From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
Regular Article
LYMPHOID NEOPLASIA
Transcriptome sequencing reveals a profile that corresponds to genomic
variants in Waldenström macroglobulinemia
Zachary R. Hunter,1,2 Lian Xu,1 Guang Yang,1,2 Nicholas Tsakmaklis,1 Josephine M. Vos,1 Xia Liu,1 Jie Chen,1
Robert J. Manning,1 Jiaji G. Chen,1 Philip Brodsky,1 Christopher J. Patterson,1 Joshua Gustine,1 Toni Dubeau,1
Jorge J. Castillo,1,2 Kenneth C. Anderson,2,3 Nikhil M. Munshi,2-4 and Steven P. Treon1,2
1
Bing Center for Waldenström’s Macroglobulinemia, Dana-Farber Cancer Institute, Boston, MA; 2Department of Medicine, Harvard Medical School, Boston,
MA; 3Jerome Lipper Multiple Myeloma Center, Dana-Farber Cancer Institute, Boston, MA; and 4Department of Medicine, Boston VA Healthcare System,
West Roxbury, MA
Whole-genome sequencing has identified highly prevalent somatic mutations including
MYD88, CXCR4, and ARID1A in Waldenström macroglobulinemia (WM). The impact of these
and other somatic mutations on transcriptional regulation in WM remains to be clarified. We
• Transcription profiles
performed next-generation transcriptional profiling in 57 WM patients and compared
associated with mutated
findings to healthy donor B cells. Compared with healthy donors, WM patient samples
MYD88, CXCR4, ARID1A,
showed greatly enhanced expression of the VDJ recombination genes DNTT, RAG1, and
abnormal cytogenetics
RAG2, but not AICDA. Genes related to CXCR4 signaling were also upregulated and
including 6q2, and familial
included CXCR4, CXCL12, and VCAM1 regardless of CXCR4 mutation status, indicating a
WM are described.
potential role for CXCR4 signaling in all WM patients. The WM transcriptional profile was
• Mutated CXCR4 profiles
equally dissimilar to healthy memory B cells and circulating B cells likely due increased
show impaired expression
differentiation rather than cellular origin. The profile for CXCR4 mutations corresponded
of the tumor suppressor
to diminished B-cell differentiation and suppression of tumor suppressors upregulated
response induced by
by MYD88 mutations in a manner associated with the suppression of TLR4 signaling
MYD88L265P and also
relative to those mutated for MYD88 alone. Promoter methylation studies of top findings
failed to explain this suppressive effect but identified aberrant methylation patterns in
G-protein/MAPK inhibitors.
MYD88 wild-type patients. CXCR4 and MYD88 transcription were negatively correlated,
demonstrated allele-specific transcription bias, and, along with CXCL13, were associated with bone marrow disease involvement.
Distinct gene expression profiles for patients with wild-type MYD88, mutated ARID1A, familial predisposition to WM, chr6q
deletions, chr3q amplifications, and trisomy 4 are also described. The findings provide novel insights into the molecular
pathogenesis and opportunities for targeted therapeutic strategies for WM. (Blood. 2016;128(6):827-838)
Key Points
Introduction
Whole-genome sequencing identified several highly recurring somatic
mutations in patients with Waldenström macroglobulinemia (WM).1,2
In over 90% of WM patients, a single point mutation at NM_002468:
c.978T.C (rs387907272) in MYD88 is found, resulting in a
p.Leu265Pro (L265P) amino acid change.1,3 MYD88 is an adaptor for
Toll-like (TLR) and interleukin 1 (IL1) receptors, and the MYD88L265P
mutation triggers constitutive activation of NF-kB through IRAK and
BTK.1,4,5 Mutated MYD88 WM patients show greater overall survival
and clinical responses to the BTK inhibitor ibrutinib.6,7
Activating CXCR4 frameshift or nonsense mutations in the
C-terminal tail are found in 30% to 40% of WM patients, are primarily
subclonal, and almost always associated with MYD88L265P.2,3,8 These
somatic mutations are similar to the causal germ line variants that
underlie WHIM (warts, hypogammaglobulinemia, infection, and
myelokathexis) syndrome.2,9 In WM, somatic CXCR4 mutations
(CXCR4WHIM) are determinants of disease presentation, as well as
resistance to ibrutinib.3,6,10 Somatic mutations in ARID1A, a member of
the SWItch/sucrose nonfermentable family and epigenetic regulator, are also found in 20% of WM patients. Gene losses affecting
NF-kB signaling (ie, HIVEP2, TNFAIP3), as well as the ARID1A
homolog ARID1B are present in most WM patients.2,11 Deletions in
chromosome 6q with and without concurrent 6p amplifications,
trisomy 3, and amplifications of 3q, as well as trisomy 4, are also
commonly found in WM.12-14 Previous array-based gene expression studies of WM were largely conducted prior to these genomic
discoveries and therefore the effects of recurrent somatic events on
transcriptional regulation remain to be clarified.15-17 We therefore
performed next-generation RNA sequencing in 57 WM patients
and compared findings to sorted healthy donor-derived nonmemory
(CD191CD272) and memory (CD191CD271) B cells. The latter
Submitted March 30, 2016; accepted June 3, 2016. Prepublished online as
Blood First Edition paper, June 14, 2016; DOI 10.1182/blood-2016-03708263.
There is an Inside Blood Commentary on this article in this issue.
The online version of this article contains a data supplement.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
The publication costs of this article were defrayed in part by page charge
payment. Therefore, and solely to indicate this fact, this article is hereby
marked “advertisement” in accordance with 18 USC section 1734.
827
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
828
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
HUNTER et al
represent the B-cell population from where most cases of WM are
thought to be derived.18,19
Methods
nonsignificant genes are distributed around zero, genes were further filtered for
those with permutation importance scores greater than the absolute value of the
lowest scoring predictor. For reproducibility, the random seed for the analysis
was set to 265. Principal component analysis (PCA) was conducted using R’s
prcomp, Pearson’s product-moment coefficient was used for correlation studies,
and cohort mean values were compared using the Wilcoxon rank sum test.
Sample selection and characterization
Gene expression, sequencing, and methylation studies
Bone marrow (BM) aspirates were collected from 57 patients with the WM
consensus diagnosis.20 Participants provided informed consent for sample
collection per the Dana-Farber/Harvard Cancer Center Institutional Review Board.
WM cells were isolated by CD191 magnetic-activated cell sorting (MACS)
microbead selection (Miltenyi Biotec, Auburn, CA) from Ficoll-Paque
(Amersham-Pharmacia Biotech, Piscataway, NJ) separated BM mononuclear
cells. Peripheral blood mononuclear cells from nine healthy donors (HDs) were
sorted for nonmemory (CD191CD272) and memory B cells (CD191CD271)
using a memory B-cell isolation kit (Miltenyi Biotec). RNA and DNA were purified
using the AllPrep mini kit (Qiagen, Valencia, CA). Most samples were previously
characterized by whole-genome sequencing and all samples were screened for
MYD88 and CXCR4 gene mutations by Sanger sequencing.2 MYD88L265P and
CXCR4 c.1013C.G and c.1013C.A mutations were analyzed by allele-specific
polymerase chain reaction (PCR) as previously described.8,21
Gene expression results were validated by using TaqMan gene expression assays
Hs01027785_m1 (DUSP4), hs00244839_m1 (DUSP5), Hs00243182_m1 (RGS13),
Hs00892674_m1 (RGS16), Hs00698249_m1 (RASSF6), Hs00396602_m1
(WNK2), Hs00924602_m1 (PRDM5) (Thermo Fisher Scientific, Waltham MA).
Methylation-specific PCR assays for RASSF6, WNK2, and PRDM5 were
conducted on bisulfite-converted DNA using previously established
protocols.31-33
Next-generation sequencing and analysis
Transcriptome profiling was conducted by the Center for Cancer Computational
Biology at the Dana-Farber Cancer Institute (Boston, MA) using the NEBNext
Ultra RNA library prep kit (New England BioLabs, Ipswich, MA). The paired-end
samples were run 2 per lane for 50 cycles on an Illumina HiSeq (Illumina, San
Diego, CA). Read-level data are available through dbGAP accession (applied).
Reads were aligned to KnownGene HG19/GRCh37 reference using STAR
(Spliced Transcripts Alignment to a Reference).22 Genes with mean raw read
counts of ,10 were not analyzed, leaving 16 888 expressed genes for analysis.
Read counts per gene were obtained using featureCounts from Rsubread, and
analyzed using voom from the edgeR/limma Bioconductor packages in R
(R Foundation for Statistical Computing, Vienna, Austria).23-27 Differential
expression models accounted for sex, prior treatment, as well as MYD88 and
CXCR4 mutation status. A false discovery rate (FDR) cutoff of 10% was used to
determine significant differentially expressed genes. Functional enrichment analysis
was conducted using Ingenuity Pathway Analysis (Qiagen). Clustering and
correlation analysis was conducted using the variance stabilizing transformation of
the count data from the Bioconductor DESeq2 package.28 In all other cases,
estimates of gene expression levels are represented in transcripts per million (TpM).
Log-fold change (LFC) listed in text and tables are derived from the limma analysis.
CXCR4 transduced cell lines and gene expression analysis
Previously described BCWM.1 and MWCL-1 cell lines transduced to
express CXCR4 with or without activating mutations observed in WM
patients were used to model CXCR4-stimulated gene expression changes in
WM.10 Briefly, CXCR4 complementary DNA (cDNA) transcripts were
subcloned into plenti-internal ribosomal entry site (IRES)–green fluorescent
protein (GFP) vectors, and stably transduced using a lentiviral system.4 In
addition to wild-type (WT) CXCR4, vectors with the mutations c.1013C.G
(p.Ser338*), c.932_933insT (p.Thr311fs), and c.1030_1041delinsGT
(p.Ser344fs) were generated. Cell lines were stimulated with 50 nM of the
CXCR4 ligand CXCL12 (SDF1A) (R&D Systems, Minneapolis MN) for
2 hours. RNA from each sample was extracted at baseline and at 2 hours.
Gene expression was assessed using Affymetrix Human Gene ST 2.0 arrays
(Affymetrix, Santa Clara, CA). Array data are archived in the Gene
Expression Omnibus database (accession number GSE83150). Array data
were analyzed using the Oligo and limma R bioconductor packages.24,29
Statistical analysis
Random forest analysis was implemented using the cforest function in the party
package in R.30 Models were constructed using the cforest_unbiased function
with 20 predictors per tree over 10 000 trees. As variable importance scores of
Results
The clinical characteristics for the 57 WM patients are presented in
Table 1. The distribution of somatic mutations and cytogenetic findings
in these patients was similar to those previously described.34 The top
500 differentially expressed genes for each comparison discussed
below are available as a part of the supplemental Data (available on the
Blood Web site).
Comparing WM to healthy donor B cells
Comparison of WM and HD samples generated a list of 13 571
differentially expressed genes. To gain further insight into these gene
expression differences, this gene list was sorted by absolute log2 fold
change, revealing a marked upregulation of DNTT, RAG1, RAG2, and
IGLL1 in WM samples. These genes have important functional roles in
VDJ recombination and B-cell receptor (BCR) signaling. Other
upregulated genes relevant to WM biology and pathogenesis included
CXCL12, VCAM1, CD5L, BMP3, and IGF1. Among the top 40
differentially expressed genes sorted by absolute expression in WM as
measured in TpM were CXCR4, the WM prognostic marker B2M, and
2 genes related to BCR signaling: CD79A and CD79B. To examine
transcriptional similarities between the samples in an unbiased manner,
multidimensional scaling of the 500 genes with the highest variance
was used to best represent the data in 2 dimensions (Figure 1A).
Although peripheral B cells (PBs) and HD memory B cells (MBs)
clustered together, separation between MYD88L265PCXCR4WHIM and
MYD88L265PCXCR4WT genotyped WM patient samples was observed.
To examine whether the patterns of expression were similar between HD
samples and patient mutational groups, a correlation coefficient based
on overall gene expression was generated for each sample-to-sample
comparison. These values were averaged together by HD PB/MB cell
type and WM MYD88/CXCR4 genotype to show the extent of
correlation between groups (Figure 1B). Strong correlations were
observed across all subsets, though mean correlation values between
HD B cells and MYD88L265PCXCR4WHIM patients were significantly
higher vs those between HD B cells and MYD88L265PCXCR4WT
patients (P 5 .025 and P 5 .036 for PB and MB cells, respectively).
No preferential association was noted between WM samples and
HD B-cell type using PCA of the 596 genes differentially expressed
between HD PB and MB cells (Figure 1C; supplemental Figure 1). To
better understand the role of B-cell differentiation in WM, we analyzed
19 genes linked to the transition from MB to plasmablasts and plasma
cells (Figure 1D).35,36 Many of these genes were not only significantly
differentially expressed between HD and WM samples, but between
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
TRANSCRIPTOME OF WALDENSTRÖM MACROGLOBULINEMIA
Table 1. Patient characteristics
Median or count Range or percentage
Sex, female
21/57
Age at diagnosis, y
b2 microglobulin at diagnosis .3, mg/L
60
19/46
Age at biopsy, y
Familial history of WM
62
36.8%
40-78
41.3%
41-83
3/57
5.3%
Splenomegaly
13/57
22.8%
Lymphadenopathy
30/57
52.6%
Bone marrow involvement, %
60
Hemoglobin, g/dL
11.1
5-95
8.2-14.5
Serum IgM, mg/dL
3750
416-8320
Serum IgG, mg/dL
517
82-3890
Serum IgA, mg/dL
40
6-516
MYD88 mutations
52/57
91.2%
CXCR4 mutations
23/57
40.0%
ARID1A mutations
5/51
9.8%
CD79B mutations
4/51
7.8%
Deletion chromosome 6q
24/53
45.3%
Amplification chromosome 6p
6/53
11.3%
Amplification chromosome 3q
11/52
19.2%
Amplification chromosome 4
10/52
21.1%
WM patients based on MYD88 and CXCR4 mutation status suggesting
reduced differentiation in both MYD88L265PCXCR4WHIM and,
in particular, MYD88WTCXCR4WT samples. Separation of patient
samples by genotype can be seen in the PCA of these 19 genes in
Figure 1E. BCL2 and BCL2L1 were upregulated in all WM samples,
however, analysis of the larger BCL2 family revealed stratification
of samples by HD cell type and MYD88/CXCR4 mutation status
(supplemental Figure 2). The proapoptotic BAX gene was downregulated in WM samples and PMAIP1 was uniquely upregulated in
MYD88L265PCXCR4WT WM.
Transcriptional effects of CXCR4WHIM
Restricting the analysis to WM patients, PCA of the top 500 high
variance genes accounted for 41% of the overall variability within
the first 2 components (Figure 2A). Tumor suppressors including
WNK2, TP53I11, PRDM5, and CABLES1 influence the second
principal component that separates MYD88L265PCXCR4WT from
MYD88L265PCXCR4WHIM and MYD88WTCXCR4WT samples (supplemental Figure 1). Supervised clustering of the 3103 genes
differentially expressed between MYD88 L265P CXCR4 WT and
MYD88L265PCXCR4WHIM revealed that most expression differences
in MYD88L265PCXCR4WHIM patients followed a pattern resembling
HD samples, despite carrying the MYD88L265P mutation (Figure 2B).
The top predicted upstream regulator for this gene signature was the
inhibition of lipopolysaccharide signaling (Figure 2C). To further
explore how CXCR4 mutations modulate MYD88 signaling in WM,
the gene signature was filtered for genes involved in TLR signaling.
MYD88L265PCXCR4WHIM patients showed downregulation of the
lipopolysaccharide receptors TLR4 and NOD2, and upregulation of
TLR7 and IRAK3 (supplemental Figure 3A). The most significant
results associated with CXCR4 mutation status were silencing of genes
in patients with MYD88L265PCXCR4WHIM that were elevated in
MYD88L265PCXCR4WT patients (supplemental Figure 3B).
Promoter methylation studies
To investigate the role of methylation in genes differentially expressed
based on CXCR4 and MYD88 mutation status, 3 such genes were
selected for screening based on their regulation by promoter
829
methylation in other malignancies: PRDM5, WNK2, and RASSF6.32,33,37
Gene expression was validated by PCR in HD B-cell samples as well as
in MYD88 and CXCR4 genotyped WM patient samples (supplemental Figure 4A). Mirroring the transcriptome data, RASSF6 expression
was elevated in MYD88L265PCXCR4WT and MYD88L265PCXCR4WHIM
samples but not in HD or MYD88WT samples whereas WNK2 and
PRDM5 were elevated only in the MYD88L265PCXCR4WT patients.
To determine relevance of these methylation sites in hematologic
malignancies, WM, B-cell lymphoma, and myeloma cell lines were
assessed for promoter methylation using methylation-specific PCR
assays (supplemental Figure 4B). Sixteen MYD88 and CXCR4
genotyped primary WM patient samples and 4 healthy donor B-cell
samples were then tested (supplemental Figure 4C). Robust promoter methylation was observed for WNK2 and PRDM5 in the
MYD88WTCXCR4WT patients but was absent in HD samples. Partial
promoter methylation of WNK2 and PRDM5 was also observed in
some MYD88L265P samples. This correlated to a relative reduction in
gene expression only for WNK2 in the MYD88L265PCXCR4WHIM
genotyped patient sample with the strongest methylation signal.
Cell line models of CXCR4WHIM signaling
To better understand interactions between CXCR4 WHIM and
MYD88L265P in WM, lentiviral transduction was performed to
overexpress CXCR4WT, or 1 of 3 documented CXCR4WHIM transcripts (c.1013C.G, c.932_933insT, or c.1030_1041delinsGT) in
MYD88L265P expressing BCWM.1 WM cells. Transduced cells were
stimulated in triplicate with CXCL12 for 2 hours and gene expression
profiling performed. The relative change in gene expression from
baseline was then compared between CXCR4WT and CXCR4WHIM
transduced cells revealing 61 differentially expressed genes. Of these
genes, 16 of 61 (26.2%) were encompassed in the gene expression
signature for MYD88L265PCXCR4WHIM patient samples (supplemental
Table 1). In 8 of 16 (50%) of these genes, the direction of the change
relative to MYD88L265PCXCR4WT was same as that observed in the
patient samples. However, it was the genes whose direction of change
was opposed that revealed the most about the underlying biology. These
discordant genes were enriched for suppressors of MAPK and G-protein
signaling downstream of CXCR4. They were preferentially induced
in CXCR4WHIM cell lines by CXCL12 stimulation and suppressed
in MYD88L265PCXCR4WHIM patient samples.10,38 Interrogation of
these dual specificity phosphatases (DUSP) and regulator of G-protein
signaling (RGS) genes revealed RGS1, RGS2, RGS13, DUSP1, DUSP2,
DUSP4, DUSP5, DUSP10, DUSP16, and DUSP22 were all significantly
suppressed in MYD88L265PCXCR4WHIM vs MYD88L265PCXCR4WT
patients. Expression in patients of DUSP4, DUSP5, RGS13, and
RGS16, which were selected for validation based on the cell line studies,
can be seen in supplemental Figure 5A. The increased expression of these
genes in CXCR4WHIM compared with CXCR4WT transduced lines was
validated by real-time PCR (supplemental Figure 5B). Dose-response
curves for these genes following CXCL12 stimulation was conducted in
transduced BCWM.1 and MWCL-1 (supplemental Figure 5C).
Gene signatures associated with other clinical and
genomic features
For the 5 patients with ARID1A mutations, 16 differentially
expressed genes were observed. The ARID1A-mutated population
also demonstrated elevated BM disease infiltration (median, 90%;
range, 70%-95%) when compared with non-ARID1A-mutated
patients (median, 58%; range, 5%-95%; P 5 .0259). No statistically
significant gene expression changes were observed in the 4 patients
with CD79B mutations. Analysis based on somatic cytogenetic
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
830
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
HUNTER et al
A
B
MYD88L265PCXCR4WT
MYD88L265PCXCR4WHIM
MYD88WTCXCR4WT
Memory B-Cell
Peripheral Blood B-Cell
Leading logFC dim 2
3
2
MC
1
M
M
M
MC
MMM
–2
M
value
1.0
MYD88L265P
CXCR4WT
0.9
0.8
Memory
B-Cell
Pheripheral
B-Cell
M
PB
–4
MYD88L265P
CXCR4WHIM
M
M
MC
MC
M
M
MCM
MC MCMC
WT
MMC
MM
MC
M
MC
MC
M
MC M MM M
MC
MC
M
M
MC
WT
M
M
MC
MC M
MC
M
WT
WT
MB
PB
MB
PB
MB PB
MB MB
PB
MBMB
PB PB
PBMB
–1
MYD88WT
CXCR4WT
M
MC
MC
WT
MB
PB
0
M
–2
0
2
Pheripheral Memory MYD88L265P MYD88L265P MYD88WT
CXCR4WT CXCR4WHIM CXCR4WT
B-Cell
B-Cell
4
Leading logFC dim 1
C
E
10
0
–20 –10
–30
–4
–2
0
2
4
–62 0
2
4
6
20
4
15
5
0
PC3
–5
PC3
MYD88L265PCXCR4WT
MYD88L265PCXCR4WHIM
MYD88WTCXCR4WT
Memory B-Cell
Peripheral Blood B-Cell
10
2
0
–2
–10
PC1
20
10
PC2
0
–10
PC2 –20
#
PC1
D
PRDM1 XBP1 IRF4 SPIB CD19 CD38 CXCR5 BCL6 CCR2 IL4R
BLK MS4A1 SDC1 CD27 PAX5 BACH2 EBF1 SOX5 IRF8
Transcripts per Million (Log 2)
12
#
#
*#
10
*#
8
#
*#
6
4
*
*
*#
#
#
#
*#
*
*
**
#
*
#*
*#
**
#
# #
*#
*#
#
*
#
#
*
*#
#
*#
**
##
*##
*
#
#
*
*#
#
*#
**
*
*
#
*#
#
*
* #*# **
*#
*
*
#
#
*
2
**
0
MYD88L265PCXCR4WT
MYD88L265PCXCR4WHIM
HD Memory B-cell
HD Peripheral Blood B-cell
MYD88WTCXCR4WT
*
#
P < .05 versus MYD88L265PCXCR4WT
P < .05 versus HD Memory B-cells
Figure 1. Comparisons of transcriptional activity for genotyped WM patient samples and healthy donor B cells. (A) Multidimensional scaling of the top 500 genes for all
HD and WM samples using Voom transformed expression values in Limma’s plotMDS function.24 (B) Pearson correlation coefficients from the correlation matrix of gene expression were
averaged over all the samples for each major group. Correlation values with the HD samples were higher for those WM patients with CXCR4 mutations (P 5 .025 and P 5 .036 for HD PB
and HD MB cells, respectively). (C) The 596 genes differentially expressed between the HD PB and HD MB cells were analyzed across all samples using PCA. No preferential relationship
between WM samples and HD B-cell type was observed. The first 3 principal components were used to plot the samples representing 54% of the overall variability. (D) Median expression
levels for 19 genes known to regulate B-cell to plasma cell transition for HD B-cell and genotyped WM samples. Error bars represent full data range per for each group. (E) PCA of the 19
genes related to B-cell to plasma cell transition shown in panel D depicting the first 3 principal components which covered over 65% of the overall gene expression variability.
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
TRANSCRIPTOME OF WALDENSTRÖM MACROGLOBULINEMIA
A
831
C
Predicted
Upstream Regulator Activation
lipopolysaccharide
Inhibited
IFNG
TGFB1
Interferon alpha
TNF
TREM1
15
10
PC2
5
Activation
z-score
P-value
0
–2.068
–0.28
1.622
0.076
0.558
–1.411
1.08E-11
1.27E-11
6.13E-11
4.22E-10
2.45E-09
3.34E-08
–5
–10
–20
–10
0
10
20
30
PC1
B
MYD88L265P
WT
CXCR4
MYD88WT
CXCR4
Nonsense
Frame Shift
CXCR4
HD B-Cells
+
CD19 CD27+
CD19+CD27–
Figure 2. Role of CXCR4WHIM mutations in MYD88L265P-mutated WM. (A) PCA of the top 500 high variance genes within the WM patient samples using the first 2 principal
components. WM patients with the MYD88L265PCXCR4WT genotype are shown in red; patients mutated for MYD88 with CXCR4 nonsense (NS) or frameshift (FS) mutations
are shown in dark and light green, respectively; and patients with MYD88WTCXCR4WT are shown in purple. (B) Heat map of the 3103 genes differentially expressed between
the MYD88L265PCXCR4WT and MYD88L265PCXCR4WHIM genotyped samples. Gene order was determined using hierarchical clustering of MYD88L265PCXCR4WT and
MYD88L265PCXCR4WHIM expression data whereas the samples where arranged by genotype. Expression data for MYD88WTCXCR4WT WM and healthy donor B-cell samples
were added to the heat map for comparison with intact clustered gene order. (C) Top predicted upstream targets of the differentially expressed genes from Ingenuity Pathway
Analysis.
abnormalities included deletion of chromosome 6q (131 genes),
amplification of chromosome 6p (65 genes), amplification of chr3q
including trisomy 3 (11 genes), and trisomy 4 (776 genes). A distinct
gene expression signature for 263 genes was also observed in the
3 patients who had first-degree relatives with WM. No statistically
significant differences in gene expression were observed based on
presence of extramedullary disease, elevated B2M levels (.3 mg/dL),
or International Waldenström Macroglobulinemia Prognostic Scoring
System score.
To identify genes associated with relevant WM clinical parameters,
all genes were tested for correlation with serum immunoglobulin
M (IgM) levels, hemoglobin, and BM disease involvement. After
filtering the results by significance (P , .05) and the absolute value of
the correlation estimate .0.3, 312 genes were found to be associated
with IgM, 965 with hemoglobin, and 3738 with BM involvement.
These genes were combined with MYD88/CXCR4 genotype, age,
sex, and treatment status in a random forest regression analysis for
the first 37 study patients, leaving the remaining 20 patients for
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
HUNTER et al
A
Serum IgM
SHF
6000
6000
6000
4000
4000
4000
2000
2000
2000
Serum IgM (mg/dl)
0
0.0 0.5 1.0 1.5 2.0 2.5
0
1.5
BDH2
2.0
2.5
1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
RSPH3
CEPZ0S
6000
6000
6000
4000
4000
4000
2000
2000
2000
0
0
1.0
1.5
TNFRSF17
2.0
ITGA3
6000
4000
4000
4000
2000
2000
2000
0
4
5
6
7
4000
2000
BMP3
6000
0
6000
0
4.4 4.6 4.8 5.0 5.2 5.4 5.6
2.5
6000
3
8000
0
1.0
2.0 2.5 3.0 3.5 4.0 4.5
Final Model
IRS1
Serum IgM (mg/dl)
AJUBA
0
0
4000
6000
8000
Adjusted R-squared: 0.4 Training RMSE: 1,270.3
0
1.0 1.5 2.0 2.5 3.0 3.5
2000
Predicted Levels (mg/dl)
2
4
6
8
P-value: 3.129e-5
Validation RMSE: 1,340.1
Gene Expression (TpM)
B
Bone Marrow Involvement
AKAP1
80
80
80
60
60
60
40
40
40
20
20
20
3.0 3.5 4.0 4.5 5.0 5.5
2.5
3.0
CXCL13
3.5
4.0
4.5
3.5 4.0 4.5 5.0 5.5 6.0 6.5
CXCR4
MYD88
80
80
80
60
60
60
40
40
40
20
20
0
1
2
3
4
5
20
10
TP53
11
12
13
5.0 5.5 6.0 6.5 7.0
TOP3A
BCL7A
80
80
80
60
60
60
40
40
40
20
Final Model
KDM1B
20
Bone Marrow Involvement (%)
Bone Marrow Involvement (%)
CDC23
100
80
60
40
20
0
0
20
40
60
80
100
Predicted Levels (%)
20
Adjusted R-squared: 0.5 Training RMSE: 16.1
3.5 4.0 4.5 5.0 5.5 6.0 6.5
2.5 3.0 3.5 4.0 4.5
2
4
6
8
P-value: 3.469e-7
Validation RMSE: 19.3
Gene Expression (TpM)
C
Hemoglobin Levels
SEC62
14
13
12
11
10
9
8
0.0 0.5 1.0 1.5 2.0 2.5
14
13
12
11
10
9
8
2.0
NOTCH1
8.0
8.5
3
4
5
14
13
12
11
10
9
8
0
1
2
3
4
5
3.5
6.0 6.5 7.0 7.5 8.0 8.5
CD46
14
13
12
11
10
9
8
3.0
14
13
12
11
10
9
8
2
CXCL13
2.5
PIK3AP1
14
13
12
11
10
9
8
7.5
15
14
13
12
11
10
9
8
5.5 6.0 6.5 7.0 7.5
OS9
14
13
12
11
10
9
8
Final Model
NKIRAS1
Hemoglobin Levels (g/dl)
AJUBA
Hemoglobin (g/dl)
832
PIK3CD
13
12
11
10
9
8
14
13
12
11
10
9
8
4.5 5.0 5.5 6.0 6.5 7.0 7.5
14
8
9
10
11
12
13
14
15
Predicted Levels (g/dl)
Adjusted R-squared: 0.3 Training RMSE: 0.95
4.5 5.0 5.5 6.0 6.5 7.0
P-value: 2.354e-4
Gene Expression (TpM)
Figure 3.
Validation RMSE: 1.492
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
TRANSCRIPTOME OF WALDENSTRÖM MACROGLOBULINEMIA
833
A
70%
Percent of Reads
Supporting Mutant Allele
60%
MYD88
CXCR4
50%
40%
30%
20%
10%
0%
WM1
WM2
WM3
WM4
WM5
WM6
WM7
WM8
WM9
WM10
WM11
S338 C/G S338 C/A S338 C/G S338 C/A S338 C/A R334 C/T S338 C/A S338 C/G S338 C/A S338 C/G S338 C/G
B
CXCR4WHIM-S338X
MYD88L265P
DNA
RNA
DNA
WM1
L265P T/C
WM1
S338X C/G
WM7
L265P T/C
WM7
S338X C/A
WM8
L265P T/C
WM8
S338X C/G
WM10
L265P T/C
WM10
S338X C/G
RNA
Figure 4. Mutant allele fraction analysis for MYD88 and CXCR4 in concurrent DNA and RNA WM patient samples. (A) Percentage of reads supporting the mutant allele
of MYD88 and CXCR4 for 11 patients with MYD88L265P and nonsense (NS) CXCR4WHIM mutations. Median coverage of the affected nucleotide was 193 (range, 38-456)
reads for MYD88 and 11 760 (range, 3318-15 883) reads for CXCR4. The amino acid location and nucleotide change for the CXCR4WHIM-NS mutation is listed for each patient
sample. (B) Sanger sequencing of paired DNA and cDNA validated the mutant allele fraction for CXCR4 and MYD88 at the RNA and DNA levels.
cross-validation. Filtering genes/predictors by variable importance
score resulted in 40, 236, and 143 associated genes for IgM, BM,
and hemoglobin, respectively (supplemental Figure 6). Functional
analysis was conducted on each gene list and top genes from each
group were selected based on biological significance. The 40 genes
associated with IgM were predicted to be downstream of IL6 signaling
(P 5 .0018) and included many genes of WM pathogenic interest
including TNFRSF17, BMP3, IRS1, and CEPZOS (Figure 3A). Many
genes relevant to WM biology, including CXCL13, TP53, CXCR4,
MYD88, CDC23, and AKAP1, were associated with BM disease
involvement (Figure 3B). Although MYD88 and CXCR4 were both
correlated to BM disease involvement (r 5 20.50; P , .0001 and
r 5 0.46; P 5 .0003, respectively), they were negatively correlated with
each other (r 5 20.46; P 5 .0004). As shown in supplemental Figure 7,
this relationship may be influenced by MD88/CXCR4 mutation status.
The gene list associated with hemoglobin included CXCL13, as well as
PIK3AP1, PIK3CD, AJUBA, and OS9 (Figure 3C). Based on these
findings, CXCL13 was included in an independent serum cytokine
profiling study of 86 WM patients that validated the CXCL13
correlation between BM (Spearman r 5 0.562; P 5 2.192 3 1028)
and hemoglobin (Spearman r 5 20.567; P 5 1.53 3 1028).39
Analysis of mutant allele burden
The percentage of reads supporting the mutant allele relative to WT
allele was calculated for MYD88, CXCR4, ARID1A, and CD79B in the
Figure 3. Random forest analysis of gene expression identifies genes predictive of important clinical features in WM patients. Gene expression from the first 37 WM
samples were analyzed for their utility in predicting serum IgM (A), BM disease involvement (B), and hemoglobin (C) levels using a random forest analysis. Of the statistically
significant genes, the top 9 genes from each group deemed the most biologically relevant are shown as single variable correlates and incorporated into a final linear model
using the full data set to demonstrate their predictive utility. The final 20 samples withheld for validation are shown in red and the root mean squared error (RMSE) of the final
model for both the training and validation subsets is shown. P values have been adjusted for multiple hypotheses testing using the Benjamini-Hochberg FDR.
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
834
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
HUNTER et al
MYD88L265P
WM Associated Genes
CXCR4WT
MYD88WT HD
CXCR4WHIM
MB
DNTT
RAG1
RAG2
IGF1
BMP3
CD5L
CXCL12
VCAM1
CXCR4
B2M
BCL2
BAX
PB
WM (All Samples)
71.146
7.0669
5.1208
3.2971
21.157
8.1506
3.44
7.2862
3607.9
11916
200.75
105.47
(0.5-2884.9)
(0.2-255.5)
(0-163.3)
(0.1-28.5)
(0.5-269.5)
(0.3-372.6)
(0.1-168.5)
(0.1-383.3)
(577.9-8578.8)
(5501.6-18419.1)
(25-392.3)
(56.1-217.5)
MYD88L265PCXCR4WT
MYD88L265PCXCR4WT
IL17RB
GPER
WNT5A
IGF1
WNK2
PRDM5
CABLES1
CXXC4
CDKN1C
LGALS3BP
PMAIP1
TP53I11
81.364
42.693
4.2895
5.9519
11.39
6.5371
17.641
2.5257
16.826
73.226
759.69
2.7809
(13.2-333.4)
(10.2-167.1)
(0-28.2)
(1.5-28.5)
(0.7-65)
(0.2-31.2)
(5.7-58.8)
(0-21.2)
(4.2-92)
(8.6-396.9)
(207.3-2711.9)
(0.6-16.4)
MYD88L265PCXCR4WHIM
MAPK/G-protien Regulators
MYD88L265PCXCR4WT
IL15
ERRFI1
NOD2
TLR4
TLR7
IRAK3
CXCR7
TSPAN33
PIK3R5
PIK3CG
RGS1
RGS2
RGS13
DUSP1
DUSP2
DUSP4
DUSP5
DUSP10
DUSP16
DUSP22
10.452
1.5737
3.2243
11.829
1.3006
1.6527
0.755
40.552
4.9129
29.595
(3.8-18)
(0.4-3.8)
(0.3-8.8)
(2.7-53.2)
(0.1-28.7)
(0.2-22.7)
(0.2-25)
(16.7-101.9)
(0.3-25.6)
(15-55.2)
MYD88L265PCXCR4WT
395.04
725.64
3.1815
556.48
46.644
2.2093
122.52
37.776
7.1218
275.97
(47.5-1870.9)
(37.4-2213.7)
(0.1-163.5)
(158.8-1554.4)
(18.2-1234.7)
(0.1-14.8)
(22.1-963.4)
(5.7-243.1)
(1.1-19.2)
(67.8-1730.3)
MYD88WTCXCR4WT
MYD88L265PCXCR4WT
IL6
IRAK2
TNFAIP3
NFKBIZ
NFKB2
TIRAP
PIM1
PIM2
CD40
PTBP3
CD86
CXCR3
IGF1R
AKT2
PIK3AP1
30.589
18.886
58.59
77.897
38.611
10.843
480.79
555.15
81.62
85.417
13.798
2.5943
0.5259
48.196
130.5
(6.6-210.8)
(8.1-59.5)
(26.7-215.2)
(26.9-271.8)
(18.7-105.5)
(4.2-26.9)
(179.9-968.5)
(289.9-981.7)
(23.6-108.1)
(46.5-115)
(0.7-41.4)
(0.2-15.6)
(0-14.6)
(36.4-67.8)
(51.9-324.2)
Healthy Donor - MB
0.023
0.038
0
0.018
0.127
0.021
0
0
1160
5563
87.24
205.3
(0-0.1)
(0-0.1)
(0-0)
(0-0)
(0-0.3)
(0-0.1)
(0-0)
(0-0.1)
(913.8-1955)
(4920.4-5862.5)
(72.4-109.4)
(182.3-244.1)
P-value
0.0005
0.0004
<0.0001
<0.0001
<0.0001
0.0001
0.0001
0.0001
<0.0001
<0.0001
<0.0001
<0.0001
MYD88L265PCXCR4WHIM P-value
19.27
3.995
0.077
1.883
0.265
0.127
2.068
0.166
4.05
6.483
177.2
(4.5-151.8)
(0.6-59.3)
(0-7.1)
(0.1-19.9)
(0-19.5)
(0-39.1)
(0.1-37.8)
(0-16.3)
(0.5-44.9)
(0.9-139.6)
(91.6-1114.7)
19.68 (0.4-36.9)
<0.0001
<0.0001
0.0001
<0.0001
<0.0001
<0.0001
<0.0001
0.0071
0.0002
<0.0001
<0.0001
<0.0001
MYD88L265PCXCR4WHIM P-value
1.158
0.665
0.958
4.55
5.162
6.08
8.855
109.6
23.61
45.76
(0.1-13.9)
(0.3-1.9)
(0.2-6.6)
(0.6-48.4)
(0.2-13.7)
(0.5-40.7)
(0.5-50.6)
(28.4-184.6)
(1.3-47.7)
(5-69.1)
MYD88L265PCXCR4WHIM
<0.0001
0.0022
0.0032
0.0075
0.0114
0.0039
<0.0001
<0.0001
<0.0001
0.0909
P-value
72.21
276.8
0.478
348.6
19.44
0.669
(16.8-646)
(63.1-1409.2)
(0-5.5)
(154.8-714.1)
(8.3-460)
(0-5.6)
0.0001
0.0175
0.0058
0.0184
0.0522
0.0497
78.88
13.46
2.489
85.08
(23.5-189)
(3.9-105.7)
(0.6-14.2)
(48.4-428.1)
0.0570
0.0125
0.0001
<0.0001
MYD88WTCXCR4WT
4.618
8.222
21.74
24.1
25.08
7.856
104.9
381.3
44.25
140.9
40.67
12.19
1.911
53.46
210.6
(0.3-13.9)
(1.2-18)
(10.5-61.3)
(10.1-88)
(9.1-41)
(3-10.8)
(50.2-309.1)
(73.9-433.8)
(5.8-98.6)
(106.6-173.8)
(15.5-214.6)
(0.6-93.8)
(1.1-4.1)
(47-85.7)
(117.7-454.2)
P-value
0.0150
0.0335
0.0231
0.0314
0.0024
0.1000
0.0002
0.0008
0.0110
0.0095
0.0179
0.0146
0.0349
0.0438
0.0342
Figure 5. Summary of key gene expression differences. Gene expression levels for all study samples are shown depicting key differences for WM vs normal B-cell
samples as well as WM patient intrapatient differences based on MYD88 and CXCR4 genotype. The heat map represents log2 transformed TpM values with median and
range TpM values listed for each relevant comparator. P values have been adjusted for multiple hypotheses testing using the Benjamini-Hochberg FDR.
RNA sequencing data. Although the relevant mutations were observed
in the RNA in all cases, many samples with CXCR4WHIM mutations
had mutant allele burdens in excess of 50%, an unexpected finding
for a typically subclonal variant.8 As there can be differences in
mapping efficiencies between indels and single-nucleotide variants,
MYD88L265P allele burden was then compared with the CXCR4WHIM
allele burden at the RNA level for the 11 patients with nonsense CXCR4
somatic mutations (Figure 4A). The median percentage of mutant reads
was 37.4% (range, 18.4%-61.4%) and 54.5% (range, 18.2%-60.9%) for
MYD88 and CXCR4, respectively (P 5 .0537) with the mutant allele
burden of CXCR4 being greater than MYD88 in 9 of 11 patients
(81.8%). Sanger sequencing of paired DNA and cDNA confirmed
the mutant allele was underrepresented in the cDNA relative to the
DNA for MYD88L265P in many patients whereas the opposite was
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
TRANSCRIPTOME OF WALDENSTRÖM MACROGLOBULINEMIA
MYD88WT
CXCR4WT
MYD88/CXCR4 Interactions
•
•
•
MYD88 and CXCR4 are negatively
correlated and expression levels
are affected by mutation status.
Overall MYD88 expression
negatively correlates with WM
bone marrow involvement. CXCR4
has a positive correlation.
•
Lowest levels of B-cell
differentiation genes.
•
Low NFkB Response genes
Increased expression of genes
associated with PIK3 signaling
•
Increased promoter.
methylation of PRDM5 and
WNK2.
All WM
•
•
MYD88L265P
•
•
•
CXCR4WT
Up regulated VDJ Genes:
DNTT, RAG1, RAG2
Role for WT CXCR4: Increased
CXCL12, CXCR4, VCAM1
Decreased BAX expression
High levels of BCL2
CXCL13 expression
correlates with BM
involvement and
Hemoglobin
•
All WM Patients
• CXCR4 (both WT & WHIM)
• CXCL13
• BCL2 and BCL2L1
IGF1/IGF1R, particularly in
MYD88L265PCXCR4WT
•
Hypomethylating agents in
MYD88WT patients
•
PIK3 delta inhibitors, particularly
in MYD88WT WM. Additional
inhibition of PIK3 gamma may be
necessary for CXCR4WHIM patients
MYD88L265P
CXCR4WHIM
•
Silencing of tumor
suppressors up regulated by
MYD88 mutations.
Associated with a
transcriptional profile that is
the most distinct from HD
samples and other WM
genotypes.
•
High IRAK3 and low TLR4
Expression.
Decreased G-protein and
MAPK signaling negative
regulators.
High levels of PMAIP1.
•
•
•
Highest levels of IGF1.
Highest expresion of B-cell
differentiation genes.
•
•
Potential Targets
•
MYD88 mutant allele expression
is often reduced versus the wild
type allele in the mRNA whereas
the mutant CXCR4 allele is
preferentially expressed.
835
•
High PIK3R5 and PIK3CG
levels.
Figure 6. Summary of key findings by MYD88 and CXCR4 mutation status. Key findings from the transcriptome analysis for all WM samples as well as findings specific to
MYD88L265PCXCR4WT, MYD88L265PCXCR4WHIM, and MYD88L265PCXCR4WT WM patients. Additional notes on potential therapeutic targets and findings regarding MYD88
and CXCR4 interactions are also listed.
true for CXCR4 mutations (Figure 4B). This reduced expression of
MYD88L265P at the RNA level was also observed in some CXCR4WT
patients.
Discussion
This study represents the first profile of the transcriptional
landscape of WM by next-generation sequencing. Among the
most striking findings was a .7.8 log-fold upregulation of VDJ
recombination related genes including RAG1, RAG2, and DNTT in
WM patients samples (Figures 5 and 6). The class switch recombination gene AICDA was not observed at meaningful levels
consistent with the lack of immunoglobulin class switching in WM. The
overexpression of RAG1 and RAG2 is notable because specific somatic
mutation patterns such as deletions in BTG1 and ETV6 are associated
with their aberrant expression, and are commonly observed in WM
patients.2,40,41 The universal expression of CXCR4 and the upregulation
of its ligand CXCL12 are suggestive of autocrine/paracrine signaling.
CXCR4 activation has been reported to increase cell adhesion to
VCAM1 via VLA4, and VCAM1 was the eighth most upregulated
gene relative vs HD samples.42,43 Taken together, these findings
suggest that VCAM1, CXCR4, and CXCL12 may facilitate the
homotypic WM cell clustering observed in the BM of many WM
patients. Likewise, IGF1, an inducer of AKT1 survival signaling in
WM, was a top hit among the most upregulated genes. IGF1 and its
receptor IGF1R may therefore be potential therapeutic targets in WM
and warrant further investigation.44
Despite previous studies suggesting that WM was derived from
MB, we did not observe increased similarity at the transcription level for
MB relative PB as shown in Figure 1B, C, and E. WM represents
activated B cells in the processes of differentiating to plasma cells,
evidenced by the high levels of XBP1, PRDM1, and SDC1, which may
obscure signaling indicative of the true cell of origin. However,
clustering by genes related to B-cell differentiation effectively
sorted samples from patients with MYD88L265PCXCR4WT vs
MYD88L265PCXCR4WHIM and MYD88WTCXCR4WT suggesting that the
latter may represent an earlier stage of B-cell differentiation, consistent
with the previously established lower rate of IgH somatic hypermutation in MYD88WT WM.45
The majority of transcriptional differences between
MYD88L265PCXCR4WT and MYD88L265PCXCR4WHIM represented
normalization of the latter for TLR4 signaling associated gene
expression. This finding may help explain why CXCR4WHIM mutations
are found almost exclusively in MYD88-mutated patients. Others have
also reported that CXCR4 activation can affect TLR4 signaling.46-48
Regardless of mechanism, the direct downregulation of TLR4, and
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
836
HUNTER et al
upregulation of IRAK3 in CXCR4WHIM patients may be the most
proximal explanation for this phenomenon. Although IRAK3 is a
negative regulator of the IRAK4/1 signaling cascade, it has been shown
that BTK activation downstream of MYD88 occurs independently
of IRAK4/1 and is likely unaffected by IRAK3 upregulation.4,49 In
addition, TLR7 is upregulated in CXCR4WHIM patients and can use the
MYD88/IRAK4/IRAK3 complex to initiate regulatory signaling
through MAP3K3-dependent activation of NF-kB.50
Compared with the other WM genotypes and healthy donor B cells,
WM patients with the MYD88L265PCXCR4WT genotype strongly
upregulate a number of surface receptors and signaling molecules
including IL17RB, GPER1, WNT5A, and IGF1. IL17RB signaling may
provide an additional source of NF-kB activation whereas GPER1 and
IGF1 activate AKT1 and MAPK signaling in marginal zone lymphoma
and WM, respectively.44,51,52 However, WNT5A is a suppressor of
B-cell proliferation. 53 Many of the genes upregulated in the
MYD88L265PCXCR4WT patient cohort such as PMAIP1, WNK2,
PRDM5, CABLES1, CXXC4, and CDKN1C are tumor suppressors
that either promote apoptosis, inhibit cell cycle, or block MAPK
signaling.54-58 Normalization of MYD88L265P-associated tumor suppressor expression in CXCR4-mutated WM patients may provide the
clearest explanation for the selective inhibition of MYD88 signaling
observed in CXCR4WHIM patients. This yin-and-yang relationship
between MYD88 and CXCR4 is further supported by the modulation
of allele specific transcription, decreasing and increasing the mutation
allele burden of each gene, respectively. CXCR4 and MYD88 total
transcription levels were negatively correlated and inversely predictive
of BM disease involvement. Although aberrant promoter methylation
of WNK2 and PRDM5 was prominent in MYD88WTCXCR4WT samples,
it did not explain the low levels of WNK2 and PRDM5 observed in
patients with CXCR4WHIM mutations, thereby strengthening the case
that these findings are related to CXCR4WHIM signaling rather than
unrelated epigenomic events.
The strongest gene markers for patients with MYD88L265PCXCR4WHIM
was the upregulation of CXCR7, which like CXCR4 is activated
by CXCL12, and TSPAN33 which is upregulated in activated
B cells.59,60 IL15 was uniquely suppressed in patients with
MYD88L265PCXCR4WHIM and warrants functional studies to assess
its role in WM microenvironment. Preclinical studies have shown
that CXCR4WHIM transduced cell lines are resistant to PIK3CD
inhibitors such as idelalisib.10,38 Both PIK3R5 and PIK3CG were
significantly higher in CXCR4WHIM patients. PIK3R5 interacts with
the b/g G-protein subunits downstream of G-protein–coupled
receptors such as CXCR4 to activate PIK3CG which may explain
the observed resistance.61 The G-protein and MAPK inhibitors
RGS1, RGS2, RGS13, DUSP1, DUSP2, DUSP4, DUSP5, DUSP10,
DUSP16, and DUSP22 as well as the MAPK-inducible tumor
suppressor ERRFI1 were significantly downregulated in patients
with MYD88L265PCXCR4WHIM vs MYD88L265PCXCR4WT.62 The fact
that DUSP2, DUSP4, DUSP5, RGS13, and ERRFI1 were all
significantly induced by CXCL12 in CXCR4WHIM vs CXCR4WTtransduced lines supports a role for these genes as negative regulators
of CXCR4 signaling and may have implications for MYD88 MAPK
signaling as well.63,64 This is similar to a number of secondary events
such as MYBBP1A mutations and deletions of TRAF3, TNFAIP3,
and HIVEP2, which are thought to represent loss of negative regulators
for NF-kB signaling in WM.2,11,65 The mechanism(s) by which
negative regulators of CXCR4 signaling are lost in patients with
CXCR4WHIM mutations remain to be clarified warranting further
investigation.
Of the 1155 genes that were differentially expressed between
MYD88L265P and MYD88WT patients, 552 were not observed in the
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
MYD88L265PCXCR4WT vs MYD88L265PCXCR4WHIM gene signature.
MYD88WT patient expression was quite heterogeneous indicating pathogenetic diversity in this population. Large-scale genomic
exploration will invariably be needed to help clarify the genetic basis
for MYD88WT WM disease. Regardless, a clear downregulation of
genes associated with NF-kB signaling including IL6, IRAK2,
TNFAIP3, NFKBIZ, NFKB2, TIRAP, PIM1, and PIM2 was observed
in these patients. Other genes of interest included the downregulation
of CD40, and upregulation of PTBP3, CD86, and CXCR3. Notably,
IGF1R, PIK3AP1, and AKT2 were all upregulated, suggesting a
reliance on PIK3 signaling. The antiapoptotic gene BCL2, known to
play an important role in WM, was overexpressed expressed in all
patient samples while the proapoptotic BAX was underexpressed,
suggesting a mechanism independent of the MYD88 and CXCR4
mutations.65-67
Somatic mutations in ARID1A were associated with increased
CXCL13, as well as increased BM infiltration. CXCL13 has not
been previously described in WM and was also a strong predictor
of BM disease involvement and hemoglobin levels. CXCL13 may
also play a role in mast cell recruitment.68 Mast cells can be
observed admixed with the WM B cells in patient BM histology
and are thought to play an important supportive role in the tumor
microenvironment of WM.69 Deletions of chromosome 6q were
associated with over 131 differentially expressed genes. Levels of
previously identified target genes such as the NF-kB negative
regulator HIVEP2, as well as BCLAF1, FOXO3 and ARID1B were
all suppressed in the presence of 6q deletions.2 Although only
3 patients had strong familial predisposition to WM, these patients demonstrated significant dysregulation of several cancerassociated genes including RB1, STAT5B, ZNF300, MAPK9,
and NFKB1.
These studies highlight the pivotal roles of MYD88 and
CXCR4 signaling in WM. CXCR4 mutations appear to function
primarily as dampeners of negative regulators that are upregulated in response to mutant MYD88 signaling. The upregulation
of the ligand, adhesion targets, and CXCR4 itself in all WM
patients provides evidence for uniform CXCR4 dysregulation in
WM and supports the development of CXCR4 antagonists for
WM therapy. The upregulation of PIK3 pathway members and the
increased promoter methylation in MYD88WT patients creates a
strong rationale for preclinical studies of PIK3 inhibitors and
demethylating agents in this population. The upregulation of
BCL2 across all WM patient genotypes also supports the
development of BCL2 antagonists. Finally, CXCL13 is highly
expressed by WM cells, and its expression correlates with
important clinical parameters. Further studies to delineate therapeutic
targeting of this cytokine in WM are warranted.
Acknowledgments
The authors acknowledge the contributions of Yaoyu Wang and John
Quackenbush at the Center for Cancer Computational Biology at the
Dana-Farber Cancer Institute for RNA Sequencing, and Mathew
Temple, Leutz Buon, Terry Haley, and the Dana-Farber Research
Computing Center for their invaluable assistance. The authors
acknowledge the generous support of the WM patients who provided
their samples.
This work was supported by Peter S. Bing, a Translational
Research grant from the Leukemia & Lymphoma Society, a grant
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
from the International Waldenström’s Macroglobulinemia Foundation, a National Institutes of Health, National Cancer Institute Spore
5P50CA100707-12 Research Development Award, and the Orzag
Family Fund for WM Research.
Authorship
Contribution: Z.R.H. and S.P.T. designed the study and wrote the
manuscript; Z.R.H. performed the data analysis and informatics
studies; X.L. and N.T. performed the validation and methylation
TRANSCRIPTOME OF WALDENSTRÖM MACROGLOBULINEMIA
837
studies; G.Y., X.L., and J.C. transduced the cell lines and ran the
stimulations; L.X., X.L., J.C., J.G.C., J.M.V., and N.T. prepared the
study samples; S.P.T., J.J.C., and T.D. provided patient care and
obtained consent and samples; R.J.M., P.B., J.G., and C.J.P. selected
samples and provided clinical data analysis; and S.P.T., N.M.M., and
K.C.A. reviewed the data and provided expert guidance.
Conflict-of-interest disclosure: The authors declare no competing
financial interests.
Correspondence: Steven P. Treon, Bing Center for Waldenström’s
Macroglobulinemia, Dana-Farber Cancer Institute, M547, 450
Brookline Ave, Boston, MA 02215; e-mail: steven_treon@dfci.
harvard.edu.
References
1. Treon SP, Xu L, Yang G, et al. MYD88
L265P somatic mutation in Waldenström’s
macroglobulinemia. N Engl J Med. 2012;
367(9):826-833.
2. Hunter ZR, Xu L, Yang G, et al. The genomic
landscape of Waldenstrom macroglobulinemia
is characterized by highly recurring MYD88
and WHIM-like CXCR4 mutations, and small
somatic deletions associated with B-cell
lymphomagenesis. Blood. 2014;123(11):
1637-1646.
3. Treon SP, Cao Y, Xu L, Yang G, Liu X, Hunter ZR.
Somatic mutations in MYD88 and CXCR4 are
determinants of clinical presentation and overall
survival in Waldenstrom macroglobulinemia.
Blood. 2014;123(18):2791-2796.
4. Yang G, Zhou Y, Liu X, et al. A mutation in
MYD88 (L265P) supports the survival of
lymphoplasmacytic cells by activation of
Bruton tyrosine kinase in Waldenström
macroglobulinemia. Blood. 2013;122(7):
1222-1232.
5. Ngo VN, Young RM, Schmitz R, et al.
Oncogenically active MYD88 mutations in human
lymphoma. Nature. 2011;470(7332):115-119.
6. Treon SP, Tripsas CK, Meid K, et al. Ibrutinib
in previously treated Waldenström’s
macroglobulinemia. N Engl J Med. 2015;372(15):
1430-1440.
7. Treon SP, Xu L, Hunter Z. MYD88 mutations
and response to ibrutinib in Waldenström’s
macroglobulinemia. N Engl J Med. 2015;373(6):
584-586.
8. Xu L, Hunter ZR, Tsakmaklis N, et al. Clonal
architecture of CXCR4 WHIM-like mutations in
Waldenström macroglobulinaemia. Br J
Haematol. 2016;172(5):735-744.
9. Hernandez PA, Gorlin RJ, Lukens JN, et al.
Mutations in the chemokine receptor gene CXCR4
are associated with WHIM syndrome, a combined
immunodeficiency disease. Nat Genet. 2003;
34(1):70-74.
10. Cao Y, Hunter ZR, Liu X, et al. The WHIM-like
CXCR4(S338X) somatic mutation activates AKT
and ERK, and promotes resistance to ibrutinib
and other agents used in the treatment of
Waldenstrom’s macroglobulinemia. Leukemia.
2015;29(1):169-176.
11. Braggio E, Keats JJ, Leleu X, et al. Identification
of copy number abnormalities and inactivating
mutations in two negative regulators of nuclear
factor-kappaB signaling pathways in
Waldenstrom’s macroglobulinemia. Cancer
Res. 2009;69(8):3579-3588.
12. Braggio E, Dogan A, Keats JJ, et al. Genomic
analysis of marginal zone and lymphoplasmacytic
lymphomas identified common and diseasespecific abnormalities. Mod Pathol. 2012;25(5):
651-660.
13. Schop RFJ, Van Wier SA, Xu R, et al. 6q deletion
discriminates Waldenström macroglobulinemia
from IgM monoclonal gammopathy of
undetermined significance. Cancer Genet
Cytogenet. 2006;169(2):150-153.
14. Treon SP, Hunter ZR, Aggarwal A, et al.
Characterization of familial Waldenstrom’s
macroglobulinemia. Ann Oncol. 2006;17(3):
488-494.
15. Chng WJ, Schop RF, Price-Troska T, et al.
Gene-expression profiling of Waldenstrom
macroglobulinemia reveals a phenotype more
similar to chronic lymphocytic leukemia than
multiple myeloma. Blood. 2006;108(8):
2755-2763.
16. Gutiérrez NC, Ocio EM, de Las Rivas J, et al.
Gene expression profiling of B lymphocytes
and plasma cells from Waldenström’s
macroglobulinemia: comparison with expression
patterns of the same cell counterparts from
chronic lymphocytic leukemia, multiple myeloma
and normal individuals. Leukemia. 2007;21(3):
541-549.
17. Poulain S, Roumier C, Venet-Caillault A, et al.
Genomic landscape of CXCR4 mutations in
Waldenström macroglobulinemia. Clin Cancer
Res. 2016;22(6):1480-1488.
25. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a
Bioconductor package for differential expression
analysis of digital gene expression data.
Bioinformatics. 2010;26(1):139-140.
26. Liao Y, Smyth GK, Shi W. featureCounts: an
efficient general purpose program for assigning
sequence reads to genomic features.
Bioinformatics. 2014;30(7):923-930.
27. Gentleman RC, Carey VJ, Bates DM, et al.
Bioconductor: open software development for
computational biology and bioinformatics.
Genome Biol. 2004;5(10):R80.
28. Love MI, Huber W, Anders S. Moderated
estimation of fold change and dispersion for
RNA-seq data with DESeq2. Genome Biol.
2014;15(12):550.
29. Carvalho BS, Irizarry RA. A framework for
oligonucleotide microarray preprocessing.
Bioinformatics. 2010;26(19):2363-2367.
30. Hothorn T, Hornik K, Zeileis A. Unbiased
recursive partitioning: a conditional inference
framework. J Comput Graph Stat. 2006;15(3):
651-674.
31. Xiong Z, Laird PW. COBRA: a sensitive and
quantitative DNA methylation assay. Nucleic
Acids Res. 1997;25(12):2532-2534.
18. Sahota SS, Babbage G, Weston-Bell NJ. CD27 in
defining memory B-cell origins in Waldenström’s
macroglobulinemia. Clin Lymphoma Myeloma.
2009;9(1):33-35.
32. Moniz S, Martinho O, Pinto F, et al. Loss of WNK2
expression by promoter gene methylation occurs
in adult gliomas and triggers Rac1-mediated
tumour cell invasiveness. Hum Mol Genet. 2013;
22(1):84-95.
19. Janz S. Waldenström macroglobulinemia: clinical
and immunological aspects, natural history, cell of
origin, and emerging mouse models. ISRN
Hematol. 2013;2013:815325.
33. Tan S-X, Hu R-C, Liu J-J, Tan Y-L, Liu W-E.
Methylation of PRDM2, PRDM5 and PRDM16
genes in lung cancer cells. Int J Clin Exp Pathol.
2014;7(5):2305-2311.
20. Owen RG, Treon SP, Al-Katib A, et al.
Clinicopathological definition of Waldenstrom’s
macroglobulinemia: consensus panel
recommendations from the Second International
Workshop on Waldenstrom’s Macroglobulinemia.
Semin Oncol. 2003;30(2):110-115.
34. Treon SP. How I treat Waldenström
macroglobulinemia. Blood. 2015;126(6):721-732.
21. Xu L, Hunter ZR, Yang G, et al. MYD88 L265P
in Waldenstrom’s macroglobulinemia, IgM
monoclonal gammopathy, and other B-cell
lymphoproliferative disorders using conventional
and quantitative allele-specific PCR [published
correction appears in Blood. 2013;121(26):5259].
Blood. 2013;121(11):2051-2058.
22. Dobin A, Davis CA, Schlesinger F, et al. STAR:
ultrafast universal RNA-seq aligner.
Bioinformatics. 2013;29(1):15-21.
23. Law CW, Chen Y, Shi W, Smyth GK. voom:
precision weights unlock linear model analysis
tools for RNA-seq read counts. Genome Biol.
2014;15(2):R29.
24. Ritchie ME, Phipson B, Wu D, et al. limma powers
differential expression analyses for RNAsequencing and microarray studies. Nucleic Acids
Res. 2015;43(7):e47.
35. Tarte K, Zhan F, De Vos J, Klein B, Shaughnessy
J Jr. Gene expression profiling of plasma cells
and plasmablasts: toward a better understanding
of the late stages of B-cell differentiation. Blood.
2003;102(2):592-600.
36. Jourdan M, Caraux A, Caron G, et al.
Characterization of a transitional preplasmablast
population in the process of human B cell to
plasma cell differentiation. J Immunol. 2011;
187(8):3931-3941.
37. Djos A, Martinsson T, Kogner P, Carén H. The
RASSF gene family members RASSF5, RASSF6
and RASSF7 show frequent DNA methylation in
neuroblastoma. Mol Cancer. 2012;11:40.
38. Roccaro AM, Sacco A, Jimenez C, et al. C1013G/
CXCR4 acts as a driver mutation of tumor
progression and modulator of drug resistance
in lymphoplasmacytic lymphoma. Blood. 2014;
123(26):4120-4131.
39. Vos JM, Tsakmaklis N, Brodsky PS, et al.
Biologically meaningful changes in cytokine and
chemokine production following ibrutinib therapy
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
838
BLOOD, 11 AUGUST 2016 x VOLUME 128, NUMBER 6
HUNTER et al
in Waldenstrom’s macroglobulinemia.
Hematologica. 2016. Abstract P312. In press.
40. Waanders E, Scheijen B, van der Meer LT, et al.
The origin and nature of tightly clustered BTG1
deletions in precursor B-cell acute lymphoblastic
leukemia support a model of multiclonal evolution.
PLoS Genet. 2012;8(2):e1002533.
41. Papaemmanuil E, Rapado I, Li Y, et al. RAGmediated recombination is the predominant driver
of oncogenic rearrangement in ETV6-RUNX1
acute lymphoblastic leukemia. Nat Genet. 2014;
46(2):116-125.
42. Ngo HT, Leleu X, Lee J, et al. SDF-1/CXCR4
and VLA-4 interaction regulates homing in
Waldenstrom macroglobulinemia. Blood. 2008;
112(1):150-158.
43. Petty JM, Lenox CC, Weiss DJ, Poynter ME,
Suratt BT. Crosstalk between CXCR4/stromal
derived factor-1 and VLA-4/VCAM-1 pathways
regulates neutrophil retention in the bone marrow.
J Immunol. 2009;182(1):604-612.
44. Leleu X, Jia X, Runnels J, et al. The Akt pathway
regulates survival and homing in Waldenstrom
macroglobulinemia. Blood. 2007;110(13):
4417-4426.
45. Jiménez C, Sebastián E, Chillón MC, et al.
MYD88 L265P is a marker highly characteristic
of, but not restricted to, Waldenström’s
macroglobulinemia. Leukemia. 2013;27(8):
1722-1728.
46. Seemann S, Lupp A. Administration of a CXCL12
analog in endotoxemia is associated with antiinflammatory, anti-oxidative and cytoprotective
effects in vivo. PLoS One. 2015;10(9):e0138389.
47. Fan H, Wong D, Ashton SH, Borg KT, Halushka
PV, Cook JA. Beneficial effect of a CXCR4
agonist in murine models of systemic
inflammation. Inflammation. 2012;35(1):130-137.
48. Triantafilou M, Lepper PM, Briault CD, et al.
Chemokine receptor 4 (CXCR4) is part of the
lipopolysaccharide “sensing apparatus”. Eur J
Immunol. 2008;38(1):192-203.
49. Kobayashi K, Hernandez LD, Galán JE, Janeway
CA Jr, Medzhitov R, Flavell RA. IRAK-M is a
negative regulator of Toll-like receptor signaling.
Cell. 2002;110(2):191-202.
50. Zhou H, Yu M, Fukuda K, et al. IRAK-M mediates
Toll-like receptor/IL-1R-induced NFkB activation
and cytokine production. EMBO J. 2013;32(4):
583-596.
51. Huang C-K, Yang C-Y, Jeng Y-M, et al. Autocrine/
paracrine mechanism of interleukin-17B receptor
promotes breast tumorigenesis through NF-kBmediated antiapoptotic pathway. Oncogene.
2014;33(23):2968-2977.
52. Rudelius M, Rauert-Wunderlich H, Hartmann E,
et al. The G protein-coupled estrogen receptor 1
(GPER-1) contributes to the proliferation and
survival of mantle cell lymphoma cells.
Haematologica. 2015;100(11):e458-e461.
53. Liang H, Chen Q, Coles AH, et al. Wnt5a inhibits
B cell proliferation and functions as a tumor
suppressor in hematopoietic tissue. Cancer Cell.
2003;4(5):349-360.
54. Moniz S, Verı́ssimo F, Matos P, et al. Protein
kinase WNK2 inhibits cell proliferation by
negatively modulating the activation of MEK1/
ERK1/2 [published correction appears in
Oncogene. 2008;27(1):155]. Oncogene. 2007;
26(41):6071-6081.
55. Deng Q, Huang S. PRDM5 is silenced in human
cancers and has growth suppressive activities.
Oncogene. 2004;23(28):4903-4910.
56. Shi Z, Park HR, Du Y, et al. Cables1 complex
couples survival signaling to the cell death
machinery. Cancer Res. 2015;75(1):147-158.
57. Lu H, Jin W, Sun J, et al. New tumor suppressor
CXXC finger protein 4 inactivates mitogen
activated protein kinase signaling. FEBS Lett.
2014;588(18):3322-3326.
60. Luu VP, Hevezi P, Vences-Catalan F, et al.
TSPAN33 is a novel marker of activated and
malignant B cells. Clin Immunol. 2013;149(3):
388-399.
61. Brock C, Schaefer M, Reusch HP, et al. Roles of
G beta gamma in membrane recruitment and
activation of p110 gamma/p101 phosphoinositide
3-kinase gamma. J Cell Biol. 2003;160(1):89-99.
62. Zhang Y-W, Staal B, Su Y, et al. Evidence that
MIG-6 is a tumor-suppressor gene. Oncogene.
2007;26(2):269-276.
63. Berthebaud M, Rivière C, Jarrier P, et al. RGS16
is a negative regulator of SDF-1-CXCR4 signaling
in megakaryocytes. Blood. 2005;106(9):
2962-2968.
64. Lang R, Hammer M, Mages J. DUSP meet
immunology: dual specificity MAPK phosphatases
in control of the inflammatory response.
J Immunol. 2006;177(11):7497-7504.
65. Wang JQ, Jeelall YS, Beutler B, Horikawa K,
Goodnow CC. Consequences of the recurrent
MYD88(L265P) somatic mutation for B cell
tolerance. J Exp Med. 2014;211(3):413-426.
66. Cao Y, Yang G, Hunter ZR, et al. The BCL2
antagonist ABT-199 triggers apoptosis, and
augments ibrutinib and idelalisib mediated
cytotoxicity in CXCR4 Wild-type and CXCR4
WHIM mutated Waldenstrom macroglobulinaemia
cells. Br J Haematol. 2015;170(1):134-138.
67. Ogmundsdóttir HM, Sveinsdóttir S, Sigfússon A,
Skaftadóttir I, Jónasson JG, Agnarsson BA.
Enhanced B cell survival in familial
macroglobulinaemia is associated with increased
expression of Bcl-2. Clin Exp Immunol. 1999;
117(2):252-260.
58. Borriello A, Caldarelli I, Bencivenga D, et al. p57
(Kip2) and cancer: time for a critical appraisal. Mol
Cancer Res. 2011;9(10):1269-1284.
68. Tripodo C, Gri G, Piccaluga PP, et al. Mast cells
and Th17 cells contribute to the lymphomaassociated pro-inflammatory microenvironment
of angioimmunoblastic T-cell lymphoma. Am J
Pathol. 2010;177(2):792-802.
59. Décaillot FM, Kazmi MA, Lin Y, Ray-Saha S,
Sakmar TP, Sachdev P. CXCR7/CXCR4
heterodimer constitutively recruits beta-arrestin to
enhance cell migration. J Biol Chem. 2011;
286(37):32188-32197.
69. Tournilhac O, Santos DD, Xu L, et al. Mast cells
in Waldenstrom’s macroglobulinemia support
lymphoplasmacytic cell growth through CD154/
CD40 signaling. Ann Oncol. 2006;17(8):
1275-1282.
From www.bloodjournal.org by guest on June 14, 2017. For personal use only.
2016 128: 827-838
doi:10.1182/blood-2016-03-708263 originally published
online June 14, 2016
Transcriptome sequencing reveals a profile that corresponds to
genomic variants in Waldenström macroglobulinemia
Zachary R. Hunter, Lian Xu, Guang Yang, Nicholas Tsakmaklis, Josephine M. Vos, Xia Liu, Jie Chen,
Robert J. Manning, Jiaji G. Chen, Philip Brodsky, Christopher J. Patterson, Joshua Gustine, Toni
Dubeau, Jorge J. Castillo, Kenneth C. Anderson, Nikhil M. Munshi and Steven P. Treon
Updated information and services can be found at:
http://www.bloodjournal.org/content/128/6/827.full.html
Articles on similar topics can be found in the following Blood collections
Clinical Trials and Observations (4553 articles)
Free Research Articles (4527 articles)
Lymphoid Neoplasia (2557 articles)
Information about reproducing this article in parts or in its entirety may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests
Information about ordering reprints may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#reprints
Information about subscriptions and ASH membership may be found online at:
http://www.bloodjournal.org/site/subscriptions/index.xhtml
Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American Society
of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.