic cells comprised ethrythoid and myeloid precursors and mature lymphocytes (Fig. 4C), although engraftment ability is low, ranging from 0.04% to 0.32% (Fig. 4C). Thus, by both hematopoietic methylcellulose colony formation and engraftment experiments, we are the first to convincingly demonstrate that HSPCs exist in human adult livers.

The Liver May Be a Good Ectopic Niche for HSCPs.

It is known that marrow HSPCs are able to mobilize to the peripheral blood in response to cytotoxic agents and cytokines23 and can home directly to inflammation sites.23, 24 More interestingly, HSPCs can enter into the circulation, even in a steady state.23, 25 Here, we demonstrate the existence of HSPCs in human adult livers, although the capacity of CFU formation and hematopoietic-repopulating potential of liver HSPCs is relatively low. The important question is whether this very small population of HSPCs was mobilized from the BM during the transplantation process or if it persistently existed in the adult liver. We hypothesize that these HSPCs were continuously present in adult livers because of the following: (1) the population was maintained in the liver graft after extensive perfusion, suggesting that the cells were not likely to have been mobilized, but, instead, were already present in the liver; (2) the liver is a hematopoietic organ in the fetal stage; therefore, it may well be a good ectopic niche for HSPCs in adults; and (3) it has been reported that stem-cell-factor–producing mesenchymal cells, essential for the generation of hematopoietic cells, are located in the sinusoids of BM.26, 27 Liver is a sinusoid-enriched organ and thus may contain niche cells capable of sustaining HSCs. Still, in this study, the formal possibility cannot be excluded that these cells were blood HSPCs adherent to the endovascular compartment of the liver, which could not be perfused out. Moreover, after LT, either donor HSPCs generate mature HSCs inside grafted liver or circulate to recipient BM for hematopoiesis. These possibilities remain to be determined in future studies.


The authors thank the Liver Transplantation Center at Queen Mary hospital of the University of Hong Kong for outstanding clinical liver transplantation care. The authors also thank Ms. Kammy Yik, Banny Lam, and Waiyee Ho for data organization of LT donors and recipients. The authors also thank Dr. Mo Yang at the Department of Pediatrics and Adolescent Medicine of the University of Hong Kong for his useful help on the experiment. The authors also thank Ms Amy Lam and Mr. Jimmy Chen of Applied Biosystems for their technical support.