Tuesday, May 17, 2016

Treating achondroplasia: NVP-BGJ398, a tyrosine kinase inhibitor, restores bone growth in a model of achondroplasia


To remind 

This review and the others in the blog try to explain the science behind the research for therapies for achondroplasia with texts adapted and simplified to help understanding, while avoiding complex technical details about discussed topics. For those who have more knowledge in the field I would recommend them to consult sources provided throughout the texts and the References’ sections of the articles to get more technical info, if needed.

For the newcomers, although I tried to simplify the text, you may find the language here a bit difficult to understand in a first read. Do not give up. We have older articles in the blog explaining achondroplasia, the growth plate, chondrocytes, fibroblast growth factor receptor 3 (FGFR3), enzymes, and other related topics using analogies, animations and pictures. You just have to browse the index page in the language of your preference to find them. Keep calm and go through them, it's good to learn about the field we are crossing, so we won't get lost, especially because it is (partially) uncharted territory. 

A bit of the basics 

I know, I know, older (?) readers of this blog won’t like it but I think it is important to highlight a simple concept about achondroplasia before continuing. Achondroplasia is caused by a change in the structure (a mutation) in an enzyme called FGFR3. FGFR3 is a natural brake of bone growth, counteracting and balancing the natural bone growth acceleration caused by several other agents working inside the growth plates. As in achondroplasia it is working excessively, the result is that bones can’t grow on their programmed pace, resulting in the clinical features of this skeletal dysplasia.
 
Now, let’s get into the subject. 

First question: what is it, a tyrosine kinase inhibitor (TKI)? 

A TKI is a small molecule designed to block the activity of a group of enzymes called tyrosine kinases (TKs), which work through chemical reactions promoted by the aminoacid tyrosine. Our familiar FGFR3 is a tyrosine kinase.

Briefly, many of these TKs are fixed across the cell membrane (Figure 1) and work transmitting, through chemical reactions in their structures, signals that come from outside the cell towards the cell nucleus. They work like TV antennas, so these TKs are also called receptor enzymes (or receptor tyrosine kinases, RTKs).

Figure 1. RTK families.

Schematic structure of the main human RTK families. Abbreviations are: TK, tyrosine kinase domain; CRD, cystein-rich domain; LD, leucine domain; FNIII, fibronectin type III-like domain; AB, acidic box; CadhD; cadherin-like domain; LRD, leucine-rich domain; IgD, immunoglubulin-like domain.Receptor protein kinases; Francesca De Bacco, Michela Fassetta and Andrea Rasola. http://www.cancer-therapy.org/

FGFR3 is part of a family of four sister RTKs (1 to 4). All FGFRs share a similar structure (Figure 2): there is a portion placed outside the cell (extracellular domain), a portion that crosses the cell membrane (transmembrane domain) and an intracellular portion (or tyrosine kinase domain). This intracellular portion is the one responsible for transmitting the signal coming from outside the cell to several enzyme networks located in the cytoplasm, allowing the cell to respond appropriately to that signal. Signals that are “managed” by FGFRs come from FGFs (1).

Figure 2. FGFR3 structure.


As just said, the intracellular domain of FGFR3 has some spots rich in tyrosine. Importantly, tyrosines are hidden in a kind of “groove” in the structure of the enzyme,  which is called “ATP pocket” (ATP: adenosine triphosphate).

Well, why are the tyrosines so important?

When a FGF connects with FGFR3 outside the cell, it promotes a shift in the shape of FGFR3 inside the cell, opening the pocket, “exposing” those tyrosines to the local environment (Animation 1). The thing is that tyrosines are very reactive aminoacids and the local ambient is rich in ATPs, molecules that carry the ion phosphorus, which is also very reactive. One attracts the other and the resultant reaction between the tyrosines and ATPs (called phosphorilation) further attracts new enzymes to the vicinity. The chemical reaction “travels” from one enzyme to another, in what is called a chemical “pathway” or “cascade”. Try to think that these are like domino chains, you (the signal) push one domino block and the others will follow in sequence, until the last ones reach some spots inside the nucleus to promote a coordinated cell nucleus response to that original signal. Watch Animation 2 to follow a signal starting outside the cell towards the nucleus (it's a bit long, but instructive). 

Animation 1. RTK activation and ATP binding (source unknown, from youTube).



Animation 2. Signaling cascade.  
http://content.dnalc.org/content/c16/16877/cell_signals.mp4
DNA Learning Center by Cold Spring Harbor Laboratory

 Blocking receptor enzymes 

Now, imagine if we could just place a tape over that “ATP pocket” to block the beginning of the chain reaction promoted by those tyrosines. If you can’t “open” the pocket, tyrosines will not be exposed and the chain reaction will stop. This is exactly what TKIs do (Animation 3). They are small molecules that can traffic easily across tissues and cell membranes and bind to those ATP pockets inside the cell, blocking the target enzyme functions.

Animation 3. Mechanism of action of lapatinib, a TKI designed to inhibit EGFR2/HER2 (a RTK).


Good, isn’t it? What are we waiting for to just put one of these TKIs to block FGFR3 and solve the bone growth arrest in achondroplasia?

The challenge is that FGFRs are not the only enzymes bearing ATP pockets. On the contrary, there are several other enzyme families containing them, too, and the pockets are really similar in all of them (Figure 1). Just to give an example, in the case of the FGFR family, the four enzymes share more than 60% of their structure, including the shape of the ATP pockets (1). This resemblance or structure similarity is called homology.

Since their inception, many TKIs have already been developed. The first generations of TKIs had a broad range of effects, binding to several different families of enzymes at the same time (Figure 3), so they are usually called multikinase drugs. The newer generations of TKIs are more specific, and developers try to find the molecule capable to block just one enzyme or at least only enzymes of a particular family. This is a tough task exactly because of the homology of the structure of these many receptor enzymes. Nevertheless, we have seen a number of new anti-FGFR TKIs being described in the last years, with less effect on other enzymes (reviewed in the blog).

Figure 3. Older
TKI and their multiple targets.
Each colored ball represents an enzyme that can be inhibited by these TKIs.

Are TKIs good for achondroplasia? 

This was the question in a recent article published in this blog. In that review we commented about the study published by Dr. Pavel Krejci's group about the use of available TKIs to treat the bone growth defect in achondroplasia (2). Dr. Krejci is one of the most enthusiastic researchers in the FGFR3 field, and have authored or co-authored several relevant studies in the last ten years or more. For instance, he was one of the researchers who worked in the study which validated C-type natriuretic peptide (CNP) as a key agent promoting bone growth (3), resulting in helping to strengthen further work on vosoritide (BMN-111), which is now in phase 2 clinical trial.

Back to the study by Gudernova et al.(2), Dr. Krejci's group frequently use a standard cell model of rat chondrosarcoma (RCS, a cancer cell of cartilaginous origin) that, under FGFR3 activation, behaves similarly to growth plate chondrocytes. Watch Video 1 to have an idea of how RCS cells behave when exposed to FGF2. This video gives a snapshot of FGFR3 signaling effects on chondrocytes.
 
Video 1. RCS cells respond to FGF2 stimulus. This cell model reacts to FGFR3 similarly to growth plate chondrocytes. From:ReACH Achondroplasia Registry.


So, the researchrs tested five available anti-FGFR TKIs in their RCS cell model, in an ex-vivo mouse model and also in newborn mice and concluded that the available TKIs would not be useful to treat achondroplasia because they blocked other enzymes and sometimes also caused cell toxicity. 

When writing the review about this study my feeling was of frustration and the reason is simple. The best approach for the treatment of achondroplasia is to beat the mutated receptor directly. The mutated receptor is the problem, so in theory, if you could put it down, the growth plate chondrocytes could resume their normal behavior and we could have bone growth rescue, minimizing or avoiding the clinical features and complications associated with this bone dysplasia. 

Just to give you an example, let's compare a direct therapy against FGFR3 with the current most advanced potential treatment for achondroplasia, the CNP analogue vosoritide. Vosoritide works activating another enzyme cascade in the chondrocytes that naturally counteracts some of the effects of FGFR3. According to available results from the ongoing phase 2 study, it has been showing efficacy in restoring the bone growth velocity of the patients included in the study (4). However, CNP compensates only for one of the chemical cascades triggered by FGFR3 (Figure 4). Conversely, a drug directly targeting the mutated receptor could potentially inhibit the excessive function of all signaling cascades working upon FGFR3 activation.

FIGURE 4. FGFR3 X CNP crosstalk.


FGFR3 signal transduction and therapeutic strategies. FGF and heparin binding to the extracellular domain of FGFR3 induce kinase activation leading to activation of downstream signaling pathways such as STAT and MAP kinase cascades. CNP binding to NPR-B induces the generation of the second messenger cGMP, which activates PKG leading to attenuation of the MEK pathway via RAF. Peptide P3 blocks ligand binding whereas meclozine is proposed to antagonize MEK activation of ERK. Modified with permission from Laederich and Horton (5). Source: Fundación Alpe.

Validating and comparing results 
 
In the last decade, several studies related to the pharmacological action of anti-FGFR TKIs had already showed that they can block several enzymes apart of FGFRs at the same time (as explained above). A small number of studies also showed conflicting effects of TKIs on bone growth because of "off-target" effects (other enzymes potentially being affected by the TKI) (6-9). So, the study by Dr. Krejci's group (2) seemed to just confirm what would be supposed to happen inside the growth plate when using anti-FGFR TKIs.

Therefore, I was (positively) surprised and relieved, too, when I read the recently published study by the group of Dr Laurence Legeai-Mallet (10). Her group tested the TKI NVP-BGJ398 (BGJ398; Figure 5) (11), one of the molecules tested by Dr. Krejci as a potential therapeutic approach for achondroplasia.

Figure 5. BGJ398 structure.


From Wikimedia


Fresh air
 
So, now that we have a basic knowledge of what a TKI is and the challenges about using them as a therapy for achondroplasia, let’s review this very interesting study with BGJ398 in achondroplasia and try to put the related conclusions in perspective. Do we have another therapeutic solution for achondroplasia in the horizon?

Briefly, in the study by Dr. Krejci group, as we already saw, they worked with in vitro and in vivo models and found that all TKIs blocked not only
FGFR3 in the cell models but also other enzymes as well, which caused toxicity. They further tested one of the TKIs, AZD4547 (12), in newborn mice, and found that instead of promoting growth, it leaded not only to growth disturbance but also to significant dose-dependent toxicity and even death of the animals. Considering all results together, their conclusion was that available TKIs could not be used to treat achondroplasia due to their lack of specificity and risk of toxicity.

However, their study brought more insight about the properties of those TKIs, regarding their ability to inhibit many RTKs. One interesting information was that BGJ398, although capable of inhibiting several enzymes, did so in doses far higher than those needed to particularly inhibit FGFR3. In other words, BGJ398 seemed to have more affinity to FGFR3 than to all the other tested enzymes, including the other FGFR members. To respect the publication copyright, I can’t reproduce the table showing this information, but you can consult their article and check Table 1.

Having these above conclusions in mind, let's now review the study by the grou
p of Dr Legeai-Mallet, which brings fresh air to the field and shows how important it is to check and verify results obtained in previous research.
 
Rescuing bone growth
 
In summary, Dr. Legeai-Mallet’s group tested BGJ398 in several cell and tissue experiments and in an animal model of achondroplasia and found that it was able to rescue bone growth and basically all the clinical characteristics of achondroplasia. For instance, the use of this TKI corrected not only the long bone growth impairment caused by the mutated FGFR3 but also the defect in skull base, vertebrae, calvaria (read synostoses) and in the intervertebral discs as well. 

Let’s look this study in more detail.
 
First, the researchers worked in human chondrocytes expressing several kinds of mutated FGFR3 and found that BGJ398 was able to inhibit the FGFR3 phosphorylation (remember what we reviewed above?).

Next, they tested whether several different doses of BGJ398 would be able to rescue bone growth in an embryonic femur culture and found that, at a concentration of 100nM (nano Molar), this TKI was able to completely rescue bone growth. They found quite similar results with the same dose in a calvaria model.
 
Time for a short pit stop here to see what does it mean nano Molar. Molar is an unit used to measure the concentration of a substance in a given solution (or medium). For pharmacological purposes, the lesser the dose concentration needed to obtain a desired effect, the better is for a potential therapy in any field. Thus, it is common to accept that a potential drug candidate should be “working” at concentrations within the nano Molar range.
 
Back to the study, the researchers tested BGJ398 in a mouse model of achondroplasia, the same one used to explore the use of vosoritide years ago (13). They tested newborn mice with a daily subcutaneous dose of 2mg/kg for 15 days and found that this dose was well tolerated, “without noticeable modification of behavior” (10).
 
According with the study, mice bearing the achondroplasia-like mutation treated with BGJ398 had a significant rescue of their bone growth compared to control animals. However, it is important to note that, comparing with non-affected animals, growth was not completely rescued as we can see in the pictures and graphics shown in the study (Figure 6; save this information for later).

Figure 6. NVP-BGJ398 improves growth of the appendicular skeleton in Fgfr3Y367C/+ mice.

(A) Radiograph of Fgfr3+/+ and treated and untreated Fgfr3Y367C/+ forelimbs. Scale bar: 1 cm. (B) Radiograph of Fgfr3+/+ and treated and untreated Fgfr3Y367C/+ hind limbs. (C) Lengths of femur, tibia, humerus, ulna, and radius (Fgfr3+/+, n = 13–14; untreated Fgfr3Y367C/+, n=9–10; treated Fgfr3Y367C/+, n = 12–13). *P < 0.05, 1-way ANOVA. (D) 2D μCT of the distal femur metaphysis cross-sectional area. Scale bar: 400 μm. (E) 3D μCT of the distal femur metaphysis cross-sectional area. Scale bar: 400 μm. (F) Total bone mineral area in the distal femur metaphysis (Fgfr3+/+, n=8; untreated Fgfr3Y367C/+, n=8; treated Fgfr3Y367C/+, n=6). *P<0.05, 1-way ANOVA. All data are from animals treated with protocol 1 (16 days old). Data are expressed as mean ± SD. From: Komla-Ebri D et al. J Clin Invest 2016;126(5):1871-84. This article has free access. Figure reproduced here for educational purposes only.

They also tested the effects of BGJ398 in spine, intervertebral discs and skull, including syncondroses and the foramen magnum, and verified that the drug was also capable to ameliorate growth in all these areas. This is relevant because we need to know the impact of any therapy on the clinical features of achondroplasia.
 
An important aspect of using an anti-FGFR TKI to block FGFR3 is the possibility of it doing the same on other FGFRs. FGFR1 is also expressed in the growth plate, so the researchers checked if BGJ398 was also causing any disruption of the FGFR1 signaling and found this was not the case. This means that with the dose used in the study BGJ398 was significantly inhibiting only FGFR3.
 
They further verified how BGJ398 was working in the growth plate chondrocytes. They wanted to check if the “domino” cascades triggered by FGFR3 were really affected by this TKI. They found that the main pathways activated by FGFR3, MAPK and STAT1 (Figure 6), were inhibited by BGJ398. They could also verify that, using this TKI, the level of signaling of these pathways in the growth plates of mice bearing the achondroplasia mutation were comparable to that in non-affected animals.
 
Figure 7. FGFR3 main signaling pathways.

Signaling pathways activated by FGF/FGFR. FGFs induce dimerization, kinase activation and transphosphorylation of tyrosine residues of FGFRs, leading to activation of downstream signaling pathways. Multiple pathways are stimulated by FGF/FGFR signaling such as Ras-MAP kinase, PI-3 kinase/AKT and PLC-γ pathways. Furthermore, FGF signaling can also stimulate STAT1/p21 pathway. FGF/FGFR signaling also phosphorylates the Shc and Src protein. FGF/FGFR play crucial roles in the regulation of proliferation, differentiation and apoptosis of chondrocytes via downstream signaling pathways. From: Su N et al. Bone Research 2014; 2:14003 (15). Article in OPEN ACCESS, figure reproduced here for educational purposes only.

It’s much about timing
 
Dr. Legeai-Mallet’s group, as we saw, tested the effects of BGJ398 for 15 days in newborn mice and the results were very relevant. However, they didn’t stop there and also tested this TKI in older mutant mice. Although the drug had positive effects in bone growth, those were far milder than what was seen in the younger animals. What was also relevant is that, as we saw above (and you can check reading their study), that even given soon after birth, the bone growth rescue with BGJ398 was not complete (Figure 6), and we are talking about a drug designed to block directly the receptor and not acting indirectly, through another pathway, as does CNP.
 
Bones start to grow early in utero and the effect of the FGFR3 mutation that leads to achondroplasia is already detectable in the third trimester of the pregnancy because the bone growth plates, instead of following the normal growth program, are set to “brake” mode. 

Most of the long term consequences of achondroplasia are already present at birth.

Why is this important? Because it is likely that if we want to reduce the effects of the FGFR3 mutation and help kids grow better, the best moment to start any therapy for achondroplasia would be at birth and, if we were capable to do that, even earlier, during the pregnancy. We already have an example that treating achondroplasia in utero might be possible. Do you remember that the Japanese group working with meclizine (14) has also tested this drug in pregnant mice and didn’t report any toxicity? So, this strategy is not impossible (but probably very challenging).
 
Time to start therapy is therefore key for achondroplasia.
 
Direct vs. indirect therapy
 
And last, but not least, as we said, the researchers used the same mouse model used to test vosoritide (13). When comparing the outcomes of both therapies they concluded that BGJ398 was superior to the CNP analogue. In other words, targeting FGFR3 directly should be better than doing that indirectly.
 
So, can we use TKIs to treat achondroplasia?
 
Based on the historical evidence we would say it would be risky to use TKIs to treat achondroplasia because of their lack of specificity. However, the study by Dr. Legeai-Mallet's group briefly summarized here shows us that as newer TKIs are being developed and carefully tested, it is possible that we could see more good news in this field in the near future. An important detail about BGJ398 is that most of the pre-clinical work with this drug has already been done, because it is already in clinical trials for cancer (ClinicalTrials.gov). 

Another very important information to have in mind about this study is that the researchers obtained remarkable results on bone growth in their achondroplasia model using doses of BGJ398 10 to 100 times lower than those used for the tests in cancer models.

Indeed, it would be interesting to understand how Novartis, the BGJ398 developer, is looking at the results obtained by Dr. Legeai-Mallet’s group and whether they are willing to perform tests in other pre-clinical models, examining specifically the effects of this TKI in younger animals, aiming to take it to clinical development for achondroplasia. It is likely that, with the wealth of information already existent, few additional tests would be needed to validate or not BGJ398 as a potential asset to enter clinical development for achondroplasia. Some questions are needed, though. For instance, are there any signs of toxicity not shown in the tests already performed? 

Go/no Go
 
In the era of targeted therapies, both Regulators and Industry as a whole must change mindsets to allow new therapies to reach untouched diseases. 

The decision to pursue clinical development of a potential drug - go/no go - is hard and based in many distinct parameters. An interesting review published a couple of years ago listed five items (called five Rs) that would need evaluation for a given asset to be successful and reach the market (16). Do we (the developer) have the:

1. Right target (is the target the right one to beat in that given disease?)
2. Right patient (is the drug being tested in the right patient?)
3. Right tissue (is the drug being tested in the right cells and tissues?)
4. Right safety (does the drug have a robust safety profile?)
5. Right commercial potential (will it allow a good ROI?)
 

However, decision making is not restricted to these five items, but also relates to how much the developer is willing to take the “risk” of failure of that new asset that could cost hundreds of millions of dollars. In other words, it may be difficult to decide whether that new potential therapy is a business risk or a business opportunity. Many major pharmaceutical industries are averse to take risks in areas where there expertise is rare or the territory is not well charted. In dubio, prohibere.

We have already reviewed challenges facing drug development in previous articles in the blog. There are potential therapies for achondroplasia in the horizon just waiting for the right mindset to be further developed and possibly reach the children in need. Who will take the lead?

Note 

In the next article we will talk a bit more about drug development and timing for therapy in achondroplasia. The promised article about the use of gene editing for achondroplasia will come after that.

References 

1. Ornitz DM and Marie PJ. Fibroblast growth factor signaling in skeletal development and disease. Genes Develop 2015;29:1463–86. Free access. 
 
2. Gudernova I et al. Multikinase activity of fibroblast growth factor receptor (FGFR) inhibitors SU5402, PD173074, AZD1480, AZD4547 and BGJ398 compromises the use of small chemicals targeting FGFR catalytic activity for therapy of short-stature syndromes. Hum Mol Genet 2016; 25(1):9-23.
 
3. Pavel Krejci at al. Interaction of fibroblast growth factor and C-natriuretic peptide signaling in regulation of chondrocyte proliferation and extracellular matrix homeostasis. J Cell Sci 2005;118 (21):5089-100. Free access.


4. Irving M et al. Vosoritide (BMN 111) in children with achondroplasia: Results from a Phase 2, open label,sequential cohort, dose escalation study. Abstract presented at the American Society of Bone and Mineral Research 2015 Meeting. Presentation Number: LB-1154. October 12, 2015. 

5. Laederich MB and Horton WA. FGFR3 targeting strategies for achondroplasia. Exp Rev Mol Med 2012;14:e11. 

6. Brown AP et al. Cartilage dysplasia and tissue mineralization in the rat following
administration of a FGF receptor tyrosine kinase inhibitor. Toxicol Pathol 2005;33: 449–55. Free access.


7. Rastogi MV et alImatinib mesylate causes growth deceleration in pediatric patients with chronic myelogenous leukemia. Pediatr Blood Cancer 2012;59:840–45.

8. Tauer JT et al. Impact of long-term exposure to the tyrosine kinase inhibitor imatinib on the skeleton of growing rats. PLoS One 2015;24;10(6):e0131192. Free access. 

9. Hall AP et al. Femoral Head Growth Plate Dysplasia and Fracture in Juvenile Rabbits Induced by Off-target Antiangiogenic Treatment. Toxicol Pathol 2016.pii: 0192623316646483.

10. Komla-Ebri Det al. Tyrosine kinase inhibitor NVP-BGJ398 functionally improves FGFR3-related dwarfism in mouse model. J Clin Invest 2016;126(5):1871-84. Free access.

11. Guagnano V et al. Discovery of 3-(2,6-dichloro-3,5-dimethoxy-phenyl)-1-{6-[4-(4-ethyl-piperazin-1-yl)-phenylamino]-pyrimidin-4-yl}-1-methyl-urea (NVP-BGJ398), a potent and selective inhibitor of the fibroblast growth factor receptor family of receptor tyrosine kinase. J Med Chem 2011;54(20):7066-83. 

12. Gavine PR et al. AZD4547: an orally bioavailable, potent, and selective inhibitor of the fibroblast growth factor receptor tyrosine kinase family. Cancer Res 2012;72(8):2045-56.
Free access.

13. Lorget F et al. Evaluation of the therapeutic potential of a CNP analog in a Fgfr3 mouse model recapitulating achondroplasia. Am J Hum Genet 2012;91(6):1108-14. Free access. 

14. Matsushita M et al. Meclozine promotes longitudinal skeletal growth in transgenic mice with achondroplasia carrying a gain-of-function mutation in the FGFR3 gene. Endocrinology 2015;156(2):548-54. Free access.

15. Su N et al. Role of FGF/FGFR signaling in skeletal development and homeostasis: learning from mouse models. Bone Res 2014; 2:14003. Free access. 

16. Cook D et al. Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nature Rev Drug Disc 2014;13:419-31. Free access.