Cord blood comprises antigen-experienced T cells specific for maternal minor histocompatibility antigen HA-1

In the last decade, umbilical cord blood transplantation (CBT) has been available as an alternative to human leukocyte antigen (HLA)-matched sibling or unrelated donor stem cell transplantation (SCT) for the treatment of hematologic malignancies.^1-7  The clinical outcome of transplanted CB grafts with 1 or 2 HLA antigen mismatches demonstrates a similar risk of developing graft-versushost disease (GvHD) as compared to HLA-matched unrelated SCT.^5,6  A significant lower incidence of acute and chronic GvHD was reported after HLA-identical CBT when compared to sibling SCT.⁴  Collectively, these clinical results point to a decreased incidence of GvHD after CBT.

GvHD is often associated with a curative graft-versus-leukemia (GvL) response. Minor H antigen disparities between donor and recipient play important roles in both the GvH and GvL reactivity after HLA-matched SCT as reviewed.⁸  One of the well-described minor H antigens is HA-1. The immunodominant minor H antigen HA-1 is encoded by a diallelic gene with a single amino acid polymorphism.⁹  The HA-1 “positive” allele (HA-1^pos) contains a histidine at position 3 (HA-1^H), whereas the HA-1 “negative” allele (HA-1^neg) contains an arginine (HA-1^R). The HA-1^H peptide is recognized by HLA-A2-restricted CD8^pos cytotoxic T cells from HA-1^neg donors.^10-12  The expression of HA-1 is restricted to the hematopoietic lineage and to epithelial carcinomas.^13,14  This restricted expression makes HA-1 an attractive target antigen for GvL and graft-versus-tumor responses.¹⁵ 

Despite the lower incidence of GvHD after CBT, there is no indication of increased leukemia relapse rates when compared with sibling or unrelated donor SCT.^3-5  Comparable survival rates point to an as-yet-unexplored GvL potential of cord blood. Relatively little is known about the development of antigen-specific T-cell responses around birth.¹⁶  Mature monocyte-derived neonatal dendritic cells (DCs) are able to efficiently prime antigen-specific cytotoxic T cells in vitro.¹⁷  In bulk cultures, cord blood T cells proliferate in response to alloantigen.¹⁸  Yet the development of functional alloreactive cytotoxic T cells is impaired.^18-20  Limiting dilution studies have, however, reported normal precursor frequencies of cytotoxic T cells specific for allo-HLA class I and class II in cord blood.^20-22  Thus, the capacity to develop allogeneic cytotoxic T cells is intact at birth, despite overall diminished magnitude of responses.²³ 

It is known that feto-maternal hemorrhage occurs during pregnancy. Fetal cells expressing paternal minor H antigens can prime maternal T cells.^12,24  Since cells of the mother also traffic into the fetus during pregnancy, we tested the hypothesis that maternal minor H antigens can prime fetal T cells. Fifteen HLA-A2^pos/HA-1^neg CB samples derived from HLA-A2^pos/HA-1^neg or HLA-A2^pos/HA-1^pos mothers were analyzed for their feasibility to generate HA-1-specific cytotoxic T cells ex vivo as well as for the presence of pre-existing HA-1-specific T cells.

After informed consent of the mother, cord blood was collected from the umbilical cord vein with the placenta still in utero. HLA-A2^pos/HA-1^neg CB samples derived from HLA-A2^pos/HA-1^neg and HLA-A2^pos/HA-1^pos mothers were selected after high-resolution HLA class I typing and HA-1 genomic typing as described previously.²⁵  Cord blood mononuclear cells (CB-MNCs) were isolated by Ficoll-Isopaque density gradient centrifugation and stored in liquid nitrogen. Approval for these studies was obtained from the Institute for Transplantation Diagnostics and Cell Therapeutics institutional review board.

Expression of the T-cell receptor specific for HLA-A2/HA-1^H peptide (VLHDDLLEA) complexes was analyzed by staining T cells with phycoerythrin (PE)-conjugated HLA-A2/HA-1 tetrameric complexes (HA-1^A2) in combination with allophycocyanin (APC)-conjugated anti-CD8 monoclonal antibody (BD Biosciences, Amsterdam, The Netherlands). Tetramers were generated as previously described.²⁶  Specificity analysis of the HA-1^A2 tetramer was performed in parallel experiments using HA-1-specific and HA-1-nonspecific cytotoxic T-lymphocyte (CTL) clones (data not shown).

CD34⁺ cells were isolated via positive selection using CD34 magnetic-activated cell sorting (MACS) beads (Miltenyi GmBH, Bergisch Gladbach, Germany). CD34⁺ cells were cultured in complete Iscove modified Dulbecco medium (IMDM) containing 10% pooled human serum (HS), 250 U/mL granulocyte macrophage colony-stimulating factor (GM-CSF) (Leucomax, Novartis, Arnhem, The Netherlands), 25 μg/mL stem cell factor (SCF) (Peprotech, London, United Kingdom), 100 U/mL tumor necrosis factor (TNF)-α (Peprotech), and 50 μg/mL FLt3 ligand (R&D Systems, Minneapolis, MN). Dendritic cells were transduced with retroviral supernatants containing HA-1^H-encoding cDNA, as previously described,²⁷  and further cultured in complete medium supplemented with 300 U/mL interleukin-4 (IL-4) (Peprotech). Maturation of HA-1^H-transduced dendritic cells was induced by co-culturing immature dendritic cells for 3 days with irradiated CD40 ligand-transfected fibroblasts.

A slightly modified protocol, originally designed for the induction of minor H antigen-specific cytotoxic T cells from adult peripheral blood mononuclear cells (PBMCs), was applied.¹¹  In short, CD8^pos T cells were isolated via positive selection using CD8 MACS beads (Miltenyi) and cultured with irradiated autologous HA-1^H-transduced dendritic cells at a CD8-to-dendritic cell ratio of 5:1 or 10:1 in IMDM supplemented with 10% HS, 0.5 U/mL IL-2 (Cetus, Emeryville, CA), and 1 U/mL IL-12 (R&D Systems). After 3 days, irradiated autologous T helper cells were added to the culture at a CD8-to-T helper ratio of 10:1. T helper cells were generated by stimulating CD34/CD8-depleted CB-MNCs with anti-CD3/CD28 beads according to supplier's protocol (Dynal Biotech, Smestad, Norway). This mode of stimulation results in at least 80% activated CD4^pos T cells that produce interferon-γ (IFN-γ), TNF-α, and IL-2 (data not shown). From day 7 onward, oligoclonal T-cell lines were restimulated weekly using irradiated dendritic cells and T helper cells at a CD8-to-T helper-to-dendritic cell ratio of 10:1:1. Fresh medium containing IL-2 was added every 3 to 4 days.

Bulk T-cell lines were tested for the presence of HA-1^A2 tetramer^pos CD8^pos T cells after 2 to 4 rounds of stimulation with T helper cells and dendritic cells. HA-1^A2 tetramer-staining cells were subsequently sorted on a FACS Vantage cell sorter (Becton Dickinson, San Jose, CA) and cloned by limiting dilution. Cloned T cells were expanded in the presence of 5 × 10⁴ irradiated allogeneic PBMCs and 5 × 10³ HLA-A2^pos/HA-1^pos Epstein-Barr virus (EBV)-transformed B cells-lymphoblastoid cell lines (EBV-LCL) and 30 U/mL IL-2.

The protocol for detection of circulating minor H antigen-specific cytotoxic T cells in PBMCs from healthy multiparous female blood donors was applied.¹²  In short, CB-MNCs were depleted for various cell subsets using CD4, CD14, CD19, CD16, and glycophorinA MACS beads (Miltenyi). The depleted fraction was subsequently stained with CD8-APC, CD45RA-FITC (BD Biosciences), and PE-conjugated HA-1^A2 tetrameric complexes. Further enrichment of HA-1^A2 tetramer^pos CD8^pos cells was performed on a FACS Vantage cell sorter using the “enrich mode,” whereby cells are sorted at 20 000 events per second. The enriched fraction was reanalyzed and resorted immediately at 10 000 events per second using the more stringent “normal-R” mode. Double FACS-sorted cells were expanded in the presence of irradiated autologous HA-1^neg CB-MNC cells, 1% phytohemag-glutinin, and 30 U/mL IL-2. Fresh IL-2-containing medium was added every 3 to 4 days.

Standard 4-hour ⁵¹Cr-release assays were performed as previously described.¹³ 

Six HLA-A2^pos CB samples, 4 HA-1^neg and 2 HA-1^pos, were used for the ex vivo generation of HA-1-specific cytotoxic T-cell lines. CD8^pos T cells were isolated and cultured with autologous HA-1^H-transduced dendritic cells and T helper cells. The generation of HA-1-specific cytotoxic T-cell lines was successful in 3 of 4 HLA-A2^pos/HA-1^neg CB samples, whereas no growth at all was observed in the 2 HLA-A2^pos/HA-1^pos CB samples. Results of 2 HLA-A2^pos/HA-1^neg CB samples (I and II) are shown in Figure 1. The percentages of HA-1^A2 tetramer and CD8 double-positive T cells are shown after 14 days of culture (Figure 1A-B). HA-1^A2 tetramer-staining CD8^pos T cells were further enriched after additional rounds of stimulation (Figure 1C-D). The latter populations were FACS sorted, expanded for 14 days, and functionally analyzed. Strong HA-1-specific lytic activity is demonstrated for both CB-derived T-cell cultures (Figure 1E-F).

Since cord blood may contain maternal cells, we determined whether the HA-1-specific T cells were child or mother derived. DNA typing of the HA-1 alleles showed unequivocally that ex vivo-generated HA-1-specific T cells are from child origin (data not shown).

T-cell clones were generated from CB I (n = 5), CB II (n = 29), and CB III (n = 8) and analyzed for HA-1 hematopoietic-restricted specificity and leukemic cell lysis. Results of 3 representative T-cell clones (clones 1-3) are shown in Figure 2. Clone 1 lysed HLA-A2^pos/HA-1^pos phytohemagglutinin (PHA) blasts, but not fibroblasts, while fibroblasts pulsed with HA-1^H peptide were lysed (Figure 2A). Clones 1, 2, and 3 were analyzed against a panel of HA-1^pos and HA-1^neg leukemic cells. Each clone recognized the 3 HLA-A2^pos/HA-1^pos leukemic targets, whereas HA-1^neg leukemic cells and autologous HA-1^neg PHA blasts, tested in parallel, were not recognized (Figure 2B-D). Thus, HA-1-specific cytotoxic T cells can be generated ex vivo from HLA-A2/HA-1^neg CB samples. These T cells display specific and hematopoietic-restricted recognition.

The presence of circulating HA-1-specific T cells was analyzed in 11 HLA-A2^pos/HA-1^neg CB samples derived from 7 HA-1^pos mothers and from 4 HA-1^neg mothers. All 11 CB samples were stained with HA-1^A2 tetramers and CD8 antibodies, FACS sorted, and nonspecifically expanded, omitting any in vitro HA-1-specific stimulation. Subsequently, HA-1^A2 tetramer analysis was performed after 21 days of expansion. Cells isolated from all 4 HA-1^neg CB samples derived from HA-1^neg mothers failed to grow in vitro, despite a few detectable tetramer-staining CD8^pos cells. The latter observed events are most likely due to background staining. Cord blood samples contain variable percentages of CD3^negCD8^pos natural killer (NK) cells and erythroblasts that display nonspecific staining in our hands. The level of background tetramer staining depends on the degree of depletion of these cells by magnetic bead separation. Yet a substantial number of cells isolated from 3 of 7 HA-1^neg CB samples derived from HA-1^pos mothers stained with the HA-1^A2 tetrameric complexes (Figure 3A-C, gate R1) and expanded to sufficient numbers for tetramer analysis. A variable percentage (1%-18%) of HA-1^A2 tetramer staining CD8^pos cells was observed at day 21 of nonspecific expansion (Figure 3E-G). Culture G was expanded nonspecifically for another 14 days, which resulted in a further enrichment of tetramer^pos CD8^pos cells (40%, insert Figure 3H). Functional analysis of the latter culture showed lysis of HLA-A2^pos/HA-1^pos leukemic cells and HLA-A2^pos/HA-1^neg EBV-LCL pulsed with HA-1^H peptide, while autologous HLA-A2^pos/HA-1^neg PHA blasts, HLA-A2^pos/HA-1^neg EBV-LCL, and K562 cells were not lysed (Figure 3H).

CD45RA expression on CB CD8^pos cells was determined directly after the first enrichment sort prior to in vitro culture (Figure 3D). The majority of tetramer^neg CD8^pos cells (Figure 3A-C, gate R2) expressed CD45RA (open histograms). In contrast, tetramer^pos CD8^pos cells (Figure 3A-C, gate R1) from 2 of the 3 CB samples clearly expressed lower levels of CD45RA (filled histograms), suggesting a primed phenotype at the time of isolation. Thus, antigen-experienced circulating T cells specific for maternal minor H antigen HA-1 can be detected in cord blood.

Our study demonstrates the presence of circulating HA-1-specific T cells in HLA-A2^pos/HA-1^neg CB samples derived from children delivered by HLA-A2-matched but HA-1-mismatched mothers. We also show that HA-1-specific T cells can be generated ex vivo from HLA-A2^pos/HA-1^neg CB samples. CB-derived HA-1-specific T cells show the expected cytotoxic activity that includes lysis of HA-1^pos leukemic cells.

The majority of unrelated CBT is performed with 1 or 2 HLA-mismatched grafts as reviewed.²⁸  An inverse correlation between the number or type of HLA disparities and relapse risk was recently found,²⁹  suggesting that alloreactivity to mismatched HLA antigens contributes to GvL responses. The fact that minor H antigen-specific cytotoxic T cells are generated across HLA haplo barriers and observed in fetal blood underlines the immunogenicity of the hematopoietic-specific minor H antigen HA-1. We speculate that pre-existing HA-1-specific T cells may contribute to GvL reactivity upon CBT in HLA-A2^pos/HA-1^pos recipients. Alternatively, HA-1-specific cytotoxic T cells can be generated ex vivo and stored for adoptive transfer in case of leukemic relapse.

The alloreactive potential of cord blood is shaped during pregnancy by fetal-maternal interactions. Both fetal and maternal HLA alloreactive T cells are, however, subjected to immune regulatory mechanisms to prevent fetal reactivity toward maternal tissues or fetal rejection.^30-32  Despite the apparent immunologic tolerance, normal precursor frequencies of cytotoxic and helper T cells specific for noninherited maternal HLA antigens (NIMA) can be detected in cord blood.^21,33  Similarly, noninherited maternal minor H antigens can prime fetal cytotoxic T cells, as shown in this study. Whether these T cells have any implications for the immunology of the maternal-fetal interface is a subject for further studies.

The tetramer^pos CD8^pos T cells directly sorted from 2 different CB samples clearly expressed lower levels of CD45RA than tetramer^neg CD8^pos T cells. Low CD45RA expression is indicative of recent antigen exposure, suggesting that HA-1-specific T-cell priming has occurred in utero. Similar fetal CD8 T-cell responses have been observed in case of maternal infections with Trypanosoma cruzi or cytomegalovirus,^16,34  as well as allergen- or Epstein-Barr virus-specific CD4^pos T-helper cells.^35-37  In line with these observations, our results demonstrate that T-cell priming for minor H antigens also occurs in utero. This is a remarkable finding, since allelic variants of minor H antigens can be considered as “modified self” in contrast to foreign viral antigens. The broadness of the autosomal encoded minor H antigen repertoire in CB samples needs to be investigated.

Maternal microchimerism is frequently found in CB samples³⁸  and in newborn tissue.³⁹  Nucleated maternal cells have been found in the fetal circulation as early as the second trimester of pregnancy.⁴⁰  Whether the presence of HA-1 cytoxic T cells is associated with the presence of maternal HA-1 microchimerism in the CB samples is as yet unknown. If so, we will analyze whether HA-1 is expressed by professional antigen-presenting cells, as we observed in another study.⁴¹  If maternal cells persist, they could provide a continuous antigen source of HA-1 that may maintain HA-1-specific cytotoxic T cells into adult life. This would explain the low but significant precursor frequencies of HA-1-specific cytotoxic T cells that we observe in some healthy blood or stem cell donors.²⁶  Pre-existing cytotoxic T-cell precursor frequencies may facilitate the ex vivo generation of HA-1-specific T cells for adoptive immunotherapy.

Recipients of HLA-identical SCT have a poorer transplant outcome if the donor is female rather than male.^42,43  The explanation of this clinical observation is that pregnancy can lead to alloimmune responses. Over decades, several types of alloimmune responses, varying from immunization against erythrocyte- and HLA-specific antibodies⁴⁴  to alloimmune responses against fetal paternal minor H antigens, have been reported.⁴⁵  With regard to the latter, both autosomal and Y chromosomes encoded minor H antigen responses have been observed.^12,24  Evidently priming of fetal cells restricts itself to the autosomal minor H repertoire, since maternal cells lack the Y-chromosome encoded H-Y antigens. The absence of fetal anti-H-Y priming, disadvantageous for female-to-male SCT, makes cord blood an attractive stem cell source for male patients.

A. Kloosterman, M. van der Keur, and R. van der Linden are acknowledged for excellent technical assistance. We thank Drs M. Oudshoorn and M.H.M. Wauben and Professor J.J. van Rood for critical reading of the manuscript.