Readers may have noticed that I have not yet delivered my promised summary of the IWCLL meeting. This is because I have been working on one of the topics that featured at the meeting, and it has taken some time to get my appreciation of this topic clear enough to write about it. The topic is CD38.
CD38 is an interesting molecule for CLL doctors and patients alike. It is found on many types of cells, but we are really interested in its role on B-lymphocytes. It is there from time to time as the B cell differentiates. It appears on bone marrow precursor cells, but is lost on mature lymphocytes; on germinal center cells it protects against apoptosis, but on leaving the germinal center, memory cells lack the antigen; on terminally differentiated plasma cells it is one of the few surface antigens present. In chronic lymphocytic leukemia (CLL) expression of CD38 signifies a poor prognosis although it does not correlate precisely with the presence of unmutated immunoglobulin variable region (IgVH) genes and it may vary during the course of the disease.
It is a molecule that lives on and in the cell membrane and it is an enzyme; that is, it catalyzes chemical reactions. It is involved in signaling through the B-cell receptor (BCR) and such signaling causes cell division. We also know that some cytokines like IL-2 and interferon gamma can increase the level of CD38 present.
The molecule that binds to CD38 (its ligand) is platelet endothelial cell adhesion molecule-1 (PECAM-1) which is also known as CD31, and it is present on CLL cells and the cells lining blood vessel walls (endothelial cells).
The microenvironment is critical for the growth of CLL cells. The cells that are in the peripheral blood deceive us; the tissue is where the action is. In lymph nodes, spleen and bone marrow are proliferation centers and in these centers the CLL cells come into contact with activated T lymphocytes and various stromal cells. The bulk of CLL lymphocytes are in the solid lymphoid organs rather than the blood. Studies using heavy water labeling of CLL cells demonstrate that they have definable and substantial birth rates varying from one cell in a hundred to one cell in a thousand of the clone per day. It is established that CLL cells circulate from blood to lymphoid organs and back, and that the solid organs provide a favorable environment for dell division, and that this is in part mediated by CD38.
A number of recent studies have examined precisely what goes on in those proliferation centers, and how CD38 is involved.
It is well established that T and B lymphocytes talk to each other and the chief means by which they do this is through the CD40-CD40 ligand system. CD40 is present on all B cells and CD40 ligand (or CD154) is present on activated T cells.
A well known culture system for B cells is to lay them on a human fibroblast cell line which expresses CD154 and feed them the cytokine IL-4. A recent paper from Willimott et al demonstrates that when CLL cells are treated in a similar way it causes them to increase the amount of CD38 and in many cases also ZAP-70. This occurred in cases with both mutated and unmutated IgVH cells. The authors speculate that CLL cells visit the lymph nodes, spleen or bone marrow and there are stimulated by CD154 which increases the amount of CD38 and ZAP-70. Some lose these molecules after leaving the proliferation center but others retain them and those that lose them are the abnormal ones that don’t divide very well and snooze most of the time. These are the ones to have, and it occurs most commonly in those with mutated IgVH genes.
Another paper from Damle et al looks at a similar question. They separated CLL cells from individual patients into CD38+ and CD38- cohorts and examined them for markers associated with cell division: CD27, CD62L, CD69; or with entry into the cell cycle: Ki-67; or with signaling: ZAP-70; or with protection from apoptosis: telomerase and Bcl-2. In general these molecules marked the CD38+ cohort of cells, no matter whether it was large or small. But both positive and negative cells had telomeres of equal length, suggesting that they belong to the same population and have undergone a similar number of divisions.
Some years ago, Tom Kipps’ lab described what he called ‘nurse-like cells’. A subset of blood cells from patients with CLL spontaneously differentiates in vitro into large, round, or fibroblast-like adherent cells that display stromal cell markers, namely vimentin and STRO-1. These cells also express stromal cell-derived factor-1 (SDF-1), a CXC chemokine that ordinarily is secreted by marrow stromal cells (it is also known as CXCL12). Leukemia B cells attach to these blood-derived adherent cells, down-modulate their receptors for SDF-1 (CXCR4), and are protected from undergoing spontaneous apoptosis in vitro.
CLL cells certainly express CXCR4, but there is no difference in expression between those with mutated or unmutated IgVH genes. However, Deaglio et al have found that the attraction to SDF-1 is greater for CD38+/ZAP-70+ cells than for CD38-/ZAP-70- cells. This paper also demonstrates that when CD38 is stimulated it results in the phosphorylation of ZAP-70 (the effect is very transient, occurring for just 5 minutes, starting 5 minutes after the stimulation). Phosphorylation is just a means of measuring that a molecule has been activated, and it so happens that by noting which tyrosines (one of the amino acids in the protein chains) has accepted the phosphate group they can determine whether the phosphorylation leads to activation (as in this case) or inhibition. Patients that are ZAP-70 negative are unable to signal via CD38.
This is not the whole story about CLL cells homing to bone marrow and lymph nodes. Nurse-like cells also express B-cell activating faction of the TNF family (BAFF), a proliferation-inducing ligand (APRIL), plexin-B1 and the chemokine CXCL13 (also known as B-cell attracting chemokine 1 – BCA-1). Its receptor, CXCR5, is expressed by CLL cells
There has been controversy over the stability of CD38 ever since we reported cases where CD38 increased during the course of the disease. It is known that exposing CLL cells to IL-2 increases the amount of CD38 they express; this paper demonstrates that similar culture conditions increase the expression of ZAP-70, and other workers have reported that in about 10% of patients ZAP-70 expression rises during the course of the disease.
This is a very busy paper, because it also looks at the gene expression profile of CLL cells that express both CD38 and ZAP-70 and are attracted to SDF-1. Remember it was gene expression profiling that first identified ZAP-70 as the gene that most comprehensively differed in IgVH mutated and unmutated CLLs. This approach identified 24 genes separating the bad prognosis cases from the good prognosis cases. Four of these, ZAP-70, LPL, ADAM29 and SEPT10 were already known. Of the others two are involved in cell-cell contacts (CD86 and CD11a – as are ZAP-70 and ADAM29), two control cytoskeletal organization (BAMBI and SEPT10), five are involved in the control of lipid metabolism (LPL, ABCA9, BCAT1, LTF and PXDNL), two are involved in transcription regulation (L3MBTL4 and HIST2H2AB) and one is thought to be a tumor suppressor gene (PTCH1). The function of the remaining eight is unknown.
The way that CD38 results are expressed – as a percentage of cells staining positively is very dependent on where the zero is set and some workers have preferred to give results as mean fluorescence intensity. But neither method truly expresses what is seen in the laboratory. In most cases there is a mixture of positive and negative cells; sometimes there is a single population of cells that looks to have a gradation of staining intensity, but in others it looks like there are two distinct populations of cells, one being CD38+ and one CD38-. Paulo Ghia has called these cases bimodal. Chris Pepper in Cardiff has been looking at these populations for some time. At first he used to talk about these cells as ‘biclonal’ but in a recent paper he has separated the positive and negative cells with a cell sorter and shown that they are part of the same clone – their IgVH gene signature is the same. However, although part of the same clone, the CD38 positive and negative populations express different levels of certain gene products. In particular, levels of vascular endothelial growth factor (VEGF) and its receptor (VEGFR-2) were higher in CD38+ cells than in their negative counterparts in the same patient. These raised levels correlated with resistance to spontaneous apoptosis in the test tube, which could be reversed by a VEGF inhibitor, SU1498, but not by anti-VEGF antibodies, implying the use of an internal autocrine VEGF survival loop. However, CD38- cells did survive better when bathed in exogenous VEGF.
One possible mechanism for this was identified. Mcl-1 (a member of the Bcl-2 family) is known to be upregulated in CLL and is thought to be responsible in part for the anti-apoptotic effect. Mcl-1 was expressed in higher levels in CD38+ cells.
Pepper’s group gave us some more information at the IWCLL meeting. The CD38+ population had higher Ki67 expression (meaning that more cells were close to division) than the CD38- population. Was the positive population dividing more rapidly than the negative population? No, when they looked at telomere length (telomeres are the ends of the chromosomes that get bitten off with every cell division) they were similarly short in CD38 positive and negative cells and moreover the levels of telomerase were the same (which ensure that this wasn’t an artifact). This means that both positive and negative cells had been through the same number of cell divisions and to me implies that the cells are cycling; that they acquire CD38 in the proliferation center, but as they leave it they lose it. The longer it is since the cell left the proliferation center, the lower will be the CD38 level.
Further elucidation of this story came form Chiorazzi’s group at the IWCLL. They have extended their heavy water experiments to study the CD38+ and CD38- populations. They demonstrated that cells that were CD38+ and that expressed brighter CD5 were more likely to have recently proliferated than those with negative CD38 and rather dimmer CD5.
Finally, some beautiful pictures of the whole process were generated by Piers Patten from Kings and presented at IWCLL. Using confocal microscopy, he demonstrated higher levels of CD38 in bone marrow than blood and in white pulp of spleen rather than red pulp – these are places where proliferation centers are likely to occur. All cells tat were Ki67+ also expressed CD38 and CD23, and all proliferating CLL cells were in contact with CD4+CD25+FOXP3- T-cells. Proliferating CD38+ CLL cells surrounded CD31+ microvessels in close proximity to areas of T-cell infiltration. In an in vitro model, contact with a CD31+ endothelial cell line and activated autologous T-cells caused an upregulation of CD38 on the CLL cells and increased viability.
From thse recent studies on CD38 we begin to understand what is happening in CLL. What we see in the blood is a reflection of what is happening in the tissue. Cell proliferation takes place in proliferation centers in lymph nodes, spleen and bone marrow. In these centers stimulation by CD31 on endothelial cells and CD154 on T-cells activates the CLL cells, upregulating CD38 and perhaps ZAP-70. This has the effect of protecting the cell against apoptosis. From the proliferation center it is sent out into the circulation where activation markers begin to decline. There is probably something intrinsically different about some CLLs compared to others (probably related to the mutational status of the IgVH genes) because some cells seem to be better at retaining some activation markers than others. At any rate the CLL cells are drawn back into the tissues by chemokines and once there repeat the whole cycle. CD38 can be seen as an index of how recently the CLL cell has visited a proliferation center.