Current therapies for transplant rejection are suboptimally effective. In an effort to discover novel immunosuppressants we used cytokine ELISPOT and ELISAs to screen extracts from 53 traditional Chinese herbs for their ability to suppress human alloreactive T cells. We identified a dichloromethane-soluble fraction (Qu Mai fraction AD [QMAD]) of Qu Mai (Dianthus superbus) as a candidate. High-performance liquid chromatography (HPLC) analysis of QMAD revealed three dominant peaks, each with a MW ∼600 Daltons and distinct from cyclosporine and rapamycin. When we added QMAD to human mixed lymphocyte cultures, we observed dose-dependent inhibition of proliferation and IFNγ production, by naïve and memory alloreactive T cells, and observed an increased frequency of Foxp3+CD4+ T cells. To address whether QMAD induces regulatory T cells we added QMAD to anti-CD3/CD28-stimulated naïve CD4 T cells and observed a dose-dependent upregulation of Foxp3 associated with new suppressive capacity. Mechanistically, QMAD did not induce T cell IL-10 or TGFβ but blocked T cell AKT phosphorylation, a key signaling nexus required for T cell proliferation and expansion, that simultaneously prevents Foxp3 transcription. Our findings provide novel insight into the antiinflammatory effects of one traditional Chinese herb, and support the need for continued isolation, characterization and testing of QMAD-derived components as immune suppressants for transplant rejection.
Transplantation is a life-saving treatment for end-stage heart, liver and lung failure, and provides a survival advantage over dialysis for patients with end-stage kidney disease [1, 2]. Over the past two decades, improvements in clinical care and the addition of multiple immunosuppressant strategies have reduced short-term morbidity following organ transplantation and lowered rates of acute rejection episodes for all types of transplants [3, 4]. Nonetheless, acute rejection continues to cause morbidity [3, 4]. Commonly used immunosuppressants have multiple toxic, off target effects [3, 4], can inhibit protective regulatory T cell (Treg) formation and function  and their use has not significantly improved the incidence or kinetics of late graft failure [6, 7]. Memory T cells reactive to donor antigens represent one important barrier to improved transplant outcomes as memory T cells are resistant to most currently used immunosuppressants  and the presence of antidonor memory confers an increased risk for poor outcome following transplantation [9-11]. Taken together, these observations support the need to identify additional immunosuppressants for use in transplantation, specifically compounds that are simultaneously Treg-protective and capable of blocking memory T cells.
Traditional Chinese herbal medicine (TCM), used for centuries to treat a wide array of illnesses, has received attention in recent years as potential therapy for a variety of immune-mediated disorders. Studies by our group and others indicate that standardized extracts of TCM, and isolated compounds derived from these extracts, have efficacy in preventing and/or treating food allergy, asthma, and rheumatoid arthritis in humans [12-15]. While several TCM-derived compounds have been shown to limit alloreactive T cell responses in animal models [16-18], whether and how TCM impacts human alloimmunity has not been carefully explored. To this end, we screened 53 Chinese herbal extracts for their potential ability to suppress human alloimmune responses and identified Qu Mai (QM, Dianthus superbus) as a candidate immunosuppressant. We report here that a nonpolar fraction of QM extracted with dichloromethane suppresses naïve and memory alloreactive T cells while simultaneously and directly generating Treg. Our findings support the need for continued isolation and testing of the active components of QM as potential immune suppressants to improve transplant outcomes.
Blood samples were obtained from normal volunteers following approval by the Institutional Review Board at the Mount Sinai School of Medicine. In some experiments PBMC were obtained from commercially purchased buffy coats from de-identified donors.
Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll separation using standard methods . Primary expanded B cell lines used as stimulator cells were produced from CD19-enriched PBMC by in vitro culture with CD40L-transfected fibroblasts and IL-4 as previously described [19, 20]. Aliquots were frozen and stored in liquid nitrogen and thawed as needed for use in various assays. CD3+ and memory CD45RO+CD8+ T cells and CD14+ cells were isolated to >90% purity using kits purchased from STEMCELL Technologies (Vancouver, British Columbia, Canada); naïve CD45RA+CD4+ T cells were isolated using MACS® cell separation (Miltenyi Biotec, Auburn, CA, USA). Dendritic cells (DCs) were induced from CD14+ monocytes by incubation with GMCSF and IL-4 as described . Cell viability was assessed by acridine orange, ethidium bromide staining.
Preparation of herbal extracts
All medicinal herbs used in this study were from the TCM antiinflammatory herbal inventory of the Botanical Chemistry Laboratory at Mount Sinai School of Medicine (New York). The inventory was established based on substantial literature search of the TCM herbs that have been reported to have antiinflammatory effects and by preliminary screening of the herbs that inhibit TNF-α (unpublished data). All herb extracts were made in a good manufacturing practice facility (Gangdong Yifang Pharmaceutical Co. Ltd., China), or from Xiyuan Chinese Medicine Research and Pharmaceutical Manufacturer, Chinese Academy of Chinese Medicine Sciences, Beijing, China. All were water extracts prepared according to the standard procedure, and then concentrated and dried. Each powdered extract was packaged and stored at room temperature under dark and dry conditions. We screened 53 distinct herbal extracts (see Table S1). All extracts were dissolved in 25 mg/mL of CTL-T serum-free media (CTL, Shaker Heights, OH, USA) and diluted in CTL-T for use in assays.
Chemical isolation of QM fractions and generation of high-performance liquid chromatography (HPLC) fingerprints
To prepare subfractions of the QM extract we dissolved 100 g of QM in 4 L distilled water, adjusted the pH to 2.0 with HCl, and extracted the solution with HPLC grade dichloromethane (Fisher Scientific, Pittsburgh, PA, USA) for 24 h in a liquid–liquid extractor (Sigma-Aldrich, St. Louis, MO, USA) at 100°C using a 2000 mL heating mantle (Barnstead International, Dubuque, IA, USA), yielding fraction Qu Mai fraction AD (QMAD). The aqueous phase was further extracted with 4 L ethyl acetate (Fisher Scientific, Fair Lawn, NJ, USA) for 30 h at 120°C (yielding QMAE), concentrated to 1 L and further extracted with 1 L of butanol (Fisher Scientific) using separation funnel to yield QMAB fraction. The butanol extraction (QMAB) was collected in a new container. All three extracts were dried with a rotary evaporator (Rotavapor R-210, BÜCHI, Switzerland) under vacuum.
Each of the three fractions, QMAD, QMAE and QMAB, were analyzed by HPLC using Alliance 2695 HPLC system coupled with photodiode array detector (Waters Corporation, Milford, MA, USA); 10 μL of each QM fraction solution (5 mg/mL) was separated with Zorbax SB-C18 column (150 × 4.6 mm; 5 μm particle size) from Agilent Technologies (Santa Clara, CA, USA). The mobile phase A was 0.10% phosphoric acid (H3PO4) and the mobile phase B was acetonitrile. The separation gradient started at 2% of B to 25% in 45 min, increasing to 35% B in 25 min, increasing to 55% B in 15 min, increasing to 75% B in 10 min and maintained at 75% for 5 min. The flow rate was set to 1.0 mL/min and the chromatograms were acquired at 254 nm.
PBMCs or enriched T cell subsets were stimulated with either soluble anti-CD3/CD28 (BD Biosciences, San Jose, CA, USA; 1 μg/mL each), beads coated with anti-CD3/CD28 (Dynabeads, Invitrogen, Grand Island, NY, USA; 25 μL/1E6 cells), HLA-disparate DCs, HLA-disparate B cells ± phorbol myristate acetate (PMA) as indicated, in CTL or RPMI (Life Technologies, Grand Island, NY, USA) plus 6–10% human AB serum (Gemini Bio-Products, West Sacramento, CA, USA) for 1–5 days at 37°C 5% CO2. QM and QMAD, QMAE, QMAB fractions or equivalent amounts of diluent (DMSO, Fisher Scientific, final concentration 0.02–0.2%) were added to the cultures where indicated. Proliferation was assessed by prestaining the responder cells with carboxyfluorescein diacetate succinimidyl ester (CFSE; Sigma-Aldrich) and determining CFSE dilution by flow cytometry 3–5 days later. Mixed lymphocyte cultures were performed in the absence (control) or presence (experiment) of QM and QM fractions and were incubated for 24–48 h. Cells enriched for CD3 or CD4CD45RA were cultured with anti-CD3/CD28 coated beads ± QMAD ± IL-2 (BD Pharmingen, San Jose, CA, USA; 100 U/mL) ± TGFβ (PeproTech, Rocky Hill, NJ, USA; 5 ng/mL) in RPMI plus 10% heat-inactivated human AB serum and assessed for Treg induction 5 days later by flow cytometry. Cells were recovered, washed and resuspended in RPMI plus 10% human AB serum with CFSE-labeled PBMCs. These cultures were then stimulated with soluble anti-CD3/CD28 and CD8 cell proliferation was measured by CFSE dilution (flow cytometry) 4 days later.
ELISPOT and ELISA
ELISPOT assays for IFNγ and IL-10 were performed as described . Briefly, ELISPOT plates (Millipore) were coated with primary antibodies overnight, blocked and washed. Assays were done with 300 K responder PBMC mixed with medium alone, 100 K stimulator B cells or phytohemagglutinin (PHA) as a positive control in CTL medium for 48 h at 37°C 5% CO2. After washing, secondary antibodies and tertiary reagents were added and the plates were developed . The resulting spots were quantified using the ImmunospotS4 Core Analyzer (CTL, Shaker Heights). ELISAs for IL-10, IFNγ and TGFβ (BD OptEIA) were performed according to manufacturer's instructions.
FITC-anti-CD3, PE-Cy5-anti-CD4, APC-H7-anti-CD8, PE-anti-CD45RO, FITC-anti-CD4 APC-anti-CD25, PE-Cy7-anti-IFNγ, PE-Cy7-anti IL-2 and Alexa-Fluor-647-anti–phosphorylated AKT (pAKT; monoclonal antibody) were obtained from BD Biosciences. PE and PE-Cy7-anti-Foxp3 was obtained from eBioscience (San Diego, CA, USA). Dead cells were excluded with crystal violet staining kit (Invitrogen). For intracellular cytokine detection, Brefeldin A (eBioscience, 1:1000, for all cultures) ± 16 ng/mL PMA/0.8 μM Ionomycin (both from Sigma-Aldrich, for detection of IFNγ in 24–48 h cultures only) was added 4 h prior to recovery. Data were acquired on a 3-laser Canto II flow cytometer (BD Biosciences), and analyzed by using FlowJo software (Tree Star, Inc., Ashland, OR, USA).
All statistical analysis was performed using GraphPad Prism (version 5 for Windows, GraphPad Software Inc., La Jolla, CA, USA). Group comparisons of paired data were analyzed by paired t-tests, and by Wilcoxon's signed rank test where applicable. One-way RM ANOVA was used for multiple comparisons among treatment groups, with pairwise multiple comparison procedures performed using Holm–Sidak method or Tukey Test; p values less than 0.05 were considered significant.
Screening assays identify QM as a candidate herbal immunosuppressant
Multiple lines of evidence from our group and others suggest that a number of herbs used as traditional Chinese therapeutics could have immunomodulatory properties [23, 24] but whether any of these herbal compounds favorably inhibit human alloreactive T cell function has not been carefully evaluated. We therefore designed a screening assay in which we tested the effects of 53 herbal preparations (Table S1) on alloreactive T cell cytokine production. We reasoned that the herbs, which most robustly inhibit allo-induced production of the proinflammatory cytokine IFNγ and simultaneously augment production of the immunoregulatory cytokine IL-10 would be the strongest candidates to have clinically favorable immunomodulatory properties. We thus initiated mixed lymphocyte cultures using responder PBMCs obtained from normal volunteers mixed with allogeneic, in vitro expanded, primary (not EBV transformed) B cells as stimulators, in the presence or absence of each herb, tested at two concentrations. We quantified the frequency of IFNγ producing cells by ELISPOT and the concentration of IL-10 in culture supernatants of the same assay wells by ELISA.
As depicted in one representative assay shown in Figure 1(A) allo-stimulation resulted in a significant increase in the detectable frequency of IFNγ-producing PBMC compared with unstimulated PBMC or B cell stimulators alone. Culture supernatants of PBMC with or without allo-stimulators contained low and not significantly different levels of IL-10 (Figure 1B). B cell stimulators produced no detectable IL-10 in the absence of responder PBMC. When we tested the herbal preparations (Figures 1 and S1) we observed that four of them inhibited IFNγ production by >50% and enhanced IL-10 production by >100%. We chose to further study the immune suppressive effects of one of these herbs, QM, which has been traditionally used as an antiinflammatory agent for “urinary tract disorders” .
We examined concentration-dependent effects of QM on allo-stimulated production of IFNγ and IL-10 by ELISPOT using 10 distinct responder/stimulator pairs. The results revealed that QM induced a dose-dependent increase in the IL-10:IFNγ ratio (Figure 2), inhibited IFNγ and increased IL-10 production by PBMC (Figure S2) by each of the responder/stimulator pairs tested (p < 0.05 at doses >20 μg/mL vs. control).
Immune effects of QM are enriched by extraction with dichloromethane
To begin to isolate active compounds from the QM herbal preparation we performed liquid–liquid extraction based on polarity, which yielded three fractions: dichloromethane soluble QMAD (mainly contains nonpolar compounds), ethyl acetate soluble QMAE (mainly contains less polar compounds) and butanol soluble QMAB (mainly contains moderate polar compounds). HPLC analysis showed that each fraction contained multiple peaks (Figure 3A). When we tested each fraction for their effects on cytokine production using our screening mixed lymphocyte reaction (MLR) assay (Figure 3B) we observed that the QMAD fraction induced the most favorable effects on MLR-induced cytokines, yielding >ninefold increase in IL-10:IFNγ ratios (p < 0.05 at 50 μg/mL QMAD vs. control). We assessed dose-dependence of the QMAD effects by IFNγ and IL-10 ELISA on 48 h MLR supernatants (Figures 4 and S3). QMAD significantly upregulated IL-10 (mean increase of 203% at 50 μg/mL, n = 13 and 305% at 100 μg/mL, n = 5, Figure 4A) and inhibited IFNγ (mean decrease 61% at 50 μg/mL, n = 13 and by 66% at 100 μg/mL, n = 5, Figure 4A), resulting in significant increases in IL-10:IFNγ ratios (Figure 4B), in all responder/stimulator pairs tested (13-fold and 32-fold increases at 50 μg/mL and 100 μg/mL, respectively, p < 0.05 for each vs. control). The number of live cells present at the end of each culture was similar regardless of QMAD concentration (Figure 4C), demonstrating that QMAD alters cytokine production without inducing cell death. In control experiments we observed that QMAD increased IL-10 production (without detectable IFNγ) when cultured with PBMC alone (in the absence of B cell stimulators, Figure 4D). QMAD had no effect on the low levels of IL-10 (and absent IFNγ) produced by B cells cultured in the absence of PBMC (Figure 4D).
QMAD inhibits proliferation and IFNγ production by memory CD8 T cells
We next used flow cytometric analysis to determine effects of QMAD on CD4 and CD8 T cell proliferation (CFSE dilution) and IFNγ production (intracellular staining). We observed a dose-dependent inhibition of CD4 and CD8 T cell proliferation in response to allogeneic DC or anti-CD3/CD28 stimulation (Figure 5A). Addition of vehicle control DMSO had no effect (Figure 5B). No IL-10 was detected in culture supernatants of these QMAD-stimulated, purified T cells (data not shown). QMAD did not alter T cell viability (Figure 5B) ruling out direct cellular toxicity. QMAD also significantly decreased IFNγ produced by CD4 and CD8 T cells in 5 days MLRs (Figure 5A) as well as in 48 h stimulations (Figures 5C and D).
Remarkably, QMAD inhibited IFNγ production by both CD45RO+ memory CD4+ and CD8+ T cells within the MLRs (mean% decreases of original values to 44% and 24.5% in CD4+CD45RO+ and to 63% and 31% in CD8+CD45RO+ cells, respectively, at 50 and 100 μg/mL, Figures 6A and B). When we isolated CD8+ CD45RO+ T cells by negative selection and stimulated them with B cell stimulators for 24 h ± QMAD (Figures 6C and D), we observed a dose-dependent inhibitory effect even in this T cell subset known to be resistant to many forms of immune suppression [26-28].
QMAD results in Foxp3+ expression by CD4+ T cells
To address whether QMAD impacts Treg we quantified the frequency of CD4+Foxp3+ T cells following 5 days stimulation of unfractionated T cells enriched from PBMC with anti-CD3/CD28 (without IL-2 or TGFβ) ± varying concentrations of QMAD. We observed a dose-dependent increase in the frequency of Foxp3+ CD4 T cells at the end of the culture period (Figures 7A and B, p < 0.05 at QMAD 100 μg/mL), without an effect on IL-2 production (using intracellular staining and flow cytometry, data not shown). To directly address whether QMAD induces Treg from naïve CD4 T cells, we enriched naïve CD45RA+ CD4 T cells, and stimulated with anti-CD3/CD28, IL-2 ± QMAD for 5 days. QMAD induced an increase in the percentage of Foxp3+ cells (Figure 7C), increased the total number of Foxp3+ cells (Figure 7D) and showed that the induced Foxp3+ cells suppressed Teff proliferation in vitro (Figure 7E). Addition of QMAD to induction cultures that included TGFβ increased generation of Foxp3+ suppressor cells over cultures containing TGFβ alone (p < 0.05, Figure S4).
QMAD does not directly induce T cell TGFβ or IL-10 but inhibits T cell AKT phosphorylation
ELISAs performed on culture supernatants from anti-CD3/28 stimulated, naïve CD4 cells showed that QMAD (+IL-2) did not induce T cell production of detectable TGFβ (Figure 8A) or IL-10 (4 individual experiments, data not shown), negating the possibility that QMAD induced Treg via amplification of an autocrine TGFβ/IL-10 loop. In contrast, when we measured phosphorylation of intracellular AKT (protein kinase B, PKB) by flow cytometry we consistently observed that QMAD lowered expression of pAKT in CD3 T cells compared to vehicle control (Figures 8B and C).
Identification of immunosuppressants from natural sou-rces has a proven track record in transplantation. Cyclosporine A was originally discovered in the fungus Tolypodadium inflatum, and sirolimus/rapamycin was first found in the bacterium Streptomyces hygroscopicus on Easter Island. Chinese herbs, which have been used for centuries to treat a wide array of ailments, represent an untapped source of potentially useful medications for use in humans. While previously published work by our lab and others showed that various extracts of Chinese herbs have favorable immunomodulatory effects on asthma and food allergy [14, 15, 23, 29-31], our current studies newly identify that a fraction of QM (D. superbus) has properties likely to favorably impact transplantation. QM has traditionally been used to treat urinary tract disorders [32-34] and “inflammation” [25, 35], can suppress IgE production by human B cells and can prevent peanut allergy in mice , but its efficacy as an inhibitor of pathogenic alloimmunity has not been previously documented. Our data show unequivocal effects of QMAD on inhibiting proliferation and IFNγ production by naïve and memory alloreactive T cells (Figures 5 and 6) while simultaneously facilitating Treg induction (Figures 7 and S4). The observed ability of QMAD to block proliferation and cytokine secretion by memory T cells is of particular interest, as memory T cells are generally resistant to immunosuppression and have been implicated as key mediators of allograft injury [36-38].
QMAD's simultaneous effect on the induction of Tregs is notable, in that many of the currently employed immune suppressants inhibit Treg , potentially limiting their long-term effectiveness. While QMAD augmented Treg induction in the presence or absence of recombinant TGFβ (Figures 7 and S4) the effects were more robust when TGFβ was present; it is likely that low levels of TGFβ known to be present in serum  is required.
Our data suggest that QMAD induces Treg via altering intracellular signaling that limits AKT phosphorylation, rather than by inducing T cell IL-10 or TGFβ. AKT is a central nidus of T cell signaling, downstream of the TCR and costimulation. When activated by phosphorylation, pAKT activates numerous substrates that exert a plethora of cellular effects . Included among the latter are enhanced T cell proliferation and survival, mediated in part by upregulating expression of the antiapoptotic molecule Bcl2 . Phospho-AKT also prevents Foxp3 transcription. Evidence indicates that prevention of AKT phosphorylation is required for induction and maintenance of the Treg phenotype . Thus, our observation that QMAD decreases pAKT in T cells provides a potential molecular link to account for the simultaneous inhibition of Teff while supporting Treg. Whether QMAD directly blocks phosphorylation of AKT, inhibits upstream signals that induce AKT phosphorylation (e.g. PI3K) and/or activates a phosphatase that dephosphorylates AKT [e.g. PHLPP ] remains to be determined.
While we have isolated the major immunosuppressive activity to the QMAD fraction, significant additional work will be required to identify the specific compound or compounds from within QMAD that mediate these effects. The HPLC analysis revealed three major peaks (Figure 3) with molecular weights of <600 Daltons each, as determined by mass spectrometry (data not shown). Based on the dichloromethane-based fractionation and isolation strategy that preferentially yields nonpolar, organic acid-rich compounds we believe the immunosuppressive molecules within QMAD are likely to be cyclopeptides, and that these differ from known immunosuppressants isolated from other “naturally occurring” sources, including cyclosporine A (MW 1203) and sirolimus (MW 912). Testing of in vivo immune suppression and potential toxicity will require compound purification.
One additional notable finding from our data is the proof of concept that ELISPOT-based testing can be employed as a high throughput screening approach for immunosuppressive drug testing (Figure 1). We rapidly screened more than 50 candidate compounds in a simple and ultimately informative functional T cell assay that guided us toward identification of a novel immune suppressant. Interestingly, while QMAD induced production of IL-10 in the screening assays (Figures 1-4) we did not detect IL-10 in culture supernatants of purified anti-CD3/CD28 stimulated T cells + QMAD, indicating that the QMAD's inhibitory effect on IFNγ production was not IL-10 dependent. The IL-10 in the screening assays likely derived from non-T cells within the PBMC (potentially monocytes).
In conclusion, we demonstrate herein that one well-known Chinese herb, QM, contains components that favorably alter alloreactive T cell immunity toward an immune suppressive phenotype potentially useful for preventing transplant rejection among other indications. These unique findings support the need for continued efforts to isolate and characterize the active compounds from within QMAD, to test their efficacy and mechanisms in vitro, and to assess their function as adjuvant immune suppressants in transplant recipients and potentially in patients with T cell-dependent, autoimmune diseases.
This publication was supported by the National Institutes of Health T32AI078892 (JRA), and in part by U01AI063594 awarded to PSH, and AT002644725-01, 2 R01 AT001495–05A1 and Winston Wolkoff Children's Holistic Medicine for Allergy and Asthma Foundation awarded to XML.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.