(2005) 19, 2264–2272 & 2005 Nature Publishing Group All rights reserved 0887-6924/05 $30.00 www.nature.com/leu CD38 expression levels in chronic lymphocytic leukemia B cells are associated with activation marker expression and differential responses to interferon stimulation

BT Pittner1, TD Shanafelt2, NE Kay2 and DF Jelinek1

1Department of Immunology and Division of Hematology, Mayo Clinic College of Medicine, Mayo Graduate School, Rochester, MN, USA; and 2Department of Internal Medicine, Mayo Clinic College of Medicine, Mayo Graduate School, Rochester, MN, USA

CD38, a surface whose expression increases upon B cells.14 Lastly, CD38 also exhibits adhesion molecule normal B-cell activation, is a marker of disease aggression in B- properties.15 cell chronic lymphocytic leukemia (B-CLL). Higher percentages of CD38-expressing CLL B cells may be found in lymphoid As we believe that the rate of disease progression might be compartments compared to peripheral blood. Therefore, it is correlated to the proportion of leukemic B cells bearing a more possible that although CLL B cells are resting, CD38 may be a activated phenotype, we have also looked at other activation marker of recent cell activation prior to entry into the periphery. markers on CLL B cells. A variety of cell surface molecules have To address this hypothesis, we examined the association of been shown to play a role in normal B-cell activation and to CD38 expression with other activation antigens identified in increase in expression upon cell activation, for example, CD38, expression profiling experiments and include CD18, 16–18 CD49d, CD20, and subunit 5 of the anaphase-promoting CD20, and integrins. Therefore, given that CD38 currently complex/cyclosome. We found that all these markers were serves as a prognostic marker of B-CLL disease aggressiveness, more highly expressed in leukemic B cells from CD38-positive we sought to determine if these or additional activation markers CLL patients. Lastly, because interferon is known to modulate correlate to CD38 expression. Exploratory experiments compar- CD38 expression, we used IFN-a to test the ability of CLL B cells ing the CLL B-cell profiles of a small cohort of to increase CD38 expression in vitro. Interestingly, IFN uniformly CD38-positive vs uniformly CD38-negative CLL stimulation only modulated CD38 expression in CLL B cells that already expressed CD38. Taken together, these data patients identified several activation marker candidates. These suggest that CD38 is a marker of a more recently activated candidates included CD20, CD18, CD49d, and subunit 5 of the CLL . This in turn may explain the biological and clinical cell cycle-associated cyclosome protein, anaphase-promoting differences between CD38-positive type B-CLL and CD38- complex/cyclosome (APC/C). Although B-CLL is characterized negative type B-CLL. by a massive accumulation of resting leukemic B cells in the Leukemia (2005) 19, 2264–2272. doi:10.1038/sj.leu.2403975; peripheral blood, examination of these activation , published online 6 October 2005 Keywords: B-CLL; CD38; interferon; cell activation; cell cycle including CD38, could shed on the developmental stage and some of the biological capabilities of these leukemic cells, including their response to environmental factors. Indeed, a recent report suggests that signaling through CD38 leads to extensive surface receptor signaling networks that may function 19 Introduction to promote a pool of proliferating CLL B cells. Since IFN-a modulates CD38 expression levels on normal B cells20 and monocytes21 and may increase CD38 expression In B-cell chronic lymphocytic leukemia (B-CLL), CD38 mem- 22 brane expression levels serve as a prognostic tool. Several levels on leukemic cells, in this study we chose to examine its reports suggest that patients with 7–30% CD38-positive ability to modulate CD38 on various CLL B-cell populations that leukemic cells exhibit a more aggressive disease course and have different levels of membrane CD38 at the onset of culture. have an inferior response to treatment.1–7 However, one study Herein, we report that leukemic clones from CD38-positive B- suggests that even lower percentages of CD38-expressing CLL patients display a phenotype congruent with a more leukemic B cells correlate with poor outcome8 and other studies recently activated B cell and it is exclusively these CD38- suggest that CD38 expression levels will change during the expressing leukemic cells that have a heightened ability to disease course.9–11 Recently, one study showed that CD38- modulate CD38 expression in response to IFN-a. expressing CLL B cells respond to antibody-induced ligation of 12 CD38 with increased cellular signaling. To date, it remains Materials and methods unclear whether CD38 signaling plays a direct physiologic role in B-CLL or serves merely as a surrogate marker of disease Patient and control specimens aggressiveness. CD38 is a type II transmembrane protein that has several Heparinized peripheral blood was obtained from B-CLL patients functions in diverse cell types. First, it acts as an ecto- under the approval of the Mayo Clinic Institutional Review catalyzing the conversion of NAD þ to cADPR (reviewed in 13 2 þ Board. Peripheral blood mononuclear cells (PBMC) were Deaglio et al ) which in turn induces Ca flux into the cell. isolated by Ficoll density-gradient centrifugation, and PBMC Secondly, upon antibody ligation, CD38 induces phosphoryla- isolated from all B-CLL patients studied were composed of tion of CD19 and prevents apoptosis of germinal center 490% CD5 þ CD19 þ B cells. All patients were assessed by flow-cytometry using PE-conjugated anti-CD38, APC-conju- Correspondence: Professor DF Jelinek, Department of Immunology, gated anti-CD19, FITC-conjugated CD5, and PE-, APC-, and Guggenheim 4, Mayo Clinic College of Medicine, 200 First St SW, FITC-conjugated isotype controls (BD-Pharmingen, San Diego, Rochester, MN 55905, USA; Fax: þ 1 507 266 0981; E-mail: [email protected] CA, USA). Normal PBMC were isolated from discarded buffy Received 5 January 2005; accepted 26 August 2005; published online coats from healthy blood donors. Freshly isolated samples were 6 October 2005 used in all experiments. CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2265 Gene expression profiling of CD38-negative and - analysis of the CD38neg and CD38pos leukemic cells revealed positive patients 499% purity (data not shown).

We performed gene expression profiling (GEP; Affymetrix U133A platform) on primary leukemic B cells isolated from Western blotting individual CD38pos and CD38neg patients and optimized identification of differentially expressed associated with Magnetic bead-purified CD19-positive CLL B cells were freshly CD38 expression by restricting the gene profiling to patient cells isolated from B-CLL patients that were either uniformly negative that were unimodally negative and positive for CD38 expres- for CD38 expression (n ¼ 5) or uniformly positive (n ¼ 5). Viable sion. In both groups of patients, CD19-positive cells were tonsillar mononuclear cells were obtained from discarded purified using magnetic bead cell sorting on a Miltenyi tonsillar tissue, and the B-CLL-derived MEC-1 cell line23 was AutoMacs prior to isolation of total RNA. To determine genes obtained from the DSMZ, Germany. Cells were directly lysed that appear to be differentially expressed in this set of into 2 Â SDS sample buffer containing 2-mercaptoethanol, and experiments, we used a recently developed model-based lysates were boiled for 5 min and then stored at À201C until algorithm (PAM analysis) that uses the normalized probe-level further use. After polyacrylamide gel electrophoresis and data as well as information about the underlying experimental transfer, the membranes were blotted with anti-APC/C subunit design. 5 (BioLegend, San Diego, CA, USA), then stripped, and subsequently blotted with anti-b-actin (Novus Biologicals, Littleton, CO, USA) for a loading control. Flow-cytometric analysis of surface marker expression

Freshly isolated PBMC from B-CLL patients were stained with APC-conjugated CD19, FITC-conjugated CD5, and either a PE- Results conjugated isotype control or PE-conjugated CD38, CD20, CD18, or CD49d (BD Pharmingen). CD38 positivity was defined Patient characteristics and CD38 expression patterns in as any patient sample where the quotient of the mean B-CLL fluorescence intensity (MFI) of the CD38 staining divided by the MFI of the isotype control was greater than 1. For patients The present study included 47 CD38neg and 43 CD38pos B-CLL displaying CD38 bimodal leukemic populations, the total MFI of patients who were untreated or had received the last dose of CD38 staining was used (ie, the fluorescence intensity was chemotherapy more than 1 year previously (Table 1). There averaged for both peaks). For CD49d staining, 47 CD38neg and were no statistically significant differences in sex distribution, 43 CD38pos patient samples were examined. For CD18 staining, age, Rai stage, time from diagnosis, or treatment status in the 33 CD38neg and 22 CD38pos patient samples were examined. two groups defined by CD38 expression. With respect to For CD20, 29 CD38neg and 22 CD38pos patient samples were CD38pos B-CLL patients, the expression patterns of CD38 are examined. The total MFI of each surface marker or isotype known to vary tremendously between B-CLL patients.8,12 control was determined by analyzing viable CD19-positive/ Indeed, the data shown in Figure 1 are illustrative of CD38 CD5-positive cells. The MFI ratio was calculated as follows: MFI expression variability on gated CD19 þ CD5 þ ratio equals the quotient of the total MFI of antisurface marker within and between B-CLL patients, for example, uniformly high staining divided by the MFI of the isotype-matched control. (Figure 1a), intermediate (Figure 1b), or low (Figure 1c) CD38 expression; uniformly negative CD38 expression (Figure 1d); and bimodal CD38 expression where patient leukemic B cells CD38 induction and expression level analysis are both CD38neg and CD38pos (Figure 1e and f).

Magnetic bead-purified CD19-positive cells or total PBMC were cultured for 48 h in 10% FCS/RPMI with or without 1000 U/ml Table 1 Patient characteristics interferon-a 2b (IFN-a) or IFN-g (Schering Corp., Kenilworth, NJ, USA. In all experiments, the results with IFN-g were congruent CD38 positivea,b CD38 negativec with the results with IFN-a; therefore, only the IFN-a results are included in this report. Fold induction of CD38 expression on Total (n)4347 viable CD19/CD5-positive cells is defined by a quotient Median age at diagnosis (range)d 65.5 (32–84) 57 (27–79) Male:female 29:13 29:18 expressed as follows: fold induction ¼ MFIIFN-a ratio/MFInil ratio, where the numerator is the MFI of anti-CD38 staining divided by Rai stage at diagnosis the MFI of the isotype-matched control when cultured with IFN- 0/I 35 41 a (MFIIFN-a ratio) and the denominator is the MFI of anti-CD38 II 6 4 staining divided by the MFI of the isotype-matched control when III/IV 1 1 cultured in media alone (MFInil ratio). Prior therapy (n)e 21 14 aDemographic data were unavailable for one of the CD38-positive Flow-cytometric sorting of CD38-negative and CD38- patients. positive cells from bimodal populations bRai stage of one CD38-positive patient has not been determined at the time of sampling. cRai stage of one CD38-negative patient has not been determined at a neg pos To examine the effects of IFN- on CD38 vs CD38 cells the time of sampling. isolated from the same patient, CD19-positive magnetic bead- dThe probability that these two populations differ based on age is: neg pos purified B-CLL cells were sorted into CD38 and CD38 P ¼ 0.17 (t-test). populations using a gating strategy shown in Figure 5. Post-sort ew2 ¼ 3.79; P40.05.

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2266 a 128 b 256

M1 M1 Events Events

0 0 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height

c 256 d 256

Events M1 Events M1

0 0 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height

e 256 f 256

Events M1 Events M1

0 0 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height

Figure 1 The patterns of CD38 expression in CLL B cells. (a–f) CD38 staining profiles (gray-shaded histogram) and isotype controls (open-line histogram) represent CLL B-cell populations that are CD38 high, CD38 moderate, CD38 low, CD38-negative, bimodal CD38 low, and bimodal CD38 high, respectively. The percentages of CD38 staining with fluorescence greater than the isotype control are indicated for each example.

CD38 expression correlates with expression of other explore this molecule, as well as the integrins and APC/C 5, at activation markers the protein level to validate their gene expression profiles. It has been well established that integrins are major Recent reports indicate that even low levels of CD38 correspond contributors to activation and survival and expres- to disease aggressiveness in B-CLL.8 Interestingly, the patient sion levels increase following cell activation.24 We next examples in Figure 1c and e also fit this definition, but they determined if CD49d and CD18 were more highly expressed clearly display different profiles; one is CD38 low with a on the leukemic cells of CD38pos B-CLL cases, and indeed this unimodal histogram peak and the other displays bimodal CD38 was observed (Figure 3a and b). Of interest, the CD38pos subset low expression, with only a small fraction of CLL-B cells of patients included those with expression of CD49d or CD18 at expressing CD38 (Figures 1c and e, respectively). To learn more or above the median value of the population, and those with about the possible role of CD38 in this disease, we performed low to negative expression. Moreover, all the patients expressing exploratory gene expression profiling on primary leukemic B low levels of CD49d and CD18 also displayed bimodal CD38 cells that were unimodally negative for CD38 and those that expression. The relative expression levels of CD49d or CD18 in expressed unimodally high levels of CD38. We initially studied this latter subgroup did not correlate to the percentage of the two CD38pos and two CD38neg patients and several genes that CD38pos leukemic cell fraction (data not shown). Furthermore, are also known as activation markers emerged from this there is no evidence of CD49d or CD18 bimodal patterns approach. These markers include the integrins, CD49d and regardless of CD38 expression patterns (data not shown). Thus, CD18, as well as CD20 and the cell cycle-related protein, even though two populations may exist within a patient as anaphase promoting complex/cyclosome subunit 5 (APC/C 5). defined by CD38 expression, both populations display uniform An additional nine patients (six CD38pos and three CD38neg) CD49d or CD18 expression. were profiled and the combined data are graphically shown in Normal B-cell expression of CD20 increases early in response Figure 2. In the first set of patients, the difference in gene to complete activation.16 However, one group showed that expression of CD20 was statistically significant; however, upon CD20 expression may decrease in response to only partial cell study of additional patients by this methodology, CD20 no activation.25 Additionally, CD20 is lost during terminal differ- longer displayed statistical significance (Figure 2c). Owing to the entiation into plasma cells.26 We next determined if CD38 important role of CD20 in B-cell biology, we continued to expression correlated to CD20 expression in order to gain

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2267 a b 590 p < 0.0002 1400 p < 0.0003

490 1200

380 910

270 660

170 400

61 140 c d 8400 p = 0.17 1800 p < 0.0006

6800 1600 1400 5200

3600 1100

2000 880

380 640

CD38neg CD38pos CD38neg CD38pos

Figure 2 Gene expression values from CD38neg and CD38pos CLL B cells. The gene expression values, determined with a model-based algorithm that uses normalized probe-set data, are depicted from representative probe-sets for CD49d, CD18, CD20, and APC/C 5 (a, b, c, and d, respectively). insight into the developmental/activation stage of CLL B cells. Figure 3c shows that there is a statistically significant higher expression of CD20 in CD38pos B-CLL. Again, CD38pos patients with CD20 expression levels below the median express bimodal CD38.

Interferon is only able to modulate CD38 expression on CLL B cells from CD38-positive patients

Since interferons are known to stimulate CD38 expression, we next examined the ability of IFN-a and IFN-g to alter CD38 expression in both CD38pos and CD38neg CLL B-cell popula- tions. B-CLL cells from Patient 1 significantly increased CD38 expression in response to IFN-a (Figure 4a) as measured by Figure 3 CD38 expression in CLL correlates to other surface activation markers. CLL patients’ B cells were triple-stained for percent CD38pos cells and MFI (4.4-fold induction). However, neg CD49d, CD18, or CD20 (a, b, and c, respectively). The mean CLL B cells from patients that were uniformly CD38 at fluorescence intensity (MFI) ratio is a quotient defined as the total MFI isolation did not acquire CD38 expression in response to IFN-a of the surface marker stain divided by the total MFI of the isotype (Figure 4a, Patient 2). Moreover, in a total of 21 CD38neg and 19 control. The left-most, center, and right-most columns of each panel pos (open circles, open triangles, and filled triangles, respectively) are from CD38 CLL patients examined pre- and post-culture with IFN- neg pos a, the inability of IFN-a to upregulate CD38 expression in CD38 B-CLL patients, CD38 B-CLL patients, or normal PBMC, neg respectively. The horizontal line in each column represents the CD38 CLL patients remained constant (Figure 4b). B cells median value of the points within that column. from healthy donors (n ¼ 4) with low to moderate levels of CD38 responded to IFN-a with a significant increase (Po0.001) in CD38 expression compared to CD38neg CLL B cells (Figure 4b). Treatment of B-CLL cells with IFN-g had similar effects for CD38 induction (data not shown). Additionally, CLL culture with either Interestingly, even minimal CD38 expression on CLL B cells IFN had no effect on the expression of CD20, CD49d, or CD18 was associated with a CD38 induction in response to IFN-a. regardless of CD38 status (data not shown). For the CD38neg CLL Thus, the CLL B cells from Patient 3 (Figure 4) had very low-level B cells, it is unlikely that these effects are due to complete IFN CD38 expression at onset of culture and IFN-a stimulation unresponsiveness or lack of receptors because IFN-a maintained increased the CD38pos fraction to 44%. We have observed CD38neg cell viability and increased expression of HLA class I similar levels of CD38 induction in six other B-CLL patients with molecules (data not shown). low numbers of CD38pos B cells, ranging from 5.8 to 16%.

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2268 Fresh nil +IFN-α a Patient 1 128 84 128

Events 92% 60% 56% Events M1 M1 M1 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height FL2-Height

Patient 2 128 128 128

0.3% 0.1% 0.8% Events Events M1 M1 M1 0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height FL2-Height

518 518 Patient 3 518

4.8% 5.7% 44% Events M1 EventsM1 EventsEvents Events M1

0 0 0 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height FL2-Height

b 100 p < 0.001

10 Fold Induction

1

CD38neg CD38pos normal B-CLL at onset B cells

Figure 4 IFN-a increases CD38 expression only on CD38pos CLL B cells. (a) Freshly isolated CLL B cells were cultured for 2 days in 10% FCS/ RPMI in the presence ( þ IFN-a) or absence (nil) of 1000 U/ml IFN-a and viable leukemic B cells were analyzed for expression of CD38 by flow- cytometry. Three representative experiments are shown by the upper panels (Patient 1), middle panels (Patient 2), or lower panels (Patient 3) displaying the CD38 staining profile (gray-shaded histogram) in each of the following conditions: freshly isolated CLL B cells (left panels), 48-h media alone (center panels), or 48-h with IFN (right panels). An isotype control staining profile (open-line histogram) is overlaid for each condition. The percent of CD38-stained cells that is captured by the marker is shown in each panel. (b) Cumulative data representing patients whose leukemic cells were either CD38neg (open circles) or CD38pos (open triangles) at onset as well as peripheral blood B cells from normal individuals (black triangles) and cultured for 2 days as in (a) above. Fold induction corresponds to the change in MFI ratios and is calculated as stated in the Materials and methods.

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2269 Despite the increase in CD38 expression, not all CLL B cells B-CLL CD38 expression correlates to the expression of become CD38pos, suggesting that a subpopulation of leukemic APC/C 5 cells are resistant to CD38 upregulation. For Patient 3 it was also conceivable that the cells that Since CD38 expression is higher in CLL cells that are more increased CD38 expression in response to IFN-a could have responsive to external stimuli and because they express other evolved from CD38neg cells prior to stimulation. To address this, activation markers, it is possible that these cells have more we sorted CD38pos and CD38neg CLL-B cells from a CLL patient, recently exited from or are better primed to re-enter the cell whose leukemic cells displayed a bimodal pattern of CD38, and cycle. In support of this concept, the exploratory gene then immediately cultured these separate populations with or expression analysis identified the expression level of APC/C 5 without IFN-a (Figure 5). Of note, the B-cell subpopulation that as higher in CD38pos B-CLL (Figure 2d). APC/C is a multisubunit expressed CD38 at the onset of culture was the primary complex that is expressed throughout the cell cycle and is population that acquired increased CD38 expression in re- responsible for degrading cell cycle-related proteins in a timed- sponse to IFN-a (3.7-fold induction; Figure 5c and e). By critical manner.27 To confirm the gene expression data, we contrast, the leukemic cells that appeared to lack CD38 assessed protein expression levels of APC/C 5 by immunoblot- expression at isolation were largely nonresponsive to IFN-a ting (Figure 6). These data show that CD19-purified, CD38neg with greater than 90% of the cells remaining CD38neg.Of CLL B cells express very little to no APC/C 5 (three interest, although a small increase was noted in the CD38neg representative patients shown, lanes 3–5). Conversely, some sorted population, this response could be due to incidental CD38pos CLL B cells express detectable levels of this subunit cosorting of cells that expressed extremely low levels of CD38 at (two representative patients shown, lanes 6 and 7). Of the CLL the onset of culture. Alternatively, it could reflect the response of B-cell patient samples tested to date, we found that none of the those cells that recently lost CD38 expression and are poised to five CD38neg CLL B-cell samples and three of five CD38pos CLL reacquire it in response to IFN-a. B-cell samples express detectable levels of APC/C 5. As

a 64

Negative Positive gate gate Events

0 100 101 102 103 104 FL2-Height

b 64 c 64

nil 4%

Events Events 12% M1 M1

0 0 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height

d 256 e 256

+ IFN-α

Events 10%Events 73% M1 M1

0 0 100 101 102 103 104 100 101 102 103 104 FL2-Height FL2-Height

Figure 5 CD38-positive vs CD38-negative cells sorted from a bimodal patient respond differently to IFN-a. The CD38 staining profile (a) of one of two representative patients shows two discernible populations of leukemic B cells in freshly isolated CD19-purified CLL B cells as well as the gate settings for the CD38-positive vs -negative sort. CD38neg-gated cells (b and d) or CD38pos-gated cells (c and e) were then cultured in the absence (b and c) or presence (d and e) of IFN-a for 48 h and then analyzed by flow-cytometry for CD38 expression (gray-shaded histogram). An isotype control staining profile (open-line histogram) is overlaid for each 48-h culture condition.

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2270 findings suggest that the regulation of these other activation markers is different from CD38 regulation in CLL B cells. Moreover, the relative abundance of the CD38pos population had no positive association with the MFI ratio of the integrin staining (data not shown). Of interest, Durig et al30 performed gene expression profiling (approximately 5600 genes) on 25 CD38pos (defined as 420% CD38pos cells) and 45 CD38neg (o20% CD38pos cells) B-CLL patients and concluded that Figure 6 APC/C 5 expression is greater in CD38pos CLL B cells. Whole cell lysates from tonsillar mononuclear cells (T), the CLL- although the vast majority of genes on these chips were derived MEC-1 cell line (M), three CD38neg CLL B-cells (À), and two expressed at similar levels in the two subsets of patients, 14 CD38pos CLL B-cells ( þ ) were analyzed by immunoblotting for APC/C genes appeared to be differentially expressed, including CD49d. 5 expression. The blots were then reprobed for b-actin as a loading Our results with CD49d are therefore consistent with this report control (lower panel). This experiment is representative of a total of neg pos and we provide validation at the protein level that this integrin is five of five CD38 B-CLL samples and three of five CD38 B-CLL more highly expressed in CD38pos CLL B cells. The results samples tested. presented in this report underscore the difficulties of determin- ing CD38 expression levels based on a fixed cutoff and highlight the need to carry out gene expression profiling on a larger cohort expected, the MEC-1 CLL cell line displayed high levels of uniformly CD38pos vs CD38neg type B-CLL (in progress) to (Figure 6, lane 2), while tonsillar mononuclear cells displayed reveal additional genetic differences that may exist between moderate levels of APC/C 5 expression (Figure 6, lane 1). these two subsets of disease that may underlie differential disease progression. Similarly, when analyzing the association of CD20 expression Discussion and CD38 expression we found that there is a significantly higher expression of CD20 in CD38-positive B-CLL. For those CD38 is expressed at higher levels in activated germinal center CLL B cells with CD20 expression levels below the median of (GC) founder B cells, GC B cells, and early memory B cells the CD38pos cohort, a bimodal expression of CD38 was within the secondary lymphoid compartment.28 The same study detected. In normal B-cell development, CD20 expression showed that GC founders and early memory B cells exist in increases early in response to activation16 and decreases in peripheral blood; a finding that could reflect the status of the response to only partial activation.25 These data suggest that the leukemic B cells in B-CLL and other hematological diseases.28,29 leukemic cells from CD38pos B-CLL patients may have arrested Peripheral blood CD38-positive B cells in B-CLL could arise and undergone tumorigenic transformation at an earlier devel- from a site of B-cell activation with a continued production and opmental stage than those from CD38neg patients. release of these GC-like cells into the periphery. Lastly, the expression of cell cycle-related protein, APC/C To further understand the underlying biology and develop- subunit 5, correlated with CD38 expression. This protein is part mental stage of CLL B cells as they correlate to CD38 of a multisubunit complex that functions at critical times expression, we initiated an exploratory analysis of genes throughout the cell cycle.27 For instance, when the APC/C is expressed by uniformly CD38pos or uniformly CD38neg CLL B bound to the activator, Cdc20, it promotes the metaphase-to- cells from a growing cohort of patients and our results identified anaphase transition. Similarly, when the complex is bound to a variety of molecules that may be expressed at higher levels in Cdh1, it promotes the G1-to-S phase transition. In eucaryotes, CD38pos B-CLL. This study focused on a small number of these APC/C 5 acts as a regulatory component of the complex, candidate genes and we validate that they are positively supported by the finding that APC/C 5 mutations cause a pre- associated with CD38 expression levels. These genes include metaphase stall in Drosophila development.31 Here, we found the integrins CD49d and CD18, as well as CD20 and APC/C 5. that CD38neg CLL B cells expressed very low to undetectable levels of APC/C 5 through immunoblotting, whereas three of five CD38pos CLL B-cell samples displayed detectable levels. Given CD49d, CD18, CD20, and APC/C 5 expression that APC/C 5 is expressed throughout the cell cycle, these data correlate to CD38 expression suggest that CD38pos cells have more recently exited from the cell cycle and/or are better primed to re-enter the cell cycle. CD49d expression is decreased on resident B cells of the Peyer’s Further exploration into the regulated expression of these and patch and other lymphoid compartments, but is upregulated other cell cycle-related proteins in B-CLL would be instructive upon activation.18 The fact that CD49d, CD18, and perhaps regarding CLL disease progression. other adhesion markers are low on CD38neg CLL B cells (Figure 3) suggests that these cells could be more likely found as residents of the peripheral blood lymphoid compartment. As The IFN responsive state of CLL B cells correlates to integrins and CD38 have been shown to play a role in CD38 expression migration, the lack of their expression on this subset of B-CLL cells may prevent these cells from recycling back into lymphoid CD38pos B-CLL cells are thought to express a more activated tissues and permanently commits them to a long-lived terminal phenotype relative to CD38neg cells.32 In addition, CD38pos CLL residency in the peripheral blood. Thus, it can be speculated B cells appear to be more responsive to stimulation through BCR that CD38pos CLL B cells may have a higher turnover rate. crosslinking than CD38neg B-CLL.33,34 One possibility for this Interestingly, although we found bimodal expression of CD38, differential response is that CD38pos cells may also express a we did not find concomitant bimodal expression of other surface higher level of other activation and/or signaling molecules and activation markers. Additionally, with the integrins, CD18 and are poised to respond to environmental stimuli, as recently CD49d, the bimodal CD38pos CLL B cells displayed lower than shown.19 Of interest, our results show that only CD38pos cells, median-level expression of these integrins (Figure 3). These even those with very low-level CD38 expression, respond to

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2271 IFN-a with increased expression of CD38 (Figures 4 and 5). References Taken together, these studies lead one to speculate that CD38pos cells are better prepared to respond to cues from their 1 Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL et al. Ig V environment, whereas CD38neg cells appear to be in a more gene mutation status and CD38 expression as novel prognostic dormant, unresponsive state. Further, the role of CD38 as a indicators in chronic lymphocytic leukemia. Blood 1999; 94: cofactor in cell activation, due to its ability to act as an adhesion 1840–1847. 2 Hamblin TJ, Orchard JA, Gardiner A, Oscier DG. Immunoglobulin molecule as well as enhance calcium mobilization,13,35 pos V genes and CD38 expression in CLL. Blood 2000; 95: 2455–2457. supports the notion that the CD38 cells are more sensitive 3 Matrai Z, Lin K, Dennis M, Sherrington P, Zuzel M, Pettitt AR et al. 33 36 to environmental stimuli, either negatively or positively. CD38 expression and Ig VH gene mutation in B-cell chronic While a small increase was noted in the CD38neg sorted lymphocytic leukemia. Blood 2001; 97: 1902–1903. population, from 4% to 10% positivity (Figure 5b and d, 4 Jelinek DF, Tschumper RC, Geyer SM, Bone ND, Dewald GW, respectively), as noted above, this response could be due to Hanson CA et al. Analysis of clonal B-cell CD38 and immunoglo- bulin variable region sequence status in relation to clinical incidental co-sorting of cells that expressed extremely low levels outcome for B-chronic lymphocytic leukemia. Br J Haematol of CD38 at the onset of culture. Alternatively, the data in 2001; 115: 854–861. neg Figure 4 (Patient 3) and Figure 5 suggest that CD38 CLL B 5 D’Arena G, Musto P, Cascavilla N, Dell’Olio M, Di Renzo N, Perla cells from CD38-bimodal patients could reflect cells that have G et al. CD38 expression correlates with adverse biological very recently lost CD38 expression but are still poised to features and predicts poor clinical outcome in B-cell chronic reacquire CD38 upon exposure to IFN. Moreover, this may lymphocytic leukemia. Leukemia Lymphoma 2001; 42: 109–114. neg 6 Del Poeta G, Maurillo L, Venditti A, Buccisano F, Epiceno AM, suggest that not all CD38 cells from bimodal profiles (Figure Capelli G et al. Clinical significance of CD38 expression in chronic 1e and f) may be as unresponsive to stimuli as the leukemic cells lymphocytic leukemia. Blood 2001; 98: 2633–2639. neg from homogeneously CD38 CLL patients (Figure 1d). Similar 7 Krober A, Seiler T, Benner A, Bullinger L, Bruckle E, Lichter P et al. neg findings were also observed in some CD38 patients whose VH mutation status, CD38 expression level, genomic aberrations, leukemic cells were cultured with IL-2 by Deaglio et al.12 Again, and survival in chronic lymphocytic leukemia. Blood 2002; 100: these data underscore the complexity of this disease and suggest 1410–1416. 8 Ghia P, Guida G, Stella S, Gottardi D, Geuna M, Strola G et al. The that bimodal CD38 leukemic cells may display different pattern of CD38 expression defines a distinct subset of chronic biological properties. lymphocytic leukemia (CLL) patients at risk of disease progression. Lastly, the failure of IFN stimulation to increase CD38 Blood 2003; 101: 1262–1269. expression in CD38neg CLL cells from patients who are 9 Ibrahim S, Keating M, Do KA, O’Brien S, Huh YO, Jilani I et al. unimodally CD38neg suggests that the CD38 interferon response CD38 expression as an important prognostic factor in B-cell locus35 may be inaccessible perhaps due to methylation,37,38 or chronic lymphocytic leukemia. Blood 2001; 98: 181–186. 10 Hamblin TJ, Orchard JA, Ibbotson RE, Davis Z, Thomas PW, that a component of the signaling pathway from the IFN-a Stevenson FK et al. CD38 expression and immunoglobulin variable receptor to the transcription of CD38 gene is missing or region mutations are independent prognostic variables in chronic dysfunctional. Of interest, recent reports state that CD38 lymphocytic leukemia, but CD38 expression may vary during the expression is higher in bone marrow and lymphoid tissues, the course of the disease. Blood 2002; 99: 1023–1029. presumed sites of CLL proliferation and cell activation, 11 Chang C-C, Cleveland RP. Conversion of CD38 and/or myeloid- compared to peripheral blood.8,39 These findings suggest that associated marker expression status during the course of B-CLL: upon CLL B-cell emigration from a lymphoid compartment into association with a change to an aggressive clinical course. Blood 2002; 100: 1106. blood, the expression of CD38 begins to decrease. In summary, 12 Deaglio S, Capobianco A, Bergui L, Durig J, Morabito F, Duhrsen we believe that it is possible that all B-CLL tumorigenic founders U et al. CD38 is a signaling molecule in B-cell chronic are CD38pos and the CD38 B-cell percentage in peripheral lymphocytic leukemia cells. Blood 2003; 102: 2146–2155. blood is directly proportional to the turnover rate with low 13 Deaglio S, Mehta K, Malavasi F. Human CD38: a (r)evolutionary turnover favoring the presence of CD38neg CLL-B cells, since story of and receptors. Leukemia Res 2001; 25: 1–12. CD38 expression is associated with a higher activation state. 14 Zupo S, Rugari E, Dono M, Taborelli G, Malavasi F, Ferrarini M. CD38 signaling by agonistic monoclonal antibody prevents The ability of CLL B cells to proliferate in vivo was recently 40 apoptosis of human germinal center B cells. Eur J Immunol examined in a study using deuterated water. In this report, 1994; 24: 1218–1222. Messmer and co-workers provide provocative evidence that CLL 15 Dianzani U, Funaro A, DiFranco D, Garbarino G, Bragardo M, B-cell turnover rates can differ among patients. This study did Redoglia V et al. Interaction between and not find a correlation to CD38 expression, but additional studies CD4+CD45RA+ lymphocytes. Role of the human CD38 molecule. using larger sample sizes are underway. Finally, it must be noted J Immunol 1994; 153: 952–954. 16 Wheeler K, Gordon J. Co-ligation of surface IgM and CD40 on that current published studies and the data presented herein do pos naı¨ve B lymphocytes generates a blast population with an not rule out that CD38 B-CLL cases could arise from ambiguous extrafollicular/germinal centre cell phenotype. Int neg transformation of CD38 precursors. These hypotheses require Immunol 1996; 8: 815–828. further investigation. 17 Dang LH, Rock KL. Stimulation of B lymphocytes through surface Ig receptors induces LFA-1 and ICAM-1-dependent adhesion. J Immunol 1991; 146: 3273–3279. 18 Postigo AA, Sanchz-Mateos P, Lazarovits AI, Sanchez-Madrid F, de Acknowledgements Landazuri MO. a4b7 integrin mediates B cell binding to fibronectin and vascular molecule-1. J Immunol We thank Bonnie Arendt, Nancy Bone, Cheryl Jankiewicz, and 1993; 151: 2471–2483. Renee Tschumper for their technical assistance, Dr Karla Ballman 19 Deaglio S, Vaisitti T, Bergui L, Bonello L, Horenstein AL, and Bruce Morlan for their assistance in analyzing the exploratory Tamagnone L et al. CD38 and CD100 lead a network of surface gene expression study data, and Dr Clive Zent for his insightful receptors relaying positive signals for B-CLL growth and survival. Blood 2005; 105: 3042–3050. comments. This study was supported by the National Institutes of 20 Galibert B, Burdin N, de Saint-Vis B, Garrone P, Van Kooten C, Health Training Grant T32-HL67742 (BTP), R01 CA91942 (NEK), Banchereau J et al. CD40 and B cell antigen receptor dual and generous philanthropic support provided by Mr Edson triggering of resting B lymphocytes turns on a partial germinal Spencer. center phenotype. J Exp Med 1996; 183: 77–85.

Leukemia CD38-positive CLL B cells display an activated phenotype BT Pittner et al 2272 21 Musso T, Deaglio S, Franco L, Calosso L, Badolato R, Garbarino G 31 Bentley AM, Williams BC, Goldberg ML, Andres AJ. Phenotypic et al. CD38 expression and functional activities are up-regulated characterization of Drosophila ida mutants: defining the role of by IFN-g on human and monocytic cell lines. APC5 in cell cycle progression. J Cell Sci 2002; 115: 949–951. J Leukocyte Biol 2001; 69: 605–612. 32 Manocha S, Matrai Z, Osthoff M, Carter A, Pettitt AR. Correlaton 22 Bauvois B, Durant L, Laboureau J, Barthelemy E, Rouillard D, between cell size and CD38 expression in chronic lymphocytic Boulla G et al. Upregulation of CD38 gene expression in leukemic leukaemia. Leukemia Lymphoma 2003; 44: 797–800. B cells by interferon types I and II. J Interferon Res 1999; 33 Zupo S, Isnardi L, Megna M, Massara R, Malavasi F, Dono M et al. 19: 1059–1066. CD38 expression distinguishes two groups of B-cell chronic 23 Stacchini A, Aragno M, Vallario A, Alfarano A, Circosta P, Gottardi lymphocytic with different responses to anti-IgM anti- D et al. MEC1 and MEC2: two new cell lines derived from bodies and propensity to apoptosis. Blood 1996; 88: 1365–1374. B-chronic lymphocytic leukaemia in prolymphocytoid transforma- 34 Lanham S, Hamblin T, Oscier D, Ibbotson R, Stevenson F, tion. Leukemia Res 1999; 23: 127–136. Packham G. Differential signaling via surface IgM is associated 24 Pribila JT, Quale AC, Mueller KL, Shimuzu Y. Integrins and - with VH gene mutational status and CD38 expression in chronic mediated immunity. Annu Rev Immunol 2004; 22: 157–180. lymphocytic leukemia. Blood 2003; 101: 1087–1093. 25 Anolik J, Looney RJ, Bottaro A, Sanz I, Young F. Down-regulation 35 Ferrero E, Malavasi F. Human CD38, a leukocyte receptor and of CD20 on B cells upon CD40 activation. Eur J Immunol 2003; ectoenzyme, is a member of a novel eukaryotic gene family of 33: 2398–2409. nicotinamide adenine dinucleotide-converting enzymes. J Immu- 26 Arpin C, Dechanet J, Van Kooten C, Merville P, Grouard G, Briere nol 1997; 159: 3858–3865. F et al. Generation of memory B cells and plasma cells in vitro. 36 Zupo S, Massara R, Dono M, Rossi E, Malavasi F, Cosulich ME Science 1995; 268: 720–722. et al. Apoptosis or differentiation of B-chronic 27 Zachariae W, Nasmyth K. Whose end is destruction: cell division lymphocytic leukemia cells induced by cross-linking of surface and the anaphase-promoting complex. Genes Dev 1999; 13: IgM or IgD. Blood 2000; 95: 1199–1206. 2039–2058. 37 Rush LJ, Plass C. Alterations of DNA methylation in hematologic 28 Bohnhorst JO, Bjorgan MB, Thoen JE, Natvig JB, Thompson KM. malignancies. Cancer Lett 2002; 185: 1–12. Bm1–Bm5 classification of peripheral blood B cells reveals 38 Fulop Z, Csernus B, Timar B, Szepesi A, Matolcsy A. Microsatellite circulating germinal center founder cells in healthy individuals instability and hMLH1 promotor hypermethylation in Richter’s and disturbance in the B cell subpopulations in patients with transformation of chronic lymphocytic leukemia. Leukemia 2003; primary Sjogren’s syndrome. J Immunol 2001; 167: 3610–3618. 17: 411–415. 29 Arce E, Jackson DG, Gill MA, Bennett LB, Banchereau J, Pascual 39 Jaksic O, Paro MMK, Skelin IK, Kusec R, Pejsa V, Jaksic B. CD38 V. Increased frequency of pre-germinal center B cells and plasma on B-cell chronic lymphocytic leukemia cells has higher expres- cell precursors in the blood of children with systemic lupus sion in lymph nodes than in peripheral blood or bone marrow. erythematosus. J Immunol 2001; 167: 2361–2369. Blood 2004; 103: 1968–1969. 30 Durig J, Nuckel H, Huttmann A, Kruse E, Holter T, Halfmeyer K 40 Messmer BT, Messmer D, Allen SL, Kolitz JE, Kudalkar P, Cesaar D et al. Expression of ribosomal and translation-associated genes is et al. In vivo measurements document the dynamic cellular correlated with a favorable clinical course in chronic lymphocytic kinetics of chronic lymphocytic leukemia B cells. J Clin Invest leukemia. Blood 2003; 101: 2748–2755. 2005; 115: 755–764.

Leukemia