In my last post I presented an exciting new approach to treating cancer and viral infections called Adoptive Cell Therapy. The technique employs engineered transmembrane proteins known as Chimeric Antigen Receptors (CAR) to direct the serial killing action of cytotoxic T cells (T lymphocytes). However, the selection of the target to which it homes is of considerable importance.

The process of tuning cytotoxic T lymphocytes (CTL) to kill cancer cells is undoubtedly a spectacular innovation but it can be precarious. The primary function of a CTL is to bind and kill any cell found bearing the ligand of the CTL’s antigen receptor (the CAR in this case) on its membrane. There are many factors that must be addressed when tailoring the therapy to treat a given pathology; the most significant of which is selecting the CAR’s ligand (its target). Other factors include how many CAR⁺ T cells should be injected and how to control their proliferation and longevity after injection. After completing the reading that went into this post I am not certain that, one, erbB-2 is a suitable CAR target, and two, I have been convinced that the use of chemotherapy as a neoadjuvant (administered just prior) to any targeted molecular therapy should be questioned.

In stark contrast to the years of safe administration and the cancer clearing success stories of late; recently there were two deaths in CAR based clinical trials. The case I’d like to focus on is one in which an on-target/off-tissue event appears to be the cause of death (1). The Cytotoxic T Lymphocyte action was on-target in that its CAR bound to the proper ligand, erbB-2; however it was off-tissue in that the ligand was expressed on normal cells in healthy tissue.

My first thought was that a poor choice was made in target selection. The CAR transfected into this person’s T cells had specificity for the extracellular domain of transmembrane receptor erbB-2 which is now commonly known as Human Epidermal growth factor Receptor 2 (HER2). However, given the fact that the CAR was built using the monoclonal antibody (mAb) Trastuzumab (Herceptin) (1) which has a stellar record of safety and efficacy, it seemed I was wrong.

ErbB-2 belongs to the ErbB family of transmembrane receptors now referred to as the Epidermal Growth Factor (EGF) receptor family which has four members. The signals generated by these receptors regulate cell proliferation, differentiation, and survival in mammalian cells (2). As their name implies they can be found in cells all over the body. Upon learning this I had two questions. First, was erbB-2 expressed in any healthy adult tissue? Perhaps it is only the other three family members that are expressed in normal cells and erbB-2 in certain types of malignant cells. And second, if it is expressed on normal cells, what is the level of its expression (how much in a given cell)?

* On a side note, during the search I found that the level of expression of erbB-2 may be the key to growth of the malignant tissue in which its found. Although it is an orphan receptor in that it does not bind any known ligand, if the expression level is high enough, it can form homodimers and produce growth and proliferation signals without one (1); thus its high oncogenic potential.

An immunohistochemical analysis (using antibodies to label a specific protein within a cell) in 1990 revealed that ErbB-2 was expressed in adult human epithelial cells “in the gastro-intestinal, respiratory, reproductive, and urinary tract as well as in the skin, breast and placenta.” (3) I also learned that erbB-2 serves as a critical signaling partner to the other erbB family members and also plays a major role in normal cardiac development (4).

In the paper presenting the women’s death, the proposed cause of death was the binding of the CTL to erbB-2 expressed on healthy lung tissue which I think is consistent with the findings of the immunohistochemical analysis. The SAE report concluded with the postulate that the woman’s death was cytokine-induced pulmonary toxicity and edema which lead to cytokine storm (essentially a massive cell-signal feedback loop) and subsequent multiple organ failure. This led me to believe I may not have been wrong after all; maybe erbB-2 is safe for mAb-based therapies, but not CAR\CTL. However, the researchers speculated that it was cytokine release, what I thought was a secondary CTL function, as the cause not the CTL’s primary effector function, cell killing. I will have to explore why that is.

I had some difficulty sifting through the papers and data to determine the levels of expression in those normal adult endothelial cells; some sets even presented conflicting data. This is due in part to the fact that I’m not independently wealthy and the cost of access to most research papers is high. Other major factors include the fact that it turns out that there are multiple splice variations (isotypes) of erbB-2. However, I did find some information on the function, structure, and expression patterns in two great papers.

From the first I learned about the structure and function of the erbB family of receptors and the a family of ligands, Neuregulin (NRG), whose Epidermal Growth Factor domains bind to erbB receptors. This paper, which also presented a summary of the NRG/erbB signaling pathway, was written by Nadia Hedhli and Kelly Russell (4). The second, from M F Press, et al, was the only paper I found which described, in depth, the expression levels of erbB-2 and offered a safety rationale for using it as a therapeutic target. The quantitative analysis went so far as to suggest a threshold for erbB-2 receptor sites per cell under which a mAb, combined with a payload that must be endocytosed, would be effective. The safety rationale stated that since within normal, adult, endothelial cell membranes there are less than the threshold, 500k sites/cell, using the erbB-2 mAb would be safe and, given that the number is higher malignant cells, effective (3).

While I found the evidence presented by M F Pass et al. compelling, they were delivering a ribosomal inhibitor that required endocytosis and migration to specific areas within the cell in sufficient quantities to be effective. The sites/cell threshold, it seems to me then, must be significantly lower when it comes to the use of cytotoxic T lymphocytes.

My search to answer the questions of erbB-2 expression distribution and levels has yielded information that seems to fit the facts. Significant cardio and pulmonary pathologies after treatment with Trastuzumab and one fatality shortly after receiving a transfusion of CTLs containing an erbB-2 tuned CAR. In fact, the NIH drug information page for Herceptin (Trastuzumab) lists serious cardio and pulmonary, along with “swelling of the arms, hands, feet, ankles or lower legs; weight gain (more than 5 lbs in 24 hours); dizziness; loss of consciousness”, in the list of side effects.

However, Trastuzumab is a great success and among those treated serious adverse effects were only reported in a minority of cases. For example, Medscape reported in 2011 that only 1-4% of patients experienced severe cardio toxicity (Herceptin as adjuvant to chemo). So the next question for me is why, given the seemingly common expression of erbB-2 would the percentage be so low. Could the methods used to perform the immunohistochemical analysis caused its expression? I will read further to investigate despite the fact that it will probably only amount to an intellectual given that there a number of far larger brains than mine already looking in to it.

When I combined what I found so far with the report that an increase in expression of ErbB-2 is linked to protective mechanisms in Hedhli and Russell’s paper, I began to question the use of chemo and radiation as a neoadjuvant treatment to erbB-2-targeted CAR and even Trastuzumab regimens. It seemed telling, given erbB-2’s role in development which I think would follow injury, not just embryogenesis, that while Trastuzumab’s efficacy has repeatedly been shown to increase with chemo as a neoadjuvant so, apparently, is the occurrence of toxicity (4).

As an example, it has been shown that activation of erbB-2 and -4 promotes the growth and survival of cardiac myocytes (cells that control heart rate) (4). Anthracycline has been shown to induce apoptosis in cardiomyocytes and the function of the erbB family with its ligand Neuregulin (NRG) has been shown to protect those myocytes (5). Further, when tissue containing endothelial cells is excised for culture and those cells are examined, erbB-2 expression is up-regulated. The reason has not been determined (1), however, given erbB-2’s established signal functions and its essential role in cardiac development, is it unreasonable to conclude that the tissue damage unavoidable in the process of excision may have induced the pathway leading to erbB-2 expression.

Could the tumor, chemo, and/or radiation sourced destruction in healthy tissue result in increased expression of erbB-2? And does this up-regulation contribute to the cardio and or pulmonary toxicity seen in Trastuzumab administration? The mechanism of action for Trastuzumab and mAb\cytotoxin-based treatments may allow for the leeway necessary for such successful treatment history. The efficiency of action of cytotoxic T lymphocytes may not.

In the end, erbB-2 is a “self” antigen. If not always, at least at certain times or under certain conditions, the receptor is expressed and functions in normal cells. This in itself doesn’t render it a questionable choice. CD19 is a self-antigen expressed on early B cells and when it was targeted in a CAR treatment, the chemorefractory patients exhibited a complete response; cancer free six months out. These patients’ healthy B cells were also targeted, however, given that lymphocytes are regenerated and that they do not form a critical structure in the body, they can be targeted with relative safety; chemo, for example kills which much less discretion. The same cannot be said with certainty, in my opinion, of erbB-2.

Works Cited

1. Case Report of a Serious Adverse Event Following the Administration of T Cells Transduced With a Chimeric Antigen Receptor Recognizing ErbB2. Morgan, Richard A, et al., et al. 4, 2010, The American Society of Gene & Cell Therapy, Vol. 18, pp. 843–851.

2. A comprehensive pathway map of epidermal growth factor receptor signaling. Oda, Kanae, et al., et al. 2005, 2005, Molecular Systems Biology.

3. Expression of the HER-2/neu proto-oncogene in normal human adult and fetal tissues. Press, M F, Cordon-Cardo, C and Stamon, D J. 7, 1990, Oncogene, Vol. 5, pp. 953-62.

4. Cytostatic Drugs, Neuregulin Activation of ErbB Receptors, and Angiogenesis. Hedhli, Nadia and Russell, Kelly Strong. 2010, Current Hypertension Reports, pp. 411-417.

5. Neuregulin-1 protects ventricular myocytes from anthracycline-induced apoptosis via erbB4-dependent activation of PI3-kinase/Akt. Fukazawa, Ryuji , et al., et al. 12, 2003, Journal of Molecular and Cellular Cardiology, Vol. 35, pp. 1473-1479.

6. Safer CARS. Heslop, Helen E. 4, s.l. : Nature, 2010, Molecular Therapy, Vol. 18, pp. 661–662.

So you’ve been told that all the cells in your body contain the same genetic material? I’ve seen it written as a general principle of Genetics and a quick search around the web will yield many sources indicating that all somatic cells contain the exact same DNA. I did find a few articles in science magazines which “revealed” that this may not be the case, but the general consensus that they do was prevalent.

From all I’ve read, research and texts included, the DNA contained in all somatic cells is the same with one exception; the hematopoietic cells. These cells are the cells of the blood and they all descend from a single progenitor, the Human Hematopoietic Stem cells which can be found in bone marrow. These stem cells give rise to a diverse population of progeny including erythrocytes (red blood cells), T Cells, B Cells, Macrophages, and Natural Killer cells. To be accurate, the stem cells and their immediate descendants contain “germ-line” DNA, which is DNA that was formed during the individual’s fertilization event and is shared with virtually every other somatic cell in the body. However, as the hematopoietic descendants start to specify into the progenitors of T and B cells or “lymphopoietic” progenitors, the process of Somatic Recombination which creates the antigen receptors I’ve been writing about cuts out and discards parts of certain chromosomes thus creating genetically distinct human cells within our own bodies.

In my last post I presented the gene loci on which this process acts and highlighted the first actors in the process; the Recombination Activating Gene 1 & 2 proteins (RAG1 & RAG2). The arrangement of the gene loci is very important in this discussion and the rest of what I’m about to present doesn’t make much sense without this knowledge, so you can click this link to read that post if necessary. The process is performed by the V(D)J Recombinase and the process progresses as the molecules that make it up are assembled, I will introduce them as I describe the process.

It begins with two molecules each containing one RAG1 and one RAG2. These complexes then randomly bind, one each, to two of the hundreds of Recombination Signaling Sequences (RSSs) found surrounding each gene segment in the target antigen locus. It should be noted that the selection is not random across the segment. For example, two V regions will not join; it must be a V and J, for example, in a light chain locus. After binding, the two complexes come together and bind resulting in the formation of a loop of DNA between them which contains all the gene segments that may have been between the two segments brought together.

With the loop formed, the RAG complex will cut the strand exactly at the points where the RSS ends and the coding segment begins (see diagram below). This happens because the RAG complex has an inherent ability to act as an endonuclease. This essentially means that it is a protein that can break a DNA molecule into parts. This is the action that excises a piece of our DNA which is called the signal joint (because the RSS of each segment are joined flush). At this point, because of reasons relating to the chemical properties of the structure of a molecule of DNA, the ends of each strand which are cut fuse to form hairpins.

At this point other proteins join the complex to complete the recombinase and the process. These proteins are actually ubiquitous in human cells and I believe they are present for every mitotic and meiotic event that occurs over an individual’s lifetime. The first two proteins to bind to the growing V(D)J Recombinase are named Ku70:80 and they hold together both the ends of the chromosomal DNA strand, and those of the signal joint. Note, that this is the last I will mention the signal joint as once it’s ends are fused, it is no longer part of our story.

It is here where the process becomes a little unnerving for me. The very precise cut made by the RAG1/RAG2 complex to setup the joining of gene segments is one thing, but what happens next, to me, is another. The following video presents an excellent visualization of the entire process with a voice-over by Dr. Julie Theriot; a brilliant mind. I will, however, go into a little more detail below.

A protein named DNA-PK, or DNA-dependent Protein Kinase, is formed when another protein named DNA-PKcs binds tightly with Ku after the strands are bound. And to this another protein named Artemis joins. Artemis is also an endonuclease without which we would be ending our story with a broken strand of DNA. Artemis cuts the hair pins formed after RAG1/RAG2 excised the signal joint, but the location of its cut appears to be random. It has been found that they occur anywhere along the hair pin and this is where the fun starts.

With the location being random, unequal ends are often left which must be equalized before the DNA can be rejoined. Enter another protein that is found in all our cells and has a great name; Terminal Deoxynucleotidyl Transferase (TdT). TdT randomly adds bases to the cut ends resulting in what is essentially a newly formed random sequence on each. At the same time, two more newcomers bind to each other and then the Recombinase; DNA Ligase IV and XCR44. There are ubiquitous DNA repair proteins which and as soon as they bind, they start trying to re-join the strands; the name Ligase actually comes from the Latin “ligare” meaning “to bind or tie”. With TdT inserting random bases, the repair/ligation complex may need to knock out and replace non-complementary pairings such as a Guanine to a Thiamine.

With the ligation effort winning out in the end and the chromatid now back in one piece, the process of Somatic Recombination is complete. With just a few more steps we have a complete gene which, when transcribed will form one of the two chains in an Immunoglobulin. The process, however, has left the chromatid irrevocably changed; it is unique, different from any other member of its chromosome pair in the body, including other B cells. I’d like to note again that this change is key to the efficacy of our adaptive immune response. Without the enormous diversity generated by this process, neither the B and T cell antigen receptors nor the antibodies secreted by the former would be capable of binding effectively to the diverse and ever changing peptides making up evolving pathogens. However, it also introduces a potential for a number of pathologies, including cancer.

It has been shown in research1 that if the DNA repair mechanisms like DNA Ligase IV\XCR44 and the DNA damage checkpoints are not functioning properly, Somatic Recombination can result in several pathologies. These include aneuploidy, an abnormal number of chromosomes, translocations, where one part of a chromosome is fused with another, and neoplastic transformation, cancer.

In fact, according to both Janeway’s Immuno Biology (p 312) and a paper published in Nature Immunology in 20071, translocations are found in most lymphoma tissues. What is compelling for me about these translocations is that they most often involve what are known as proto-oncogenes which are genes encoding proteins that, when not functioning properly, cause cancer.

What causes DNA checkpoints and the DNA repair proteins to malfunction? I’ve seen the research showing the obvious conclusion; mutations forming in the genes coding for the proteins which carry out the functions. However, I suspect epigenetics may play a role in at least some of the scenarios. What causes these translocations? Given the fact that many lymphomas are age onset, I would conclude that probability must be considered, but the translocations of proto-oncogenes suggest to me that epigenetic factors should be examined. There are genes such as Myc (Myelocytomatosis)that regulate the expression of up to 15% of our entire genome and do so through epigenetic factors such as HAT (Histone Acetyltransferase). It should not be surprising that Myc is an oncogene and it is, in fact, one that is involved in the translocations found in lymphoma tissues.

1. Nature Immunology 8, 801 – 808 (2007)
Published online: 19 July 2007 | doi:10.1038/