Monthly Archives: December 2013

Cells underlying fibrosis: emerging themes

by Paul D. Rennert, 28 December 2013
A few weeks ago a paper describing the origin of cells responsible for dermal wound healing showed up in Nature, with a beautiful cover picture showing distinct cell lineages in skin. I actually missed the paper when it came out, so thanks to my friend Katherine Turner (ex-GI, ex-Biogen, now at Scholar Rock) for pointing it out.
The abstract is here, the full text requires a subscription. The paper is from Fiona Watt’s group at the Center for Stem Cells and Regenerative Medicine, King’s College, London.
The paper convincingly shows that the dermal layer of the skin is comprised of two distinct fibroblast lineages. Some context: the dermis is the layer of the skin that lies below the epidermis and includes the upper dermis (the dermal papilla) and the hypodermis (lower dermis). The major cells types residing in the dermis are fibroblasts, which along with the extracellular matrix they produce, form the connective tissue. Other prominent cells include macrophages and adipocytes (fat cells). There are also resident dendritic cells, Mast cells, T cells and other immune system cell types. Other structures include the hair follicles, sweat glands, oil glands, blood vessel endothelium, lymphatic vessels, and nerve endings.
Ok, so why is this paper interesting?
What the authors demonstrate is that the upper and lower dermis originate from distinct lineages of fibroblasts. This is shown using lineage tracing techniques and cell transfer assays. The lineage tracing assays required the use of multiple promoter/expression transgenic mice, and is elegantly and robustly done. The data show that late in mouse embryonic development (day 16.5) the dermis becomes fate-restricted: the Lrig1+ fibroblasts become the cells of the upper dermis and Dlk1+ cells give rise to the lower dermis. Here I’ve just picked a few of the markers they used and have grossly simplified the developmental story because I’m more interested in what they did next.
Having established that there are 2 different fibroblast lineages in the mouse dermis, the investigators asked a pretty simple question: what happens when you wound the skin in the adult mouse? Cellular wound-healing models have been widely used as in vitro models of fibroblast migration, ECM production and differentiation into myofibroblasts (so called because they express smooth muscle actin, a marker of myocytes/smooth muscle cells). Wound healing more broadly is a process of local repair and remodeling. “Deranged” or unregulated wound-healing is one model of fibrosis.
Skin or dermal fibrosis is a highly active area of study. Important fibrotic diseases of the skin include systemic sclerosis (scleroderma), amyloidosis, nephrogenic systemic fibrosis, mixed connective tissue disease, wound-healing fibrosis and others. Scleroderma is modeled in mouse by creating chemical insults on the skin, for example with bleomycin. Such models mimic the effect of chronic wounding, thus the connection to wound healing assays. It has been further demonstrated that many of the processes and pathways involved in dermal wound-healing are dysregulated in skin fibrosis, for example ECM production, TGFbeta production, and CTGF activity among many others.
While there is a good deal of consensus as to the mediators of the fibrotic process, the source of the activated cells that drive fibrosis has been controversial. This is a critical question to address, as effective anti-fibrotic therapies should target the pathogenic cell types. Without getting into immune cells, resident macrophages and Mast cells and the like, we can now at least carefully address this question regarding the source of activated fibroblasts. One model proposes that migrating fibrocytes traffic into wounded/fibrotic sites from the circulation. Another model, widely held, proposes that pericytes, which are mesenchymal cells that support blood vessels, somehow migrate away from endothelium and become myofibroblasts, thus supporting wound healing (and by extension contributing to fibrotic diseases). Finally, it has been proposed that myofibroblasts are derived from local (resident) fibroblasts that migrate into the wound and differentiate.
The Watt’s paper shows clearly that the Dlk1+ fibroblasts of the lower dermis migrate into the wound and that a proportion of these cells express smooth muscle actin and are therefore myofibroblasts. These cells produce collagen, the classic ECM protein that characterizes fibrosis, in this case simply deployed to help close the wound. After the wound is closed the epidermal layer is reformed and only then does the upper dermis reform.
These findings allow for the formulation of specific questions regarding the lower dermal fibroblasts:
– what signals activate and then shut down the lower dermal fibroblast to myofibroblast differentiation process?
– are these signals dysregulated in fibrosis?
– do these same cells deploy in response to fibrotic injury (e.g. with bleomycin)?
– do these cells interact with inflammatory cells, thought to contribute to the pathogenic fibrotic response?
– can we identify actionable targets on this cell type to prevent or decrease the fibrotic response?
Importantly this paper encourages us to ask whether similar resident fibroblast or other mesenchymal cell populations contribute to fibrotic disease in other organs such as the kidney, liver, heart and perhaps even lung. If so, the validity of the circulating fibrocyte and differentiating pericyte models might be reevaluated. Notably, linage tracking was used in support of a pericyte model of myofibroblast differentiation in a similar dermal/muscle injury model as detailed in a paper published in 2012 in Nature Medicine (abstract here). I suspect that squaring these studies will take a careful review of the markers used to identify different cell populations and a detailed look at the injury protocols. I’ll update as more info becomes available.
This is a classic example of the beautiful work that comes from an active and competitive research area and we can and look forward to further studies that will illuminate the processes of wound healing and fibrosis. Such work will certainly support and advance successful fibrosis drug development.
you may leave a comment or contact me:
on Twitter @PDRennert

Cool Science: Pathways Driving Fibrosis and Related Drug Development

December 10, 2013, by P.D. Rennert
Some recent interesting papers reveal new mechanisms for the regulation of fibrotic pathways, and suggest drug discovery opportunities.
Fibrosis is a huge medical problem and encompasses many different diseases across most organs and tissues: lung, kidney, liver, gut, vasculature, skin, ocular, cardiac, etc. Lets just assume upfront that we recognize that TGFbeta (TGFb) signaling is a critical pathologic pathway in the induction and persistence of the fibrotic state. I want to take that as a given. TGFb is also showing up more and more often in the context of tumor genetics, as befits a critical growth factor. From the drug development perspective, TGFb antagonism not yet been successful due to the toxicities encountered: TGFb is very pleiotropic and controls many normal functions as well. As a result several alternative means to target the TGFb pathway are under development for the treatment of fibrosis.
A very clever strategy for targeting TGFb was developed by researchers at Biogen, along with a top notch group of outside collaborators (see for example Horan et al. 2008). It was determined that activation of TGFb at sites of inflammation and injury, which are sites for the initiation or promotion of fibrosis, depended on an integrin, namely alphavbeta6 (avb6). Biogen developed a monoclonal antibody (mAb) to avb6 and subsequently outlicensed this mAb to Stromedix. I helped bring that mAb back into Biogen 5 years later (early 2012). By then the mAb, now called STX-100, was Phase 2 ready, a testament to the strength of the Stromedix team. Stromedix was acquired for 75MM upfront, plus milestones: a good deal for both companies. 
On the science side, how avb6 activated TGFb was not well understood, but this was thought to involve a conformational change in the beta6 strand of the integrin.
So lets jump ahead, or sideways anyway.
A parallel basic research and drug discovery effort had uncovered an interesting role for the bioactive lipid lysophosphatidic acid (LPA), and for the LPA receptors, in fibrosis. LPA receptor gene-deficient mice (LPAR knock-out or KO) were resistant to the development of fibrosis in a variety of models, and across diseases in different organ systems. This was interesting since it suggested that somehow LPA and the LPARs were regulating the TGFb pathway (see here, and here). How this occurred was not known, and the available data were correlative: when LPARs were blocked, disease was lessened and there was less TGFb, but was this specific to the LPA pathway or was it due to reduced disease activity? It turned out that the LPA/LPAR pathway interacted with integrin biology, in a couple of interesting ways. The first concerned signaling through one of the LPARs, LPAR2. LPARs are G-protein-coupled receptors, that is, they signal using G proteins. We won’t go into these, but its enough to say that signaling from LPAR2 through a specific G-protein induced a conformational change in the beta chain of avb6, and that this conformational change was associated with the ability of avb6 to activate TGFb. The second way has to do with the manner in which LPA is produced by the enzyme autotaxin, and we’ll come back to that later. This is not the only manner in which LPA impacts fibrosis, as was demonstrated by the Amira drugs that are selective antagonists for LPAR1 and perhaps also LPAR3 (see for example, here). This drug also blocked the development of fibrosis in multiple models, and showed activity in early clinical development for idiopathic pulmonary fibrosis (IPF). The data were suggestive enough that Bristol-Myers Squibb bought the company for $325MM upfront in 2011.
Now lets jump ahead.
In 2011 Tim Springer’s lab published a remarkable paper (Shi et al). What this paper explored, using structural biochemical and biophysical techniques, was how TGFb was kept in an inactive state by its own protein structure (called the “straightjacket”). Breaking this inactive state required distinct and critical interactions, one in which TGFb was embedded into the cell membrane in complex with latent TGFb binding proteins (LTBP), which in turn bound to the extracellular matrix, and the other by which TGFb interacted with the avb6 integrin. Productive binding of the integrin required interaction with cell’s the actin filaments, the protein fibers that hold the cell together and also rearrange themselves in order to allow the cell to move. We should note here that integrin:actin interaction is most often mediated by talins, the proteins that sort of cross link the actin filaments (think of rungs on a ladder as a simple picture). This complex interaction of TGFb, avb6, the cell membrane and the actin fibers imposed enough force on TGFb to break its lock, and free it for activation and release from the cell surface. Its as if you had to rearrange the walls of your house in order to open a window. This is a highly regulated system. So, how did avb6 accomplish this trick? Spinger’s lab didn’t give an answer – they were intently focused on the TGFb structure. But a recent paper on a different integrin provides a clue, and maybe brings us back to the LPAR pathway.
A few weeks ago Shen et al (in Nature) elucidated differential signaling by the integrin alphaIIb-beta3 (aIIb-b3), focusing on the interaction of the b3 subunit with talin and with a G-protein, Galpha13 (Ga13). They found a specific amino acid motif – EXE – mediated binding to the G-protein. This motif is found in many beta subunits, including b6. The protein domain in which this motif is found is within a region known to mediate binding to talin, and Shen et al. go on to show that the two binding phenomena are mutually exclusive. Importantly, interaction with the G-protein required binding to fibrinogen, a ligand for the aIIb-b3 integrin. Integrins transmit signals bidirectionally. Inside-out signalling, the process by which an integrin undergoes conformational change in response to ligand binding, is a talin-dependent mechanism. aIIb-b3 integrin ligation mediates platelet aggregation and triggers outside-in signalling, which requires Ga13 and greatly expands thrombus formation. Switching back to interaction with the G-protein corresponds to clot retraction. So this model proposes waves of signaling controlled in part by which component of the cellular machinery – talin or G-protein – is engaged.
It has been demonstrated in lung and kidney fibrosis models that avb6-mediated activation of TGFb can be triggered by the LPAR2 signaling through Galphaq (Gaq) to the Rho/Rac kinase pathway (see here, and here). The connecting dots from Rho/Rac kinase to avb6 activation are not yet described, by may work by regulating the interaction of avb6 with a G-protein (eg. Ga13) v talin or actin. In KO mouse experiments, Ga12/13 were shown to be required for basal TGFb activation in embryonic fibroblasts (MEFs), but not in LPAR2 transfected MEFs. The relevance of this data to the fibrotic condition however is unclear, and I think it remains worth exploring this link further.
There is an independent link between the LPA/LPAR2/avb6/TGFb pathway and integrin biology. The production of LPA is controlled by the enzyme autotaxin. In settings of tissue injury and inflammation, autotaxin and the LPARs are upregulated by various cells. In this way the LPA production machinery (autotaxin) and the responding receptors are operating in tandem. More remarkably, autotaxin is carried into the inflamed or injured tissue by activated cells – platelets, T cells, and likely other cells types. The manner in which autotaxin is transported is by binding to integrins with b1 or b3 subunits, carried on the cell surface and maneuvered into place by the coordinated action of the activated cell (see a review here). This has likely evolved to maximize the production of LPA at sites where it is required – the half life of LPA in circulation is very short, minutes only. Indeed the half-life of TGFb is even shorter, less than one minute.

So, this is a highly coordinated and highly regulated system, as might be expected when dealing with a potent growth factor like TGFb. The implications of the model for drug development are several. First it seems clear that LPA receptor antagonists that can target LPAR2 would be therapeutic and differentiated from the Amira assets now owned by BMS. The benefit of targeting avb6 was discussed earlier in the context of the Stromedix/Biogen deal. Another target that appears here is autotaxin, and several companies are pursuing autotaxin antagonists – I developed a beautiful compound series at X-Rx Discovery, as one example. Finally, there is the very exciting prospect of using the new structural information about TGFb and related proteins (its a big family) to design drugs that selectively block TGFb activation in the context of an activated cell response. This may be one of the things that Scholar Rock Therapeutics, a venture-company founded by Tim Springer and colleagues, is up to now. We’ll have to wait and see.
Bottom line: we are learning a lot about fibrosis pathophysiology, only a little of which is touched on here. With that knowledge will come a new wave of drug development, and ultimately we hope to provide relief for patients suffering from these chronic and very serious diseases.
Contact info 
Twitter:      @PDRennert

Final ASH13 SnapShot: AbbVie’s ABT-199

Update on ABT-199. 12-3-2013, by @PDRennert 

I reviewed the status of ABT-199 back in August, following ASCO (see link). At the time I felt the drug was being overlooked in the hype around ibrutinib and idelalisib. This was based on impressive response rates and the sense that AbbVie had gotten a handle on how to dose safely. After all, tumor lysis syndrome (TLS) is not uncommon in the treatment of lymphomas, and can be dealt with by dose modification or intervention.

At ASH we get a further look at this drug. Abstract #872, to be presented by J. Seymour, introduces a modified dosing regimen. Patients (rrCLL/SLL, n = 56) were given single doses of ABT-199 at day -3 or day -7, then started on daily dosing after that. This is a Phase 1 monotherapy trial designed to determine MTD. The efficacy readouts were pretty dramatic. ORR = 84% with CR = 21% and PR = 63%. Within the CR group (n = 12), 8 patients had no or very low minimal residual disease. There were 12 discontinuations due to progressive disease (21%). The response rate was not related to del(17p) status but the PR rate was lower in fludarabine-refractory patients (these patients are all at high risk).

Notably, among the usual AEs (diarrhea, neutropenia, URIs, etc) there was an 11% grade 3/4 TLS rate. This is 6 patients. Problematically, TLS was not clearly dose associated: 2 @ 50mg, 1 @ 100mg, 2 @ 200mg and 1 @ 1200mg, this last one resulting in sudden death. So dose optimization remains in progress.

The observation that response was not related to del(17q) status or other aberrations in the TP53 locus in rrCLL patients will be discussed in more detail by M.A. Anderson (Abstract #1304).

M. Davids (Abstract #1789) will present a similar study in a variety of rrNHL patients, including MZL, MCL, DLBCL, FL and WG. Of the 32 patients enrolled and followed for a median time of 6 months, 18 discontinued due to progressive disease (that’s 56%). AEs were typical (nausea, diarrhea, cytopenias, URIs). There were 2 mild episodes of TLS. For FL and DLBCL, doses of 600mg or higher were required for efficacy. The responses were good given this difficult mix of patients. Note that the text and table in this abstract don’t entirely line up, so best to hear the current results at the meeting. The implication here however is that ABT-199 is safer in these patient populations than in rrCLL, a theme we also heard at ASCO.

Aside from the clinical data there is an awful lot of preclinical modeling using cell lines or patient derived cells in vitro or in vivo (mouse). The point of all this nice work is to show the potential of combination therapy using ABT-199 along with other drugs. A few examples:

–  2-DG + ABT-199 kills all Myeloma subtypes (#1921)

–  ibrutinib + ABT-199 is effective against MCL cells and CLL cells (#645 and # 3080)

–  imatinib + ABT-199 kills chronic phase CML cells

–  BTK inhibitor R406 + ABT-199 kills DLBCL cells

   etc, etc

The CML observation points to another trend, which is efficacy of ABT-199 in settings beyond NHL, including ALL (#3919), AML (#885) and MM (#4453). There are others…

On balance this remains an exciting and potentially important drug. The issue of TLS in certain subtypes of NHL remains to be solved, while in other, difficult NHLs there appears to be clear and compelling efficacy with less toxicity.


ASH13 previews

Part 8.   ABT-199
Part 7.   CAR-T tech                        

Part 6b. new targets for Myelofibrosis           

Part 6a. Jak inhibitors in Myelofibrosis                       

Part 5.   Biologics for Non-Hodgkin Lymphomas              

Part 4.   New & noteworthy: CLL etc             

Part 3.   Btk and PI3K inhibitors for CLL      

Part 2.   Ibrutinib                              

Part 1.   Idelalisib

pre-ASH post on ADC technology:  here                         

ASH13 just around the corner – quick update of CAR-T technology

December 2, 2013.
by Paul D Rennert 

Part 7. Chimeric Antigen Receptor T cell technology (CAR-T) in the treatment of hematopoietic malignancies.

The American Society of Hematology Meeting will take place in New Orleans, December 7 – 10, 2013. The abstracts are available at
Having detoured briefly into myelofibrosis (see parts 6a and b), there are just a few more subjects to try to cover this week. With luck and time, I’ll get through this bit today and then maybe on to lymphoma genetics, we’ll see.
This is from the introduction to Carl June’s seminal 2011 NEJM case report:
“We designed a lentiviral vector expressing a chimeric antigen receptor with specificity for the B-cell antigen CD19, coupled with CD137 … (4-1BB) and CD3-zeta … signaling domains. A low dose (approximately 1.5×10^5 cells per kilogram of body weight) of autologous chimeric antigen receptor–modified T cells reinfused into a patient with refractory … CLL expanded to a level that was more than 1000 times as high as the initial engraftment level in vivo, with delayed development of the tumor lysis syndrome and with complete remission. Apart from the tumor lysis syndrome, the only other grade 3/4 toxic effect related to chimeric antigen receptor T cells was lymphopenia.” (Porter et al. 2011. NEJM 365: 725-733). The therapy induced long term remission is a patient who had failed 4 rounds of rituximab+chemo, and then had failed alemtuzumab, anti-CD52, therapy. Pretty amazing.
The anti-CD19 CAR is essentially an antibody fragment containing a single chain Fv (antigen binding domain). The CD3-zeta chain induces T cell activation and the addition of the 4-1BB cytoplasmic domain ensures prolonged and robust response – 4-1BB is in immune checkpoint activator, and is gaining some favor in its own right in immunotherapy, through the development of agonist anti-4-1BB antibodies. The CAR-T components are introduced to the patient’s own T cells ex vivo via lentivirus transduction, then given back to the patient in hopes of inducing a T cell mediated immune response to the cancer (e.g. a CD19+ CLL). The original case reports were followed for up to 3 years, as reported in Abstract #4162. Of 14 patients treated in the pilot studies, the ORR = 57% (21% CR and 30% PR). 43% of patients did not respond. Of the PR cohort, 40% progressed within 4 months. So that’s about 1/3 of patients with a durable response.
Additional clinical trials have been funded via collaboration with Novartis, who has bought the technology and patents. A few of these are updated at ASH. The CLL and ALL data for patients treated with the anti-CD19 CAR T cells (CTL019) are summarized in Abstract #163. 24 rrCLL patients have been treated using 2 different protocols that vary by the number of CTL019 cells given back to the patient. The response rates were CR = 21%, PR = 29% (so ORR = 50%) and non-responders = 50%. In pediatric ALL (n=14) the CR = 57%; the rest of the patients (43%) either did not respond or progressed. In adult ALL, all 3 patients had a CR (=100%). CRs were always accompanied by in vivo expansion and persistence of CTL019 cells. Tumor cells were eliminated from circulation and also, importantly, from bone marrow. Molecular analyses showed that tumor cells were essentially eliminated in patients with CR – this is defined as minimal residual disease (i.e. not detectable). Additional data specific to these studies are reported in Abstract #873 (CLL) and Abstract #67 (ALL) – the latter study reports persistence up to 15 months. Another group at U Penn reported similarly high RRs in ALL. Lee et al (Abstract #68) report the use of an CD19-CD28-CD3zeta CAR construct to engineer t cells for use in ALL, with an initial CR (n=7) of 71.5%, with other 2 PR responders and 2 non-responders. These are impressive early data from multiple studies.
Steve Rosenberg’s group is also reporting use of anti-CD19 CAR-T cells, these made using a gamma-retrovirus construct to genetically modify the T cells. The technology differs also by use of the CD28 signaling domain instead of 4-1BB, along with CD3-zeta. Of 14 patients with rrCLL, rrDLBCL or primary mediastinal BCL, 36% achieved a CR and 43% a PR, the rest being non-responders or SD. All responders (PR + CR = 79%) were ongoing at the time of abstract submission. The study will be further updated at the meeting. The trial was funded under a CRADA-based collaboration between the NCI and Kite Pharma, a private biotech company.
Given the compelling response rates observed, it is unclear whether the ex vivo selection and expansion methods employed by the MD Anderson group will add benefit. Laurence Cooper and colleagues will present a CD19 CAR technique that utilizes artificial antigen-presenting cells to select the T cell population that is then given following hematopoietic stem cell transplantation in ALL and NHL patients (Abstracts #166 and #4208). Their very early results will be updated at the meeting. Additional efforts targeting CD19 include the trial in rrCLL patients who have received only 1 prior chemotherapy regimen; the idea is that these patients are earlier in the disease course and may have better response rates. It is not possible to tell from the Abstract (#874) if this effort is succeeding, but an update is promised at the meeting.
Turning from the CD19-directed technologies, Carl June’s group is presenting the first clinical data on the use of engineered T cells in multiple myeloma. The T cells are expanded using CD3/CD28 beads (a technology I worked on 20 years ago at Repligen, in the context of HIV therapy) and are engineered to express a modified TCR that recognizes the MM antigens NY-ESO-1/LAGE1. The recognition of this peptide complex is HLA-class restricted, so the patients are screened in advance for responsive HLA haplotypes. The T cells are infused followed depletion and stem cell transplantation, so CAR-T is used here in the context of adjunct therapy. Best response ORR = 100%, although some patients have since progressed. An update will be given at the meeting. Also of note is a trial in which this novel CAR-T therapy is used in a non-transplant setting (no data available yet). An interesting twist is the use of kappa-light chain of surface immunoglobulin expressed on malignant B cells (as opposed to lambda light chain expressed by most normal B cells? – I guess that’s right). A group from Baylor, funded by Celgene, will present Phase 1 data (Abstract #506). Another CAR antigen technology in preclinical development at U Penn targets CD123 for AML (Abstract #143). Preclinical data from the OSU group show that a different MM antigen can be used in the CAR-T setting. Abstract #14 shows that a CS-1 directed CAR works in a mouse xenograft model. I like the straightforward description of the technology: “We successfully generated a specific CS1-CAR construct with a lentiviral vector backbone, sequentially containing a signal peptide (SP), a heavy chain variable region (VH), a linker, a light chain variable region (VL), a hinge, CD28 and CD3epsilon.” Simple, right? Finally, Haso et al from the NIH compare CD22-targeting CAR constructs using different signaling chains (4-1BB v CD28) in preclinical mouse models of ALL, and report superior results using the 4-1BB construct (Abstract #1431). This is nice as they used a humanized mouse model, the NOD/SCID/Common gamma chain KO mouse (NSG), engrafted with a human ALL line. Love the humanized mouse technology, right up my alley.

A persistent theme in the evolving treatment of leukemias and lymphomas is the use of combination therapies. We see a similar trend developing with CAR-T technologies. Paolo Ghia et al combine a CAR directed to CD23 along with low dose lenalidomide treatment using the Rag2/Common gamma chain KO humanized mouse model and cell from CLL patients – nice work (Abstract #4171). A second study evaluated the use of mTOR modulation in the context of CAR-T therapy (Abstract #4488). As this technology continues to advance we can expect to see additional uses of targeted or other therapies in combination.