Latent Membrane Protein 1, Tumor Necrosis Factor Receptor–Associated Factor (TRAF) 1, TRAF-2, TRAF-3, and Nuclear Factor Kappa B Expression in Posttransplantation Lymphoproliferative Disorders Open Access

Posttransplantation lymphoproliferative disorders (PTLDs) are a heterogeneous group of lesions that occur in patients following bone marrow and solid organ transplantation.1–3 The PTLDs are commonly associated with Epstein-Barr virus (EBV) infection, which likely plays a pathogenic role in this lymphoproliferative disorder.1,4–6 Latently infected lymphocytes express only a limited set of viral proteins of which latent membrane protein 1 (LMP-1) has been consistently implicated in the transformation of B lymphocytes.7,8 LMP-1 aggregates in the plasma membrane of an EBV-infected cell, and its cytoplasmic domain specifically binds intracellular proteins such as tumor necrosis factor receptor (TNFR)–associated factors (TRAFs).9 TRAFs are a family of signal transduction molecules that activate downstream signaling pathways such as nuclear factor kappa B (NFκB)10 and the Jun-N-terminal kinase pathway,7 leading to inhibition of apoptosis. Thus, LMP-1 appears to act as a constitutively activated TNFR.11–13 

In fact, in vitro studies have demonstrated that the mechanism of B-lymphocyte transformation by LMP-1 is mediated through TRAFs, such as TRAF-1, TRAF-2, and TRAF-3, which have all been shown to physically associate with LMP-1 in vitro.14 However, the expression of TRAFs in both normal tissues and lymphoid malignancies has not been extensively tested.15–18 Recently, TRAF-2 has been shown to be important in the activation of NFκB in Reed-Sternberg cells in Hodgkin lymphoma, perhaps via CD30 signaling.15 Specifically studying a small set of EBV-associated PTLDs, Liebowitz6 examined the interaction among LMP-1, TRAF, and NFκB. He demonstrated a physical association between LMP-1 and TRAF-1 and TRAF-3, using immunoprecipitation studies and double-immunofluorescence microscopy. NFκB was activated in these tumors, suggesting that LMP-1 was responsible for the transformation of B-lymphocytes through its interaction with TRAF molecules. The purpose of our study was to further analyze the immunohistochemical expression of LMP-1, TRAF-1, TRAF-2, TRAF-3, and NFκB in a larger cohort of PTLDs to determine whether a correlation exists between their expression and with outcome.

Twenty-seven formalin-fixed, paraffin-embedded tissue sections from 25 patients (2 recurrences) were retrieved from the surgical pathology files of the Cleveland Clinic Foundation. All cases were reviewed and classified according to the World Health Organization classification for lymphoid malignancies.19 Nineteen specimens were monomorphic, 5 were polymorphic, 2 were plasmacytoma-like, and 1 was a Hodgkin disease–like PTLD. These PTLDs were the subject of a recently characterized series.20 

Immunohistochemical analysis for TRAF-1, TRAF-2, TRAF-3, NFκB, and LMP-1 was performed on cases for which paraffin-embedded tissue was available. Briefly, 4-μm-thick sections from each case were deparaffinized in xylene and hydrated in graded alcohol. Immunohistochemical analysis, using a standard streptavidin-biotin complex technique, was performed on paraffin-embedded sections and cell blocks, using an automated stainer (Ventana, Tucson, Ariz). The clone and source of the antibodies, positive controls, and antigen retrieval conditions are summarized in Table 1.

All stains were considered positive when at least 10% or more of the tumor cells demonstrated staining. TRAF-1 and TRAF-2 had a diffuse cytoplasmic staining pattern. TRAF-3 had a staining pattern similar to TRAF-1 and TRAF-2, but some cases had a Golgi accentuation.16 NFκB immunohistochemical staining was performed at the Armed Forces Institute of Pathology. Staining for NFκB was performed using a mouse monoclonal anti-human p65 antibody that recognizes the unbound, active form of p65 (Rel A).21,22 The ability of the antibody to recognize the active (p65) form of NFκB indicates the presence of constitutively activated NFκB in the tumor cells.23 A NFκB immunohistochemical stain showed nuclear and cytoplasmic staining.

EBV-encoded RNA (EBER) by in situ hybridization (ISH) was performed on formalin-fixed, paraffin-embedded sections on all cases except for the cytology specimens (patients 24 and 25). Unstained paraffin sections on electrostatically charged slides were baked overnight at 60°C and deparaffinized in 2 changes of xylene, 5 minutes each, followed by absolute ethanol (two 10-minute changes each). Slides were then washed in distilled water for 5 minutes. Sections fixed in B5 or Hollande's fixative were treated with Lugol iodine (Fisher Scientific, Pittsburgh, Pa) and sodium thiosulfate (Allegiance, McGaw Park, Ill) to remove heavy metal deposits. The sections were then enzyme digested in 100 μL of Protease I (Ventana Medical Systems, Tucson, Ariz) at 37°C for 8 minutes. The reaction was stopped in distilled water for 5 minutes, and the section was dehydrated with graded alcohols. A total of 10 μL of biotin-labeled oligonucleotide (EBER) probe (5′ AGA CAC CGT CCT CAC CAC CCG GGA CTT GTA 3′), in hybridization solution (Sigma) consisting of 50% deionized formamide, 20× SSC, and dextran sulfate, was applied to the section and hybridized overnight at 37°C in a humidified chamber. Stringency wash of 2× SSC for 5 minutes at room temperature was performed. Hybridization detection was performed according to manufacturer's instructions. This consisted of sequential application of 100 μL of mouse antibiotin antibody and Ventana-enhanced, alkaline phosphatase–labeled streptavidin followed by color development with a 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (Ventana) chromogen/substrate system. The chromogen reaction was stopped by immersing slides in distilled water for 5 minutes and then counterstaining with nuclear fast red (Newcomer Supply, Middletown, Wis) for 8 minutes. Slides were then dehydrated with alcohol and xylene and mounted with Cytoseal 60 (Stephens Scientific, Riverdale, NJ).

A few cases that were negative on initial EBER testing but had positive LMP-1 staining were then checked with an RNA-positive control oligo-dT probe (Ventana) to evaluate if sufficient RNA was present in the specimen and retested with INFORM EBER probe (Ventana). The staining was performed on the Discovery automated ISH instrument (Ventana). The following steps were performed by the instrument: slides were deparaffinized, enzyme digested, probe added, probe and target codenatured, hybridized, stringency washed, and detection reagents dispensed per manufacturer's protocol. After washing, sections were dehydrated in graded alcohol and xylene and mounted with Cytoseal 60 (Stephens Scientific).

Tissue microarray blocks were constructed in cases for which adequate tissue was available. The microscopic slides were reviewed, and blocks with sufficient tumor were arrayed using two 1.5-mm-diameter tissue cores per case (Beecher Instruments, Silver Spring, Md). Two such blocks with appropriate controls incorporated in them were used for immunohistochemical analysis. In cases where adequate tissue was not available or immunostaining was difficult to interpret, routine unstained sections were used.

The Mantel-Haenszel χ² test (2-tailed test) was used to determine if there was a linear association between the immunohistochemical stains. Cox proportional hazards analysis was used to determine if immunostains correlated with prognosis, specifically death. Additionally, the Kaplan-Meier method was used as a second way of examining the correlation of immunostains and prognosis. Survival was estimated using this method and compared between positive and negative immunostains using the log-rank test. The findings were consistent with the results obtained from Cox analysis.

Expression of EBER was tested on 27 specimens and was positive in 21 (78%). LMP-1 was expressed in 20 (77%) of 26 specimens. In 1 specimen, the LMP-1 was uninterpretable. Nineteen specimens coexpressed EBER and LMP-1, and their correlation was, as expected, statistically significant (P = .001). Five specimens that were EBER negative were also LMP-1 negative. However, one specimen was EBER negative but LMP-1 positive. LMP-1 expression did not correlate with TRAF-1 (P = .17), TRAF-2 (P = .22), TRAF-3 (P = .68), or NFκB (P = .58) (Table 2).

Twenty-six (96%) of the 27 specimens tested for NFκB showed diffuse moderate-to-strong nuclear and/or cytoplasmic staining. Most PTLDs (96%) in our study expressed at least 1 of the 3 TRAFs. Eighty-two percent of the specimens expressed TRAF-1, 84% expressed TRAF-2, and 74% expressed TRAF-3. Twenty-five specimens were tested for both TRAF-1 and TRAF-2 (one case was not tested for TRAF-2 due to insufficient tissue, and the other case was uninterpretable due to excessive background). Seventeen cases (68%) showed coexpression of TRAF-1 and TRAF-2. Four TRAF-1–negative specimens were TRAF-2 positive. Only one case was negative for both TRAF-1 and TRAF-2, and, interestingly, this specimen was also negative for EBER, LMP-1, and TRAF-3 but was positive for NFκB. The one case that was negative for NFκB was positive for all the other markers except TRAF-2. The Figure shows examples of EBER, NFκB, TRAF-1, and TRAF-2 staining.

TRAF-2 and NFκB expression showed a statistically significant (P = .02) correlation; however, there was no correlation between NFκB expression and either TRAF-1 (P = .63) or TRAF-3 (P = .55) expression. Because TRAF-1 and TRAF-2 heterodimers can activate NFκB,10 we attempted to see if a correlation between cases that coexpressed TRAF-1 and TRAF-2 and NFκB existed. However, no correlation was found (P = .14). When looking at overall survival, we found no statistically significant correlation with any of the markers (Kaplan-Meier, log-rank test, and Cox proportional hazards analysis).

The PTLDs are a heterogeneous group of lymphoproliferative disorders with respect to their morphologic, molecular genetic, and clinical features. Despite this, most PTLDs are EBV associated, suggesting an important pathogenic role for this virus. Much attention has been focused on the transforming properties of LMP-1. This molecule mimics an activated TNFR via recruitment of TRAFs with the attendant downstream signaling events, including NFκB activation. In support of this mechanism, Devergne and colleagues14 demonstrated that LMP-1 has the potential to transform B cells through interactions with TRAF molecules such as TRAF-1, TRAF-2, and TRAF-3. Liebowitz6 demonstrated physical association of LMP-1 and TRAF-1 and TRAF-3 in a group of EBV-associated lymphoid neoplasms such as PTLD, Burkitt lymphoma, and human immunodeficiency virus–associated lymphoma. We examined a larger number of PTLDs to further characterize the expression pattern of these molecules in PTLD.

NFκB expression, as recognized by an antibody to the nuclear localization signal, appeared to be present in almost all cases. This antibody has been shown to specifically recognize activated p65/RelA.24,25 Unfortunately, we were unable to confirm this with alternate methods, such as the electrophoretic shift mobility assay or confocal fluorescent microscopy, due to lack of fresh material. Expression of this important transcription factor has been recognized in increasing numbers of lymphoid malignancies, including cutaneous T-cell lymphoma, B-chronic lymphocytic lymphoma/leukemia, multiple myeloma, and Hodgkin lymphoma.22,23,26,27 Thus, its activation in PTLDs is not unexpected and is consistent with the recent study by Liebowitz.6 

Few studies analyzing TRAF expression patterns in lymphoid malignancies have been published. A recent paper by Zapata et al18 analyzed TRAF expression patterns in normal tissues and other lymphoid malignancies, but PTLDs and Hodgkin lymphoma were not part of their analysis. The study showed that TRAF-1 was most commonly associated with non-Hodgkin lymphomas, using both paraffin-embedded tissue and tumor cell lines. Durkop et al28 and Izban et al26 studied TRAF expression in Hodgkin lymphoma and demonstrated expression of both TRAF-1 and TRAF-2 in Reed-Sternberg cells.

In a recent paper, Murray et al17 studied the immunohistochemical expression of TRAF-1 in EBV-associated lymphomas, which included 22 PTLDs. They found that TRAF-1 was expressed in most (72%) of their EBV-positive PTLDs and our results are similar (83%), supporting its role in the pathogenesis of PTLDs.6,29 TRAF-1 and TRAF-2 have been shown by Devergne et al14 to activate NFκB via LMP-1 in vitro, usually in conjunction as heterodimers. In our study, coexpression of TRAF-1 and TRAF-2 was present in two thirds of the cases. The association of NFκB p65 and TRAF-2 expression in this series further suggests that TRAF-2, in addition to TRAF-1, may also play a role in the pathogenesis of PTLDs. A direct association with LMP-1, however, was not found. Although EBV was present in most of our cases, TRAF molecules in PTLDs might also be activated by other TNFR members, resulting in NFκB activation. Interestingly, TRAF-1/TRAF-2 heteroaggregates also recruit inhibitors of apoptosis proteins,30 which may enhance cell survival and facilitate B-lymphocyte transformation. Further studies of freshly isolated PTLD tissue would be required to study the physical association of these molecules in vivo.

In summary, we found that most PTLDs were EBV associated, which is consistent with previous reports.5,31,32 We show that TRAFs and NFκB p65 are commonly expressed in both EBV-positive and EBV-negative PTLDs. Our preliminary data suggest that, in addition to TRAF-1, TRAF-2 may also play a role in the constitutive activation of NFκB in PTLDs. We recognize the limitations of this immunohistochemical-based study, and further studies on a well-characterized group of PTLDs are warranted to gain further understanding of these complex pathogenetic mechanisms.