Summary:

Even in diffuse large B-cell lymphoma (DLBCL), a cancer of professional antigen-presenting cells, response rates to immune checkpoint blockade therapy have been limited. One reason for DLBCL immune evasion is epigenetic repression instead of activation of the antigen-presenting MHC—a dissection of mechanisms underlying this repression suggests an opening for restoring B-cell maturation and, along the way, MHC expression as a novel modality of cytoreducing DLBCL and simultaneously augmenting possibilities for immunotherapy.

See related article by Ennishi et al., p. 546.

New, mechanistically rational ways for complementing immune checkpoint blockade therapy (ICB), to extend its durable benefits to more patients, are needed. Even in diffuse large B-cell lymphoma (DLBCL), a cancer of professional antigen-presenting cells (APC), response rates to ICB have been <40% (1). One reason for DLBCL immune evasion, first described more than 10 years ago, is attenuated expression of MHC by the DLBCL cells (2). MHC is how cells present self antigens and non-self antigens to T cells. MHC Class II (MHC-II) is normally expressed by B cells and monocytes, and presents antigens to Th cells (CD4+), key to engendering a broad-based immune response to antigen. MHC Class I (MHC-I) is expressed by all mature nucleated cells and presents antigen, including invading non-self antigen, to cytotoxic T cells (CD8+), identifying for these effectors what to attack.

In this issue, Ennishi and colleagues noted that membranous MHC-I and MHC-II were absent in approximately 40% to approximately 30%, respectively, of 347 DLBCL cases (3), extending and confirming prior observations (2). MHC-II–deficient DLBCL also had fewer tumor-infiltrating T cells and less T-cell cytolytic activity (2, 3). In some cases, MHC deficiency was due to deletion of MHC genes (2). But this was <10% of the time, and in most DLBCL, as shown also by others, MHC gene loci were present, and the suppression was epigenetically mediated (2, 3). This is important, because epigenetic repression can in principle be reversed, to potentially restore MHC expression, trigger immune recognition, and perhaps improve responses to ICB. Such an endeavor would of course be aided by a better understanding of how the epigenetic repression occurs in the first place.

One clue noted by Ennishi and colleagues was that in DLBCLs subclassified as being germinal center B cell–like (GCB-DLBCL) in B-cell lineage maturation stage, suppression of MHC-II was one facet of a broader (664-gene) transcriptomic profile shift versus other cases (3). This broader gene expression pattern indicated that the GCB-DLBCL cells with low MHC-II expression were akin to centroblasts, whereas the GCB-DLBCL cells with higher MHC-II expression resembled centrocytes (3). Indeed, MHC-II upregulation is a late event in B-cell maturation (Fig. 1A). What are the genetic events that stall B-cell maturation at different points, to create these discernible DLBCL subtypes? Ennishi and colleagues observed that activating mutations in EZH2, first reported in DLBCL more than 10 years ago, were significantly enriched in GCB-DLBCL with low MHC-II. EZH2 belongs to the polycomb group (PcG) of chromatin-modifying proteins. PcGs were identified originally in fruit flies with confused body segmentation—PcG mediates the repression, and trithorax proteins the activation, of body pattern—determining homeobox transcription factors; loss-of-function alterations to PcG upset this dynamic balance and resulted in confused lineage-commitment decisions. Specifically, EZH2 is the enzyme module in PcG repressor complex 2 (PRC2) that trimethylates lysine 27 on histone 3 (H3K27me3) to create this epigenetic repression mark. Commitment of stem cells or multipotent progenitors toward specific lineages involves removal of the H3K27me3 mark at the loci of master transcription factor regulators of lineage specification. Indeed, the B-cell lineage master transcription factor BCL6 directs EZH2 to key B-cell lineage differentiation gene loci (4). It would seem then, that the gain-of-function mutations in EZH2 in lymphomas should turn “off” lineage differentiation genes and mediate maturation arrest. Ennichi and colleagues treated EZH2-mutated DLBCL cells with an EZH2 inhibitor, tazemetostat, which decreased H3K27me3 marks and increased MHC-II expression. The investigators did not evaluate effects in conjunction with ICB in vivo. Nevertheless, this observation is of a piece with extensive literature spanning decades that inhibitors of enzyme mediators of transcription repression, for example the drugs decitabine and 5-azacytidine that deplete DNA methyltransferase 1 (DNMT1), induce terminal differentiation of cancers of various histologies including lymphomas, increase MHC expression in the process, increase or induce immune recognition of cancers of various histologies, and potentially complement ICB (reviewed in ref. 5).

Figure 1.

Lineage and maturation-stage roles of EZH2/PRC2 dictate selection by neoplastic evaluation for gain-of-function (GOF) versus loss-of-function (LOF) genetic alterations. A, MHC (HLA) genes are activated at late stages of B-cell maturation. Public gene expression by microarray data (GSE14714). HSC, hematopoietic stem cells, n = 6; CLP, common lymphoid progenitors, n = 6; Pro-B, B-cell progenitors, n = 6; Pre-B, early B cells, n = 6; Immature B cells, n = 6; naïve B cells, n = 1; centroblasts, n = 1; centrocytes, n = 1. B,EZH2 expression increases with lymphoid commitment and in early B-cell differentiation (GOF EZH2 alterations characterize lymphomas), but decreases with megakaryocyte lineage commitment and differentiation (LOF EZH2 alterations characterize myeloid malignancies with megakaryocytic features). Fold change versus average expression in HSC. Box, median ± interquartile range; whiskers, range. Public gene expression by microarray data (GSE14714 and GSE24759: HSC, n = 14; MEP, megakaryocyte–erythroid progenitor, n = 9; CFUMK, colony-forming unit megakaryocyte and megakaryocytes, n = 12). C, The H3K27me3 mark in embryonic stem cells (ESC), HSC, monocytes, and B cells, measured at genes that are expressed specifically at the indicated phases of hematopoietic differentiation (listed on the right). CMP-GMP, common myeloid and granulocyte–monocyte progenitors. The gene expression modules were identified, and the H3K27me3 chromatin immunoprecipitation sequencing data was analyzed, as described previously (5, 7). D, Selection by neoplastic evolution for PRC2 GOF versus LOF alterations dictates, and is dictated by, cancer lineage and maturation stage. TSS, transcription start site.

Figure 1.

Lineage and maturation-stage roles of EZH2/PRC2 dictate selection by neoplastic evaluation for gain-of-function (GOF) versus loss-of-function (LOF) genetic alterations. A, MHC (HLA) genes are activated at late stages of B-cell maturation. Public gene expression by microarray data (GSE14714). HSC, hematopoietic stem cells, n = 6; CLP, common lymphoid progenitors, n = 6; Pro-B, B-cell progenitors, n = 6; Pre-B, early B cells, n = 6; Immature B cells, n = 6; naïve B cells, n = 1; centroblasts, n = 1; centrocytes, n = 1. B,EZH2 expression increases with lymphoid commitment and in early B-cell differentiation (GOF EZH2 alterations characterize lymphomas), but decreases with megakaryocyte lineage commitment and differentiation (LOF EZH2 alterations characterize myeloid malignancies with megakaryocytic features). Fold change versus average expression in HSC. Box, median ± interquartile range; whiskers, range. Public gene expression by microarray data (GSE14714 and GSE24759: HSC, n = 14; MEP, megakaryocyte–erythroid progenitor, n = 9; CFUMK, colony-forming unit megakaryocyte and megakaryocytes, n = 12). C, The H3K27me3 mark in embryonic stem cells (ESC), HSC, monocytes, and B cells, measured at genes that are expressed specifically at the indicated phases of hematopoietic differentiation (listed on the right). CMP-GMP, common myeloid and granulocyte–monocyte progenitors. The gene expression modules were identified, and the H3K27me3 chromatin immunoprecipitation sequencing data was analyzed, as described previously (5, 7). D, Selection by neoplastic evolution for PRC2 GOF versus LOF alterations dictates, and is dictated by, cancer lineage and maturation stage. TSS, transcription start site.

Close modal

This story so far is intuitive, but the link between EZH2 mutations with a particular B-cell maturation phenotype (centroblast-like), and not other maturation phenotypes, underscores vital larger realities about PRC2, epigenetics, and epigenetically directed therapy. Simply stated, the consequences of altered function of a key epigenetic protein such as EZH2 depend on lineage, and maturation stage within a lineage, in other words, the master transcription factor context of a cell. For example, myeloid malignancies share a hematopoietic origin with lymphomas, but are characterized by recurrent loss-of-function alterations to EZH2 and other PRC2-linked modules (e.g., ASXL1 and DNMT3A). Reminiscent of body segmentation confusion in fruit flies, PRC2 loss-of-function alterations in myeloid malignancies are strongly linked with specific lineage biases of the malignancy (6). In normal hematopoiesis, EZH2 expression is high in hematopoietic stem cells (HSC) and increases into the early phases of B-cell differentiation (Fig. 1B). In contrast, EZH2 expression is preserved from HSC into megakaryocyte–erythroid progenitors, but decreases with commitment into the megakaryocyte (platelet) lineage (Fig. 1B), and interestingly EZH2 loss-of-function genetic alterations are linked with megakaryocyte lineage features of myeloid malignancy, including high platelet counts (6). In other words, that EZH2 loss-of-function alterations skew myeloid lineage differentiation in specific directions is also intuitive. Perplexing is that the same loss-of-function genetic alterations are strongly linked also with transformation of myeloproliferative neoplasms into acute myeloid leukemias (AML), which like DLBCL are defined by arrest at specific points along lineage-differentiation axes (6, 7). How can loss-of-function in PRC2 both release cells into a specific lineage and stall maturation in that lineage? Examination of the H3K27me3 mark at genes highly expressed sequentially during hematopoietic differentiation is informative: In monocytes, the H3K27me3 mark is enriched at genes highly expressed in HSCs and in genes that are highly expressed in committed myeloid progenitors (common myeloid progenitor/granulocyte–monocyte progenitor genes). That is, the H3K27me3 mark was enriched at genes that were transiently expressed then shut down in the cells as they continued to advance to a terminal lineage fate (Fig. 1C; refs. 5, 7). This “closing the doors behind you” on stem cell and early lineage-commitment programs seems necessary for further forward differentiation (reviewed in ref. 5), potentially explaining how PRC2 loss-of-function alterations can both favor lineage commitment and impede subsequent maturation (Fig. 1D). Notably, the H3K27me3 mark does not seem to accumulate at early lymphoid-commitment genes, and this characteristic of lymphoid maturation, different from myeloid maturation, may contribute to untempered advantage of EZH2 gain-of-function alterations specifically in lymphomagenesis and in some other cancer histologies (Fig. 1C; ref. 8). Neither does H3K27me3 accumulate at MYC target genes during normal lymphocyte or monocyte differentiation (Fig. 1C), which may be another reason why EZH2 gain-of-function can be nondetrimental to oncogenesis, and why pharmacologic inhibition of EZH2 is not expected to directly favor cell growth and division (8).

This biology can inform how we approach therapy with EZH2 inhibitors or other epigenetic drugs targeting epigenetic repression. A therapy is about not just the molecular target, but also the pathway downstream of the target. Historically, in oncology, this pathway intent has been apoptosis (cytotoxicity). The report from Ennichi and colleagues is a reminder that engaging mediators of epigenetic repression can be for other purposes, for example, releasing cancer cells to the lineage-maturation fates intended by their lineage master transcription factor content—activating MHC along the way. Ennichi and colleagues did not formally evaluate for B-cell lineage maturation of lymphoma cells by EZH2 inhibitors, but this does not diminish the importance to the wider drug-development community of considering this as a pathway goal—terminal maturation avoidance is so commonplace in cancer as to lull attention. Importantly from a clinical perspective, cell-cycle exits by terminal maturation do not require the frequently genetically deleted p53/p16 apoptosis axis that characterizes chemorefractory DLBCL and other cancers with the worst prognoses (5).

Thus, the results from Ennichi and colleagues add significantly to an expanding body of work prompting drug development for specific epigenetic targets, guided by specific cancer lineage and mutation considerations, bearing in mind the biological lesson that the consequences of inhibiting EZH2 depend very much on lineage and maturation context (Fig. 1D). From an immunotherapy perspective, it is pertinent to recognize that inhibition of specific epigenetic proteins might also affect T-cell maturation and phenotype in ways that might influence cancer immunotherapy; for example, tumor-infiltrating regulatory T cells have been shown to have higher levels of EZH2 and H3K27me3 than tumor-infiltrating cytotoxic T cells (9). Further, these epigenetic targets can be engaged not necessarily for cytotoxicity, but to cytoreduce tumors by terminal maturation instead; this would simultaneously spare immune-effectors while increasing immune visibility of the cancer, and thus more effectively enhance the efficacy of ICB than broadly cytotoxic treatments (10). Preclinical studies can identify minimal drug concentrations and doses needed to engage the epigenetic target, avoiding the usual instinct to escalate toward cytotoxicity, measuring instead for cytoreduction via terminal maturation and for accompanying MHC upregulation; this can entail multiday time scales, substantially longer than needed to observe relatively rapid cytotoxic effects. Separation of epigenetic target engagement from cytotoxicity may not always be possible, because some epigenetic targets, for example histone deacetylases, may have pleiotropic functions, such that even “on-target” drug effects cause antimetabolite effects and cytotoxicity to normal cells. And of course, drugs themselves may not be so specific, causing antimetabolite effects and cytotoxicity from “off-target” actions. Because there are only a few clinical drugs targeting transcription-repressing enzymes, there are good reasons to identify and develop new compound scaffolds (8). Finally, for small-molecule drugs in which preclinical studies demonstrate the potential to separate epigenetic from nonspecific antimetabolite effects, clinical dose and schedule can also be configured for these goals, recognizing that in some treatment contexts, for example combination with ICB, or treatment of chemorefractory p53- or p16-null cancers, cytotoxicity may be less a virtue than it is a vice (5, 10).

V. Velcheti is a consultant/advisory board member for Genentech, Bristol-Myers Squibb, AstraZeneca, Reddy Labs, Takeda Oncology, Foundation Medicine, Alkermes, Nektar Therapeutics, and Amgen. K.-K. Wong reports receiving commercial research grants from MedImmune, Takeda, Targimmune, Bristol-Myers Squibb, AstraZeneca, Janssen, Pfizer, Novartis, Merck, and Ono; has ownership interest (including stocks, patents, etc.) in G1 Therapeutics; and is a consultant/advisory board member for Array. Y. Saunthararajah is a consultant/advisory board member/owner for EpiDestiny.

1.
Lesokhin
AM
,
Ansell
SM
,
Armand
P
,
Scott
EC
,
Halwani
A
,
Gutierrez
M
, et al
Nivolumab in patients with relapsed or refractory hematologic malignancy: preliminary results of a phase Ib study
.
J Clin Oncol
2016
;
34
:
2698
704
.
2.
Rimsza
LM
,
Roberts
RA
,
Campo
E
,
Grogan
TM
,
Bea
S
,
Salaverria
I
, et al
Loss of major histocompatibility class II expression in non-immune-privileged site diffuse large B-cell lymphoma is highly coordinated and not due to chromosomal deletions
.
Blood
2006
;
107
:
1101
7
.
3.
Ennishi
D
,
Takata
K
,
Béguelin
W
,
Duns
G
,
Mottok
A
,
Farinha
P
, et al
Molecular and genetic characterization of MHC deficiency identifies EZH2 as therapeutic target for enhancing immune recognition
.
Cancer Discov
2019
;
9
:
546
63
.
4.
Beguelin
W
,
Teater
M
,
Gearhart
MD
,
Calvo Fernandez
MT
,
Goldstein
RL
,
Cardenas
MG
, et al
EZH2 and BCL6 cooperate to assemble CBX8-BCOR complex to repress bivalent promoters, mediate germinal center formation and lymphomagenesis
.
Cancer Cell
2016
;
30
:
197
213
.
5.
Velcheti
V
,
Schrump
D
,
Saunthararajah
Y
. 
Ultimate precision: targeting cancer but not normal self-replication
.
Am Soc Clin Oncol Educ Book
2018
;
38
:
950
63
.
6.
Saunthararajah
Y
,
Maciejewski
J
. 
Polycomb segment myeloid malignancies
.
Blood
2012
;
119
:
1097
8
.
7.
Gu
X
,
Ebrahem
Q
,
Mahfouz
RZ
,
Hasipek
M
,
Enane
F
,
Radivoyevitch
T
, et al
Leukemogenic nucleophosmin mutation disrupts the transcription factor hub that regulates granulomonocytic fates
.
J Clin Invest
2018
;
128
:
4260
79
.
8.
Zhang
H
,
Qi
J
,
Reyes
JM
,
Li
L
,
Rao
PK
,
Li
F
, et al
Oncogenic deregulation of EZH2 as an opportunity for targeted therapy in lung cancer
.
Cancer Discov
2016
;
6
:
1006
21
.
9.
Wang
D
,
Quiros
J
,
Mahuron
K
,
Pai
CC
,
Ranzani
V
,
Young
A
, et al
Targeting EZH2 reprograms intratumoral regulatory T cells to enhance cancer immunity
.
Cell Rep
2018
;
23
:
3262
74
.
10.
Helleday
T
. 
Making immunotherapy “cold” tumours “hot” by chemotherapy-induced mutations - a misconception
.
Ann Oncol
2019 Jan 30
[Epub ahead of print]
.