Research ArticleAutoimmunity

Pituitary Expression of CTLA-4 Mediates Hypophysitis Secondary to Administration of CTLA-4 Blocking Antibody

See allHide authors and affiliations

Science Translational Medicine  02 Apr 2014:
Vol. 6, Issue 230, pp. 230ra45
DOI: 10.1126/scitranslmed.3008002


Hypophysitis is a chronic inflammation of the pituitary gland of unknown (primary forms) or recognizable (secondary forms) etiology, such as the use of ipilimumab in cancer immunotherapy. Ipilimumab, which blocks the T cell inhibitory molecule CTLA-4 (cytotoxic T lymphocyte antigen-4), induces hypophysitis in about 4% of patients through unknown mechanisms. We first established a model of secondary hypophysitis by repeated injections of a CTLA-4 blocking antibody into SJL/J or C57BL/6J mice, and showed that they developed lymphocytic infiltration of the pituitary gland and circulating pituitary antibodies. We next assessed the prevalence of pituitary antibodies in a cohort of 20 patients with advanced melanoma or prostate cancer, 7 with a clinical diagnosis of hypophysitis, before and after ipilimumab administration. Pituitary antibodies, negative at baseline, developed in the 7 patients with hypophysitis but not in the 13 without it; these antibodies predominantly recognized thyrotropin-, follicle-stimulating hormone–, and corticotropin-secreting cells. We then hypothesized that the injected CTLA-4 antibody could cause pituitary toxicity if bound to CTLA-4 antigen expressed “ectopically” on pituitary endocrine cells. Pituitary glands indeed expressed CTLA-4 at both RNA and protein levels, particularly in a subset of prolactin- and thyrotropin-secreting cells. Notably, these cells became the site of complement activation, featuring deposition of C3d and C4d components and an inflammatory cascade akin to that seen in type II hypersensitivity. In summary, the study offers a mechanism to explain the pituitary toxicity observed in patients receiving ipilimumab, and highlights the utility of measuring pituitary antibodies in this form of secondary hypophysitis.


The pituitary gland integrates nervous and peripheral signals to orchestrate the control of key vital functions (such as growth, sexual drive and reproduction, lactation, stress response, and basal metabolism) via the production of growth hormone (GH), follicle-stimulating hormone (FSH), luteinizing hormone (LH), prolactin (PRL), adrenocorticotropic hormone (ACTH), and thyroid-stimulating hormone (TSH) (1, 2). Numerous pathological conditions can affect the pituitary and expand its size, forming a mass in the sella turcica. Some sellar masses secrete one of the hormones indicated above; others, like autoimmune hypophysitis, are non–hormone-secreting. Hypophysitis is a chronic inflammation of the pituitary gland characterized pathologically by infiltration with hematopoietic cells (mainly lymphocytes) and destruction of endocrine cells (3). There are primary forms of hypophysitis, where the causative agent is unknown, and secondary forms, where instead an initiating event is clearly traceable. Emerging among the secondary forms is the one that occurs in association with blockade of CTLA-4 (cytotoxic T lymphocyte antigen-4) (4). CTLA-4 is an inhibitory molecule expressed on activated and regulatory T cells: when bound to its cognate receptor on antigen-presenting cells, it initiates a signaling cascade that ultimately dampens T cell activation (5, 6). If CTLA-4 is blocked, therefore, T cells remain active, a feature that has been exploited as a form of immunotherapy in patients with advanced melanoma or other cancers (7, 8). The reagent most commonly used to block CTLA-4 is ipilimumab (Yervoy), a humanized immunoglobulin G1 (IgG1) κ monoclonal antibody (mAb) produced by Bristol-Myers Squibb and approved by the U.S. Food and Drug Administration in March 2011 for the treatment of metastatic melanoma, and now being tested for several other types of advanced cancer. Ipilimumab is effective in tumor control but also induces a variety of autoinflammatory responses collectively referred to as immune-related adverse events (irAEs) (9). The most common irAEs are dermatitis, enterocolitis, hepatitis, and hypophysitis (10).

Hypophysitis secondary to ipilimumab administration can be difficult to diagnose and severe enough to require hospitalization. Presenting symptoms may include those of mass effect (headache) and glucocorticoid deficiency [easy fatigability, weakness, anorexia, nausea, and occasionally hyponatremia (11, 12)]. Hypophysitis usually develops 6 to 8 weeks after ipilimumab initiation (13) and often requires prolonged or permanent hormonal replacement. Twenty-one clinical trials (summarized in table S1) have used ipilimumab and reported hypophysitis as a side effect, with a mean incidence of 4% (106 of 2853 patients) and a range from 1% (14) to 25% (15).

It remains unknown why hypophysitis, traditionally considered a rare disease, is seen with such high frequency in patients treated with CTLA-4 blockade. CTLA-4 is expressed predominantly in T lymphocytes (16, 17), although it has also been reported on murine embryonic stem cells (18), human muscle cells (19), placental fibroblasts (20), monocytes (21), leukemia cells (22), various cell line and tumor cells (23), and dendritic cells (24). Here, we hypothesized that CTLA-4 is expressed on pituitary endocrine cells and that this noncanonical expression forms the basis of the hypophysitis secondary to CTLA-4 blockade.


Repeated injections of a CTLA-4 blocking antibody induce a murine model of hypophysitis

To recapitulate the pituitary disease seen in patients receiving ipilimumab, we injected SJL/J mice with a hamster IgG1κ mAb blocking murine CTLA-4 (UC10-4F10-11), using a dose regimen comparable to the one used in humans. As control, we injected SJL/J mice with hamster IgG using the same dose and administration schedule. A distinct infiltration of the pituitary gland with hematopoietic mononuclear cells was seen in mice injected with anti–CTLA-4 (Fig. 1A, white arrows) but not control (Fig. 1B) antibody, with average numbers of 30 ± 9 cells per high-power field versus 3 ± 1 (P = 0.001; Fig. 1C). The infiltrating cells were composed mainly of CD45-positive lymphocytes (Fig. 1D), which were instead rarely found in mice injected with the control IgG (Fig. 1E) or left uninjected (Fig. 1F; P = 0.001), as well as F4/80-positive macrophages (Fig. 1G), again rarely seen in controls (Fig. 1, H and I; P = 0.001). The inflammatory infiltrate in the anti–CTLA-4 group was focal, distributed to the interstitial space, and not disrupting the normal pituitary architecture (Fig. 1A), suggesting that it represented an initial phase of hypophysitis. Accordingly, pituitary function, as assessed by serum PRL levels (table S2), remained normal in this model of hypophysitis secondary to injections of a CTLA-4 blocking antibody. No hematopoietic infiltration was seen with either antibody in thyroid gland (Fig. 1, J to L), liver (fig. S1, A and B), colon (fig. S1, C and D), or skin (fig. S1, E and F), suggesting that the mechanisms of ipilimumab toxicity may be different between pituitary and other organs.

Fig. 1. Hematopoietic cell infiltration of the pituitary gland in mice injected with a CTLA-4 blocking antibody.

(A and B) Histopathological appearance [by hematoxylin and eosin (H&E) microscopy] of the pituitary gland in mice injected with a mAb blocking CTLA-4 (A) or a control IgG (B). (C) Infiltrating cells [white arrows in (A)] were significantly more abundant in the CTLA-4 group than in the control IgG or uninjected groups. P = 0.001. Magnification, ×256. Scale bars, 50 μm. (D and E) Immunohistochemical staining for CD45-positive hematopoietic cells in pituitary glands from mice injected with CTLA-4 mAb (D) or control IgG (E). Magnification, ×160. Scale bars, 100 μm. (F) CD45 cells were significantly more numerous in the CTLA-4 group. (G and H) Immunohistochemical staining for F4/80-positive macrophages in pituitary glands from mice injected with CTLA-4 mAb (G) or control IgG (H). Magnification, ×160. Scale bars, 100 μm. (I) F4/80 macrophages were significantly more numerous in the CTLA-4 group. (J and K) Histopathological appearance (by H&E microscopy) of the thyroid gland in mice injected with CTLA-4 (J) or a control IgG (K). Scale bars, 250 μm. (L) The histology was normal in all groups, and no infiltrating cells were found. Numbers in the bar graphs represent mean counts ± SD. **P = 0.001 by Wilcoxon rank sum test.

In addition to the cellular immune events indicated above, antibodies directed against the anterior pituitary gland were absent before injections in both groups (Fig. 2A) and developed in mice receiving the CTLA-4 blocking antibody (Fig. 2B), recognizing about 8 ± 2% of the total adenohypophyseal cell population. Pituitary antibodies were instead not found after injection of the control IgG (Fig. 2C). Double immunofluorescence stainings showed that these pituitary antibodies recognized a subset (3 ± 1%) of PRL-secreting cells (Fig. 2D) and a subset (2 ± 1%) of ACTH-secreting cells (Fig. 2E). Overall, results suggest that both cellular and humoral immune responses against the pituitary gland can be induced in SJL/J mice by repeated injections of a CTLA-4 blocking antibody.

Fig. 2. Development and cell specificity of circulating pituitary antibodies in mice injected with a CTLA-4 blocking antibody.

(A) Before injections, pituitary antibodies were not found in the serum of either experimental group. (B and C) These antibodies developed after injection of the CTLA-4 blocking antibody (B) but not the control IgG (C). Fluorescent cells represented 8 ± 2% of the total pituitary cell population, as assessed by analysis of 10 microscopy fields per slide. (D and E) Double immunofluorescence showed that these antibodies recognized 3 ± 1% of PRL-secreting cells (D) and 2 ± 1% of ACTH-secreting cells (E). Magnification, ×40. Scale bars, 100 μm.

Patients with advanced cancer treated with ipilimumab and diagnosed with secondary hypophysitis develop serum autoantibodies against the anterior pituitary gland

We next assessed the presence of pituitary antibodies in 20 patients with advanced cancer (12 melanoma and 8 prostate) treated with ipilimumab, 7 with a clinical diagnosis of hypophysitis and 13 without it (Table 1). Pituitary antibodies, negative at baseline, developed in the 7 patients with hypophysitis but not in the 13 without it (Table 1). The antibodies recognized predominantly TSH-secreting cells and, less frequently, FSH- or ACTH-secreting cells (Table 1). This recognition was associated with functional defects in the thyrotroph (all seven patients), gonadotroph (three of seven), corticotroph (four of seven), or somatotroph (three of seven) axes (Table 2).

Table 1 Serum pituitary antibodies in 20 cancer patients with or without a clinical diagnosis of hypophysitis before and after administration of ipilimumab (Ipi).

na, sera that were “not assessed” for cell-specific antibodies because the overall pituitary antibodies were negative.

View this table:
Table 2 Clinical, magnetic resonance imaging (MRI), and endocrine features of 20 patients with advanced cancer (melanoma or prostate) treated with ipilimumab administration.

IGF-1, insulin-like growth factor 1; irAEs, immune-related adverse events; na, not applicable because these patients had no clinical evidence of hypophysitis; nd, not done.

View this table:

CTLA-4 is expressed in the pituitary gland

To dissect the mechanism underlying the pituitary toxicity secondary to ipilimumab administration, we used our murine model and hypothesized that CTLA-4 antigen is expressed “ectopically” (meaning outside of the canonical T cell compartment) in the pituitary. CTLA-4 mRNA was clearly detected by reverse transcription polymerase chain reaction (RT-PCR) in the murine pituitary gland, featuring a migration pattern similar to the one obtained from the spleen positive control (Fig. 3A). It was also detectable using two different CTLA-4 primer pairs (Fig. 3A). No CTLA-4 signal was found in the thyroid gland, used here as negative control (Fig. 3A). By quantitative real-time PCR, all 10 murine pituitary glands analyzed showed CTLA-4 expression that, although significantly lower than that observed in the lymphoid-rich spleen (P = 0.0002) and thymus (P = 0.0002), was clearly detectable and higher (P = 0.0001) than that observed in the thyroid (Fig. 3B).

Fig. 3. CTLA-4 expression in the pituitary gland.

(A) RT-PCR showing CTLA-4 mRNA expression in pituitary, spleen, and thyroid using two different primer pairs (primers 1 and 2). Upper and lower bands indicate full-length and soluble CTLA-4, respectively. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used as loading control. (B) Quantitative real-time PCR for CTLA-4 in pituitary, spleen, thymus, and thyroid. Ct values for CTLA-4 are expressed as means ± SE after adjusting for Ct for GAPDH. **P < 0.0001; *P = 0.001. (C and D) Protein extracts from murine pituitary (male and female) and spleen (C), as well as from human pituitary (D), were analyzed by Western blotting with an antibody directed against CTLA-4. The chimeric protein CTLA-4 Ig served as the experimental control. (E and F) Protein extracts from murine pituitary and spleen were probed with an antibody against CD3ε (E) or CD45 (F). (G) Actin loading control.

CTLA-4 protein was detected by Western blotting in murine (male and female, Fig. 3C) and human pituitary glands (Fig. 3D), and appeared as two dominant bands slightly different from those found in the spleen (Fig. 3C). No CTLA-4 was detected in cerebral cortex, thyroid, or adrenal gland (fig. S2). Because CTLA-4 is expressed mainly in T cells, we blotted the same murine pituitary and spleen protein lysates with antibodies against a pan–T cell marker (CD3ε) or a pan-hematopoietic marker (CD45) and found no signal in the pituitary (Fig. 3, E and F), even after adjusting for protein loading (Fig. 3G). Similar results were obtained in the human pituitary gland. Blotting pituitary lysates with a CD3ε antibody showed the absence of signal in the pituitary (fig. S3A). In addition, CD3 immunohistochemical staining of normal human pituitary glands collected at autopsy (n = 7) showed that parenchymal T lymphocytes are rare (fig. S3B), with an average of four cells (range, 0 to 8) in four microscopy fields analyzed at ×20 magnification. Overall, these results indicate at both mRNA and protein levels that CTLA-4 is expressed in nonhematopoietic cells of the pituitary gland.

CTLA-4 is expressed predominantly in PRL- and TSH-secreting cells

To identify the pituitary cell type(s) expressing CTLA-4, we performed double immunofluorescence using antibodies to CTLA-4 and anterior pituitary hormones. In mice, CTLA-4 was found in a subset of PRL-secreting (2 ± 1%, Fig. 4A) and TSH-secreting (3 ± 2%, Fig. 4B) cells, whereas it was undetected in cells producing GH (Fig. 4C), ACTH (Fig. 4D), or FSH (Fig. 4E), or in folliculo-stellate cells (labeled by the S100 staining, Fig. 4F). In humans, these stainings yielded similar results (fig. S4).

Fig. 4. Double immunofluorescence to assess the expression of CTLA-4 in the major pituitary cell types.

Each triptych represents the staining obtained using an antibody against CTLA-4 (left panels), a pituitary protein (middle panels), and the merging of the previous two images (right panels). CTLA-4–positive cells within each pituitary population were quantified on the basis of fluorescence signal as the percentage of positive cells per microscopy field (using 10 fields per slide). (A and B) CTLA-4 accounted for 2 ± 1% of the PRL-secreting cells (A) and 3 ± 2% of the TSH-secreting cells (B). (C to F) No CTLA-4 signal was detected in cells expressing GH (C), ACTH (D), FSH (E), or S100 (F). Magnification, ×40. Scale bars, 100 μm.

To confirm and better quantify the results obtained by microscopy, we performed flow cytometry on pooled (between 6 and 10) pituitary glands removed from C57BL6/J mice (Fig. 5A). After gating on live cells (68% of the total single-cell suspension, Fig. 5B) and CD45-negative cells (91% of live cells, Fig. 5C), PRL- and TSH-containing cells nicely separated in two distinct populations, representing about 20 and 2% of the total nonhematopoietic cells (Fig. 5D). CTLA-4 was expressed in about 3% of the PRL-stained cells (Fig. 5E) and in 3% of the TSH-stained cells (Fig. 5F). Considering the background staining given by isotype controls (Fig. 5, G to I), the cytometric CTLA-4–positive population accounted for about 3% of PRL-secreting cells and 1% of TSH-secreting cells.

Fig. 5. Flow cytometry of mouse pituitary glands to quantify the expression of CTLA-4 in PRL- and TSH-secreting cells.

(A to C) Single-cell suspensions of murine pituitary gland pools (A) were first gated on live cells on the basis of exclusion of the Live/Dead Aqua dye (B) and then on cells not of hematopoietic origin on the basis of the negativity for CD45 (C). (D) Gated cells were then analyzed for the expression of PRL and TSH. (E to I) Finally, the expression of CTLA-4 in cells positive for PRL (E) or TSH (F) was quantified, after adjusting for the expression obtained in the same gates using isotype controls (G to I). The percentage of positive cells in each gate is indicated.

Injection of a mAb blocking CTLA-4 leads to complement deposition on PRL- and TSH-secreting cells

Considering these findings of CTLA-4 expression in pituitary endocrine cells, we envisioned a mechanism through which ipilimumab could cause pituitary toxicity: the formation of immune complexes made of CTLA-4 antigens and CTLA-4 antibodies at the pituitary level, followed by binding of C1q to the Fc fragment of the CTLA-4 antibody and activation of the complement classical pathway. This pathway leads to the production of C3, C3d, C4d, and, ultimately, tissue damage. To test this hypothesis, we injected SJL/J mice intraperitoneally once with 1 mg of the hamster CTLA-4 blocking antibody or control IgG. Two days after the injection, mice were sacrificed and pituitary glands were collected to assess the presence of C3, C3d, and C4d by double immunofluorescence.

We found deposition of C3 (Fig. 6A) and C3d (Fig. 6B) onto a subset of PRL-secreting cells (4 ± 1% and 7 ± 2%, respectively), and deposition of C4d onto a subset (6 ± 2%) of PRL-secreting (Fig. 6C) and TSH-secreting (Fig. 6D) cells, in mice injected with CTLA-4 antibody but not in those injected with the control IgG (Fig. 6E). These findings were repeated and confirmed in the C57BL/6J strain (fig. S5). The complement deposition was specific to the pituitary because it was not found in thyroid glands of mice injected with either CTLA-4 or control antibody (fig. S6). Incubating the murine pituitary glands removed from CTLA-4 antibody–injected mice with a fluorescein isothiocyanate (FITC)–conjugated anti-hamster Ig was not able to reveal directly the binding of the CTLA-4 antibody to the pituitary substrate. On the basis of the notion that C4d binds covalently to the surface where complement activation is initiated, its deposition on the pituitary (but not the thyroid) after injection of the CTLA-4 antibody (but not of the control IgG) suggests that a pituitary-specific inflammation is being generated, which in time could lead to the development of hypophysitis. Overall, these results highlight a type II hypersensitivity reaction in which the CTLA-4 antibody binds to the cognate antigen expressed on pituitary cells, activates complement, and initiates tissue destruction.

Fig. 6. Double immunofluorescence to assess the deposition of complement components (C3, C3d, and C4d) in murine pituitary after injection of CTLA-4 mAb.

Each triptych represents the staining obtained using an antibody against complement components (left panels), PRL or TSH (middle panels), and the merging of the previous two images (right panels). Complement-positive cells within each pituitary population were quantified on the basis of fluorescence signal as the percentage of positive cells per microscopy field (using 10 fields per slide). (A) C3 accounted for 4 ± 1% of the PRL-secreting cells. (B) C3d accounted for 7 ± 2% of the PRL-secreting cells. (C and D) C4d accounted for 6 ± 2% of PRL-secreting (C) or TSH-secreting (D) cells. (E) No complement deposition was found in pituitary glands from mice injected with the control IgG. Magnification, ×40. Scale bars, 100 μm.


The study reports that CTLA-4 is expressed in pituitary endocrine cells and, when blocked by administration of a specific mAb, leads to site-specific deposition of complement components, pituitary infiltration, and antibody formation. The study also shows that patients receiving ipilimumab develop pituitary antibodies.

CTLA-4, a member of the Ig superfamily, is classically found in T lymphocytes upon activation, first inside cytosolic vesicles and then on the plasma membrane (25). CTLA-4’s main function is to inhibit the T cells that express it, in part by outcompeting the stimulatory molecule CD28 for binding to CD80/CD86 (25), but is likely endowed with other roles. Although lacking intrinsic catalytic activity, CTLA-4 provides scaffolding sites for intracellular molecules, often kinases, and has thus the potential of influencing numerous cellular processes (25). Our findings of CTLA-4 expression in a small fraction of PRL and TSH cells support a noncanonical role for CTLA-4, which could participate to differentiation of these lineages and mark a hitherto uncharacterized subset of immature PRL and TSH cells.

Establishing a murine model of secondary hypophysitis based on injections of a CTLA-4 blocking antibody gave us the opportunity to define a mechanism of pituitary toxicity akin to type II hypersensitivity. We showed in fact that injected mice developed mononuclear cell infiltration of the pituitary gland, pituitary antibodies, and complement activation with C4d deposition onto PRL- and TSH-secreting cells. These findings suggest that CTLA-4 blocking antibody could bind its cognate antigen CTLA-4 on pituitary cells, triggers complement activation, and then recruits inflammatory leukocytes, further amplifying the pituitary-specific immune response. The combination of these site-specific immune mechanisms with systemic T cell activation, as seen in patients receiving ipilimumab, may result in full-blown, clinically evident hypophysitis. It remains unclear why patients receiving the same immunotherapy develop different types of autoimmune damage (the conundrum of tissue/organ specificity), but expression levels of CTLA-4 antigen in the various tissues could be part of the explanation.

It is interesting to reflect on the physical properties of the CTLA-4 blocking antibody because this might have practical consequences. The antibody most commonly used, ipilimumab, is of the IgG1 subclass, a subclass known to activate the classical pathway of complement, although not as potently as IgM and IgG3. The other commercially available CTLA-4 blocking antibody, tremelimumab, is of the IgG2 subclass, which can activate the classical pathway of complement, but less potently than IgG1. In keeping with our findings, hypophysitis secondary to tremelimumab (table S4) has been reported at lower frequencies (10 of 917 patients, 1%) than hypophysitis secondary to ipilimumab (106 of 2853 patients, 4%). It is therefore conceivable that development of CTLA-4 blocking antibodies that lack the ability to fix complement could decrease the incidence of irAEs.

The study also reported the development of pituitary antibodies in patients treated with ipilimumab and a clinical diagnosis of hypophysitis, and identified TSH-, gonadotropin-, and ACTH-secreting cells as the most common antibody targets. These results, in keeping with the notion that most patients with hypophysitis secondary to ipilimumab have defect in TSH, ACTH, and gonadotroph axes (4, 12, 26, 27), should provide clinicians with an additional tool to establish a diagnosis of hypophysitis.

One limitation of our study pertains to the lack of direct evidence for the binding of the injected anti–CTLA-4 antibody to its cognate antigen in the pituitary. Although we obtained indirect evidence, given that these injections led to pituitary-specific activation of complement, finite resources did not allow us to visualize binding in vivo. The CTLA-4 blocking antibody we used (clone 4F10), however, has been extensively characterized for its ability to bind CTLA-4 and block its function (28). A second limitation relates to the extent to which this mouse model of secondary hypophysitis mimics the human counterpart. We noted discrepancies between the toxicities observed in mice, mainly focused on the pituitary, and the more complex spectrum of irAEs seen in patients. In addition, considering that no pituitary biopsy has ever been published in patients receiving ipilimumab, it remains to be ascertained how the pituitary pathology observed in mice closely replicates the human lesions. A final limitation relates to the lack of uniformly accepted criteria to establish a diagnosis of hypophysitis in patients receiving cancer immunotherapy. Symptoms resulting from expansion of the local inflammation (headache and visual disturbances) or anterior pituitary hormone deficiencies (fatigue, decreased libido, nausea, and orthostatic hypotension) may be subtle or masked by the severity of the underlying neoplastic disease. Patients, including the ones reported in this study, do not currently undergo systematic assessment of pituitary function so that subclinical deficiencies may be missed, especially when dynamic endocrine testing is not performed. Refinement and implementation of hypophysitis algorithms for patients treated with CTLA-4 blocking antibodies (12) should overcome this limitation.

In summary, our study suggests a mechanism to explain the pituitary toxicity observed in patients receiving ipilimumab and provides tools to improve the diagnosis of secondary hypophysitis and refine the understanding of the irAEs observed with cancer immunotherapy.


Study design

The study included 20 patients with advanced metastatic melanoma or prostate cancer enrolled in a multicenter, Bristol-Myers Squibb–sponsored trial, opened at the Memorial Sloan Kettering Cancer Center (MSKCC) between 2007 and 2012 (Tables 1 and 2). After approval of an MSKCC Institutional Review Board protocol, serum samples were collected from melanoma patients treated with ipilimumab (trials NCT00495066, NCT00623766, and NCT00920907) and from prostate cancer patients treated with ipilimumab combined with radiotherapy (trial NCT00323882). Ipilimumab was administered to all patients at least once at a dose of either 3 or 10 mg/kg (Table 2). The patients were selected on the basis of the following criteria: participation in one of the four clinical trials indicated above, availability of banked plasma samples, and documented clinical or laboratory evidence of hypophysitis. On the basis of these criteria, seven patients (IDs 1 to 7 in Table 1) were classified as having hypophysitis. The most common endocrine deficiencies were in the thyroid and/or adrenal axes (Table 2), reflecting, in part, the current clinical practice of testing predominantly for TSH and ACTH when considering a diagnosis of hypophysitis. Testing for the gonadal axis (testosterone, LH, and FSH) or PRL provided supportive evidence in some cases. The remaining 13 ipilimumab-treated patients had either other irAEs (n = 9) or no evidence of irAEs (n = 4) (Table 2). The sera from these 20 patients were studied at baseline (n = 20) and at several time points after ipilimumab treatment (n = 101).

Purification of a mAb blocking murine CTLA-4

The hybridoma cell line UC10-4F10-11 (American Type Culture Collection) produces a hamster IgG1κ mAb directed against the extracellular portion of murine CTLA-4. Hybridoma cells were first cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich) (supplemented with 10% fetal bovine serum, 0.01 mM nonessential amino acid solution, and 0.05 mM 2-mercaptoethanol) and then adapted to a serum-free medium (CD Hybridoma Medium, Gibco) (supplemented with 8 mM l-glutamine) to remove interferences caused by Igs present in the bovine serum. Supernatants from serum-free cultures were then concentrated and purified by passage through protein G columns (Pierce Protein G Chromatography Cartridges, catalog no. 89926, Thermo Fisher Scientific) and used for the experiments.


Mice (SJL/J and C57BL/6J strains) were purchased from The Jackson Laboratory (stocks #000686 and #000664); housed in pathogen-free facility with a 12-hour light/12-hour dark cycle, free access to food and water, and a room temperature of 22°C; and used when 8 weeks old. Care of the mice was in accordance with the standards established by the U.S. Animal Welfare Acts, set forth in the National Institutes of Health guidelines, and approved by the Johns Hopkins University Animal Care and Use Committee.

Induction of hypophysitis by injections of a CTLA-4 blocking antibody

SJL/J mice (n = 9) were injected intraperitoneally with 100 μg of the CTLA-4 blocking antibody purified as above three times a week (days 0, 1, and 3) for a total of 5 weeks. This dose regimen (1.5 mg per mouse over a period of 6 weeks) corresponds to four ipilimumab injections of 3 mg/kg into patients, after the dose conversion across species reported by Reagan-Shaw et al. (29). To enhance systemic immune response, we also injected the mice on the first day of the weekly protocol, except on the fourth week, with complete Freund’s adjuvant without any specific antigens. As control, SJL/J mice (n = 6) were injected intraperitoneally with a commercial hamster polyclonal IgG (Bio X Cell, BE0091).

Analysis of pituitary histopathology by H&E staining and immunohistochemistry

Upon sacrifice, murine pituitary glands were collected into Beckstead’s zinc fixative (30), cut at 5-μm thickness (collecting one section every five cuts to encompass the entire pituitary gland), and then stained with H&E for light microscopy evaluation as described (31). The number of infiltrating hematopoietic cells was estimated by counting the mononuclear cells per high-power field in the section with the most severe pathology. As control tissues, thyroid glands, liver, colon, and skin (which are classic targets of ipilimumab toxicity) were also analyzed. The thyroid was processed as the pituitary. Liver, colon, and skin were analyzed by sampling three different parts of each organ and evaluating them for hematopoietic cell infiltration. The type of infiltrating hematopoietic cells was evaluated by immunohistochemistry using the technique reported in (31) and the primary antibodies listed in table S3.

Detection of serum pituitary antibodies, adenohypophyseal cell types, CTLA-4, and complement components (C3, C3d, and C4d) by immunofluorescence

We used indirect immunofluorescence on both murine and human pituitary gland substrates to assess the presence and cell specificity of serum pituitary antibodies, as well as the expression of CTLA-4 and complement components. To this end, murine and human pituitary glands were embedded, frozen in OCT compound (Sakura Finetek), and cut with a Leica CM-1950 cryostat at 5-μm thickness. The staining procedure was then performed as described in detail in (32), and the primary antibodies can be found in table S3.

RNA extraction, reverse transcription, and PCR amplification of CTLA-4

Pituitary, spleen, thyroid, and thymus were removed from 10 SJL/J mice, snap-frozen in liquid nitrogen, and processed with the RNeasy Mini Kit (Qiagen) to extract total RNA. Total RNA (1 μg) was reverse-transcribed into complementary DNA (cDNA) with the QuantiTect Reverse Transcription kit (Qiagen) for RT-PCR, or the RT2 First Strand Kit (Qiagen) for quantitative real-time PCR. CTLA-4 was amplified using two primer pairs. Pair 1 consisted of forward 5′-GTAGCCCTGCTCACTCTTC-3′ primer in exon 1 and reverse 5′-CATGCTCCTTAGCTTTAAATTG-3′ primer in exon 4, which theoretically yield amplicons of 791, 681, 443, and 333 base pairs (bp) according to the four isoforms of CTLA-4: full-length, soluble, ligand-independent, or exons 1 and 4 only. Pair 2 consisted of forward 5′-GACATTCACAGAGAAGAATAC-3′ primer in exon 2 and reverse 5′-CATGCTCCTTAGCTTTAAATTG-3′ primer in exon 4, which are able to detect full-length and soluble isoforms of 598 and 488 bp, respectively. As control for RNA content, cDNAs were also amplified with the GAPDH housekeeping gene using forward 5′-AGGTCATCCCAGAGCTGAAC-3′ and reverse 5′-TCTCTTGCTCAGTGTCCTTG-3′ primers, which yield an amplicon of 379 bp. RT-PCR was performed with 35 cycles of denaturation (30 s at 94°C), annealing (30 s at 54.5°C), and extension (45 s at 72°C), followed by a final extension of 7 min at 72°C. Quantitative real-time PCR was performed with the RT2 SYBR Green qPCR Mastermix (Qiagen), a commercial CTLA-4 primer pair targeting exon 4 (Qiagen, PPM03217E-200), and a Bio-Rad CFX96 modular thermal cycler platform (Bio-Rad Laboratories), according to the manufacturer’s specifications. Expression level (L) of CTLA-4 mRNA was calculated by L = 2Ct, where Ct is the number of cycles at which the amount of amplified target reaches a fixed threshold (the lower this number, the higher the expression). Then, it was adjusted for the expression level of GAPDH to obtain the ΔCtCt = 2−[Ct(CTLA-4) − Ct(GAPDH)]), and for the expression levels of CTLA-4 and GAPDH in the thyroid gland, used here as negative control, to obtain the ΔΔCt.

Western blotting

Murine pituitary, spleen, heart, thyroid, cerebral cortex, and adrenals, as well as human pituitary, were homogenized on ice in radioimmunoprecipitation assay buffer (R0278, Sigma-Aldrich), supplemented with a cocktail of protease inhibitors (P8340, Sigma-Aldrich). Homogenates were cleared by centrifugation at 600g to pellet insoluble debris. Supernatants were then aliquoted and stored at −80°C. Fifty micrograms of protein lysates was separated by 10% SDS–polyacrylamide gel electrophoresis under reducing conditions and transferred to nitrocellulose membranes (ECL Hybond, GE Healthcare Life Sciences). Membranes were blocked in 4% skim milk and incubated overnight at 4°C with the primary antibodies listed in table S3. After three washes in phosphate-buffered saline (PBS)–Tween 20 (0.1%), membranes were incubated for 1 hour at room temperature with a peroxidase-conjugated secondary antibody. Subsequently, blots were developed with ECL Plus (GE Healthcare) and exposed to x-ray films (BioExpress). As positive control, gels were loaded (0.1 μg per lane) with a CTLA-4–Fc fusion dimer (97 kD, Abcam, ab59654), which consists of the extracellular domain (160 amino acids) of CTLA-4 and the mouse (mutated) IgG2a Fc domain, and migrates at around 50 kD when monomer on denaturing gels.

Pituitary flow cytometry

Murine pituitary glands (pools of 6 to 12 glands) were removed from C57BL/6 mice and digested for 30 min at 37°C in DMEM containing 0.25% collagenase II and 0.1% dispase II with gentle agitation. Single cells were then rinsed twice with PBS supplemented with 0.05% bovine serum albumin (BSA) (Sigma-Aldrich) and 2 mM EDTA (Corning Cellgro). As positive control, we prepared single-cell suspensions from murine spleens. Neubauer chamber microscopy was used to assess cell number, viability, and digestion completeness. Cells (3 × 105 per tube) were first incubated for 30 min at 4°C with the Live/Dead Aqua dye (1 μl per tube, from Invitrogen) to exclude dead cells, and then for 10 min at 4°C with an Fc block (anti-CD16/32) to decrease background. Cells were then surface-stained for 30 min at 4°C with fluorochrome-conjugated antibodies against CD45 and CTLA-4 (eBioscience), washed once in PBS–0.5% BSA, and incubated overnight at 4°C in fixation and permeabilization buffer (BD Biosciences). For intracellular staining, cells were incubated for 30 min at 4°C with antibodies to TSH (from A. F. Parlow, National Hormone and Peptide Program; previously labeled with Zenon FITC anti-rabbit IgG, Life Technologies), PRL, and CTLA-4 (table S3). Cells were finally washed once in permeabilization buffer (BD Biosciences), resuspended in 200 μl of PBS–0.5% BSA, and counted with an LSR II quad-laser cytometer running FACSDiva 6 (BD Immunocytometry). Data were analyzed with FlowJo 8.6 (Tree Star Software), gating first on live cells and then on CD45-negative (nonhematopoietic) cells.

Measurement of circulating murine PRL by enzyme-linked immunosorbent assay

The serum levels of PRL in murine sera were measured with the Prolactin Mouse ELISA Kit (Abcam, ab100736) following the manufacturer’s instruction. Blood samples were collected before (day 0) and at the end (day 42) of the injection protocol with CTLA-4 blocking antibody or the control IgG.


Statistics was used to calculate the sample size of mice injected with CTLA-4 blocking antibody, control antibody, or uninjected controls; to compare the number of infiltrating lymphocytes in the pituitary glands of both groups; and to compare the mRNA expression of CTLA-4 among four murine organs (spleen, thymus, thyroid, and pituitary). Power calculations used a two-sided α value (probability that a difference could have arisen by chance alone) of 0.05 (and therefore a z value of 1.96) and a power (probability of detecting a significant difference when there is really one) of 0.8 (and therefore a z value of 0.84). Considering that by H&E microscopy “lymphocytes are not normally present in the anterior pituitary gland” (33), and that by specific immunostaining “only very rare lymphocytes are detected” (33), we chose a difference (δ) of 4 and an SD (σ) of 2 lymphocytes. Finally, we chose an experiment-to-control ratio (λ) of 1. Entering these five numbers in the formula reported below, we obtained a number of at least four mice per group.

Embedded Image

To take into account accidents occurring during the experimental protocol, we increased this sample size by about 50%. To compare counts of pituitary lymphocytes among the three experimental groups (CTLA-4–injected, control injected, and uninjected), as well as to compare the CTLA-4 expression levels among organs, we first used the Kruskal-Wallis test and then performed pairwise assessments with the Wilcoxon signed-rank test. P values smaller than 0.05 were considered statistically significant. Statistical analyses were performed with Stata Statistical Software, Release 13 (Stata Corp.).


Fig. S1. H&E staining of liver, colon, and skin from mice injected with a CTLA-4 blocking antibody or control IgG (polyclonal hamster IgG).

Fig. S2. Western blotting analysis of CTLA-4 expression in control murine tissues.

Fig. S3. Analysis of CD3 expression in the normal human pituitary gland.

Fig. S4. Cell-specific expression of CTLA-4 in the human pituitary gland.

Fig. S5. Complement deposition in pituitary glands of C57BL/6J mice after injection of a CTLA-4 blocking antibody.

Fig. S6. Assessment of complement deposition in murine thyroids after injection of a CTLA-4 blocking antibody.

Table S1. Clinical trials using ipilimumab and reporting autoimmune hypophysitis as an irAE.

Table S2. Serum levels of PRL (ng/ml) in mice before and after injection of a hamster monoclonal IgG1 against CTLA-4 or a hamster polyclonal control IgG.

Table S3. Primary antibodies used for immunohistochemistry, immunofluorescence, Western blotting, and flow cytometry.

Table S4. Clinical trials using tremelimumab and reporting autoimmune hypophysitis as an irAE.

References (3462)


  1. Funding: The study was supported by NIH grant DK080351 to P.C. S.I. was supported in part by a fellowship from the Uehara Memorial Foundation. Author contributions: S.I. designed and performed most of the experiments and contributed to manuscript writing. A.D.R. performed some of the experiments and contributed to manuscript writing. M.K.C. conducted trials NCT00495066, NCT00623766, and NCT00920907; provided sera from the patients; and reviewed the manuscript. J.D.W. directed trials NCT00495066, NCT00623766, and NCT00920907. S.F.S. directed trial NCT00323882. P.C. designed and sponsored the study and wrote the manuscript. Competing interests: J.D.W. has had a consultancy or advisory role with Bristol-Meyers Squibb and has received research funding from Bristol-Meyers Squibb. M.K.C. has received research funding from Bristol-Meyers Squibb. The other authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article