Fourth Addendum to Research and License Agreement by and between Technion Research and Development Foundation Ltd. and Eloxx Pharmaceuticals Ltd., dated April 29, 2015

Exhibit 10.8

Execution Copy

FOURTH ADDENDUM

TO THE RESEARCH AND LICENSE AGREEMENT

This Fourth Addendum to Research and license Agreement (the “Fourth Addendum”) is made by and between the Technion Research and Development Foundation Ltd. (“TRDF”) and Eloxx Pharmaceuticals Ltd. (“Licensee” or “Eloxx”).

Whereas, TRDF and Eloxx are parties to a Research and License Agreement with an effective date of August 29^th 2013 (the “License Agreement”), as amended on November 26^th, 2013 (“First Amendment”), January 14^th, 2014 (“Second Amendment”), June 9^th, 2014 (“Third Amendment”), August 3^rd 2014 (“First Addendum”), January 21^st, 2015 (“Second Addendum”) and February 9^th 2015 (“Third Addendum”) (collectively, the “Agreement”); and

Whereas, according to Section 1.26 to the License Agreement, the first Research Period has ended on September 30th, 2014 (“First Research Period”) and the First Research Plan was completed respectively (“First Research Plan”);and

Whereas, the parties desire to extend and continue the Research Period and the Research for a second year; and

Whereas, the parties desire to continue the relationship contemplated by the Agreement and therefore to further amend the Agreement as set forth herein;

NOW, THEREFORE, the parties hereby agree as follows:

Execution Copy

Exhibit D

Timor Baasov

While these data support the likelihood that the compounds 1-9 of set1 structures (Fig. 1) with 6’-OH and unsaturation at ring I may perform better than previous leads with the saturated ring I, we want to introduce additional structural motifs at the C6’- region that would further increase this expectation. We plan to replace the CH₂OH (and/or CH(CH₃)OH) at 6’-position with the vicinal 6’,7’-diol (compounds 10-11, Fig.1). The replacement of C6’-OH with 6’,7’-diol is expected to generate stronger H-bond interactions of ring I with A site residues G1408 and C1409 resulting in better pseudo-disaccharide scaffold of set1 (Fig. 3). Following important points in Fig. 3 are of note. First, unlike our solved structure of G418 bound to Leishmania A-site (Fig. 2B and 2C) in which ring I of G418 makes two H-bonds with G1408, the structure of G418 bound to the entire yeast ribosome recently reported by Yusupov and co-workers⁹, shows only one H-bond between O6’ and G1408 nitrogen (Fig. 3). Modelling ring I of the 6’,7’-diol 10 onto this structure of Yusopov, clearly shows that O7’-OH can make additional H-bond with the oxygen of C1409.

To test this issue, we initially synthesized compound 10 (Fig. 1) as a single diastereomer at C6’ with unknown configuration (the complete assignment of the absolute configuration of 10 at C6’, along with the synthesis of its C6’- distereomer are underway). Preliminary in-vitro suppression tests demonstrated that 10 exhibits superior readthrough to that of 12, and paromamine (Fig. 4). Interestingly, installation of 6’,7’-diol on the paromamine scaffold (10) dramatically increases its in-vitro readthrough activ ity. It is of note that the observed activity increase in 10 is far higher than the impact of the chiral (R)-6’- methyl group in 12, which was to date our most favorable pharmacophore at C6’- regon^7,8. With compound 10 we are already one step ahead! Further installation of chiral (R)-6’-methyl group and/or N1-AHB group on compound 10 will likely provide additional improvement of this performance, and these potentials will be tested in the subsequent steps.

In addition to 10 we also synthesized the simplest representative of set1 structures, compound 1 (R=R₁=R₂=H, Figs. 1 and 2). Preliminary comparative in-vitro suppression tests demonstrated that 1 exhibits superior readthrough to that of its parent paromamine and is very similar to that of compound 12 (Fig. 4). These data suggest that the anticipated p-p stacking interaction of the unsaturated ring I of 1 with A1491 is much stronger than the CH-p interactions of the saturated ring I of paromamine with A1491. The data in Fig. 4 also suggests that the advantage of such stacking interactions in 1 (paromamineg1) is similar to that of (R)-6’-methyl group in 12 (paromamineg12), and thus identifies unsaturation at ring I as a potential new pharmacophore.

Aim #1- b) rational design of pseudo-trisaccharide scaffolds: It should be noted that the set1 structures (Fig. 1) are not expected to possess significant PTC suppression activity (Fig. 4); rather they will be used as scaffolds for the construction of new efficacious structures by attaching various sugar and/or acyclic appendages. For a systematic study, the same modifications will be performed on all pairs of set1, with the aim to reach the best lead compound. As a ring III we have developed a new structure of (S)- 5’’-amino-5’’-methyl-D-ribose, and its advantage over 5”-aminoribose was evident in our previous leads^7,8. Therefore, we will use this novel ring III for the generation of new pseudo-trisaccharides (set2-set3, Fig. 5). Encouraged by the preliminary data obtained with 10 (Fig. 4), we will scrutinize the stereochemistry at C6’ of 10 and will assemble the corresponding pseudo-

6

Timor Baasov

trisaccharides 13 and 14 (set2, Fig. 5), along with all the variations of pseudo-trisaccharides with unsuterated ring I (compounds 15-22, set3, Fig. 5). In addition, based on the observed strong impact of the vicinal 6’,7’-diol in 10 (Fig. 4), this variable (6’,7’-diol) will further be added to the planned pseudo-trisaccharides of set3 structures in Fig. 5, to yield additional sets of compounds incorporating the vicinal 6’,7’-diol and unsaturation at ring I.

Aim #2. a) Modification of lead structures for improved cell permeability and oral bioavailability: Because AGs are highly charged, water soluble compounds, they have poor intestinal absorption and are usually injected. In addition, AGs are short-lived in the circulatory system, rapidly eliminated by glomerular filtration in the kidney. They also exhibit poor permeability into eukaryotic cells, which requires higher dosages that often cause harmful side effects that limit their use in therapy. In order to overcome these limitations we designed different sets of compounds. One of such sets (Fig. 6) considers that all hydroxyls of the AG are replaced by esters. The rationale is to increase the lipophilicity and cell-permeability of the drugs, improve in-vivo half-life and oral bioavailability. These changes should provide important therapeutic benefit over treatment with the initial lead alone.

The suggested modifications on NB124 and G418 (Fig. 6) to yield the corresponding per-benzoate esters (OBz, 51 and 52, respectively) are based on the following observations. (i) In our most recent comparative study of our lead compounds using an extended repertoire of CF cell lines, reporters, assays and animal models, comp. NB124 was able to most effectively rescue CFTR function, was superior to gentamicin, exhibited favorable pharmacokinetics, and was less cytotoxic than gentamicin in the explant model of ototoxicity¹⁰; (ii) the benzoate esters 51 and 52 are chemically more stable than the corresponding simple alkyl asters (like acetate, isobutirate, isopropionate); (iii) the poly-esters 51 and 52 can easily be prepared in good overall yields from the corresponding AGs in three steps (Boc₂O, Et₃N, H₂O/MeOH; BzCl, Py, 4DMAP; TRA, CH₂Cl₂); (iv) to test our hypothesis we initially used 52 as a prototype derived from commercial G418.

Preliminary tests of 52 against its parent G418 demonstrated that 52 lacks any readthrough activity both in cell free (see the data in the 1^st year research report) assays and HEK293 cells stably expressing UGAC/CGAC dual luciferase (Fig. 7A) even when the incubation time was extended up to 72 hrs. Interestingly, when we tested the antibacterial activities of G418 and 52, we found that while the activity of 52 was similar to that of G418 against various G^- and G⁺ WT bacteria, the MIC values of 52 were significantly lower than that of G418 in bacterial strains harboring plasmids of AG resistance determinant enzymes (see the data in the 1^st year research report). Encouraged, we argued that HEK293 cells, being engineered cell lines, might not contain the required esterases to hydrolyze the benzoate esters in 52. Indeed, using MEFs derived from homozygous Idua-W402X mice¹¹, we found that the activity of 52 significantly exceeds that of G418 at all incubation times and concentrations tested. At a dose of only 13.6 micromolar, treated cells had already reached the glycosaminoglycan (GAG) levels found in WT cells (Fig. 7B). These encouraging data with 52 prompted us to prepare compound 51 (see the data in the 1^st year research report), which is currently under similar tests in Idua-W402X MEFs (by our collaborator D. Bedwell at Univ. of Alabama, Birmingham) and CF cell lines (primary human bronchial epithelial, HBE cells, G542X/delF508, S. Rowe at the UAB). In-vivo tests in the Idua-W402X mice (D. Bedwell) will follow these experiments.

7

Timor Baasov

Aim #2 b) Modifications for improved capacity to pass the blood-brain barrier (BBB). While we have been successful in identifying promising nonsense suppression agents (NB84) for the Idua-W402X mutation and partially alleviating the primary GAG storage defect associated with MPS I-H within a variety of tissues (including brain), we have been unable to completely restore GAG levels in the mouse model, even in a long-term, 28-week drug-treatment^12,13. Even though the newly proposed sets of compounds might prove better than the previously examined leads (especially the structures with elevated lipophilicity like 51) for treatment of MPS I-H, we propose to also test a brain-targeted chemical delivery system (CDS)¹⁴ which is expected to produce a more robust therapeutic response. The approach is based on the attachment of a dihydropyridine moiety as a carrier for delivering drugs through the BBB (Fig. 8). This site-selective delivery will reduce the exposure of the body to high drug levels and consequently increase their therapeutic index. Since NB84 has already been proven to significantly moderate disease progression in-vivo in our MPS I-H mouse model ,¹² we will now synthesize the new derivative of NB84, the compound NB84-CDS (Fig. 11). We (our collaborator at UAB, D. Bedwell) will then determine whether delivery of NB84-CDS through the BBB is better than NB84 in a long term study using established biochemical and immunohistochemical assays as previously described¹².

REFERENCES

8

Timor Baasov

III. Proposed Budget (in US Dollars) and Time Frame

9

Timor Baasov

NOTE the budget does not include ex vivo (MEFs) and in vivo studies required for the project completion and only can be done with additional funding of collaborators labs (Bedwell and Rowe at UAB)!

IV. Detailed Budget (in US Dollars)

Personnel:

Supplies:

Budget Justification:

The requests for laboratory assistant and postdoctorant for the duration of the grant period are in recognition of the amount of work required in this project. Considerable effort will be expended in the syntheses of various lead compounds discussed in the proposal, their structure determination and analysis, assays for their activity. The request for materials, supplies, and chemicals/biochemicals is an important part for a successful development and completion of the project.

V. Investigators’ Curriculum Vitae

(a) Education Background

Major research interest: Carbohydrate chemistry, Bioorganic and medicinal chemistry, Drug design and development, Rational design of substrate and inhibitors, Mechanistic enzymology.

(b) Employment

10

Timor Baasov

VI. T. Baasov - List of Publications last three years (2011-2014)

11

Timor Baasov

Chapters in Books and other Publications

12