Seeding Open Innovation Drug Discovery and Translational Collaborations to Leverage Government Funding: A Case Study of Strategic Partnership between Sanford-Burnham and Mayo Clinic
1Sanford-Burnham Medical Research Institute, La Jolla, CA, USA
2Mayo Clinic, Rochester, MN, USA
Pharmaceutical Industry Stagnation and the Clarion Call for New Business Models
Over the last decade and especially in the past few years, there has been increasing scrutiny of the stagnant productivity of the pharmaceutical industry. In a recent provocative opinion on “crowdsourcing” of pharma drug development, Mintz [1] suggests that thinning drug pipelines, burgeoning development costs, and a lack of innovation condemn the “traditional” pharmaceutical drug development process as too costly and inefficient to survive in its current form. According to Munos [2], the costs of pharmaceutical research and development (R&D) to generate new drugs has dramatically increased to ∼$50 billion per year, while the number of new drugs have remained invariant at 25–30 per year.
Pharmaceutical executives have blamed this poor productivity on increased regulatory scrutiny, chronic indications leading to longer R&D cycles, escalating labor costs, and patent expirations leading to increased generics competition (estimated at 17% of all sales by 2014 compared with 10% in 2008), which have led to massive job cuts (>200,000 in past 4 years), corporate restructuring, and mergers and acquisitions. While these will control costs and provide products in the short term, none improves long-term productivity or fosters the spark of innovation needed for future advances. Dire forecasts estimate loss of sales from patent expirations between 2010 and 2014 of $204 billion [3] and between 2011 and 2015 to exceed $250 billion [4] (cutting revenues by as much as 41% [1]). However, Munos [2] and Kaitin [5] have suggested that big pharma’s current woes are rather the result of fundamentally flawed business practices established in the 1960s that have guided drug development and that have remained largely unchanged for the past half a century.
The current estimates for the R&D cycle time (10–15 years [3–6]) and costs for taking a drug from discovery to market now range widely, depending upon indication complexity and amortization of failures: estimates range from $1.32 billion [5] to $1.3–2.35 billion [3] to $3.7–12 billion [4], though most of the costs (75% [7]) are due to high failure rates. Indeed, Abbasi et al. [4] suggest the high R&D cost structure (>18% of sales), high failure rate [3, 8], long cycle time, and high R&D spend per drug make it a dismally performing, high-risk low-reward industry. When compared with orthogonal high-technology businesses such as fabrication of semiconductor devices, where the average complexity has grown in 20 years from each device containing 25,000 to more than 2 million transistors while development time and costs have been dramatically reduced with greater than 98% success of devices post fabrication, pharma has shown little progress. They cite a fundamental reason for pharma’s failure is a lack of extensive qualification and lack in validation of designs by running the massive simulations and “what-if” scenarios that are the norm in industries with high success rates.
Clearly, industry, disease advocacy groups, government, and investment communities have sounded clarion calls to find new innovative business models for drug discovery and development [6]. These include private–public partnerships such as the formation of pharma-sponsored incubators and centers of innovation consortia [9], deeper and more translational collaborations [10, 11], and strategic alliances [5] with academic [12] and hospital centers. Drug repurposing (or repositioning, reprofiling) [13–17] has also been proposed as solution to the low productivity, and this angle has been pursued by several biotechnology companies as well as by multiple academic groups that have arranged selective access to sets of “shelved development compounds” from pharma companies [18, 19]. More recently, open source drug discovery [20–22], open access [23], and crowdsourcing, as well as a number of nonprofit pharmaceutical development corporations [24, 25] have made their mark in the less profit-driven pursuit of therapies for rare diseases and neglected diseases of the developing countries. The productivity of the Office of Orphan Products Development, in which 41 FDA-approved therapies have advanced with only a budget of $14 million, is noteworthy.
NIH Roadmap Molecular Libraries Initiative Seeds Open Innovation for the Discovery of Chemical Probes
Sanford-Burnham’s Experience and Lessons Learned
It was in the context of a perceived lack of innovation and unsustainability in the pharmaceutical industry, as well as a poor track record of basic research discoveries yielding meaningful translation into new therapies for human diseases, that Elias A. Zerhouni, MD, Director of the National Institutes of Health (NIH), announced the Molecular Libraries Initiative (MLI) creation on June 15, 2005. With $88.9 million dollars of funding, a Molecular Libraries Screening Centers Network (MLSCN) comprising nine institutions from seven states was launched. This was a component of the NIH Roadmap’s “New Pathways to Discovery” initiative to advance the understanding of biological systems and to build a better “toolbox” for twenty-first-century medical researchers [26, 27]. This “open innovation” initiative was controversial: it mandated that the primary and confirmatory activity data generated from high-throughput screens conducted on a publicly disclosed Molecular Libraries Small Molecule Repository (MLSMR) of ∼100,000 compounds by the MLSCN would be deposited and publically accessible in a PubChem database, managed by the National Library of Medicine at the NIH [28]. Assays for the program would be sourced from academic investigators through an abbreviated grant application and peer review process. Successful proposals were assigned to an MLSCN center that subsequently collaborated with the investigator to complete assay development, resulting in robust, scalable assays that could be used to screen the entire MSLMR.
The Burnham Institute for Medical Research was selected as an MSLCN center based on the singular vision of the Chief Executive Officer, Dr. John C. Reed, who had invested and established an automated chemical screening core lab, the San Diego Center for Chemical Genomics, around several integrated Beckman FX workstations, as well as strategically recruiting dedicated staff with the functional skill set and specific experience in high-throughput screening (HTS) and drug discovery. The MLI was a heuristically valuable 3-year pilot whereby the individual center learned to function as a network, the NIH Roadmap managers tested and developed operational management processes and performance metrics and clarified deliverables and definitions of success, and the NIH learned what needed further improvement. The key collective lessons learned during this “pilot” phase were (1) the need for the network to provide significant assay development mentoring and educational outreach to improve the quality and technical feasibility of assays, as well as supportive preliminary results for quality grant applications, (2) development of clear work plans and go/no go decision gates during progression of hits toward selection of a chemical probe, and (3) the need for additional synthetic chemistry resources and access to new screening modalities. Through participation in the MLSCN, the Burnham Chemical genomics team was able to optimize processes, solidify broad expertise in all signal generation and detection technologies, complete hiring and training of staff, and most importantly learn to manage a portfolio of collaborations with principal investigators (PIs) of varied temperaments and styles of engagement and communication. During this pilot period core competencies in managing projects were developed, specifically dealing with the significant diversity among the PIs with respect to their knowledge of or interest in assay development and screening and their overall willingness to deliver to a timeline. These accumulated core competencies and operational experiences were key factors in our ability to attract future collaborations as well as enable effective execution and delivery on future milestone-driven collaborations.
In the next sections, we describe where we envision the renamed Conrad Prebys Center for Chemical Genomics (Prebys Center) at the (also renamed) Sanford-Burnham Medical Research Institute (SBMRI) would most contribute in new models of academic drug discovery and development. We will describe the capabilities and infrastructure we have developed and invested in that align with our ambitions and validate our ability to contribute meaningfully to new processes by which the early stages of drug discovery are executed. We will additionally describe some cutting edge approaches the Prebys Center has invested in with regard to creating “disease in a dish” models that may well provide future translational technologies and physiologically relevant yet renewable and propagatable model systems of disease. We will also provide an example where we have advanced the translation of an early MLPCN (Molecular Libraries Probe Production Centers Network) probe into a potential proof-of-concept preclinical lead candidate. We will articulate how the Prebys Center’s accumulated experiences and core competencies developed and refined with the NIH Roadmap’s Molecular Libraries Initiative and Programs poise us to leverage NIH funds toward advancing probe and drug discovery initially in a precompetitive environment through strategic collaborations.
Finally, we will describe Mayo Clinic’s experience, whose initial focus has been at the other end of the NIH Roadmap initiative. They were an early awardee of the Clinical and Translational Science Awards (CTSAs), and we will discuss how this award and the announcement of the National Center for the Advancement of Translational Sciences (NCATS) provided the context and motivation for the Mayo-Sanford-Burnham strategic collaboration. We will share how we selected each other as partners, the key cultural alignments and complementarities in core competencies that yield a mutually beneficial alliance whose impact is greater than the sum of the individual parts. And finally we will share some of the operational and governance models that have worked well in our translational research collaboration.
The Prebys Center as the Engine of New Drug Development Beyond Basic Research
Traditionally pharmaceutical companies and academia have sought collaborations that generally provided research funds in the form of sponsored reach agreements that typically focused on “Target Identification or Basic Research” as shown in the lowermost bracket “Academia, Scientific Institution, Not-for-profit” to the right of the leftmost stack of activity boxes entitled “The Recent Past” in Figure 27.1. In the future model of drug discovery and development, academic or not-for-profit organizations will have to extend their contribution further along the process, advancing compounds through more costly investment phases to larger, preclinical animal studies where proof-of-concept in an organism can be attained. This is shown in the lowermost right bracket, which extends upwards now close to the Preclinical development box, and into the region of the vertical gray-shaded oval, which indicates where a project must now advance (“The Future”) for a venture capitalist and/or a pharma company to take on a project.
An approximate timeline and attrition rate for compounds (sideward funnel) as they transition through the drug discovery and development phases or pipeline [29] is also shown in Figure 27.1, and the left box indicates where drug discovery needs to proceed to a proof-of-concept in an animal model to become valuable enough for a pharma company, venture capitalist or clinical partner to invest to bring the compound into a proof of clinical concept, or first in man stage (second box). The Prebys Center has the competencies to advance compounds into at least small animal studies; however, we lack clinical trial capabilities and the institutional funding to advance projects this far.
Prebys Center’s Key Strengths, Expertise, and Experience
Through its participation in the NIH Roadmap, the Prebys Center has almost 8 years of experience in conducting and completing chemical probe and drug discovery projects from target/assay concept, through assay development, HTS, and hit validation, “hit-to-lead,” and “lead optimization” to yield optimized drug leads or preclinical development candidates.
- More than 50 professional staff and management comprising pharmaceutical and biotechnology veterans with 450+ person-years of industrial experience in HTS and high-content screening (HCS), as well as all associated activities of drug discovery and early development.
- Performed >350 assay development projects (1° and 2° assays) and completed 125 HTS campaigns on large chemical libraries (>350,000 compounds) for both NIH- and NCI-sponsored as well as pharma/biotech collaborators. These assays represent
- A large diversity of target classes and assay formats (Figure 27.2a)
- With 39% cell-based versus “simple” biochemical formats (including primary cells and organelles)
- With 11% phenotypic HCS automated imaging formats (including induced pluripotent stem cell (iPSC) and Caenorhabditis elegans) (Figure 27.2b)
- All signal generation and detection modalities (except radiometric).
- A large diversity of target classes and assay formats (Figure 27.2a)
- Designated as a “comprehensive” center (capable of both screening and lead generation chemistry) for NIH MLPCN, delivering 52 new chemical probes and depositing more than 33.6 million test data to PubChem and the NCI Chemical Biological Consortium (CBC).
- Proven track record of successfully collaborating with >130 PIs, representing >42 academic, nonprofit medical research, and commercial institutions in >31 geographic locations and 3 foreign countries (Figure 27.3), including several of the Mayo Clinic centers (Scottsdale, AZ; Rochester, MN; and Jacksonville, FL).
- Proven capability and infrastructure to manage 50 concurrent projects, to meet/exceed timelines, milestones, metrics, and deliverables while providing timely communication, data loads to PubChem, project reviews/meetings, and reports to stakeholders.
- One of most advanced state-of-the-art laboratory automation platforms and bio-coastal infrastructure in a not-for-profit institution to support drug discovery and innovation that represent a combined $44 million investment in the Prebys Center (Figure 27.4).
- Workstation and modular “plug & play” fully integrated 1-arm (La Jolla, CA) and 3-arm (Orlando, FL) robotic HTS platforms
- The largest installation among academic institutions of five fully integrated and two offline noncontact nanoliter acoustic drop ejection devices and capabilities (Figure 27.4 and Figure 27.5)
- A greater than 900,000 small molecule chemical and natural product extract collection
- Medicinal chemistry, cheminformatics, and exploratory pharmacology
- Tissue culture facilities and advanced image algorithm development for high-content iPSC assays.
- The largest installation among academic institutions of five fully integrated and two offline noncontact nanoliter acoustic drop ejection devices and capabilities (Figure 27.4 and Figure 27.5)
- Proven track record of successfully collaborating with >130 PIs, representing >42 academic, nonprofit medical research, and commercial institutions in >31 geographic locations and 3 foreign countries (Figure 27.3), including several of the Mayo Clinic centers (Scottsdale, AZ; Rochester, MN; and Jacksonville, FL).
Overview of the Screening Resources at the San Diego Campus.
In San Diego, the drug discovery facilities are located in approximately 20,000 sq. ft. of lab space. A major portion of the space is dedicated to central assay development and HTS lab space, liquid handler workstations, and various plate readers (as well as an image-based high-content screening system). The centerpiece of the screening facility is a HighRes Biosolutions (HighRes) 1-POD system (Figure 27.4), as well as two Beckman Coulter core systems. The ultrahigh-throughput screening (uHTS) system in San Diego has a theoretical maximum throughput of ∼320,000 wells per day. When combined with the two Beckman Coulter core systems, an overall theoretical maximum throughput of approximately 475,000 wells/day is achievable for plate reader-based assays at the San Diego site.
Overview of the Screening Resources in the Orlando Campus.
Sanford-Burnham has established an uHTS screening system in Orlando, Florida, occupying a 175,000-sq ft building since April of 2009. This uHTS system has a theoretical maximum throughput of ∼2.2 million wells/day, with the capability to perform almost any assay, whether cellular or biochemical plate reader, or cellular image-based assay. This system currently exceeds the capability of most HTS systems in industry or academia, and provides a unique resource. Approximately 6500 sq ft has been provided for the uHTS facility in Orlando, including biosafety level 2 (BSL-2) tissue culture labs and compound management space. The facility is adjacent to the Exploratory Pharmacology and Medicinal Chemistry Cores. The HighRes Biosolutions 3-POD-based system at Sanford-Burnham’s Orlando campus consists of three linked robot PODs. At the center of each POD is a Stäubli RX-160L robot with a CS8 controller.
Hit Confirmation and SAR Support Capabilities.
The Echo555 (Labcyte) acoustic dispensers (Figure 27.5) available at both the San Diego and Orlando sites are invaluable for hit picking during reconfirmation, as well as for direct serial dilutions (dose–response) in structure–activity relationship (SAR) studies. Echo instruments perform contact-free transfer of 2.5-nL droplets directly from a source to a destination plate using focused energy of acoustic waveforms. Solution from any well within a source plate can be transferred into any well of the destination plate. The majority of our assays are performed in 1536-well plates with the assay volumes between 3 and 8 μL that require withdrawals of low-nanoliter volumes and which consume only a negligible portion of the 3- to 5-μL aliquots of compounds in the source plates. This helps to overcome the uneven depletion of wells associated with conventional liquid handling cherry-picking instruments. As a working copy of source plates is kept online for less than 6 months, they are usually replaced with a fresh copy prior to a noticeable depletion of compound wells. The ability to dispense small-volume aliquots into the assay plates enables cherry-picking of hits directly from compound source plates containing 2- or 10-mM stock solutions of library compounds without an intermediate dilution into aqueous solution. This shortcut helps save on costs associated with intermediate plates and reduces potential compound solubility issues. A majority of compounds with limited solubility in aqueous solution are known to come out of solution during the intermediate-dilution stage required with conventional liquid-handling instrumentation. Dispensing nanoliter volumes directly into assay plates circumvents this step, maintains high dimethyl sulfoxide (DMSO) concentration, and helps prevent precipitation.
Furthermore, the wide range of transfer volumes possible with acoustic dispensing allows dose-dependent (over 3 logs) confirmation of hits directly from source compound plates as part of our workflow. This approach is especially valuable for culling out nonselective or assay-interfering compounds through side-by-side tests in primary and counterscreen assays. The relative ease of use and precision of dispensing inherent to Echo® instruments led us to establish acoustic compound dispensing as the main workflow for compound handling within SAR studies. This approach allows us to dramatically reduce compound consumption extending the use of stock solutions through multiple secondary assays. The Labcyte Access workstation significantly simplifies the logistics of handling serial dilution of large number of compounds. Another benefit of Echo dispensers, their ability to dispense aqueous solutions, is instrumental for performing mechanism-of-action studies in which both compound and a known ligand for the target are cross-titrated in the assays plates to determine their interaction, while minimizing depletion of these often critically limited reagents.
High-Content Screening Capabilities, Facilities, and Image Management Infrastructure.
We provide herein a rather extensive exposition of our capabilities in this area, as this is a particular strength and differentiator of the Prebys Center. As Figure 27.2b exemplifies, the 14 HCS campaigns yielded a disproportionately high number of 11 novel probes. In particular, novel nonpeptide small molecule probe agonists for the orphan GPR55 and GPR35, as well NTR1 agonists were significant developments. This finding was in agreement with a recent claim by Swinney and Anthony [30] that more first-in-class drugs have arisen from phenotypic and image-based assays. The Prebys Center regards intimate association of HCS with patient-derived induced pluripotent cells to a strategic focus toward assays that reflect relevant translational biology. Among the current academic/nonprofit screening centers, we are the only center that has completed multiple complex HCS campaigns in both 384-well and 1536-well formats of large compound libraries (Figure 27.6) and deposited those data into PubChem.
Assay development for HCS is a multidisciplinary endeavor combining cell-based assays, multistep fixation and staining protocols, multicolor imaging, and custom algorithm development for image segmentation, feature extraction, and correlation of the multiparametric data to meaningful biology. The expert HCS staff, located at the San Diego site while supporting both the San Diego and Orlando locations, executes and/or supports assay development, screening, and data analysis/mining for image-based high-content primary and secondary assays, where the readout is typically based on images from high-throughput microscopy systems. Due to the variety of image-based assay projects encountered at CPCCG, the HCS facility’s systems provide capabilities covering fluorescence confocal, fluorescence wide-field, and trans-illumination imaging at magnifications ranging from 3.5× to 60× air and water immersion objectives (0.15 to 1.2 NA). Specifically, the HCS equipment in San Diego includes a spinning-disk confocal Opera QEHS system (PerkinElmer) with on-the-fly image analysis and an InCell1000 imaging system (GE). Examples of a wide range of full MLSMR phenotypic HCS campaigns and representative images are shown in Figure 27.6.
A variety of image processing, data analysis, and visualization tools are available for multiparametric imaging data analysis and mining. With these software packages, the HCS team can rapidly apply available HCS solutions and algorithms, as well as develop new HCS solutions, algorithms, and tools for the wide variety of HCS projects to generate image montages as previously mentioned. HCS assays are suitable to support SARs by quantitative determination of compound potency from phenotypic dose–response curves that the software can aggregate for presentation. An example of an IC50 curve for a phenotypic G-coupled protein receptor assay based on the redistribution of GFP-tagged β-arrestin (Figure 27.7) also illustrates the unexpected finding that at the highest concentrations of agonist, the β-arrestin while remaining condensed, redistribute into smaller punctae compared with larger granules at the initial saturating doses.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
