Mixed chimerism renders residual host dendritic cells incapable of alloimmunization of the marrow donor in the canine model of allogeneic marrow transplantation

ABSTRACT

This study tested whether an alloimmune response can occur in the marrow donor when infused or injected with leukocytes from their mixed chimeric transplant recipient. Two mixed chimeras were produced after conditioning with three Gray total body irradiation, donor marrow infusion, and post-grafting immunosuppression. The marrow donors were then repeatedly infused and injected with leukocytes from their respective chimeric recipient. A donor lymphocyte infusion (DLI) into their mixed chimeras had no effect, even after the experiments were repeated. The presence of blood dendritic cells (DCs) of recipient origin was confirmed in chimeric recipients, as well as the presence of microchimerism in the marrow donors. Donor sensitization did occur following placement of a recipient skin graft that was confirmed following DLI into recipients that changed the mixed chimeras into full donor chimeras. These observations suggest that mixed chimerism renders recipient peripheral blood DCs incapable of inducing a donor T cell response.

KEYWORDS:

• alloimmunization
• dendritic cells
• microchimerism
• minor histocompatibility antigens
• mixed chimerism
• tolerance

Introduction

Alloimmunity refers to various cellular and humoral immune responses against genotypic differences between individuals, whereas immunological tolerance refers to the ability to prevent these immune responses.¹ For major histocompatibility matched (MHC) allogeneic haematopoietic cell transplantation (HCT), the majority of alloimmunity refers to adaptive T cell-mediated responses against minor histocompatibility antigens (miHAs). MiHAs are MHC presented peptides derived from endogenous polymorphic gene differences between the transplant donor and host.² T cell responses against miHAs are implicated in the non-relapse mortality caused from host versus graft rejection³ and graft-vs.-host disease (GVHD),⁴ as well as in the curative graft-vs.-tumor (GVT) responses for patients with hematologic malignancies.⁵ Therefore, improving outcomes following allogeneic HCT is directly dependent on refining the ability to control alloimmunity and immunological tolerance.

Boosting miHA-mediated immune responses depends on understanding the initiating events. The 1960 Nobel Prize in Physiology or Medicine highlighted the work of Owen,⁶ Burnet,⁷ Medawar⁸ and colleagues who formulated a “nonself” model to explain alloimmunity. Building on that foundation, a more nuanced “danger” model emerged supported by the discovery of pattern recognition receptors,^9,10 that provide evidence of alarm signals that link the innate activity of dendritic cells (DCs)¹¹ with the adaptive responses of T cells.¹² According to the danger model, inducing a T cell response to miHAs depends on the presence of alarm signals, and an immune response may not occur even when exposed to miHAs if the tissue, or more aptly the DCs in the tissue, lack these alarm signals. It is now apparent that alloimmunity or tolerance depends upon a series of receptor-ligand interactions between various T cell and DC subsets under the influence of the local cytokine milieu.¹³

MiHA-mediated immune responses can occur when the transplant recipient's DCs present antigens through direct antigen presentation to donor T cells or when DCs present recipient miHAs through indirect antigen presentation to donor T cells. With this biological understanding, the canine model of allogeneic HCT, specifically mixed chimerism, was used as a platform to further characterize donor T cell responses to miHAs. Stable mixed haematopoietic chimerism can be achieved following low dose total body irradiation (TBI), dog leukocyte antigen (DLA)-identical marrow infusion, and a short course of post-grafting immunosuppression with mycophenolate mofetil and cyclosporine.¹⁴ T regulatory cells induce an active cellular immunological tolerance allowing donor and host T cells to co-exist,^15,16 despite numerous miHA differences. “Unsensitized” donor lymphocyte infusions (DLI) do not overcome the T regulatory-mediated tolerance of mixed chimerism, producing no significant change in mixed chimerism.^17,18 However, if the marrow donor is first “sensitized” to miHAs of the recipient via recipient skin grafts,¹⁷ recipient organ grafts,¹⁹ or infusion of leukocytes from DLA-identical littermates,²⁰ then a “sensitized” DLI breaks tolerance changing the mixed chimera to 100% donor chimerism. Thus, extrapolating from these earlier studies, it was hypothesized that intravenous and subcutaneous infusions of viable recipient cells from mixed chimeras into their respective marrow donors would sensitize the donors to the miHAs of the recipients. In this setting, a DLI converting the mixed chimera to full donor chimera would provide an in vivo readout and confirm the hypothesis. However, the marrow donors did not become sensitized to recipient miHAs despite multiple exposures to viable recipient leukocytes present within the blood of the mixed chimeras. The studies were repeated and the ability of the marrow donor to respond to recipient miHAs was verified by skin grafting.

Results

First attempt at alloimmunization of the donor

To test the hypothesis whether remaining recipient leukocytes in a stable mixed chimera could successfully alloimmunize the marrow donor, two mixed chimeric recipients, H550 and H552, were produced with three Gray TBI, marrow donor infusion, and post grafting immunosuppression with cyclosporine and mycophenolate mofetil. At the time of the first DLI, both recipients had been off all immunosuppression for 12 months. The percentage of donor-recipient mixed chimerism for the eight months preceding the first DLI is shown as a baseline for H550 and H552, with the eight-month time point occurring two weeks prior to the first DLI (black lines, Fig. 1). For H550, granulocytes ranged from 90 to 97% donor and PBMC ranged from 62 to 86% donor, while H552 granulocytes ranged from 26 to 68% donor and PBMC ranged from 45 to 64% donor for the eight months preceding the first DLI (black lines, Fig. 1).

Figure 1. Summary of the chimerism results before and after three DLIs into two stable mixed chimeric recipients. The y-axis shows the percentage of donor chimerism in the granulocyte fraction for H550 (A) and H552 (B), and PBMC fraction for H550 (C) and H552 (D). The x-axis shows times in months. Baseline reported chimerism values for each recipient for the 8 months preceding the first DLI with the last time point taken two weeks prior to the first DLI. The rest of the chimerism values are reported following each DLI with zero representing the time of the DLI.

For the first attempt, marrow donors were infused intravenously and injected subcutaneously with leukocytes harvested from 60 mL of blood from their respective mixed chimeric recipient on day −30, day −20, and day −10, followed by a DLI into the recipient on day 0. The donor for H550 received a total dose of 1.64 × 10⁸ viable mixed chimeric leukocytes, while the donor for H552 received a total dose of 2.51 × 10⁸ viable mixed chimeric leukocytes in the first attempt (Table 1). Taking into account a mixed chimera with 80 to 90% donor chimerism resulting in only 10 to 20% of the cells of recipient origin, the number of recipient cells injected into the donor reached at least 1–2 × 10⁷ cells per attempt.

Table 1. Summary of the donor sensitization attempts followed by DLI into the mixed chimeric recipient.

Marrow donor (sex)	Recipient (sex)	Marrow donor sensitization attempt	Recipient antigen source	Route of delivery	Viable cell number ×10^8 leukocytes	DLI T cells ×10^7/kg	Persistent change in chimerism
H554 (female)	H550 (female)	1st	leukocytes	½ IV and ½ SQ every 10 days × 3	1) 0.6	3.7	No
					2) 0.35
					3) 0.69
					Total dose = 1.64
		2nd	leukocytes	½ IV and ½ ID every 10 days × 3	1) 029	3.9	No
					2) 0.34
					3) 0.52
					Total dose = 1.15
		3rd	skin	1) 2×2 cm full thickness	N/A	7.2	Yes
				2) modified
H555 (male)	H552 (female)	1st	leukocytes	½ IV and ½ SQ every 10 days × 3	1) 0.7	3.4	No
					2) 1.05
					3) 0.76
					Total dose = 2.51
		2nd	leukocytes	½ IV and ½ ID every 10 days × 3	1) 0.53	6.7	No
					Two) 0.49
					3) 0.95
					Total dose = 1.97
		3rd	skin	1) 2×2 full thickness	N/A	5	Yes
				2) Modified

Intravenous (IV), subcutaneous (SQ), or intradermal (ID) route, as described.

To minimize the effects of route of administration on the interpretation of the results, half the cell dose was infused intravenously and the other half was injected subcutaneously.^21,22 A DLI was then performed into H550 and H552 and the level of donor chimerism was assessed over time (Fig. 1). There was a transient increase in the percent donor granulocyte levels for H550 ranging from 73 to 100% donor (blue line, Fig. 1A). However, there was no change in the percent donor PBMC for H550, and by eight months after the first DLI, the percent donor chimerism for the granulocyte and PBMC fractions was essentially unchanged from pre-DLI levels in both chimeras.

Second attempt at alloimmunization of the donor

The second attempt at marrow donor sensitization was performed to verify the results obtained following the initial attempt. The antigen source and the timing of three separate infusions of mixed chimeric recipient leukocytes were kept the same as the initial attempt. To further eliminate route of administration effects on the interpretation of these results the cells were again split, half were infused IV, and the other half injected via an intradermal route instead of a subcutaneous route in order to deliver recipient antigens directly into dermal DCs of the donor. Using trypan blue exclusion, the donor for H550 received a total dose of 1.15 × 10⁸ viable mixed chimeric leukocytes, while the donor for H552 received a total dose of 1.97 × 10⁸ viable mixed chimeric leukocytes (Table 1).

To further confirm the viability of the cell injections, cell dosages were also counted using acridine orange/ethidium bromide (AO/EB) double staining.²³ The viable cell counts using AO/EB were on average 80%, range 57–106%, of the value obtained using trypan blue exclusion (data not shown). The difference in viability was consistent with the underestimation of the cell death assessment of trypan blue staining by light microscopy as compared with fluorescent exclusion dyes detected by fluorescence microscopy. A DLI was then performed into H550 and H552 and the level of donor chimerism was assessed over time (Fig. 1). Again there was a transient increase in the percent donor granulocyte levels for H550 ranging from 80 to 97% donor (green line, Fig. 1A). However, there was no change in the percent donor PBMC for H550, and by eight months after the second DLI, the percent donor chimerism for the granulocyte and PBMC fractions was essentially unchanged from pre-DLI levels in both chimeras.

Donor microchimerism studies and third attempt at alloimmunization

The two marrow donors were evaluated for the presence of recipient cells. Microchimerism, or the presence of small amounts of cells from another individual, is a well-recognized phenomenon of pregnancy, as well as increasingly appreciated following blood transfusions.²⁴ A PCR signal of recipient DNA was detected from the peripheral blood of both marrow donors one month after the last injection of mixed chimeric cells. This signal was not present for marrow donor H555 (Fig. 2A), and was only half the signal strength for marrow donor H554 (Fig. 2C) prior to the infusion of mixed chimeric cells. A standard curve was created with mixtures of marrow donor and recipient DNA collected prior to transplant. The standard curve was accurate up to a mixture of 97% donor DNA for H555 (Fig. 2B) and 99.7% donor DNA for H554 (Fig. 2D).

Figure 2. Variable number tandem repeat (VNTR) assay. Genomic DNA of H555 and H552 (A) H554 and H550 (C) were isolated and amplified by PCR for VNTR. The data are represented by peaks with the shaded areas representing area under the curve (AUC). The top panel shows the marrow donor VNTR prior to any cell injections from the mixed chimera. The middle panel shows marrow donor VNTR one month after the last set of cell injections from the mixed chimera. The bottom panel shows the VNTR of the transplanted recipient prior to haematopoietic cell transplant as a comparison. Titrated mixtures of pre-transplant donor and recipient DNA shown for H555 and H552 (B) and H554 and H550 (D) were shown in a logarithmic scale to determine the limits of detection.

The third attempt at marrow donor sensitization was performed to confirm that both marrow donors remained capable of responding to miHAs. Past studies have already shown that miHA sensitization of the marrow donor occurs following mixed chimeric recipient skin graft placement.¹⁷ Each donor underwent placement of a two by two centimeter full thickness skin graft from their respective mixed chimeric recipients. The skin grafts were viewed every other day to confirm adequate surgical technique and assess for signs of rejection. The recipient skin graft on the marrow donor for H550 became erythematous on day +16 after placement; the skin graft on the marrow donor for H552 became erythematous on day +13 after placement. Both skin grafts then blistered and were obliterated over the next seven days. Six weeks after the full thickness skin graft, the donors were further sensitized with a “modified” skin graft placement consisting of a subcutaneous placement of a minced square centimeter of recipient skin. DLIs were performed 10 days after the modified skin graft. H550 rapidly converted to 100% donor granulocytes and PBMC within four weeks (red lines, Fig. 1). H552 converted more slowly, but ultimately achieving 100% donor granulocytes and PBMC (red lines, Fig. 1). Neither recipient demonstrated any signs of GVHD at necropsy.

Evaluating dendritic cell and T regulatory cell populations in mixed chimeras

The active cellular tolerance that prevents marrow donor sensitization by mixed chimeric leukocytes likely depends on regulatory T cells and tolerogenic DCs, and these cell populations were further evaluated. Consistent with previous reports,²⁵ the level of T regulatory cells in the peripheral blood of H550 and H552 prior to donor conversions were within the range of normal controls, as assessed by CD4+, CD25+, and FoxP3+ expression by flow cytometry. Percent CD4+, CD25+, and FoxP3+ cells from within the lymphocyte gate ranged between 0.26 to 0.8 for H550 and H552, which was comparable to five normal controls that ranged from 0.06 to 0.83.

Next, studies were performed to eliminate the possibility that the results from the first two sensitization attempts were due to the lack of recipient-derived blood DCs in the mixed chimera capable of sensitizing the marrow donor. Functional, peripheral blood DC population in canines were previously defined as cells that express MHC class II and canine CD11c that were not granulocytes or monocytes (Fig. 3).²⁶ At the time of this analysis, no cells were available from H550 and H552 prior to conversion to full donor; instead, blood DCs were flow sorted from frozen PBMC samples from three mixed chimeras H382, H519, and H597, previously banked from past studies. The purity of the sorted cells ranged from 91 to 92%, representing approximately 100-fold enrichment considering blood DCs consist of less than 1% of the PBMC population.²⁶ Donor chimerism percentages of the sorted DC as compared with the PBMCs populations were 76% vs. 62% for H382, 31% vs. 47% for H519, and 63% vs. 65% for H597, confirming that the sorted blood DCs contained both recipient and donor cells in nearly the same ratio as the PBMC fraction.

Figure 3. Fluorescence-activated cell sorting of peripheral blood dendritic cells. The first gate utilized forward scatter and side scatter properties to eliminate cell debris (left plot). The second gate was then used to exclude CD14+, DM5- monocytes and the DM5+, CD14- granulocytes from the sorted population (middle plot). MHC class II+ cells determined using an antibody against DLA-DR, and canine CD11c+ cells were then sorted for use in chimerism analysis or in MLRs (right plot).

Functional analysis of the sorted DCs from these mixed chimeras was then pursued to eliminate the possibility that the lack of sensitization was due to non-functional recipient-derived DCs. However, an in vitro test to determine the responsiveness of recipient-derived DCs to miHAs was not possible, given that miHAs await molecular characterization in the dog model. Instead, the functional characterization of the sorted DCs was limited to testing their ability to respond to major MHC mismatches in a mixed leukocyte reaction (MLR). The sorted DCs from all three mixed chimeras were capable of stimulating third party responders, demonstrating that these DCs were functional (Fig. 4). However, there was no way to further narrow the contribution of this response to either the donor- or recipient-derived DCs.

Figure 4. Functional characterization of sorted cell populations assayed in MLR. The monocyte, granulocyte, and peripheral blood DCs populations were sorted as shown for Figure 3 and used as stimulator cells. Constant numbers of third party responding PBMC (1×10⁵ cells per well) were mixed with varying numbers of irradiated stimulator cells (5–20×10³). The responders and stimulators were MHC-mismatched. The stimulation index was calculated as the average counts per minute of the responder cells with stimulator cells added divided by the average counts per minute of responder cells only plus or minus the standard deviation (SD). The solid line and filled in data points were for the normal control, while the three samples from mixed chimeric recipients H382, H597, and H519 were shown with dotted lines and open data points.

Discussion

Conclusions

This report represents the first study in the dog HCT model to ask the question whether an alloimmune response can develop when a marrow donor is infused or injected with leukocytes from their respective mixed chimeric transplant recipients. The small numbers of transplant pairs used in these studies prevent definitive conclusions. However, the sensitization experiments were repeated and miHA sensitization of the marrow donor did not occur despite repeated exposures to recipient blood cells from their respective mixed chimera. These experiments disprove the initial hypothesis that miHA sensitization of the marrow donor was going to occur following exposure to recipient leukocytes from a mixed chimera. The results of these experiments suggest that direct miHA presentation from the mixed chimeric recipient's DCs and indirect miHA presentation from the marrow donor DCs did not occur in spite of the presence of the genetically disparate cells.

The lack of miHA sensitization following the infusion and injection of mixed chimeric leukocytes into the marrow donor was not due to insufficient numbers of viable recipient cells. It is known that leukocyte infusions between MHC-matched sibling transplant pairs in outbred populations leads to rapid miHA sensitization.²⁷ For example, a single 100 mL, non-irradiated marrow donor blood transfusion into an MHC-matched sibling recipient 24 hours prior to transplant resulted in six of seven graft rejections despite ablative 9.2 Gy TBI conditioning and DLA-identical marrow infusion.²⁸ MiHA sensitization occurs due to a small number of blood DCs present within the infused leukocytes that sensitize the transplant recipient's T cells.²⁹ It takes only a small number of infused leukocytes, and hence an exceedingly small number of blood DCs, between DLA-identical siblings to induce miHA sensitization^27-39 From past experiments, it was shown that infusing the transplant recipient with as few as 5 × 10⁴ donor leukocytes^33,40,41 or PBMC²⁹ on day −24, −17, and day −10 prior to transplant for a total dose of 1.5 × 10⁵ cells resulted in 40% graft rejection despite myeloablative 9.2 Gy TBI conditioning. Compared to these past studies and taking into account that a percentage of the mixed chimeric cells were donor, each marrow donor in this study received at least 1–2 × 10⁷ total recipient leukocytes per sensitization attempt. Therefore, each marrow donor was exposed to at least 100-fold more recipient leukocytes than what can induce miHA sensitization, and each marrow donor then underwent a second sensitization attempt.

Future experiments will need to test whether infusion of mixed chimeric cells into the marrow donor could be used as a platform to study the still poorly understood danger signals that govern alloimmunization vs. tolerance. These studies can then further test the hypothesis that mixed chimerism renders recipient blood DCs tolerant and incapable of sensitizing the marrow donor. A hypothesis consistent with the underlying notion that mixed chimerism is maintained by T regulatory cells,^15,16 and it is increasingly recognized that the presence of T regulatory cells is synonymous with the presence of tolerogenic DCs.⁴² If future studies confirm the hypothesis, then it remains a testable question whether infusion of mixed chimeric cells into the marrow donor can be used to achieve microchimerism or even mixed chimerism with the associated potential of a transfer of tolerance. With respect to this concept, this study did report some preliminary evidence of microchimerism in the marrow donors following the infusion of mixed chimeric cells. However, a larger cohort is needed to confirm the presence and persistence of microchimerism, and determine if this can be increased or even converted to mixed chimerism with infusions of mixed chimeric cells that contain haematopoietic stem cells. As for this study, the marrow donors did not demonstrate any evidence of tolerance.

Materials and methods

Experimental animals and DLA typing

Random-bred beagles and mini-mongrel cross breeds were raised at the Fred Hutchinson Cancer Research Center (FHCRC), Seattle, WA or purchased commercially. Animals were housed in kennels certified by the American Association for Accreditation of Laboratory Animal Care. All study designs were approved by the Institutional Animal Care and Use Committee. Dog leukocyte antigen (DLA)-typing used highly polymorphic microsatellite markers within DLA class I (DLA 88)⁴³ and class II regions (DLA-DRB1).^44,45 The DLA typing for donor-recipient transplant pair H554-H550 were DRB1 6/15 and DLA-88 00402/03201, and for donor-recipient transplant pair H555-H552 were DRB1 6/17 and DLA-88 00402/04901.

Minimal intensity DLA-identical HCT and chimerism analysis

HCT recipients were treated with 300 centigray (cGy) TBI at a dose rate of 7 cGy/minute from a high-energy linear accelerator (6 MV Elekta Precise; Stockholm, Sweden). Within four hours of TBI, bone marrow was collected and infused from their respective DLA-identical siblings as described,⁴⁶ H550 received 5 × 10⁸ nucleated cells/kg and H552 received 4.8 × 10⁸ nucleated cells/kg. Post-grafting immunosuppression consisting of oral cyclosporine at 15 mg/kg orally twice daily started the day before the marrow infusion and continued for 36 days, and mycophenolate mofetil at 10 mg/kg twice daily injected subcutaneously started on the day of marrow infusion for 28 days, as decribed.⁴⁷ Chimerism analyses were done on PBMCs and granulocytes following separation of blood on Ficoll (density = 1.074), and quantified by fluorescent variable number of tandem repeat PCR analysis, as described.⁴⁸ Microchimerism analysis required pre-transplant samples from the donor in order to compare the PCR signal for the recipient, and the robustness of the PCR reaction was further validated using a standard curve created through mixture of donor and recipient DNA.

Marrow donor sensitization

Buffy coat cells from their respective mixed chimera were isolated from 60 mL of heparinized blood. The buffy coat was lysed with ammonium chloride lysing solution (155 mM ammonium chloride, 10 mM sodium bicarbonate, 0.1 mM EDTA), washed twice with phosphate-buffered saline (PBS). Viable infused cells were determined using a hemocytometer and Trypan Blue 0.4% staining (Thermo Fisher Scientific, Waltham, MA). For the first sensitization attempt, half of the cells were suspended in 10 mL of PBS and infused intravenously. The other half was suspended in 200 microliters of PBS and injected subcutaneously into the flank of their respective donors. This was repeated every 10 days for a total of three injections. For the second attempt the same experimental protocol was followed with the exception that half of the cells were injected into the dermis instead of subcutaneous tissue of the flank. Viable cells were determined by trypan blue and acridine orange/ethidium bromide per Cold Spring Harbor Protocols (cshprotocols.cshlp.org). For the third attempt at donor sensitization, subcutaneous placement of a full-thickness 3 × 3 cm recipient skin graft onto the marrow donor was performed under general anesthesia, as described.^49,50 Six weeks after the full thickness skin graft, a “modified” skin graft was placed via a small incision in the donor to create a subcutaneous pocket following conscious sedation and local anesthetic as follows: several 8 mm punch biopsies that did not exceed one square centimeter were obtained from the recipient following conscious sedation and local anesthetic;the biopsies were minced and placed into a subcutaneous pocket in the marrow donor that was then stapled closed, as described.⁵¹

Donor lymphocyte infusion

Ten days after the last cell infusion or skin graft, marrow donor dogs underwent general anesthesia, central venous catheter placement in the external jugular vein, and collection of 200 mL leukapheresis product, as described (COBE 2997; Blood Component Technology, Lakewood, CO).^52,53 Cell counts with differentials were obtained using an ADVIA 2120i (Siemens, Deerfield, IL). Recipients were pre-treated with diphenhydramine and the entire leukapheresis product was slowly infused. Cell counts with differentials were obtained using an ADVIA 2120i (Siemens, Deerfield, IL).

Flow cytometry, cell sorting, and mixed lymphocyte reaction

CD3 cell counts were determined using anti-canine CD3 (17.6F9), as described,¹⁷ and reported in Table 1. T regulatory cells were evaluated using anti-canine CD4-FITC (CA13.1E4) (Bio-Rad, Hercules, CA), anti-canine CD25-PE (P4A10) (Thermo Fisher, Waltham, MA), and intracellular stained using anti-FOXP3-AF647 (150D) Fix/Perm buffer set (Biolegend, San Diego, CA). For the T regulatory cell analysis, the flow cytometry gates were set based on isotype controls for mouse IgG1-PE and mouse IgG1-AF647 (Southern Biotech, Birmingham, AL). Peripheral blood DCs were defined using anti-DM5-FITC,⁵⁴ anti-human CD14-PerCP (TuK4) (Thermo Fisher, Waltham, MA), anti-canine CD11c-PE (6A1),⁵⁵ and anti-canine DLA-DR-AF647 (H81.9).⁵⁶ Flow cytometry was performed using a FACS Canto II and cell sorting with a FACS ARIAII (BD Biosciences, San Jose, CA). Data was analyzed using FlowJo Software (Treestar Ashland, OR). MLRs were performed with MHC-mismatched stimulator and responder cells as previously described.^57,58 The DLA typing for the stimulator H382 was DRB1 1/1 and DLA-88 50801/50801, H597 DRB1 9/15 and DLA-88 01101/01201, H519 DRB1 9/22 and DLA-88 03801/50101, and the normal control stimulator H810 DRB1 1/3 DLA-88 03202/03401. The DLA typing for the normal responder H750 was DRB1 17/19 and DLA-88 04501/04901.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Author contributions

SLR designed experiments, performed the data analysis, and drafted the article. SSG and RS designed the experiments and drafted the article. DAH reviewed the article and assisted in data acquisition.

Funding

The authors are grateful for research funding from the National Institutes of Health, Bethesda, MD, grants P01 CA078902, P30 CA015704, K12 CA076930, P30 DK056465, support from Gabrielle's Angel Foundation for Cancer Research (SLR), and by awards from the Joseph Steiner Krebsstiftung, Bern, Switzerland and Lupin Foundation, Metairie, LA (RS).