Germinal Center B-Cells Resist Transformation by Kras Independently of Tumor Suppressor Arf

Activating mutations in Ras (N- and K-) are the most common point mutations found in patients with multiple myeloma (MM) and are associated with poor clinical outcome. We sought to directly examine the role of Ras activation in MM pathogenesis and used two different tissue-specific Cre recombinase mouse lines (Cγ1-Cre and AID-Cre), to generate mice with mutant Kras (KrasG12D) activated specifically in germinal center B-cells. We also generated mice with activation of the KrasG12D allele in a tumor-prone Arf-null genetic background. Surprisingly, we observed no significant disruption in B-cell homeostasis in any of these models by serum immunoglobulin ELISA, SPEP, flow cytometry and histological examination. We observed development of non-overlapping tumor types due to off-target Cre expression, but despite successful recombination in germinal center and later B-cell populations, we observed no B-cell phenotype. Together, these data demonstrate that Ras activation is not sufficient to transform primary germinal center B-cells, even in an Arf-null context, and that the temporal order of mutation acquisition may be critical for myeloma development. Specific pathways, yet to be identified, are required before Kras can contribute to the development of MM.


Introduction
Multiple myeloma (MM) is an incurable malignancy of antibody-secreting plasma B-cells, whose etiology remains poorly understood. Mutations in Ras genes, encoding key proteins regulating cell growth, differentiation and survival, occur commonly in MM with a prevalence of 20-39% [1][2][3]. Indeed, using a targeted sequencing approach to screen highly expressed tyrosine kinase and cytokine signaling genes in primary human patient myeloma, we previously identified mutations at codon 12 and 61 in N-and KRAS as being the only recurrent variation in our sample set [4]. Recent genome sequencing efforts also found Ras mutations to be the most common single nucleotide variant (SNV) in MM [4], suggesting that Ras activation is an important event in MM pathogenesis. The somatic SNVs found most frequently in MM are gain-of-function mutations in Ras oncogenes (Kras and Nras), causing constitutive activation of the Ras protein [5].
Despite the genomic evidence for Ras pathogenesis, the functional role of Ras activation in MM has not previously been tested. This issue is not trivial as the induction of neoplasia by Ras activation is highly dependent on cellular context [6]. Understanding the effects of Ras activation in mature B-cells will allow us to better define the downstream pathways critical for development of MM. Moreoever, pharmaceutical approaches to target cancers with mutant Ras are underway [7][8][9][10], and a pre-clinical model faithfully replicating Ras-driven myeloma would be critical in evaluating the therapeutic potential of these agents in myeloma.
Post-germinal center (GC) B-cells are strongly implicated as the cell of origin in MM by demonstration of stable immunoglobulin (Ig) switch clonotypes over the course of disease [11,12]. To test if expression of oncogenic Ras in GC B-cells was sufficient to induce myeloma, we utilized transgenic mice harboring a constitutively active Kras (G12D mutation) knocked-in to the endogenous Kras locus and flanked by a Lox-Stop-Lox cassette [13]. The Kras mouse model has been successfully used in several labs in developing cancer models [14,15] [13,16]. These mice were crossed with two different mature B cell-specific Cre recombinase (Cre) mouse strains (Cc1-Cre and AID-Cre) to definitively test the effects of Ras activation in post-GC B-cells, including downstream memory B and plasma cells [17,18]. As Ras activation can induce cellular senescence [19] and often requires cooperating mutations to induce transformation, so we also generated a strain of triple transgenic mice by crossing Kras G12D mice with mice null for the P19 ARF tumor-suppressor gene (Arf 2/2 ) [20]. Arf (P14 ARF in humans) is a potent tumor suppressor gene that cooperates with Ras activation in cellular transformation and carcinogenesis [21,22]. In patients with myeloma, the P14/P16 locus is methylated in 42% [23], although the biological significance of this epigenetic modification is contested [24]. Surprisingly, in these settings we found B-cell development to be only subtly perturbed, even in the setting of Arf deficiency. Conversely, mice frequently developed tumors harboring Crerecombined Ras alleles in non-B-cell tissues due to small amounts of off-target Cre expression. These data demonstrate that post-GC B-cells are resistant to transformation by mutations that are strongly oncogenic in other cellular contexts and that Ras activation must likely cooperate with tissue-specific mutations or epigenetic events to induce myeloma.

Cc1-Cre Kras G12D Mice Develop Thymic Lymphomas and Lung Adenomas but not Myeloma
To examine the effect of Kras in plasma cells, we generated double transgenic mice. In Kras G12D mice, the G12D mutation is knocked-in to the endogenous Kras locus, upstream of the Lox-Stop-Lox cassette ( Figure 1A,B). Kras G12D mice were crossed with mice expressing Cre recombinase (Cre) under control of the Ig heavy chain locus (Cc1-Cre) reported to express Cre selectively in a subset of germinal center B-cells ( Figure 1C).
We first confirmed that wild type Kras is strongly expressed in murine B-lineage cells; naïve splenic B-cells, germinal center Bcells, memory B-cells and plasma cells from C57BL/6 mice ( Figure 2A) [25]. As expected, Cre-mediated excision of the Kras allele stop cassette was robust and specific to B-lineage cells undergoing class-switch recombination in vitro ( Figure 2B and Figure S1). We also confirmed Cre-recombination in vivo in mature B-cell populations isolated from Cc1-Cre Kras G12D mice by fluorescence associated cell sorting (FACS). Splenic germinal center B-cells (B220 + /IgM 2/ GL7 + ) and class switched memory/ plasma cells (IgG1 + ) demonstrated clear, albeit low-level recombination, as did bone marrow plasma cells (B220 lo /CD138 + , Figure 2C).
We aged Cc1-Cre Kras G12D mice, both naïve and immunized with chicken gamma globulin to expand plasma cells, to monitor the development of disease. After 100 days, 58% (n = 12) of naïve mice developed weight loss, ruffled fur and shortness of breath and were found on necropsy to have thoracic cavity tumors. Unexpectedly, these tumors were T-lymphoblastic in phenotype (CD4 + CD8 + ) by flow cytometry ( Figure S2A). Additionally, 42% (n = 12) of naïve Cc1-Cre Kras G12D mice and 66% (n = 7) CGG-immunized Cc1-Cre Kras G12D mice were found to have lung nodules at autopsy (300 day endpoint). Sections of lung from immunized Cc1-Cre Kras G12D show well-demarcated nodules composed mostly of sheets of bronchial epithelial cells and some ''signet ring'' cells with bland nuclear features and absence of mitotic figures consistent with adenomas or low-grade adenocarcinomas (Figure S2B-E). Tissue from lung tumors in two independent Cc1-Cre Kras G12D mice shows partial recombination of the Kras allele ( Figure S2F). The immunized and unimmunized negative control Cc1-Cre mice showed no evidence of disease ( Figure 3A). Tissue from T-cell lymphomas found in two separate unimmunized Cc1-Cre Kras G12D mice showed complete Kras allele recombination, suggestive of loss of the wild-type allele, whereas spleen showed a partial recombination pattern consistent with infiltration of the spleen with these same cells ( Figure 3B). Despite extensive analysis, no B-lineage oncogenic transformation was observed in any Cc1-Cre Kras G12D mice. Bcell subsets in spleen and bone marrow and serum immunoglobulin levels were all normal (data not shown). Taken together, these data suggest that Kras G12D allele activation in germinal center B-cells failed to perturb B-cell homeostasis in Cc1-Cre Kras G12D mice.

AID-Cre-YFP Kras G12D Mice Develop Focal Epidermal Papillomas
Noting the low level of in vivo recombination in Cc1-Cre Kras G12D mice ( Figure 2C), and the lack of appreciable B-or plasma cell phenotype, we generated a second strain of mice using an independent tissue specific Cre allele. We crossed the Kras G12D mice with mice expressing Cre recombinase under the control of the activation-induced cytosine deaminase (AID) gene ( Figure 1D). AID is expressed with exquisite specificity in Bcells undergoing the germinal center reaction where it mediates class switch recombination and somatic hypermutation. To facilitate our analysis, this strain of mice also included the Rosa26-EYFP reporter allele, which allowed us to effectively track B-cells where recombination had occurred (AID-Cre-YFP Kras G12D ). Upon cre-mediated recombination, YFP marks cells where Kras G12D is also expressed. In an attempt to stimulate malignant B-cell transformation in AID-Cre-YFP Kras G12D mice, vitamin D deficient chow and/or sub-lethal radiation was given to cohorts of mice after immunization.
Robust Kras G12D allele recombination was induced in AID-Cre-YFP Kras G12D splenic B-cells undergoing plasmacytic differentiation and class switch recombination ex vivo ( Figure 4A). In contrast to the weak levels of in vivo recombination observed in Cc1-Cre Kras G12D mice, germinal center splenocyte populations and post germinal center cells isolated from AID-Cre-YFP Kras G12D mice showed robust Cre-mediated recombination at both the Kras G12D locus ( Figure 4B) and the YFP reporter in the spleen and to lesser extent in the bone marrow ( Figure 4C).
At 3 weeks of age, 100% (n = 20) AID-Cre-YFP Kras G12D mice lacked fur on the ventral neck and developed small growths, compared to control mice ( Figure 5A,B). Radiation and Vitamin D deficient chow (RV) treatments increased the number and size of growths on AID-Cre-YFP Kras G12D mice as early as 17 weeks, compared to AID-Cre-YFP Kras G12D given neither ( Figure 5C,D). By 26 weeks of age, all AID-Cre-YFP Kras G12D mice receiving both irradiation and vitamin D deficient chow (100%, n = 5) were hunched with ruffled fur and had infected lesions over the cutaneous growths with a median survival of 196 days ( Figure 5E, F). AID-Cre-YFP Kras G12D mice with no treatment (besides immunization) at 26 weeks had an increase in the number of growths similar in appearance to that at 17 weeks. At 17 weeks, Kras G12D mice given both irradiation and vitamin D deficient chow appeared healthy without growths, similar to the 26 week timepoint (data not shown). All AID-Cre-YFP Kras G12D mice regardless of irradiation or vitamin deficient chow subsequently died or were sacrificed due to persistent skin infections associated with fungating skin lesions ( Figure 6A). The cutaneous lesions were identified by histological examination to be benign papillomas (data not shown). Papillomas from 3 separate AID-Cre-YFP Kras G12D mice showed strong Cre-mediated recombination by PCR ( Figure 6B). A small increase in total serum gamma region protein level achieved statistical significance in AID-Cre-YFP Kras G12D mice fed vitamin deficient chow ( Figure S3A, middle panel), however the increase was not maintained over time, and mice treated with radiation, or no treatment at all had no significant changes in total serum gamma protein levels at any time point ( Figure S3A). Serum ELISA showed small changes among the antibody subtypes in AID-Cre-YFP Kras G12D mice, but no evidence of plasma cell transformation or any B-cell malignancy was found ( Figure S3B and data not shown). We reasoned that a lack of a detectable B-cell phenotype in Cc1-Cre Kras G12D and AID-Cre-YFP Kras G12D mice was most likely due to a requirement for a cooperating ''second hit'' to induce cellular transformation. Therefore, to test the effects of a second mutation known to cooperate with Kras G12D , we crossed AID-Cre-YFP Kras G12D mice into a tumor-prone Arf-null background (Arf 2/2 ) ( Figure 1). All AID-Cre-YFP Kras G12D Arf 2/2 mice developed rapidly progressive papillomas and by 13 wks, 66% of AID-Cre-YFP Kras G12D Arf 2/2 mice (n = 3) developed cutaneous sarcomas ( Figure 7A), while AID-Cre-YFP Arf 2/2 control mice remained disease-free ( Figure 7B). Histopathological sections of spleen from control mice show typical red pulp, white pulp and germinal center structures ( Figure 7D); whereas AID-Cre-YFP Kras G12D Arf 2/2 spleen showed defacement of splenic architecture with loss of distinction between red and white pulp and a paucity of germinal centers ( Figure 7C). Sections of sarcomas from AID-Cre-YFP Kras G12D Arf 2/2 showed characteristic undifferentiated spindle cells ( Figure 7E), consistent with tumors previously described in Arf-deficient mice (10). The only abnormalities attributable to Bcells that we identified were small but significant increases in polyclonal antibody responses over time. The gamma protein fraction by SPEP was higher in AID-Cre-YFP Kras G12D Arf 2/2 at 12 weeks compared to AID-Cre-YFP Arf 2/2 controls ( Figure  S4C), but none of the mice developed multiple myeloma or monoclonal gammopathy. AID-Cre-YFP Kras G12D Arf 2/2 and AID-Cre-YFP Kras G12D Arf +/2 mice also showed significant differences in total serum gamma region protein levels between baseline and 12 weeks ( Figure S4A). Serum ELISA of antibody subtypes from AID-Cre-YFP Kras G12D Arf 2/2 , AID-Cre-YFP Kras G12D Arf +/2 , and control AID-Cre-YFP Arf 2/2 also showed small but significant changes between baseline and 12 weeks in IgM and IgG isosubtypes ( Figure S4B), perhaps related to infected, fungating papillomas in these mice. Flow cytometric immunophenotyping of bone marrow and splenic mononuclear cells failed to detect the abnormal growth in any B-cell populations in AID-Cre-YFP Kras G12D Arf 2/2 mice.

Discussion
Kras is the oncogene most frequently mutated in MM, yet its role in the pathogenesis of the disease has yet to be elucidated. Here, we used a mouse model of activated Kras to directly test the effect of activated Kras in post-germinal center B-cells using two different Cre recombinases reported to be specific to germinal center B-cells. These mice developed T-cell lymphomas, lung For malignant transformation in many contexts, activated Ras requires cooperation with additional mutations [27] and we tried several strategies to accelerate disease in AID-Cre-YFP Kras G12D mice. Cohorts of AID-Cre-YFP Kras G12D mice were subjected to vitamin D deficient chow or sub-lethal radiation or both in an attempt to generate additional mutations and increase the proliferation of pre-malignant B-cells. The combination of vitamin D deficiency and radiation significantly accelerated and worsened the development of skin tumors in AID-Cre-YFP Kras G12D mice, but we observed no B-cell phenotype in any of these mice, despite extensive analysis.
Lastly, we engineered mice with a specific cooperating mutation, germinal center expression of Kras G12D in an Arf-null background. The Ink4a gene locus encoding both Ink4a and Arf is frequently silenced by hypermethylation in MM [28][29][30] and mutated in some cases of MM ( [31] and COSMIC database). Germline mutations in INK4a affect predisposition to plasmacytomas in mice [32] and to MM in people [33].
We observed significant acceleration of skin tumors and progression to invasive carcinomas, demonstrating the successful cooperation between the Kras and Arf pathways, but again, these mice failed to demonstrate a significant B-cell phenotype. The development of non-overlapping off-target tumors demonstrates that Kras G12D can mediate oncogenicity, but germinal center Bcells seem to possess an inherent resistance to its oncogenic effects. We conclude that activation of Kras alone or in the context of Arf pathway inactivation is insufficient to disrupt B-cell homeostasis. These negative data demonstrate that GC B-cells are refractory to mutations which are sufficient to transform other murine tissues, and suggest that distinct tumor suppressor pathways may be active in post-GC B-cells.
The temporal order of acquisition of mutations is likely to be important in the development of some cancers. Observational Figure 5. Gross appearance of cutaneous papillomas in AID-Cre-YFP Kras G12D mice is enhanced by tumor-promoting treatments. A) By 3 weeks of age, AID-Cre-YFP Kras G12D mice uniformly have hair loss and a single papilloma localized to the ventral neck; B) control Kras G12D mouse shows normal hair pattern and no papilloma; C) By 17 weeks, AID-Cre-YFP Kras G12D mice given radiation and vitamin D deficient chow (RV) had numerous fungating papillomas and more hair loss at the same site on the ventral neck; D) AID-Cre-YFP Kras G12D mice without tumor-promoting treatments also had progressive papillomas but much fewer and with less hair loss associated; E) AID-Cre-YFP Kras G12D +RV mice aged to 26 weeks showed confluent fungating and ulcerated masses at the ventral neck with spread to paws; F) age-matched control Kras G12D +RV mouse shows no similar signs. doi:10.1371/journal.pone.0067941.g005 studies have suggested that Ras activation is a ''late event'' in myeloma pathogenesis [34]. Ras mutations are significantly less common in patients with monoclonal gammopathy of uncertain significance (MGUS), and are not found in the memory B-cell population of patients with MM, [34] and our data directly supports the model that the temporal order of these events is important to the development of myeloma disease. The lack of a significant B-cell phenotype in our mice is reminiscent of the intrinsic resistance to the effects of Kras G12D displayed by intestinal cells. Intestinal homeostasis is unperturbed in mice by expression of Kras G12D alone [35], but carcinogenesis occurs with concurrent inactivation of the adenomatous polyposis coli (APC) tumor suppressor gene [36]. Mutations in APC do not occur with significant frequency in MM, and it remains unclear what specific mutations cooperate with Ras in myeloma development. In on going work, it will be important to determine the pathways that cooperate with Ras activation to transform germinal center B-cells.

Ethics Statement
This study was performed in strict accordance with animal use protocols approved by the Washington University Institutional Animal Care and Use Committee (IACUC, protocol number 20120152). Mice were euthanized if they met any early removal Figure 6. Cutaneous papillomas in AID-Cre-YFP Kras G12D mice and acceleration of lethality by tumor-promoting treatments. A) Kaplan-Meier survival curves of AID-Cre-YFP Kras G12D mice and control Kras G12D mice. Cohorts of AID-Cre-YFP Kras G12D and Kras G12D mice were subjected to vitamin D deficient chow continuously from 8.5 weeks of age or a single dose of sub-lethal gamma irradiation or given both. All AID-Cre-YFP Kras G12D mice developed progressive cutaneous papillomas that were made more extensive/aggressive with radiation or vitamin D deficiency. Mice were sacrificed when morbidity developed, defined by weight loss, unkempt coat, hunched posture, and lethargy. Each AID-Cre-YFP Kras G12D group had (n = 5) and developed papillomas, leading to infection, whereas every Kras G12D (n = 5) survived to day 352 endpoint. No B-cell phenotype was observed in any cohort. B) Cre-mediated recombination of Kras locus in DNA from papillomas was detected by PCR in three separate papilloma samples from AID-Cre-YFP Kras G12D mice. WT, wild-type control. doi:10.1371/journal.pone.0067941.g006 criteria (weight loss, lethargy, hunched posture, and/or ruffled coat) to limit suffering, in accordance with NIH-approved institutional animal care guidelines.

Kras Expression
Raw dataset files of Kras, Hras, and Nras expression in murine B cells was accessible through www.ncbi.nlm.nih.gov/geo (accession GSE4142).

Stimulation of Germinal Centers
1 mg of (4-hydroxy-3-nitrophenyl) acetyl conjugated chicken gamma globulin (NP-CGG; Biosearch Technologies, Novato, California) was mixed with Freud's Adjuvant, Complete (Sigma-Aldrich, St. Louis, Missouri) for primary immunization or Freud's Adjuvant, Incomplete (Sigma-Aldrich) for boosting immunization in 100 ul to inject intraperitoneally at specific time points ( Figure  S5). Where indicated, AID-Cre-YFP Kras G12D mice were given a Vitamin D deficient diet (Harlan, Madison, Wisconsin) beginning at 8.5 weeks of age and/or 4 Gy of sub-lethal ionizing radiation at 12 weeks of age.

Molecular Genotyping of Mouse Strains
For genotyping by PCR, genomic DNA was extracted from tail tissue using Extract-N-Amp Tissue Kit (Sigma-Aldrich). All primer sequences used for genotyping are available upon request. To detect Cre-mediated somatic recombination in bone marrow, spleen, and tumors, genomic DNA was extracted using DNeasy Blood and Tissue Kit (Qiagen, Germantown, Maryland), then amplified using PCR to yield expected 622-bp WT and 650-bp loxP, signifying recombination of the Kras G12D allele. Mature B cell populations sorted from spleen and bone marrow DNA was extracted using prepGEM Tissue Kit (ZyGEM, New Zealand).

Histopathology
Mouse tissues were fixed in 10% neutral buffered formalin for at least 48 hours, dehydrated in an alcohol gradient, cleared in xylene, and infiltrated and embedded in paraffin. Sections were stained for hematoxylin/eosin (H&E).

Flow Cytometric Analysis
Single cell suspensions of bone marrow and spleen briefly underwent red blood cell lysis.

MoFlo
Two 8-week-old Cc1-Cre Kras G12D or Kras AID-Cre mice were stimulated with 100 ug CGG intraperitoneally and sacrificed 14 days later. Preparation of spleen and bone marrow was previously described. Flow cytometric analysis was performed with the MoFlo single-cell sorter (Becton-Coulter, Brea, California).