Cutting the brake lines: Unlocking cancer immunity?

Unlocking cancer immunity with T-cells

Editor's Introduction

Enhancement of Antitumor Immunity by CTLA-4 Blockade

annotated by

Our immune system protects us daily against thousands of pathogens that try to invade. But when our own cells go rogue, the immune system is not always capable of protecting us. Have you ever wondered how cancer can go undetected by the immune system’s T cells? Nobel laureate James Allison and the co-authors of this paper take a deeper look at CTLA-4, a molecule found on the surface of T cells. CTLA-4 is one of the brakes of the immune system. Is it the key to developing an immune response against cancer?

Paper Details

Original title
Enhancement of Antitumor Immunity by CTLA-4 Blockade
Original publication date
Vol. 271, Issue 5256, pp. 1734-1736
Issue name


One reason for the poor immunogenicity of many tumors may be that they cannot provide signals for CD28-mediated costimulation necessary to fully activate T cells. It has recently become apparent that CTLA-4, a second counterreceptor for the B7 family of costimulatory molecules, is a negative regulator of T cell activation. Here, in vivo administration of antibodies to CTLA-4 resulted in the rejection of tumors, including preestablished tumors. Furthermore, this rejection resulted in immunity to a secondary exposure to tumor cells. These results suggest that blockade of the inhibitory effects of CTLA-4 can allow for, and potentiate, effective immune responses against tumor cells.


Despite expressing antigens recognizable by a host's immune system, tumors are very poor in initiating effective immune responses. One reason for this poor immunogenicity may be that the presentation of antigen alone is insufficient to activate T cells. In addition to T cell receptor engagement of an antigenic peptide bound to major histocompatibility complex (MHC) molecules, additional costimulatory signals are necessary for T cell activation (1). The most important of these costimulatory signals appears to be provided by the interaction of CD28 on T cells with its primary ligands B7-1 (CD80) and B7-2 (CD86) on the surface of specialized antigen-presenting cells (APCs) (2-4). Expression of B7 costimulatory molecules is limited to specialized APCs. Therefore, even though most tissue-derived tumors may present antigen in the context of MHC molecules, they may fail to elicit effective immunity because of a lack of costimulatory ability. Several studies support this notion. In a variety of model systems, transfected tumor cells expressing costimulatory B7 molecules induced potent responses against both modified and unmodified tumor cells (5-8). It appears that tumor cells transfected with B7 are able to behave as APCs, presumably allowing direct activation of tumor-specific T cells.

Recent evidence suggests that costimulation is more complex than originally thought and involves competing stimulatory and inhibitory signaling events (3, 9-12). CTLA-4, a homolog of CD28, binds both B7-1 and B7-2 with affinities much greater than does CD28 (13-16). In vitro, antibody cross-linking of CTLA-4 has been shown to inhibit T cell proliferation and interleukin-2 production induced by antibody to CD3 (anti-CD3), whereas blockade of CTLA-4 with soluble intact or Fab fragments of antibody enhances proliferative responses (17, 18). Similarly, soluble intact or Fab fragments of anti-CTLA-4 greatly augment T cell responses to nominal peptide antigen or the superantigen Staphylococcus enterotoxin B in vivo (19, 20). It has also been suggested that CTLA-4 engagement can induce apoptosis in activated T cells (21). Finally, mice deficient in CTLA-4 exhibit severe T cell proliferative disorders (22). These results demonstrate that CTLA-4 is a negative regulator of T cell responses and raise the possibility that blockade of inhibitory signals delivered by CTLA-4-B7 interactions might augment T cell responses to tumor cells and enhance antitumor immunity.

We first sought to determine whether CTLA-4 blockade with nonstimulatory, bivalent antibody (18, 20) would accelerate rejection of B7-positive tumor cells. Previously, we showed that B7-1 expression was partially successful at inducing rejection of the transplantable murine colon carcinoma 51BLim10 (23). We reasoned that CTLA-4 blockade would remove inhibitory signals in the costimulatory pathway, resulting in enhanced rejection of the tumor cells. We injected groups of BALB/c mice with B7-1-transfected 51BLim10 tumor cells (B7-51BLim10) (23). Two groups were treated with a series of intraperitoneal injections of either anti-CTLA-4 or anti-CD28 (18, 24). Treatment with anti-CTLA-4 inhibited B7-51BLim10 tumor growth as compared with the antiCD28-treated mice or the untreated controls (Fig. 1). All mice in the untreated and antiCD28-treated groups developed small tumors that grew progressively for 5 to 10 days and then ultimately regressed in 8 of the 10 mice by about day 23 after injection. The two small tumors that did not regress remained static for more than 90 days. In contrast, three of five mice treated with anti-CTLA-4 developed very small tumors, all of which regressed completely by day 17. Although these results were encouraging and were consistent with our hypothesis, they were not very dramatic because B7-1 expression resulted in fairly rapid rejection of transfected 51BLim10 cells even in the absence of CTLA-4 blockade; however, these results confirmed that anti-CTLA-4 did not inhibit tumor rejection.


Fig. 1 Treatment with anti-CTLA-4 accelerates rejection of a B7-1-positive  colon carcinoma (23). A volume of 100μl of cell suspension (4 x 106 cells) was injected subcutaneously into the left flanks of five female BALB/c mice. Two of the groups received three intraperitoneal injections of either anti-CTLA-4 or anti-CD28 (18). Injections of 100, 50, and 50 μg of antibody were given on days 0, 3, and 6, respectively, as indicated by the arrows. Control animals received no injections. Data points represent the average of the products of bisecting tumor diameters. Error bars represent standard error of the mean.


Does administering anti-CTLA-4 or anti-CD28 enhance tumor rejection?


For each group of BALB/c mice, the authors injected 4 x 106 cells of B7-positive colon carcinoma just under the skin of the animal (subcutaneous injection). This means that the tumor will develop as a raised sphere and be measurable using calipers. The size of the tumor can be tracked at different time points.

The tumor is injected on day 0. For the mice treated with different antibodies, the treatment is given on days 0, 3, and 6, with 100, 50, and 50 μg being injected each day, respectively.

Graphical Representation

The three groups of mice are represented with different symbols. The y-axis represents the tumor size, which is calculated by measuring the diameter of the tumor in two different directions. The x-axis shows time in days. A double dash along the x-axis shows that the scale is broken: the authors do not show measurements between day 35 and day 60. The three arrows along the x-axis show the three days at which antibodies were injected for the anti–CD28 and anti–CTLA-4 groups. The data points are the average across all mice in each group, and the error bars represent standard error.

Experimental groups

There are three groups consisting of five female mice each.

The control group was injected with 4 x 106 cells of B7-positive colon carcinoma but was not given any treatments.

The anti-CD28 group was injected with the tumor cells and with antibodies against CD28.

The anti-CTLA-4 group was injected with the tumor cells and with antibodies against CTLA-4.


For all three groups, a tumor is established and grows in size, followed by a regression in size. However, the time it takes for this to occur in the anti–CTLA-4 group is shorter than the time it takes for the tumor to begin shrinking in the other two groups. Also, the tumors in the mice treated with anti–CTLA-4 completely shrink to a size of 0 mm2 by day 17. (Notice that there are no error bars on those points, meaning that every one of the five mice achieved a tumor size of zero.)

We next examined the effects of CTLA-4 blockade on the growth of V51BLim10, a vector control tumor cell line that does not express B7 (23). All mice either injected with 4 x 106 V51BLim10 tumor cells and left untreated, or treated with anti-CD28, developed progressively growing tumors and required euthanasia by 35 days after inoculation (Fig. 2A). In contrast, all mice treated with anti-CTLA-4 completely rejected their tumors after a short period of limited growth. Similarly, control mice injected with 2 x 106 tumor cells developed rapidly growing tumors and required euthanasia by day 35 (Fig. 2B). Anti-CTLA-4 treatment had a dramatic effect on tumor growth, but one mouse did develop a tumor quickly (accounting for a majority of the growth indicated in Fig. 2B) and another developed a tumor much later (Fig. 2C). Anti-CTLA-4 appeared to be less effective at a tumor dose of 1 x 106 cells, where treatment resulted in significantly reduced tumor growth rates, but four of five mice developed progressively growing tumors (25). Thus, although curative responses were not obtained in all cases, it is clear that CTLA-4 blockade significantly enhanced rejection of B7-negative tumor cells.

We next sought to determine whether tumor rejection as a consequence of CTLA-4 blockade was associated with enhanced immunity to a secondary challenge. Mice that had rejected V51BLim10 tumor cells as a result of treatment with anti-CTLA-4 were challenged with 4 x 106 wild-type 51BLim10 cells 70 days after their initial tumor injections. These mice showed significant protection against a secondary challenge as compared with naive controls (Fig. 2D). All control animals had progressively growing tumors by 14 days after injection, developed massive tumor burdens, and required euthanasia by day 35. Only one of the previously immunized mice had a detectable tumor by day 14, and growth of this tumor was very slow. Ultimately, two more tumors developed in the immunized mice 42 days after challenge. Two mice remained tumor-free throughout the course of the experiment. These results demonstrate that tumor rejection mediated by CTLA-4 blockade results in immunologic memory.


Fig. 2 Treatment of anti-CTLA-4 enhances rejection of B7-negative colon carcinoma cells and results in protection against subsequent challenge with wild-type colon carcinoma cells. Groups of BALB/c mice were injected with B7-negative 51BLim10 vector control cells (V51BLim10), left untreated, or treated wtih anti-CTLA-4 or control antibody. Mice were euthanized when tumors reached a size of 200mm2 or became ulcerated. If individual mice within a group were euthanized, the final measurement was carried over to subsequent time points. (A) Average tumor size in mice injected with 4 x 106 tumor cells. Groups of five mice were injected with 4 x 106 V51BLim10 tumor cells. Treated groups were injected three times with 100 μg of anti-CTLA-4 or anti-CD28 as indicated by the arrows. All untreated control and anti-CD28-treated mice were killed by day 35. Mice treated with anti-CTLA-4 remained tumor-free for more than 90 days. Error bars represent standard error of the mean. (B) Average tumor size in mice injected with 2 x 106 V51BLim10 cells and treated with anti-CTLA-4. Three of the mice remained tumor-free beyond 80 days. (C) Individual tumor growth in mice injected with 2 x 106 V51 BLiml 0 cells and treated with anti-CTLA-4. Three of the mice remained tumor-free beyond 80 days. (D) Challenge tumor growth in anti-CTLA-4-treated mice. Five anti-CTLA-4-treated mice that had completely rejected V51BLim10 tumor cells were rechallenged 70 days later with 4 x 106 wild-type tumor cells injected subcutaneously in the opposite flank. Five naïve mice were also injected as controls. All control mice developed progressively growing tumors and were euthanized on day 35 after inoculation. Three of five previously immunized mice remained tumor-free 70 days after rechallenge.

Panel A

The authors now investigate the effects of anti–CTLA-4 on growth of a non-B7–expressing tumor. Groups of five mice were challenged with 4 x 106 V51BLim10 tumor cells. The anti–CD28 and anti–CTLA-4 groups were given 100 μg of antibody in each of three injections on days 0, 3, and 6 as indicated by the arrows. The control group was untreated with any antibody. Based on this figure, what is the effect of administering anti–CTLA-4?

Panel B

The authors decreased the inoculating dose of V51BLim10 tumor cells. In this experiment, two groups of five mice were injected with 2 x 106 cells, half the number used in previous experiments. The control group in this experiment was injected with irrelevant antibody instead of being untreated.

Why do you think the authors chose to inject antibody into all the experimental groups? Does the effect of anti–CTLA-4 differ based on inoculating dose?

Notice that the error bars for the anti–CTLA-4 group are quite large compared to what we have seen in Figures 1 and 2A. Think about one possible reason that might be the case before you read about the data in Panel C.

Panel C

Panel C shows a different representation of the data from the experiment shown in Panel B. This time, we are looking at the tumor sizes for each individual mouse in the anti–CTLA-4 group instead of seeing the average. How many mice had tumors after day 70? Is the anti–CTLA-4 treatment always effective?

Panel D

The experiment showed here aims to test whether successful rejection of tumor cells as a result of anti–CTLA-4 treatment can result in immunologic memory. For this experiment, the authors use wild-type, or naturally occurring, tumor cells rather than cells with any modifications made in the lab. The authors inject 4 x 106 wild-type tumor cells into groups of five mice each. One group has previously been challenged with V51BLim10 tumors and treated with anti–CTLA-4, while the other has not encountered the tumors in the past. This secondary challenge was given 70 days after the initial tumor rejection and on the opposite flank of the mouse. During the course of this experiment, neither group of mice was given any treatment.

Approximately how long does it take, on average, for the tumors to be established (i.e., have a size greater than zero) in the first group of mice? In which group do the tumors grow faster, if at all?

Importance of statistics

It is very important for scientists to report some metric of error when they calculate averages. As you can see from Panel B, if the scientists had only reported the average tumor size of the five mice in the anti–CTLA-4 group, the reader could have been led to believe that all the mice were able to control the growth of the tumor as a result of anti–CTLA-4 treatment. In reality, one of the mice seems not to have responded at all to the treatment, as seen in Panel C, and its tumor grew at a similar rate and to a final size comparable to those of the mice in the control group.

To determine whether anti-CTLA-4 treatment could have an effect on the growth of established tumors, we injected groups of mice with 4 x 106 wild-type 51BLim10 tumor cells and treated them with anti-CTLA-4 beginning on day 0 as before, or beginning 7 days later at which time most mice had palpable tumors. Mice treated with anti-CTLA-4 at either time period had significantly reduced tumor growth compared with untreated controls (Fig. 3). In fact, delaying treatment appeared to be more effective, with two of five mice remaining tumor-free beyond 30 days after inoculation.


Fig. 3 Treatment with anti-CTLA-4 reduces the growth of established tumor. Groups of mice were injected subcutaneously with 2 x 106 51BLim10 tumor cells. Control animals (n=10) were injected intraperitoneally with 100 μg of irrelevant hamster antibody on days 0, 3, 6, and 9, as indicated by the upward-pointing arrows. One anti-CTLA-4 treatment group (n=10) received intraperitoneal injections on the same days. The other treated mice (n=5) were given intraperitoneal injections of anti-CTLA-4 beginning on day 7 and subsequently on days 10, 13, and 16 (downward-pointing arrows).

Questions and purpose

Is it necessary to administer anti–CTLA-4 at the same time that a tumor is established?

The purpose of this experiment is to determine whether anti–CTLA-4 treatment is effective against pre-established tumors.

Graphical representation and experimental groups

As in previous figures, the three groups of mice are represented with different symbols and the y-axis represents the tumor size at different time points, averaged for the mice in each group.

All mice were injected with 2 x 106 wild-type tumor cells on day 0.

The control group consists of 10 mice which were given 100 μg of irrelevant antibody on days 0, 3, 6, and 9, while the “anti–CTLA-4, day 0” group consists of 10 mice given 100 μg of anti–CTLA-4 antibody on those same days. Those days are represented by the upward-pointing arrows.

The “anti–CTLA-4, day 7” group consists of five mice given treatment on days 7, 10, 13, and 16, represented by the downward-pointing arrows.


Treatment with anti–CTLA-4 is effective in reducing the size of tumors pre-established for 7 days prior to beginning the course of treatment.

The effects of anti-CTLA-4 treatment were not limited to variants of the murine colon carcinoma 51BLim10. Similar results were obtained with a rapidly growing fibrocarcinoma of A/JCr mice, SA1N (26) (Fig. 4). All control mice injected subcutaneously with 1 x 106 Sa1N cells developed measurable, rapidly growing tumors within 7 days, whereas only two mice treated with anti-CTLA-4 had tumors by day 30, and one additional mouse developed a tumor around day 40 after injection. The remaining mice were still tumor-free 70 days after injection. In another experiment, control mice injected with 4 x 105 Sa1N tumor cells also developed rapidly growing tumors, whereas 7 of 10 mice treated with anti-CTLA-4 were tumor-free by day 25 after injection (25).


Fig. 4 Treatment with anti-CTLA-4 reduces the growth of the murine fibrosarcoma Sa1N. Groups of five mice were injected subcutaneously in the flank with a suspension of 1 x 106 Sa1N fibrosarcoma cells. Treated groups were injected intraperitoneally with 100 μg of anti-CTLA-4 or irrelevant hamster control antibody at days 0, 3, and 6 as indicated by the arrows. All control animals were killed by day 30. Two of five animals treated with anti-CTLA-4 remained tumor free at day 55.

Questions and purpose

Is anti–CTLA-4 able to control growth of tumors other than murine colon carcinoma?

Graphical representation and experimental groups

This experiment looks at another strain of mice, A/JCr, injected with Sa1N fibrocarcinoma cells. Each group consists of five mice each injected with 1 x 106 Sa1N cells on day 0. The “Sa1N only” group was untreated. The “control” group was given 100 μg of irrelevant mouse antibody on days 0, 3, and 6, indicated by the arrows. The final group was given 100 μg of anti–CTLA-4 antibody on the same days.


The treatment is also effective at treating this more rapidly growing type of cancer. Since the treatment is independent of the type of cancer and instead acts directly on the immune system, it is showing promise to be widely applicable to many types of cancer.

Our results indicate that removing inhibitory signals in the costimulatory pathway can enhance antitumor immunity. Although it has been shown that anti-CTLA-4 interferes with signals that normally down-regulate T cell responses in vivo (17, 18), the exact mechanisms of antitumor immunity elicited by CTLA-4 blockade are not clear. In the case of B7-negative tumors, antigens are most likely transferred to and presented by host APCs (27), where CTLA-4 blockade might effect T cell responses in two nonexclusive ways. First, removal of inhibitory signals may lower the overall threshold of T cell activation and allow normally unreactive T cells to become activated. Alternatively, CTLA-4 blockade might sustain proliferation of activated T cells by removing inhibitory signals that would normally terminate the response, thus allowing for greater expansion of tumor-specific T cells.

Regardless of the mechanism, it is clear that CTLA-4 blockade enhances antitumor responses. Most importantly, we have observed these effects against unmanipulated, wild-type tumors. Current methods of enhancing antitumor immunity generally require the engineering of tumor cells (8). Some of these methods, such as the induction of B7 expression, rely on enhancing the costimulatory activity of the tumor cells themselves. Others, such as engineering tumor cells to express MHC class II molecules (26, 28, 29) or to produce granulocyte-macrophage colony-simulating factor (27, 30, 31) or pulsing dendritic cells with antigen ex vivo (32, 33), seek to enhance antigen presentation, antigen transfer, or both. Thus, CTLA-4 blockade, by removing potentially competing inhibitory signals, may be a particularly useful adjunct to other therapeutic approaches involving the costimulatory pathway.



1. D. L. Mueller, M. K. Jenkins, R. H. Schwartz, Ann. Rev. Immunol. 7, 445 (1989).

2. P. S. Linsley and J. A. Ledbetter, ibid. 11, 191 (1993).

3. C. H. June, J. A. Bluestone, L. M. Nadler, C. B. Thompson, Immunol. Today 15, 321 (1994).

4. J. P. Allison, Curr. Opin. Immunol. 6, 414 (1994).

5. L. Chen, S. Ashe, W. A. Brady, l. Hellstrom, K. E. Hellstrom et al., Cell 71, 1093 (1992).

6. S. E. Townsend and J. P. Allison, Science 259, 368 (1993).

7. S. Baskar et al., Proc. Natl. Acad. Sci. U.S.A. 90, 5687 (1993).

8. J. P. Allison, A. A. Hurwitz, D. R. Leach, Curr. Opin. Immunol. 7, 682 (1995).

9. M. K. Jenkins, Immunity 1, 443 (1994).

10. J. A. Bluestone, ibid. 2, 555 (1995).

11. P. S. Linsley, J. Exp. Med. 182, 289 (1995).

12. J. P. Allison and M. F. Krummel, Science 270, 932 (1995).

13. J. F. Brunet et al., Nature 328, 267 (1987).

14. K. Harper et al., J. Immunol. 147, 1037 (1991).

15. P. S. Linsley et al., J. Exp. Med. 174, 561 (1991).

16. P. S. Linsley et al., Immunity 1, 793 (1994).

17. T. L. Walunas et al., ibid., p. 405.

18. M. F. Krummel and J. P. Allison, J. Exp. Med. 182, 459 (1995). Antibodies used in these studies were protein G-purified from hybridoma supernatants and quantified by ultraviolet spectrophotometry.

19. E. R. Kearney et al., J. Immunol. 155, 1033 (1995).

20. M. F. Krummel, T. J. Sullivan, J. P. Allison, Int. Immunol., in press.

21. J. G. Gribben et al., Proc. Natl. Acad. Sci. U.S.A. 92, 811 (1995).

22. P. Waterhouse et al., Science 270, 985 (1995).

23. S. E. Townsend, F. W. Su, J. M. Atherton, J. P. Allison, Cancer Res. 54, 6477 (1994). 51BLim10 colon carcinoma cells were transfected with a plasmid construct containing the gene for murine B7-1 and cloned by limiting dilution. Fresh cultures of tumor cells were established from early passage frozen stocks and maintained in culture for no more than 30 days before use. Tumor cells were harvested by trypsinization from tissue culture plates, washed three times in serum-free medium, and suspended at 4 x 107 cells per milliliter. Expression of B7-1 molecules on transfected cells was verified by flow cytometry before injection. V51 BLim10 and wild-type 51BLim10 tumor cells do not express detectable amounts of B7-1, B7-2, or CTLA-4 as determined by flow cytometric analyses.

24. J. A. Gross, E. Callas, J. P. Allison, J. Immunol. 149, 380 (1992).

25. D. R. Leach, M. F. Krummel, J. P. Allison, data not shown.

26. S. Baskar, L. Glimcher, N. Nabavi, R. T. Jones, S. Ostrand-Rosenberg, J. Exp. Med. 181, 619 (1995).

27. A. Y. C. Huang et al., Science 264, 961 (1994).

28. S. Ostrand-Rosenberg, A. Thakur, V. Clements, J. Immunol. 144, 4068 (1990).

29. D. R. Leach and G. N. Callahan, ibid. 154, 738 (1995).

30. G. Dranoff et al., Proc. Natl. Acad. Sci. U.S.A. 90, 3539 (1993).

31. H. I. Levitsky, A. Lazenby, R. J. Hayashi, D. M. Pardoll, J. Exp. Med. 179, 1215 (1994).

32. V. Flamand et al., Eur. J. Immunol. 24, 605 (1994).

33. S. Grabbe, S. Beissert, T. Schwarz, R. D. Granstein, Immunol. Today 16, 117 (1995).

34. We thank S. Ostrand-Rosenberg and R. Warren for providing tumor lines. Supported by NIH grants CA57986, CA09179, and CA40041.