F344 (RT1^lvl) rats were purchased from the Animal Resource Centre (Murdoch, WA, Australia). Lewis (RT-1^l), PVG (RT1^c) and DA(RT-1a) rats were bred and maintained in the animal house, Liverpool Hospital. All animals were fed standard chow and given water ad libitum. The housing and experiments were in accordance with the Australian Code for the Care and Use of Animals for Scientific Purposes and approved by the Animal Ethics Committee of the UNSW Sydney. Rats that received standard care in the animal house and not given any treatment or alloantigen were considered naïve.

F344 male rats of 200g or more were anesthetized with isoflurane and grafted with heterotopic adult Lewis hearts from 180-230g donors, as described (40). Graft function was monitored daily during the treatment period then two to three times per week. Graft function was scored as 4. for a strong and fast beat similar to an isograft, 3. for mild graft swelling and slowing of graft contraction, 2. for moderate swelling and slowing of graft heartbeat, 1. for marked swelling and slowing of contraction, 0.5. for marked bradycardia and minimal and variable contraction, and 0. if no beat was detected. Total rejection was defined as a score of 0.5 or 0 observed for 10 days. In some tolerance models, graft function can improve days after what appears to be complete rejection. Thus, we observed graft function for weeks after major rejection. Some animals were sacrificed at the end of rIL-5 treatment for histology, as described (17).

Rat rIL-5 and rat rIL-4 were produced as supernatant from a transfected CHO-K1 cell line that was cultured in serum free medium and activity assessed in bioassays as described (41, 42). Supernatant was concentrated and rIL-5 quantified in a bioassay using the IL-5 dependent cell line B13 (a gift of Dr C. Sanderson, Curtin University, Perth WA, Australia), as described (43–46). 5000 Units of rIL-5 in 0.5 ml was given ipi as a daily dose. 5000 units of rIL-5 per day is well tolerated by rats, induces Ts2 cells to reverse autoimmunity and induces eosinophilia (22).

To deplete CD25⁺cells, the mAb NDS61 (gift of M Dallman, Imperial College London, UK) was given ipi at 7mg/kg daily from day 3 to 17 post-transplantation (30, 47). To block IL-4, 7mg/kg MRCOX81 (gift of N Barclay, Sir William Dunn School of Pathology, Oxford, UK) was given ipi on days 3-8 post-grafting, then every second day until day 15, as described (30, 41). These mAb were produced as described (15).

Five groups of F344 rats with heterotopic Lewis heart grafts (n=4-5) were studied and animals were monitored for heart allograft function. A sham treated group received saline injections daily from Day 7-16 post-transplant and four groups were given rIL-5 daily for 10 days from day 7 to 16 post-transplant. One of these four groups, the short-term rIL-5 treated, had rIL-5 therapy stopped after day 16, this group was repeated three times with results of all animals combined (n=12). Another rIL-5 treatment group, the long term treated group, had rIL-5 therapy continued as three times a week after the day 16. For the remaining two groups that received rIL-5, one was also treated with anti-CD25 mAb and the other with anti-IL-4 mAb, as described above. Some animals were used for histology of the heart graft and/or collection of spleen and lymph node cells for enrichment of CD4⁺CD25⁺T cells for MLC. At the end of experiments, at about 60 days post-transplant, all graft recipients in groups 1, 2 and 3 were sacrificed for FACS, RT-PCR and MLC studies on enriched CD4⁺CD25⁺T cells.

Donor heart sections were paraffin fixed and stained with hematoxylin and eosin. The histology images shown in Figure 5B were taken by a Leica DFC 450C camera with 20x magnification on a Leica DM 2000 LED microscope as we have described in Figure 5B legend. For quantification of areas of myocyte necrosis and mononuclear cell infiltration these paraffin sections were assessed in multiple images taken at 400x magnification on a Zeiss Axioscope A1 microscope (Zeiss, North Ryde, Australia). Image Pro Plus 6.2 software (Media Cybernetics, Rockville, MA) was used to estimate the area of myocytes necrosis and mononuclear cell infiltration, which were expressed as pixels per high power field (HPF).

Immunohistology was performed on 5μM sections of frozen heart allografts cut on a cryostat. Sections were air dried after fixation with acetone for 10min, then stained with a two-step indirect immunoperoxidase technique, as previously described (5, 48). The primary mAb used were W3/25 to CD4, MRCOx8 to CD8 (BD), FJK-16 to Foxp3 and ED1 to CD68 on macrophages (Abcam, Cambridge,cUK), as described (49). The secondary antibodies were HRP labelled anti-mouse Ig (Dako A/s, Glostrup, Denmark). Positive staining was assessed in multiple images taken at 400 X magnification on a Zeiss Axioscope A1 microscope. Image Pro Plus 6.2 software was used to estimate the area of positive staining and was expressed as pixels per high power field (HPF).

FITC labeled anti-rat mAb used were G4.18 (CD3), W3/25(CD4), MRCOx8 (CD8α), MRCOx39 (CD25, IL-2R alpha chain), MRCOx33 (CD45RA)(BD) and FJK-165 (anti-mouse/rat Foxp3) (eBioscience, San Diego, CA). Staining and analysis of lymphoid cells using a FACScan (BD, San Jose, CA) was as described (19, 50, 51).

Single cell suspensions from spleen and lymph node were prepared as described (50, 52) and RBCs were lysed with a buffer of 0.83% NH₄Cl, 0.1%KHCO₃ and 10mM EDTA at pH 7.2. Cells were re-suspended in PBS/0.4% BSA (MultiGel, Biosciences, Castle Hill, NSW, Australia). Spleen and lymph node cells from three or more animals were pooled to provide sufficient CD4⁺CD25⁺ cells for cultures.

An indirect panning technique was used to deplete CD8⁺T and B cells, as described (14, 50). Briefly, cells were incubated for 30 minutes at 4°C with optimized concentrations of MRCOx8 (an anti-rat CD8α mAb) and MRCOx33 (a rat CD45RA mAb that binds B cells and other cells but not T cells). All mAb were purchased from ThermoFischer. Cells were washed with PBS/0.4%BSA, re-suspended at 2x10⁷ cells/ml and incubated for an hour on Petri dishes (Greiner Bio-one, Kremsmuenster, Austria) coated with both rabbit anti-mouse Ig and rabbit anti-rat Ig (Dako). The unbound CD4⁺ cells were collected and incubated at 4°C for 20 min with PE conjugated MRCOx39 (BD) (an anti-rat CD25 mAb), then washed twice before 8μl/10₆ cells were incubated for 15 min at 4°C with of anti-PE microbeads (Miltenyi). Enriched CD4⁺CD25⁺ cells were then eluted through a LS MACS column (Miltenyi) and were re-suspended in RPMI 1640 media with 20% Lewis rat serum for culture. Cell subsets were subjected to immunostaining with mAb. Enriched cells were 97-99% CD4⁺ and 80-95% CD25⁺. 60-80% of these CD4⁺CD25⁺T cells were Foxp3⁺.

For RT-PCR and cell culture in MLC, CD4⁺CD25⁺ T cells were re-suspended in PBS/0.4%BSA.

Stimulator cells were prepared from irradiated (25 gray) thymus cells, as described^19. In each experiment parallel cultures with self (F344), specific donor (Lewis), third party (PVG) stimulator cells or no stimulator cells were performed. Cell culture medium was RPMI 1640 (GIBCO, Grand Island, NY) supplemented with 100 ng/ml penicillin, 100 U/ml streptomycin (Glaxo, Boronia, Victoria, Australia), 2 mM L-glutamine, 5x10^-5M 2-mercaptoethanol (Sigma), and 20% Lewis rat serum. 20% Lewis rat serum produces low background stimulation in autologous controls (19). Cultures with 5-6 replicates for each experimental group were set up in U-bottom micro-titer plates (Linbro, Flow Labs, VA) containing 2 x 10⁴ stimulators cells and 1 x 10⁵ CD4⁺CD25⁺ cells/well in a total volume of 200 μl. To assess the effects of rIL-5 on proliferation of these cells, 200 U/ml of rIL-5 was added to some cultures, as described (30, 32). Where stated CD4⁺CD25⁺T cells from naïve animals were cultured with rIL-4 (200 units/ml) as described (29).

Cells were cultured at 37°C in humidified air containing 5% CO₂ for 4 days, the peak of CD4⁺CD25⁺T cell proliferation (26). 0.5µCi ³H-thymidine (TRK-120, Amersham, Arlington Heights, IL) was added 16hr prior to harvesting with a Tomtec Cell Harvester (Flow Lab, Ayrshire, Scotland). Proliferation was assayed by adding liquid scintillation fluid before counting on a beta counter (1450 Microbeta Plus, Beckman Instruments, Palo Alto, CA). Each experiment has 5-6 replicates and results were expressed as cpm and presented as mean +/- standard deviation (SD). Counts of <400/min were considered within the range of background

The effect of rIL-5 on CD4⁺CD25⁺T cells proliferation in culture was calculated as a Stimulation Index using the formula: proliferation of cells with rIL-5 to a defined antigen divided by proliferation to the same defined antigen without rIL-5.

mRNA extraction from cells and reverse transcription to DNA were as described (21). Primers for rat Foxp3, Gata3, Tbet, Il2, Il4, Il5, Ifng, Ifngr, Il5ra, Il12rb2 and Gapdh were as previously reported (28–30, 53). The primers for Irf4 were F-TGTCCTCCGTGAGCTGTCT; R- CCTGGATCGGCTCCTCTATG, as described (49). The panel of molecules examined were selected for their relevance to Treg activation. Real-time PCR was performed as described (54) with a Rotorgene (Corbett Research) and SYBR Green I detection. Sensimix Taq polymerase (BioLine) was used according to manufacturer’s instructions. Copy numbers of each gene was derived from a known standard curve performed in parallel and normalized against Gapdh expression.

Parametric data were expressed as mean ± standard deviation. Results from repeat experiments were pooled, with replicates of ≥ 3 in each experiment. Means were compared using t test with GraphPad Prism (Graphpad Software Inc, La Jolla, CA). Statistical significance was set at p< 0.05.

To establish the changes in alloantigen activated Treg during Type 2- responses, we examined cytokines, cytokine receptors and transcription factors that are induced after naïve/resting CD4⁺CD25⁺T cells are cultured first with alloantigen and rIL-4 and later in a second culture with specific alloantigen and rIL-5. The experimental protocol is illustrated in Figure 2A . The hypothesis was that rIL-4 and alloantigen would activate naïve/resting CD4⁺CD25⁺Treg to Ts2 cells expressing IL-5Rα that would proliferate when stimulated by specific-alloantigen and rIL-5 (29, 30, 32) and develop into Th2-like Treg.

CD4⁺CD25⁺cells from naïve DA rats were cultured for 4 days with fully allogeneic PVG stimulator cells and 200 units of rIL-4 as described (29) to induce Ts2 cells. These Ts2 cells were washed and re-cultured with 200 units of rIL-5 and the same alloantigen to induce Th2-like Treg. Combined results from three separate experiments of RT-PCR of mRNA of these cells are shown in Figure 2B . The Th2-like Treg had increased expression of mRNA for the transcription factors Foxp3, Irf4, and Gata-3, but had no induction of tbet. il5 but not il4, il2 or ifng was induced in these cells. Compared to starting CD4⁺CD25⁺ cells from naïve rats where naïve/resting tTreg (CD4⁺CD25⁺Foxp3⁺Treg) forms a major part, Il5ra expression was increased in Ts2 cells, but this increase was not sustained in Th2-like Treg. We used expression of Irf4 and Il5 as markers of Th2-like Treg induction.

Our hypothesis is that during a rejection response, some Th2 cells will be activated to produce IL-4 that with donor antigen would activate naïve/resting CD4⁺CD25⁺Foxp3⁺Treg to Ts2 Treg as proposed in Figure 1 . Lewis heterotopic cardiac allografts into F344 rats are slow to reject as there is only a single class I MHC and multiple minor incompatibilities with no class II MHC incompatibility (27, 28). The model is delayed acute rejection with T cell activation and infiltration. The experimental plan is shown in Figures 3A and 4A . Graft function was scored using a semi-quantitative scale described in methods and mean heart graft function score are presented on y-axis ( Figures 3B , 4B ).

Rejection in sham treated hosts (n=8) caused a decline in graft function after day 10, with complete rejection by day 31 ( Figure 3B ). No rats in this sham treated control group recovered to have significant function, with graft function scores of 0.5 or 0.

Short-term rIL-5 treatment (n=12) was 5000 units ipi daily for 10 days between 7 and 16 days post-transplantation ( Figure 3A ). In the long-term treatment group (n=5), rIL-5 therapy was continued (ipi) three times a week immediately following the daily rIL-5 from 7 to 16 days post-transplant ( Figure 3A ). In both rIL-5 treated groups, graft function scores were higher than in sham treated rats at all time points beyond day 10 post-transplant (p<0.01)( Figure 3B ). rIL-5 treatment preserved graft function, with all grafts scoring ≥2 until cessation of rIL-5 treatment at day 16. The graft function score was significantly higher than sham treated group, p<0.01 at day 16 and p<0.001 at day 17 ( Figure 3B ).

Heart graft function in both rIL-5 treatment groups stabilized around 20 days post-transplant then improved. The group that received long-term treatment with rIL-5 therapy had more rapid improvement in graft function, with scores significantly higher than sham treated controls at all time points beyond day 22 (p<0.05). Compared to 10-day treatment group, the long-term rIL-5 treated group had higher graft function scores on day 35 (p=0.05) and 40 (p<0.01) ( Figure 3B ). By day 43, both rIL-5 treated groups had a mean graft function score of 3. At the end of monitoring on day 60, 3 of 5 long-term rIL-5 treated rats had an excellent graft function score of 4 and another rat had a score of 3 ( Figure 3B ). This level of heart graft function is consistent with operational transplant tolerance and similar to long-term syngeneic heart graft function in this rat allograft model, as observed in previous studies.

The rationale for rIL-5 treatment was to expand CD4⁺CD25⁺Foxp3⁺Treg that had been activated by IL-4 produced in the early rejection response. NDS61, a mAb to rat CD25, depletes CD4⁺CD25⁺T cells in rats (47) and prevents rIL-5 treatment inhibiting autoimmune responses (30, 32). To demonstrate a role for CD25⁺ cells, we depleted these cells by ipi of NDS61 daily from 3 to 17 days post-transplantation, as illustrated in Figure 4A . Hosts depleted of CD25⁺T cells and treated with rIL-5 rejected their allografts faster, with all allografts fully rejected at day 19 (n=4) ( Figure 4B ). No graft function was detected in any animal treated with NDS61 and rIL-5 after day 20 and there was no recovery in graft function in the next 10 days. Graft rejection was more severe in anti-CD25mAb and rIL-5 treated rats ( Figure 4B ) than in sham treated from day 14 to 19 (p<0.05) ( Figure 3B ). This suggests CD25⁺cells are activated during rejection and slow the progress of rejection.

MRCOx81, a mAb that blocks IL-4 (40, 41), was administered daily from day 3-8 then on days 10,12,14 post-transplantation, as illustrated in Figure 4A . Anti-IL-4 mAb treatment also led to accelerated rejection and abolished the effect of rIL-5 treatment on allograft survival (n=4) ( Figure 4B ). All rats totally rejected their heart grafts by day 17 and there was no recovery in graft function over the next 10 days. All MRCOx81 treated rats rejected faster than sham treated controls ( Figure 3B and Figure 4B ). MRCOx81 and rIL-5 treated group had significantly lower graft function scores ( Figure 4B ) compared to those from rats treated with rIL-5 alone, on all monitoring days from day 11 (p<0.01)( Figure 4B ).

The experimental protocol for obtaining Lewis heart graft tissue from F344 rats for histology is illustrated in Figure 5A . Heart grafts from rIL-5 treated rats taken at day 16 post-transplant had good cardiac myocyte preservation with scattered mononuclear cells infiltration ( Figure 5B ). In contrast, grafts from sham treated rats had wide-spread myocyte necrosis and large infiltrates of mononuclear cells. Additional examples are in Supplementary Figure 1 . Donor Lewis hearts in F344 recipients treated with rIL-5 and either anti-CD25 or anti-IL-4 mAb had massive areas of myocyte necrosis with dense infiltrates of mononuclear cells, however these heart grafts were taken at the end of the experiment at day 30 not at day 17 post-transplant ( Figure 5B ).

Image analysis of donor hearts showed the pixels occupied by necrotic myocytes was less in rIL-5 treated 139,475 ± 35,078 than in sham treated controls 474,969 ± 154,423 (p=0.00011); anti-IL-4 mAb and rIL-5 treated 436,217 ± 138,148 and anti-CD25 mAb plus rIL-5 treated 536,889 ± 272,577 ( Figure 5C ). The area of mononuclear cells measured by pixels was less in rIL-5 treated 193,883 ± 108,701 than in sham treated controls, 311,4112 ± 124,968 (p=0.03), anti-IL-4 mAb and rIL-5 treated 269,521 ± 35,636 and anti-CD25 mAb plus rIL-5 treated 249,281 ± 102,820 ( Figure 5C ). The mAb treated animals grafts were collected two weeks longer post-transplant and had established rejection, thus this data is not directly comparable.

Characterization of the mononuclear cell infiltrate in heart grafts using immunostaining with mAb, ( Figure 5D ), showed that compared to grafts from sham treated hosts that were rejected, rIL-5 treated grafts had significantly fewer CD8⁺ cells (p=0.02), CD4⁺ cells (p<0.05) and Foxp3⁺ cells (p= 0.05). 44% of CD4⁺ cells in IL-5 treated expressed Foxp3, whereas 41.9% expressed Foxp3 in rejected grafts. Representative sections are shown in Supplementary Figure 2 . There was no difference in the infiltrate of ED1⁺ macrophages between sham treated and rIL-5 treated.

Thus, rIL-5 treatment preserved the heart graft from injury and markedly reduced the mononuclear cell infiltrate compared to grafts from sham-treated rats. The benefits of rIL-5 treatment were abolished by treatment with anti-CD25 mAb or anti-IL-4 mAb ( Figures 5A–D ).

The source of lymphocytes for these studies is illustrated in Figure 6A . We examined CD4⁺CD25⁺ cells from graft bearing hosts to examine if IL-5 administration resulted in in vivo activation of Ts2 cells and/or Th2-like Treg as assessed by in vitro proliferation with rIL-5 ( Figures 6B–D ) and RT-PCR of key markers ( Figures 7A, B ).

Spleen and lymph node cells from rats treated with rIL-5 for 10 days were examined either at the end of rIL-5 treatment on day 16 ( Figure 6B ), or on day 66 ( Figure 6C ) post-transplantation. Cells from sham treated F344 rats with Lewis heart graft were assessed at 56 days post-transplantation ( Figure 6D ). The proportion of CD4⁺CD25⁺cells in rIL-5 treated rats was 6.3-7.8% ( Figures 6B, C ) compared to 4% in sham-treated controls ( Figure 6D ). Foxp3⁺ cells were 2.8% - 4% in rIL-5 treated rats and 6.6% in sham-treated controls (data not shown).

The enriched CD4⁺CD25⁺ cells in all three groups were 80-88% CD4⁺CD25⁺cells ( Figures 6B–D left panel), 94-99% CD3⁺, <2.4% CD8⁺, and 61-70% Foxp3⁺ cells (data not shown). This is within the standard enrichment of murine Treg using CD25. Thus, 30-40% of cells in the enriched CD4⁺CD25⁺T cells were Foxp3^- and not necessarily T regulatory cells. They may include activated effector CD4⁺T cells.

Enriched CD4⁺CD25⁺ cells were tested for proliferation in MLC ( Figures 6B–D , middle and right panel) and their mRNA tested by RT-PCR ( Figures 7A, B ). Enriched CD4⁺CD25⁺T cells from rIL-5 treated hosts taken 16 days post-transplant ( Figure 6B , middle panel), in absence of rIL-5 in culture, had a greater response to Lewis, than to self (F344) or third party. Such proliferation to graft alloantigen suggested increased numbers of cells activated by graft alloantigens. rIL-5 in cultures partially enhanced responses to specific donor Lewis, but this was not significant, as seen in the Stimulation Index ( Figure 6B , right panel). rIL-5 in culture also did not enhance the response to self (F344) or third party (PVG) ( Figure 6B , middle and right panel).

Enriched CD4⁺CD25⁺ cells (>88%) from F344 rats with Lewis heart grafts, treated for 10 days with rIL-5, at day 66 post-transplant had no proliferation to specific-donor in the absence of rIL-5 ( Figure 6C , middle panel). This was consistent with our previous observations that Treg from rats with transplant tolerance do not proliferate to specific-donor in the absence of key cytokines (26, 31). rIL-5 in culture enhanced their proliferation to specific-donor Lewis with a Stimulation Index that was signifiantly greater than that to self (p<0.01) and third-party PVG (p<0.001) ( Figure 6C right panel). Proliferation to self or to third-party was not enhanced by addition of rIL-5. ( Figure 6C , right panel).

CD4⁺CD25⁺ cells from sham-treated hosts, alone or with rIL-5 had no proliferation to specific donor, self or third party simulators ( Figure 6D , middle panel). These animals had rejected their grafts and would not be expected to have activated Treg surviving 56 days post-transplant. Cultures of CD4⁺CD25⁺ cells from rats receiving long-term rIL-5, taken at day 60 post-transplantation, failed due to malfunction of an incubator and could not be repeated due to animal ethics issues.

RT-PCR of mRNA from CD4⁺CD25⁺T cells from peripheral lymphoid tissue of heart grafted animals taken on day 16 at the end of treatment for 10 days with rIL-5 was performed. Controls were CD4⁺CD25⁺ cells from naïve F344 animals that had not been transplanted with a heart graft and had no treatment ( Figure 7A ). CD4⁺CD25⁺T cells from heart graft recipients treated with rIL-5 had significantly greater expression of foxp3 (p<0.001), Irf4 (p<0.05), Il5 (p<0.001), and higher Il5Ra (not significant) compared to CD4⁺CD25⁺T cells from naive F344 rats. These findings suggested that rIL-5 treatment activated Th2-like Treg, which expressed Irf4 and Il5. There was also induction of Th1-like Treg markers Ifngr (p<0.05), Il12rb2 (p,0.001) and Ifng (p<0.01), showing Th1-like Treg were also present.

CD4⁺CD25⁺ cells from rIL-5 and sham treated rats were also compared at around 60 days post transplantation ( Figure 7B ). CD4⁺CD25⁺ cells from long-term rIL-5 treated rats on day 60 post-transplantation, expressed more mRNA for Foxp3 (p<0.01), Irf4 (p<0.05) and Il5 (p<0.01) than those from sham-treated rats with heart grafts, at 56 days post-transplantation. Sham treated rats did not receive rIL-5 and had rejected by day 31 post-transplant. The cells from short-term rIL-5 treated group (66 days post-transplantation), also had an increase in Foxp3 (p<0.001), Irf4 (p<0.01) and il5 (p<0.001) compared to sham-treated at day 56 post-transplantation. Expression of Th1-like Treg marker Ifng (p<0.01) was also increased, but Ifngr and Il12rb2 were not ( Figure 7B ).