One siRNA pool targeting the λ constant region stops λ light-chain production and causes terminal endoplasmic reticulum stress

The immunoglobulin light chains in systemic light-chain amyloidosis (AL) have a κ-to-λ case rate of 1 to 4 and cause cardiac-related deaths in up to 25% of patients within months of diagnosis.^1,2  Current therapies for AL aim to reduce light-chain production and include steroids, alkylating agents, immunomodulatory drugs, and proteasome inhibitors.³  Although responses are notable, incurability and early cardiac death are more the norm, and more effective therapies are needed.^4,5 

Although the unfolded protein response (UPR) is constitutively active in plasma cells,⁶  and although the pairing of light and heavy chains is under strict intracellular quality control in the endoplasmic reticulum (ER),^7,8  we learned in initial experiments that variable region–targeted λ-light-chain knockdown could trigger a reactive UPR and a terminal ER stress response, whereas knockdown of immunoglobulin G (IgG) heavy-chain expression had no impact on cell growth or viability. We then hypothesized that λ-light-chain (IgL) production could be disrupted by short interfering RNA (siRNA) targeting consensus sequences in the IgL constant region (CR) messenger RNA (mRNA) with 1 siRNA pool. We now report that this pool, siRNA targeting the CR of λ light chains (si[IGLC_CR]), significantly reduces IgL production and secretion by human plasma cells without regard for unique variable region gene sequences, and that, in human plasma cells making intact antibodies, treatment with si[IGLC_CR] results in the intracellular retention of unpaired immunoglobulin heavy chains (IgH), the activation of the UPR, and upregulation of genes involved in ER stress signaling⁹  that can cause NOXA-mediated mitochondrial depolarization and caspase-dependent apoptosis. In addition, si[IGLC_CR] treatment can substantially reduce IgL message and intracellular IgL in AL patient plasma cells producing monoclonal IgL and can also increase caspase 3/7 activity in clones making intact antibodies.

ALMC1 and ALMC2 cells (gift of Diane Jelinek) were cultured as described.¹⁰  MM.1S and MM.1R cells from American Type Culture Collection (Manassas, VA), and EJM and OPM-2 cells from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany) were cultured as directed. These cell lines were chosen because they are human myeloma cell lines that produce λ light chains (IgL). Z-VAD-FMK, l-phenylalanine mustard, and arachadonic acid were from Sigma-Aldrich (St. Louis, MO).

Clinical specimens obtained under an institutional review board–approved protocol from patients with confirmed diagnosis of systemic AL and monoclonal IgL disease were used for CD138 selection as previously described.¹¹  This study was conducted in accordance with the Declaration of Helsinki.

Monoclonal IgL and IgH genes for ALMC1, ALMC2, and EJM cells were identified and sequenced as previously described.¹² 

Individual cell lines were evaluated for tolerance to streptolysin-O (Sigma-Aldrich) reversible permeabilization for transfection of siRNA, and the optimal effective concentrations of streptolysin-O and siRNA were determined. A similar approach testing tolerance was used for patient plasma cells (supplemental Figure 1; available on the Blood Web site).¹³ 

All siRNA agents were obtained from Dharmacon (Thermo Scientific, Lafayette, CO) using www.thermo.com/sidesign with modifications to minimize seed-region off-target effects (ON-TARGETplus SMARTpool). The siRNA pools we used are in the supplemental Data, and custom-designed siRNA pools are defined in supplemental Table 1.

Luciferin-based caspase 3/7 activity assays (Promega, Madison, WI) were performed following the manufacturer’s instructions on a Promega GloMax microplate luminometer in triplicate for each situation with 5 × 10³ cells per well and reported as relative luminescence units.

The MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] cell proliferation and viability assays (Invitrogen) were performed according to the manufacturer’s instructions in triplicate with 2.5 × 10⁴ cells per well and read on a microplate absorbance reader (BIO-RAD; Hercules, CA). Cell viability was also assessed by trypan blue staining where indicated.¹⁴ 

All antibodies used are listed in supplemental Table 2. Antibodies were titrated for optimal use and used with appropriate isotype controls. Flow cytometry was performed in our core facility on a FACSCalibur Cytometer (Becton Dickinson, Franklin Lakes, NJ). Mean fluorescence intensity (MFI) in each case was computed minus that of isotype control. Flow cytometry for intracellular immunoglobulin was performed with fluorescein isothiocyanate–conjugated anti-human IgG heavy chain and phycoerythrin-conjugated anti-human immunoglobulin λ-light-chain antibodies titrated for optimal use with appropriate isotype controls. Cells were permeabilized by using CytoFix/CytoPerm Fixation/Permeabilization kit (BD Pharmingen, Franklin Lakes, NJ), then stained with antibodies and acquired. MFI was analyzed with FlowJo (Tree Star, Ashland, OR).

The annexin-V/propidium iodide (PI) kit was from BD Pharmingen (San Jose, CA). As a control for annexin-V/PI staining, melphalan (l-phenylalanine mustard; Sigma-Aldrich) was dissolved in acid alcohol and used fresh at a concentration of 25 µM. For this assay, cells were harvested, washed twice, and suspended in annexin-V labeling buffer with fluorescein isothiocyanate–annexin V and PI as described in the kit and acquired. The percentage of apoptotic cells was computed with FlowJo.

All antibodies used are listed in supplemental Table 2. Cells were washed with phosphate-buffered saline, pelleted, and lysed in modified radioimmunoprecipitation assay buffer (Pierce/Thermo Scientific, Rockford, IL) with protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN) and sodium pervanadate (Sigma-Aldrich) and MG132 (Millipore, Billerica, MA). After protein determination with bicinchoninic acid kit (Pierce/Thermo Scientific), equal amounts of denatured protein were subjected to 4% to 20% gradient sodium dodecyl sulfate–polyacrylamide gel electrophoresis and electroblotted onto Immobilon-P polyvinylidene difluoride transfer membrane (Millipore). After blocking nonspecific binding, blots were probed with appropriate primary antibodies and corresponding horseradish peroxidase–conjugated anti-rabbit or anti-mouse secondary antibodies. Signal was revealed with SuperSignal West Pico chemiluminescence reagent (Pierce/Thermo Scientific), detected by using ImageQuant LAS4000 mini and analyzed with ImageQuant TL (GE Healthcare Life Sciences, Pittsburgh, PA).

For immunoprecipitation, 200 μg of total lysate protein was precleared, antigen-specific antibodies were added, and the mix was incubated overnight at 4°C with rotation. Then 20 μL of protein A/G agarose beads was added, and the mix was further incubated for another 4 hours. After spinning down the beads and washing 3 times with modified radioimmunoprecipitation assay buffer/IP buffer, 25 µL of reducing sampling buffer was added to the pellets, and the suspension was reduced and denatured through boiling, ice chilling, and spinning at 15 000g for 10 minutes at room temperature. Equal volumes of cleared supernatants were subject to gradient sodium dodecyl sulfate–polyacrylamide gel electrophoresis, further blotting, and probing with antibodies.

RNA was extracted and complementary DNA (cDNA) generated using standard methods.¹⁵  Reverse-transcription quantitative polymerase chain reaction (qPCR) was performed in our core facility using TaqMAN Gene Expression Assays with all primers and probes from Applied Biosystems (Foster City, CA) on an Mx 3000P platform and related software (Stratagene, La Jolla, CA). For qPCR, expression levels were calculated using the 2^-∆ΔCt method. Primers and probes are listed in the supplemental Data.

All antibodies used are listed in supplemental Table 2. Cultures were inoculated at a cell density of 10⁶/mL of complete medium in all experiments for comparison and incubated for 1 day, and then supernatants of the suspensions were obtained from 3 independent repeat experiments for ELISA. For cells producing IgH and IgL, the same supernatants were used for measurements of both proteins. To measure the amounts of immunoglobulins in supernatants, quantitative ELISA was done by using anti-human IgG heavy chain and anti-human immunoglobulin λ-light-chain antibodies with sandwich enzyme-linked immunosorbent assays (Bethyl Laboratories, Montgomery, TX) according to the manufacturer’s protocols. Optical density was read on a microplate reader (BIO-RAD), and protein quantity was calculated according to standard curves.

Gene expression studies were performed using cDNA from 3 paired specimens of ALMC1 cells treated with si[IGLC_CR] or si[-] on the Illumina platform using the standard protocols for HumanHT-12 v4 Expression BeadChip at the Yale Center for Genome Analysis (New Haven, CT). The BeadStudio suite of programs was used to calculate the expression values for probe sets (Illumina Inc., San Diego, CA). These studies can be accessioned with National Center for Biotechnology Information Gene Expression Omnibus number GSE54507.

PRISM (GraphPad, San Diego, CA) was used for descriptive statistics and analyses. All experiments were repeated a minimum of 3 times in triplicate wells unless otherwise noted. For microarray, Bioconductor packages Lumi/Limma were used to calculate the fold changes and P values.^16,17  TmeV was used to produce the heatmap of gene expression values for the regulated genes¹⁸  whose expression differed by >1.5-fold with multiple hypothesis testing with P < .10. Web-based resource DAVID was used to calculate the enrichment of functional categories (DAVID Bioinformatics Resources 6.7, National Institute of Allergy and Infectious Diseases, National Institutes of Health).¹⁹  The combined heatmap and Gene Ontology (GO) enrichment categories were generated using the GeneAnswers package in Bioconductor.²⁰ 

With siRNA targeted to the IgL or IgH variable region mRNA of ALMC1 (and ALMC2) or EJM cells,^10,21  at 24 hours expression of IgL was reduced as appreciated by flow cytometry dot plot (Figure 1A), and expression of either IgL or IgH was reduced as shown by flow cytometry histograms (Figure 1B) and by differences in MFI (Figure 1C). With this approach to knockdown of IgL, IgH, or IgL and IgH, reductions in cell viability and proliferation and increases in caspase 3/7 activity were observed in cells with IgL but not IgH or IgL and IgH knockdown (Figure 1D-E). With IgL knockdown, increased intracellular staining for IgH was observed (Figure 1A,F).

After knockdown of IgL in clonal cells making IgGλ, IgH production was maintained, whereas intracellular IgL was diminished (Figure 2A). From lysates of control si[-] and si[IGLC]-treated cells, IgH was pulled down to assess IPs for IgL and for the chaperone glucose-regulated protein 78 kDa (GRP78). The results (Figure 2B) show that IgH is associated with GRP78. Moreover, IP of GRP78 shows association with IgH as well, although we note that in the IBs from both cell lines there may be contamination in the IgH lanes by the IP antibody for GRP78. Nevertheless, the relevant features of IgL knockdown, GRP78 induction, and IgH/GRP78 association can be appreciated.

The UPR is triggered by dissociation of GRP78 from the activators inositol-requiring enzyme 1 alpha (IRE1α), protein kinase RNA–like ER kinase (PERK), and activating transcription factor 6.²²  In Figure 2C, prompt activation of the UPR is seen with expression of CHOP,²³ GRP78, and XBP1s. In Figure 2D, increased production of IRE1α, GRP78, and C/EBP homologous protein (CHOP) is seen within hours of UPR activation. Given the evidence for likely caspase-dependent apoptosis with IgL knockdown, the timeline of changes in expression of proapoptotic Bcl-2 homology 3 domain only family members was studied, and fourfold upregulated expression of NOXA and upregulated expression of PUMA but not BIM in ALMC1 cells, and to a lesser degree of PUMA and NOXA in EJM cells, was seen (Figure 2E). Moreover, by 20 hours after IgL knockdown, 25% of ALMC1 cells showed evidence of mitochondrial depolarization reversible by pan-caspase inhibition, consistent with the results in Figure 1D-E, and therefore caspase dependent (Figure 2F). EJM cells treated in the same way did not display increased levels of mitochondrial depolarization. By annexin-V/PI staining, in ALMC1 cells there was an average of 29.8 ± 6.8% specific apoptosis associated with IgL knockdown (n = 3) and none with IgH or combined IgL and IgH knockdown. Pan-caspase inhibition reversed this effect (Figure 2G). In EJM cells, despite the modest increase in caspase 3/7 activity appreciated by bioluminescence (Figure 1E) and the transient upregulation of NOXA (Figure 2E), evidence of mitochondrial depolarization by 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimi-dazolylcarbocyanine iodide (JC-1) staining (Figure 2F) and of early or late apoptosis by annexin-V/PI staining was not found (data not shown).

The differences between ALMC1 and EJM cells were then examined further. ALMC1 cells produce an intact IgGλ with excess IgL, whereas EJM cells have 2 populations by flow cytometry, including a minor one that makes IgL only (Figure 1A). The efficiency of IgL knockdown with variable region siRNA was lower in EJM than in ALMC1 cells (Figure 1C), and the relative decrease in cell viability and proliferation (Figure 1D) and increase in caspase 3/7 activity with IgL knockdown were also smaller in EJM than ALMC1 cells (Figure 1E). (Of note, the efficiency of GAPDH knockdown was best in ALMC1 cells; supplemental Figure 1). Both cell lines, however, have similar timelines of IgL knockdown (Figure 2A), and both show the coassociation of IgH and GRP78 (Figure 2B) and activation of the UPR (Figure 2C-D). Unlike the CHOP preeminent pattern of UPR activation in ALMC1 cells, a GRP78 preeminent pattern occurs in EJM cells (Figure 2C).

In ALMC1 cells, CHOP is significantly increased compared with EJM cells (Figure 2H), and there was more NOXA, an MCL-1 antagonist, pulled down with MCL-1 from ALMC1 than from EJM cells (Figure 2I). CHOP is viewed as a mediator of a terminal UPR, and MCL-1, an antiapoptotic Bcl-2 family member, as a critical comediator of apoptosis in human myeloma cells.^24,25  These results are consistent with a terminal ER stress response in ALMC1 but not EJM cells.

Multiple factors may contribute to this difference, including the presence of an IgL-only subpopulation of EJM cells and the lower efficiency of IgL knockdown in EJM cells, as well as the differences between ALMC1 and EJM cells with respect to the ratios of IgL and IgH production (shown subsequently) and the production by EJM cells of IgL of 2 different sizes (supplemental Figure 2). Although the process of intact antibody formation is disrupted by IgL knockdown, the amount of IgL available in EJM cells may be adequate to match the amount of intracellular IgH waiting to be paired.

In order to design si[IGLC_CR], 1 pool of siRNA specific for IgL CR, 4 consensus region targets in the IgL CR were identified. Each target is listed below the relevant CR consensus sequence in Table 1 (the pool is in supplemental Table 1). The availability of 1 pool would in theory allow reduction of IgL production and secretion in multiple specimens without regard for variable region sequences; it would also allow an evaluation of the differences in secretion between ALMC1 and EJM cells using the same reagent. (Subsequently, the IgL CR of both cell lines was determined to be concordant with the siRNA targets; data not shown.)

Five human myeloma cell lines that produce IgL were treated with si[IGLC_CR], demonstrating a 45% average reduction in intracellular IgL MFI after less than a day (Figure 3A) and by ELISA an overall reduction of IgL secretion of 45% (Figure 3B-C). IgH secretion was markedly reduced as well at 1 day after IgL knockdown (Figure 3D). EJM cells secreted the most IgL (Figure 3B). When treated with si[IGLC_CR], EJM cells secreted ∼35% less IgL in 1 day than control cells (Figure 3C). In comparison, ALMC1 cells treated with si[IGLC_CR] secreted ∼60% less (Figure 3B-C). EJM cells also secreted about half the amount of IgH produced by ALMC1 cells, and when treated with si[IGLC_CR] secreted only 14% less IgH; in contrast, ALMC1 cells treated with si[IGLC_CR] secreted almost 50% less IgH (Figure 4D). This difference led to the concept that the ratio of postknockdown IgL and basal IgH secretion might be related to the triggering of the terminal UPR. This ratio was greater in EJM cells at 3 (30 μg/mL ÷ 10 μg/mL per day per million cells) vs 0.5 (10 μg/mL ÷ 20 μg/mL) in ALMC1 cells. There likely were lower levels of unpaired IgH in EJM than in ALMC1 cells with IgL knockdown, a difference that may contribute to the lack of a terminal ER stress response in EJM cells.

To examine further the activation of the UPR and changes in NOXA expression with CR knockdown, ALMC1 cells that secrete both intact IgGλ and IgL were compared with MM.1S cells that secrete only IgL. As shown in Figure 3E, CHOP, GRP78, XBP1S, and NOXA were all substantially upregulated in ALMC1 but not in MM.1S cells, whereas the efficiency of IgL knockdown was the same for both. Caspase 3/7 activity and cell viability and proliferation were then studied after si[IGLC_CR] treatment, and the former was increased and the latter decreased significantly in cells making intact IgGλ (Figure 3F-G).

The heatmap of genes significantly regulated in ALMC1 cells by si[IGLC_CR] treatment is shown in supplemental Figure 3. CHOP (also known as DDIT3) is the most highly upregulated gene. The detailed depiction of the most upregulated genes (Figure 4A) and the integrated view provided by a GO heatmap (Figure 4B) of the functional classes of activated genes demonstrate that a high level of ER stress is induced by the load of unpaired IgH and that genes involved in the UPR and endoplasmic-reticulum–associated protein degradation (ERAD) are upregulated in concert.

The circuitry of intrinsic terminal ER stress responses in clonal plasma cells remains unclear. The IRE1α pathway and the downstream activation of c-Jun N-terminal kinase by phosphorylated apoptosis signal-regulating kinase 1 (ASK1)²⁶  can lead to apoptosis of plasma cells. Arachadonic acid can inhibit ASK1 phosphorylation in plasma cells by activating protein phosphatase 5. In cells treated with si[IGLC_CR], however, there was no increase in phosphorylated c-Jun N-terminal kinase at 24 hours by IB and no effect of arachadonic acid at 72 hours by MTT assay, suggesting that IRE1α-mediated activation of ASK1 did not play a significant role in si[IGLC_CR]-related apoptosis (data not shown).

A simultaneous double knockdown technique was then employed, asking whether simultaneous knockdown of IgL expression along with CHOP or NOXA would change the level of caspase 3/7 activity. There was no difference in the levels of caspase 3/7 activity between si[IGLC_CR]- and si[IGLC_CR+CHOP]-treated cells at 169 ± 31% and 170 ± 34% of controls, respectively (P = .87, n = 5), whereas si[CHOP]-treated cells had levels that were 102 ± 43% of controls. Also, cells treated with si[IGLC_CR+CHOP] showed increased levels of NOXA as did si[IGLC_CR]-treated cells (Figure 4C). In contrast, si[IGLC_CR+NOXA] treatment significantly reduced caspase 3/7 activity (Figure 4D-E). Although NOXA upregulation had been noted in EJM cells (Figure 2E), increased expression began earlier, was greater, and was observed for a longer time in ALMC1 than in EJM cells (Figure 2E), consistent with the comparative increase of NOXA in ALMC1 cells in the IP of MCL-1 (Figure 2I).

With CD138-selected specimens from patients with AL, highly enriched suspensions of plasma cells were secured.¹¹  The patient characteristics, number, and uses of specimens are detailed in supplemental Table 3. In Figure 5A, an example of prompt and effective treatment with si[IGLC_CR] causing an 85% reduction in intracellular IgL immunofluorescence is shown. In supplemental Figure 4, IBs of 2 specimens are shown, demonstrating significant reductions in IgL with si[IGLC_CR] treatment. As shown in Figure 5B, in 16 specimens evaluated by qPCR, the average reduction in IgL message exceeded 70%, and as shown in Figure 5C, in 13 specimens evaluated by flow cytometry the average reduction in MFI was 51%. In 10 instances for which there are both qPCR and flow cytometry data, as shown in Figure 5D, the reductions in MFI were significantly correlated with the reductions in message by linear regression analysis (r² = 0.56, P < .01). In 5 instances, the plasma cells made both an intact immunoglobulin and IgL, and in those cases, a pattern of increased intracellular IgH immunofluorescence related to the degree of reduction of IgL immunofluorescence could be seen (Figure 5E).

Caspase3/7 activity was evaluated with si[IGLC_CR] or si[-] in 6 specimens making IgL only, giving relative luminescence units of 24 186 ± 19 114 and 22 999 ± 15 772 in si[IGLC_CR]-treated and si[-] control aliquots, respectively (P = .57, 2-tailed paired Student t test). In contrast, when caspase 3/7 activity was evaluated in 5 specimens making intact antibodies, as shown in Figure 5F, there was a significant increase in caspase 3/7 activity with si[IGLC_CR] treatment (P = .04, 2-tailed paired Student t test). These results support the conclusion that 1 siRNA pool targeting consensus sequences in the IgL CR message can effectively reduce both IgL message and IgL protein production without regard for the diversity of IgL variable region sequences and can also trigger a terminal ER stress response in cells making intact antibodies.

We report that the targeting of consensus sequences in the CR of λ-light-chain genes with 1 pool of siRNA (si[IGLC_CR]) can rapidly and substantially reduce λ-light-chain (IgL) production in numerous clones of λ plasma cells without regard for the diversity of the variable regions. Moreover, in λ plasma cells making intact immunoglobulin, si[IGLC_CR] also can activate both UPR and ERAD with terminal potential because of the stress associated with unpaired IgH in the ER. Preliminary data indicate that the terminal potential is inversely related to the ratio of residual IgL to basal IgH, a hypothesis amenable to further testing. The ratio depicts the relative availability of a light chain to pair with a heavy: with a ratio of <1, both the excess of unpaired IgH and the ER stress-related terminal potential may be higher.

Activation of the UPR is a complex interplay of pathways that can be associated with either restored homeostasis or apoptosis.²⁷  The UPR is constitutively activated in plasma cells as they differentiate from B cells with increased demand for protein folding and ER trafficking capacity; it is particularly the activity of spliced form of X-box-binding protein 1 that escorts the metamorphosis of B cells into morphologically distinct antibody-producing cells.⁶  In myeloma, the UPR has been a major theme in the study of the mechanisms of action of proteasome inhibitors.^28-31  The importance of the PERK–activating transcription factor 4–CHOP arm of the UPR as a potential indicator of caspase-dependent apoptosis highlights a paradoxical aspect of the role of the UPR in plasma cell biology.³² 

In this report, knockdown of IgL production and accumulation of unpaired intracellular IgH activate the UPR and ERAD in plasma cells making intact immunoglobulins but not in plasma cells making only IgL. Of note, the loss of IgH production with manufacture of only IgL and the possible toxicity of IgH to plasma cells are fundamental themes in plasma cell biology.^7,33-36  After passage in culture, hybridomas often lose IgH production and make IgL only, and the majority of human myeloma cell lines produce IgL without IgH partners.³³  IgL-deficient mice attain a complete block in B-cell development at the stage when light-chain rearrangement should occur, resulting in surface IgM deficiency, retention of unpaired IgH in the cytoplasm, and lack of plasma cells.³⁷  Normal human plasma cells make more IgL than IgH, possibly to minimize intracellular accumulation of unpaired IgH.³⁸  In 20% of cases of multiple myeloma, only IgL is made.³⁹  Moreover, when myeloma relapses after the plateau phase, we often observe “light-chain escape,” the emergence of a modified phenotype in which IgL replaces, or is produced far in excess of, IgH.^40,41 

The model of terminal UPR and ERAD activation that we report is caspase and NOXA dependent. Regulation of NOXA has been related to DNA damage as well as ERAD and epigenetic changes that enhance the activity of specific transcription factors; moreover, in vitro, bortezomib-induced apoptosis has been shown to depend on NOXA.^42-45  The model we report provides the opportunity to investigate in greater detail how plasma cells may overcome, adapt, or succumb to intrinsic terminal ER stress signals.

siRNA therapeutics are in clinical trials for transthyretin-related amyloidosis, delivered via lipid nanoparticles targeting hepatic cells producing transthyretin⁴⁶ ; moreover, investigators interested in AL have recently demonstrated that RNA interference with IgL production is feasible and merits further study.^47,48  Targeting specific types of cells, such as clonal marrow plasma cells, for RNA interference poses additional major challenges. The significance of the results with si[IGLC_CR] with respect to turning off IgL production and secretion may prove relevant to light-chain-mediated diseases. Some of the AL patient samples treated with si[IGLC_CR] were obtained from patients whose plasma cell disease had not responded to conventional therapies including bortezomib; the reduction of IgL message in si[IGLC_CR]-treated cells may reflect the importance of RNA degradation to the malignant plasma cell phenotype,⁴⁹  and, although the reductions were in some cases substantial, they were also notably variable, accenting the need for skepticism as this work continues. Further study of si[IGLC_CR] must seek to optimize siRNA design, packaging, and delivery for in vivo testing,⁵⁰  and this report encourages those ongoing efforts.

Contribution: P.Z. and X.M. conceived, designed, and conducted experiments and wrote the manuscript; C.C. conducted clinical research, obtained patient marrows, and wrote the manuscript; L.I. oversaw microarrays, analyzed and presented gene expression studies, and wrote the manuscript; and R.L.C. conceived the design of the research and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Raymond L. Comenzo, Tufts Medical Center, Box 826, 800 Washington St, Boston, MA 02111; e-mail rcomenzo@tuftsmedicalcenter.org.

The authors thank the Division of Hematology-Oncology and Departments of Medicine and Pathology at Tufts for their continued support, Andrew Evens for helpful reading of the manuscript, and the Tufts Medical Center and Tufts University School of Medicine core facilities and their staffs for assistance.

This work was supported by the Amyloidosis and Myeloma Research Fund at Tufts, the Cam Neely and John Davis Myeloma Research Fund, the Sidewater Family Fund, the Lavonne Horowitz Trust, the Werner and Elaine Dannheiser Fund for Research on the Biology of Aging of the Lymphoma Foundation, the Amyloidosis Foundation (P.Z.), and the Demarest Lloyd Jr Foundation, with its continuing commitment to “shutting down the factory” in AL.