The role of second signals delivered through B7/CD28 interactions in T cell activation is well documented. However, once CTLs are elicited, TCR-mediated cytotoxicity appears to be uncoupled from the requirement for costimulatory signals. In this study, we show an uncoupling across a broad range of concentrations of peptide, thus demonstrating that cytolysis is a TCR-mediated response that is fully independent of costimulatory signals. However, the same T cell effectors remain fully responsive to B7 engagement, which is able to amplify Ag-mediated proliferation and cytolytic capacity. B7 expression by targets results in an IL-2-mediated proliferative expansion of the effectors concurrent with the elimination of the targets. Thus, costimulation of effectors results in a vast expansion in lytic units over time, which does not occur in the absence of IL-2 or B7. Both TCR-derived and second signals appear to be necessary to achieve this result. These results suggest that B7-expressing APC or a cohort of IL-2-producing helper cells would functionally extend the duration and effectiveness of the cytotoxic response occurring in localized immune responses.
The importance of costimulatory signals in the activation of naive T cells is now well established (1, 2). Signals through B7/CD28 interactions appear to result in increases in TCR-induced proliferation (3, 4), cytokine production (5), and a concomitant lowering of the threshold for MHC/Ag concentrations required to achieve an activated state (6, 7). In addition, studies using CD4+ T cell clones have suggested that these signals prevent the induction of clonal anergy in T cells receiving antigenic signals (4, 8). Finally, these signals appear to augment T cell survival through the modulation of the expression of various members of the bcl family (9, 10). The requisite role of costimulatory signals in the function of effectors is somewhat less clear.
Based on the phenotype of CD28-deficient animals, it appears that, in vivo, these molecules serve primarily to prolong or amplify the initial response of T cells to Ag (7, 11, 12). Single injections of immunogen into CD28-deficient animals resulted in cytotoxic killing identical to wild-type controls when assayed 1 day after injection, but greatly impaired function when assayed after 3 days (12). Similarly, the proliferative responses of CD28-deficient T cells to their cognate Ag in vitro was similar to wild-type T cells at early time points, but was only weakly sustained over time (11). Together, this suggests that CD28 costimulation plays a key role in prolonging the T cell proliferative response.
Studies on the effect of B7 costimulation on CD8+ T cell effector function have suggested that B7 interactions are not necessary once T cell effectors have been generated. In naive CD8+ T cell precursors, costimulation lowers the level of Ag required to achieve an activated state, much as for CD4+ T cells (13). Further in vitro studies have clearly demonstrated that elicited effectors are capable of killing B7-negative targets, although B7 expression or exogenous IL-2 was necessary for the elicitation of the effectors from unactivated precursors (14, 15, 16, 17). In vivo experiments have also demonstrated that immunization with costimulation-competent tumor cells results in cross-protection of mice against B7− parental tumor challenge (18, 19, 20). Taken together, this has suggested that B7-mediated costimulation is needed for the elicitation of effectors but is not required to sustain cytotoxic effectors.
In a somewhat puzzling contradiction, a number of recent models have presented in vivo examples in which B7-negative targets escape cytotoxicity, while their B7-positive counterparts are efficiently depleted (21, 22, 23, 24). This has suggested that T cell effectors might indeed remain sensitive to costimulation, even once elicited. In one study, immunization with B7-positive transfectants resulted in protection against subsequent challenge with B7-positive tumors, but not against B7-negative tumors (22), and selective outgrowth of B7-negative tumor variants appears to occur in animals initially given B7-positive tumors (21). In addition, animals bearing both Ld skin grafts and Ld tumors rejected the skin grafts while ignoring the tumors. This tumor outgrowth could be reduced by transfection of the tumor cells with B7-1 and CD48 (23).
We have recently described C57BL/6 transgenic mice that expressed B7-1 in either all or only some of their pancreatic islet β cells (“confluent” or “patchy” rat-insulin promoter (RIP)3-B7-1 mice) (24). The islets of these mice were normal, but aged mice (>225 days) developed spontaneous autoimmune diabetes at significant frequencies (10%). Although all of the β cells in islets with “confluent” B7-1 expression were killed, there was selective killing of B7-1+ β cells in islets with “patchy” expression, and B7-1− islet cells survived or regenerated. These results are analogous to those described above for tumor cells expressing B7-1 and indicate a requirement for B7-1 costimulation during the effector (tissue localized) phase of the T cell response. Further support for this idea was observed when transferring CD8+ T cells, specific for an OVA-derived peptide, into mice that did or did not coexpress B7-1 with OVA in their β cells. The number of specific T cells needed to elicit diabetes was reduced by 100-fold if B7-1 was also present.
In this study, we have reexamined the role of B7-derived second signals in vitro in elicited cytotoxic effectors. Ag-induced cytotoxicity by these effectors is independent of B7, even across a broad range of peptide concentrations. Thus, the actual Ag-mediated signals leading to the killing of targets appears to be entirely uncoupled from these second signals. However, in following the fate of these effectors beyond the typical 4-h 51Cr release assay, we observed that effectors remained highly sensitive to costimulatory signals for both further proliferation and, in part, for the maintenance of their cytotoxic capacity. As a result, B7 expression during cytotoxicity results in a long-term increase in cytotoxic potential. The proliferation-promoting activity can be mediated by either B7-1 or B7-2 and can be substituted by addition of exogenous IL-2. We suggest that, although B7-derived second signals are not directly necessary for cytotoxicity, they may nevertheless play an important role in regulating and extending cytotoxicity in vivo.
Materials and Methods
OT-I TCR transgenic mice producing CD8+ T cells with specificity for OVA-SL8 (SIINFEKL) (25) were bred on a RAG1−/− C57BL/6 background. C57BL/6 were obtained from the breeding facility in Kew (Victoria, Australia). All experimental mice were between 6 and 12 wk of age.
Abs and reagents
Abs used included anti-B7-1 from clone 1610A (27), anti-B7-2 from clone GL1 (28), anti-Vβ5 from clone MR9.4 (29), anti-CD25 from clone PC61 (30), anti-IL-2 from clone S4B6 (31), anti-CD44 from clone IM7.81 (32), and anti-CD62L from clone Mel-14 (33). Where indicated, FITC conjugates of these Abs were prepared in our laboratory, and anti-CD8.PE was purchased from PharMingen (San Diego, CA). OVA peptide SL8 (SIINFEKL) was a kind gift of Dr. Frank Carbone (Monash University, Melbourne, Australia).
A total of 2 × 105 cells was suspended in 100 μl ice-cold PBS/1% calf serum/0.05% sodium azide. Abs were added for 30 min on ice, followed by a single 4-ml wash in PBS/calf serum/NaAzide. Data was acquired on a Becton Dickinson (Mountain View, CA) FACScan, and the LYSIS II program was used to electronically gate on relevant populations.
Generation of OVA-specific effectors
Peripheral lymph nodes were isolated from RAG1−/− OT-I mice and viable lymphocytes were isolated by mincing, followed by density centrifugation of the resulting cell suspension over Ficoll 1.119 (Sigma). Stimulators were irradiated (1500 Rad) spleen cells from unprimed C57BL/6 mice, which had been pulsed with 100 ng/ml SL8-peptide for 30 min at 37°C, followed by extensive washing. Bulk activation cultures consisting of 5 × 106 OT-I lymph node cells plus 5 × 107 irradiated SL8-pulsed spleen cells in 50 ml of complete RPMI-10 (containing 10% FCS, 50 μM β-mercaptoethanol, 2 mM glutamine, and 50 μg/ml gentamicin) were incubated for 3.5–4 days at 37°C, 5% CO2. Viable effectors were isolated from these cultures by density centrifugation over Ficoll 1.119, followed by two washes in PBS. Flow cytometric analysis indicated that these cells were routinely >99% Vβ5+ CD8+, and typical preparations yielded a 3- to 6-fold increase in OT-I T cell number after the 3-day incubation period. Expression of markers, such as CD25, CD62 ligand, and CD44, as well as increased cell size and granularity, indicate that all recovered cells had encountered Ag. These priming conditions were used for all experiments shown here.
Primary cytotoxicity assays
EL4, EL4B7-1, or EL4B7-2 (5 × 106/ml) were incubated with (51Cr) sodium chromate and 100 μg/ml (or the indicated concentration) of SL8-peptide for 1.5 h, followed by extensive washing. These targets were plated at 104/well in round-bottom 96-well dishes, and serial dilution of effectors were added to give a final volume of 200 μl. The plates were incubated for 4 h or 16 h at 37°C, 5% CO2, the plates were subsequently centrifuged, and supernatants (100 μl/well) were harvested for counting. The percent specific lysis was calculated as: [(experimental lysis) − (spontaneous release)]/[(maximal lysis) − (spontaneous release)] × 100, and SEs were derived from the SDs of duplicate samples. Spontaneous release and maximal release were determined in the presence of RPMI-10 and 2% Triton X-100, respectively. High rates of spontaneous release after 16 h (>30% of maximal) confounds the analysis, as experimental lysis must be significantly greater than this spontaneous release to be significant (see also Ref. 34). This is particularly a problem at low E:T ratios where experimental lysis is low.
Proliferative response to targets and secondary cultures
SL8-pulsed EL4 cells were irradiated (10,000 Rad) and prepared identically to the cytotoxicity assay with the omission of the 51Cr labeling. (If nonirradiated targets were used, the experimental outcome was similar, except when targets were not loaded with SL8-peptide. This resulted in increased background [3H]thymidine incorporation and cell expansions due to the additional unrestricted proliferation of the EL4 cells.) Effectors (105/well) and targets (104/well) were mixed and plated identically to cytotoxicity assays and, where indicated, human rIL-2 (100 U/ml; Chiron, Emeryville, CA), rIL-7 (5 U/ml; Immunex, Seattle WA), anti-B7-1, or anti-IL-2 (10 μg/ml S4B6) were added at the beginning of the culture. For proliferation analysis, wells were pulsed after 16 h with 1 μCi of [3H]thymidine and incubated for an additional 8 h before harvesting and counting. For analysis of viability and use in secondary cytotoxicity assays, viable cells were reisolated from these cultures after 16 h by density centrifugation over Ficoll 1.119, followed by washing in PBS. Viable cell counts were performed in duplicate from the resulting populations using trypan blue exclusion. Values were converted to percent input by the formula: % input = [(viable count after 16 h)/(initial viable cell count)] × 100. SDs of means were propagated to derive the SE for the percent input.
Secondary cytotoxicity assays
Effectors derived from secondary cultures were tested for cytotoxicity in an identical manner to the primary cytotoxicity assays. Where indicated, data is represented as lytic units 20 (LU20), where one lytic unit is the number of cells derived from the secondary culture necessary to achieve 20% specific lysis of targets. This measure is most amenable for data, such as ours, in which some populations are not efficient killers. This value was obtained by interpolation or extrapolation using the slope of the graphical data of the specific lysis in the region of interest. The total number of lytic units was obtained by the formula: [total no. of effectors obtained after the secondary culture]/[no. of effectors required to achieve 20% lysis (LU20)]. The error for this value was derived from the error in cell yield combined with the error in extrapolation (derived from the maximum and minimum slopes of lines through the mean ± SDs of the percent specific lysis in the specific region of curve).
Different requirements for B7 costimulation by effector T cells for cytotoxicity and proliferation
To assess the role of B7 in regulating the cytotoxicity of CD8+ T cells, we reexamined the cytotoxic potential of effectors on B7+ and B7− targets. In this study, we used CD8+ T cells from OT-I TCR transgenic mice (25) as a source of effector cells. These expressed a Vβ5/Vα2 TCR specific for OVA peptide SL8. As shown in Fig. 1, the effectors generated in vitro by incubation with peptide-pulsed splenocytes were uniformly bright for Vβ5 (indicative of the TCR transgene), were large in both forward and side scatter, and were CD25high CD62Llow, CD44high, indicating the isolation of a homogeneous population with high levels of the molecules characteristic of activated effector cells. As shown in Fig. 1 B, the target populations for this study all expressed similar levels of the restricting element, H-2Kb, but only the B7-1 transfectants expressed detectable levels of B7-1, and only the B7-2 transfectants expressed detectable levels of B7-2.
When target populations were pulsed with high concentrations of OVA peptide and loaded with 51Cr, the OT-I effectors lysed the B7-1+ and B7-2+ targets with similar efficiency in a standard 4-h cytotoxicity assay (Fig. 2, A and B). However, when these assays were repeated over a 16-h release period (Fig. 2, C and D), we observed significant differences in killing for B7-negative and -positive targets, as assessed by comparing the SDs of the means. As expected, we observed saturating levels of lysis at lower E:T ratios under these conditions, as compared with 4-h incubations. However, as effector numbers decreased below this saturation level, a clear 1.5- to 5-fold difference in lysis of B7-positive and -negative targets was observed. While technical problems associated with the spontaneous release over a 16-h Cr-release period make it difficult to accurately measure lysis at low E:T, even more distinct effects are seen in this range for B7+ vs B7− targets.
One possible explanation for these results would be a B7-mediated shift in the dose-response curve for Ag. To examine whether the B7-1 expression by targets was a neutral event for the unleashing of CTL cytotoxicity across the entire dose-response, we pulsed the target populations with varying quantities of SL8-peptide and used these as targets in a 4-h Cr-release assay. As shown in Fig. 3,A, the elicited effector population killed B7− and B7+ targets equally, even when low concentrations of peptide were loaded onto targets; implying that B7-derived costimulation was not acting to alter the threshold of Ag-concentration required for lysis. When we assayed for the ability of the OT-I effectors to proliferate in response to these MHC/Ag signals from the targets, we observed that the effectors remained sensitive and dependent on B7 signals for proliferation (Fig. 3 B). When targets were pulsed with the SL8 peptide, the B7-1-transfected EL4 yielded 5- to 7-fold greater levels of proliferation, and B7-2-transfected EL4 yielded 5- to 10-fold greater levels of proliferation relative to targets lacking these costimulatory molecules. In the following studies, we focused on B7-1, as no obvious qualitative distinctions were observed between these ligands.
B7-1 engagement during target lysis induces proliferation via an IL-2-dependent mechanism
One obvious explanation for the proliferative effect is the production of autocrine IL-2. In Fig. 4,A, the effect of exogenous IL-2 and B7-1-expression on proliferation of effectors was examined. Proliferation of effector OT-I T cells was limited, except under those conditions when targets expressed B7-1 or when exogenous IL-2 was added. The interaction with a B7-1+ target resulted in a 10-fold increase in thymidine incorporation, and the addition of exogenous IL-2 had similar effects. The B7-1-dependent effect was mediated mostly by the autocrine production of IL-2, as the addition of saturating quantities of anti-IL-2 Abs resulted in an ∼70% reduction in the B7-1-mediated proliferation (Fig. 4 B). The proliferation was not due to B7 engagement alone, as EL4-B7-1 cells (lacking OVA peptide) failed to induce detectable proliferation. This costimulation appeared to occur through CD28 and not CTLA-4, as blocking monovalent Ab Fab fragments to the former were capable of modulating the effect, while similar Fab fragments against the latter had no effect (data not shown).
To observe the effects of costimulation on the effector population, we enumerated the number of viable CD8+ T effector cells generated by a 16-h exposure to targets. As shown in Fig. 4 C, we observed a 50–100% increase in cell numbers during the 16-h culture period when B7-1-bearing effectors or exogenous IL-2 were present in the secondary culture. In contrast, culture with peptide-pulsed B7-negative targets resulted in only 60% of the input number of effectors being recovered, which was similar to the number obtained with no added effectors or unpulsed effectors alone. The addition of exogenous IL-2 to effectors plus B7-negative targets resulted in a restoration of the expansion, although, as noted below, these cells had less cytotoxic potential. The effector populations were >98% OT-I Vβ5+, CD8+ following the exposure to targets and reisolation over Ficoll, as assessed by FACS (data not shown).
B7-1 engagement during target lysis maintains cytotoxic effector functions and results in higher lytic capacity
The data obtained thus far indicated that interactions with B7-1+ targets, in contrast to B7-1-negative targets, led to an expansion in the number of effectors. We next sought to determine whether the costimulatory signals affected the cytotoxic capacity of the effectors. To achieve this, we restimulated our effectors with conditions identical to those for Fig. 4, reisolated OT-I effectors, and tested the ability of these cells to lyse fresh Ag-pulsed B7-1− targets. As shown in Fig. 5 A, effectors stimulated with B7-1-positive targets lysed subsequent targets with high efficiency. This efficiency was similar in magnitude to the lysis achieved with freshly elicited effectors (data not shown). Similarly efficient lysis was observed by those effectors whose primary encounter with B7-1− targets was supplemented with exogenous IL-2. However, T cell populations reprimed on B7-negative targets lysed targets with 25–50% the efficiency of those that were exposed to B7-1-positive targets. Lysis by effectors that had received either no stimulation or unpulsed targets for 16 h before the secondary encounter was further reduced. Notably, only intermediate levels of lysis were observed from effectors that had been cultured in IL-2 alone, indicating that antigenic encounter, in addition to division, was necessary for the high level of effector function. The cells given IL-2 alone also began to assume a less-blast-like and less-granular appearance consistent with a requirement for continued Ag encounter to maintain cytotoxic capabilities (data not shown).
When the data from Figs. 4,C and 5A were combined to represent the resulting lytic capacity of the effectors, the differences in the overall lytic potential following the various encounters became even more prominent. As shown in Fig. 5 B, 6-fold higher lytic units resulted following 16 h of stimulation with B7+ targets, as compared with B7− targets. Addition of IL-2 to peptide-pulsed targets produced results similar to those observed with B7+ targets. In contrast, exposure of effectors to IL-2 alone resulted in less lytic capacity than when both Ag and costimulatory signals were present, due in great part to the loss of cytotoxic potential on a per-cell basis.
This report provides evidence that elicited effectors require stimulation through B7 interactions or the addition of exogenous cytokine stimulation to continue expanding and to maintain high levels of cytotoxicity. In agreement with previous reports, we observed an absence of B7 discrimination for cytotoxic killing of targets. We have extended this observation with the formal demonstration that cytotoxicity against targets appears to depend only upon the concentration of Ag expressed. This data supports a theory in which cytotoxic effectors primed by “professional” (i.e., B7-expressing) APC may be subsequently unleashed on virally infected or other non-B7-expressing (e.g., tumor cells) (14, 15, 19). However, this study provides additional evidence that the cytotoxic response remains subject to regulation by costimulatory signals at the effector phase.
Our study demonstrates that cytotoxicity and proliferation are differentially regulated by costimulatory signals. The cytotoxic function of T cell effectors requires only recognition of MHC/Ag, whereas second signals are needed to generate proliferation and accompanying long-term cytotoxicity. Although costimulation has been previously suggested to result in a shift in the TCR dose response (6), our study shows that cytotoxicity is fully uncoupled from B7-mediated costimulation. Any effect of costimulation on direct killing would require integration of CD28/B7 signals with those necessary for cytotoxic granule release. The generation of calcium signals from intracellular stores appears to be the earliest form of calcium mobilization and appears necessary for perforin granule release (35, 36, 37). Downstream from this calcium release, the generation of calcium signals from extracellular stores is necessary to generate the de novo synthesis of Fas ligand (37). The initial calcium signals have been shown to be generated by TCR/MHC/Ag signals alone (38, 39), and, indeed, B7/CD28-mediated costimulation of proliferation and IL-2 production occurs without influencing any detectable aspect of calcium release (40, 41, 42). It is perhaps not surprising, therefore, to find that the effects just after calcium release (such as killing in the standard 4-h cytotoxicity assay with elicited effectors) would be independent of B7 engagement.
Our dose-response data using varying concentrations of OVA peptide suggests that B7 is not capable of shifting the dose response for cytolysis, though it does function to augment the proliferative response. In support of our conclusion, other studies have demonstrated that cytolysis of targets bearing weak peptide variants is identical in the presence and absence of CD28 (7). As noted previously, most other early responses, such as TCR down regulation (43) and early calcium mobilization (40, 41, 42), are independent of B7/CD28 engagement. Nonetheless, it remains possible that very borderline signals might still be modulated by costimulatory signals to produce cytolysis.
The amplification in the overall lytic capacity of effectors that we observed was due to two separate factors. As previously discussed, the engagement of costimulatory receptors augmented the actual number of effectors generated by target encounter. In addition, our data suggests that the maintenance of cytotoxicity in effectors requires the continuous provision of both Ag-dependent and costimulatory signals. In the absence of these signals, the ability of elicited effectors to lyse targets over time was compromised. Notably, the supply of IL-2 can substitute for Ag- and B7-mediated signals for proliferation of effectors, but, nonetheless, results in the loss of cytotoxic capacity. It remains probable, therefore, that IL-2 alone is not sufficient for maintaining CTL activity. Overall, this data implies that the maintenance of a cytotoxic response would require the provision of a continuous exposure to antigenic stimulation and either expression of B7 by a cohort of cytotoxic targets or exogenous supplies of paracrine IL-2 and additional factors.
Throughout these experiments, it has appeared that B7 was either neutral for effectors or else produced positive effects (proliferation) upon the T cell responses under study. Although B7 is capable of binding to both stimulatory CD28 receptors and inhibitory CTLA-4 receptors (44), it remains possible that even more complex B7 interactions may occur in other stages of effectors. It is currently poorly understood how CTLA-4 and CD28 compete for B7 occupancy, but it remains a possibility that CTLA-4 functions to regulate CD8 T cell effectors in some as-yet-undetermined capacity. Recent evidence from CTLA-4 knockouts has implied that the predominant defect in CTLA-4 deficient mice lies in the regulation of CD4+ T cell responses and less so in the regulation of CD8+ T cells (45). Our experiments are consistent with this finding, as we have only observed positive effects of B7 engagement on CD8+ T cell effectors.
Perhaps the most important implications of this study are for the lysis of targets during ongoing immune responses. Studies of the presentation of tissue Ags in vivo have suggested that initial priming of effectors occurs by Ags trafficking to the local draining lymph nodes (46). For cytolysis of targets, these activated effectors would then be required to traffic to the affected tissue. Our data suggests that T cell effectors, though capable of killing a variety of targets, are nevertheless quite sensitive to their costimulatory environment during this effector phase. Based on the data provided here, it would be predicted that a large number of targets (relative to the number of effectors initially elicited in the draining lymph nodes) might be incapable of lysing the entire target population without significant concurrent costimulation. In the absence of such costimulation, CTL activity might dissipate, and the significant numbers of targets may remain. In contrast, small numbers of B7-negative targets might be effectively eliminated by the initial cohort of elicited effectors and, thus, might not require the presence of costimulation for their effective removal.
It is notable that such a prediction is at least consistent with the data generated in vivo. For example, pancreatic tissue expression of IL-2 or B7-1 costimuli results in early and increased levels of destruction of pancreatic islet β cells resulting in diabetes in mice predisposed to this autoimmune disease. Nonobese diabetic (NOD) RIP IL-2 and NOD RIP B7-1 mice both have increased incidence of autoimmune diabetes after 200 days, and the double transgenic RIP IL-2 × RIP B7-1 develop diabetes with 100% incidence within 60 days (47, 48, 49). This indicates a function for B7 at the site of autoimmune killing, as it seems unlikely that the B7+ islets would traffic intact to the draining lymph nodes. The “patchy” RIP B7-1 phenotype provides further compelling evidence for localized effects of costimulatory signals in the control of cytotoxic effectors in tissues, as the clusters of islet cells that expressed B7-1 were more rapidly eliminated than the B7-1 negative counterparts. Although this systems does not reveal the mechanism by which B7-1 functions to enhance cytotoxicity, it is clear that costimulation contributes to timely and complete tissue destruction. It also remains possible that NK-mediated killing might be affected in a similar manner, as B7 presentation has been implicated in amplifying the killing of targets by NK cells (50, 51).
Little is known of the factors that contribute to extending a preexisting autoimmune response. One in vitro model has been previously been described in which CD8+ T cells clones that do not receive costimulation undergo “split anergy,” characterized by the ability to lyse targets but not proliferate (52). However, it may be that activated CD8+ T cell effectors have curtailed life spans in vivo, thus contributing to peripheral tolerance. Numerous reports have implicated CD28 costimulation in prevention of cell death during primary T cell activation (9, 53, 54). One system has been described that suggests such a role for costimulation during the cytotoxic phase (55) in preventing Fas-mediated death. In preliminary assays, we found no evidence for increased levels of actively induced cell death in effectors associated with the absence of costimulatory signals. This implies that most cell death in T cell effectors in vitro occurs “by neglect,” but this will be a subject for much further study. For the regulation of T cell tolerance, the maintenance of effector function within the elicited population, as well as augmentation of the size of this population in situ, may ultimately prove to be as important as the factors initiating proliferation.
We thank David Vremec and Jo Butler for assistance with tissue culture and numerous lively discussions, Dr. K. Shortman for support, and Dr. A. Strasser for helpful discussions.
This work was supported by a Postdoctoral Fellowship from the Juvenile Diabetes Foundation International (to M.F.K.) and National Health and Medical Research Council block grants.
Abbreviation used in this paper: RIP, rat-insulin promoter.