The panel met on December 11-12, 2018, at the National Academies of Sciences, Engineering, and Medicine (National Academies) facility in Washington, D.C., to review the In-House Laboratory Independent Research (ILIR) program projects in life sciences conducted in 2018 at the U.S. Army Armament Research, Development, and Engineering Center (ARDEC) and the U.S. Army Edgewood Chemical Biological Center (ECBC). The panel received overview presentations on the ILIR programs conducted at these two RDECs and technical presentations describing the projects. During each presentation, the panel engaged in question-and-answer sessions with the presenter, and a general discussion with RDEC staff after the panel had formulated initial impressions and developed additional questions during its closed-session deliberations, conducted after the RDEC staff had concluded their presentations.
Project: Human Electrophysiology for Soldier-Armament Integration
The overarching purpose of this project is to use a variety of psychophysiological measures to index covert psychological decision-making processes. Specifically, the project incorporates a variety of neurological and motor markers of behavior into the study of a shooting task. Two experiments were designed that used a go/no-go task. In both experiments, participants were expert marksmen who were to withhold a response (i.e., a no-go trial) when they were presented with friendly targets and to pull the trigger of their firearm when presented with foe targets (i.e., a go trial). In the first experiment, participants viewed 1,000 targets—700 friendly and 300 foe—presented on a viewing screen in sequence for 50-150 minutes each. Foe targets required a trigger pull response as quickly as possible. In the second experiment, 150 friendly and 150 foe targets were placed in motion for 5 seconds, and participants were instructed to withhold trigger pulls until foe targets entered a predesignated area (i.e., a bounding box) on
the screen. Together, these experiments are capable of revealing the ways in which a total body response may differentiate target identification processes separately from patterns of psychophysiological activity related to response preparation. Moreover, the project seeks to utilize nonlinear analytics, data fusion, and supervised machine learning techniques to analyze the recorded total-body responses to predict future behavior. The principal investigator (PI) has completed 1 year of work, which has culminated in the successful recording of data from 10 participants. These data were not presented during the review in processed or summary form, although a few illustrative examples of raw data patterns were provided. The eventual outcomes of the project, therefore, could not be assessed or predicted.
Discovering the links between covert psychological/psychophysiological events and observable measures of behavior—whether the firing rates of individual neurons or the gross actions of an organism—is the bedrock of basic psychological science. As new investigative technologies and dependent variables have been developed, however, they have tended to be used in relative isolation, leaving gaps in knowledge regarding the interrelatedness of both the measures and the underlying processes they purport to reveal. The more innovative multichannel approach used in this project simultaneously records activity within the brain (electroenchephalography), autonomic nervous system (electrocardiography), sympathetic nervous system (impedance cardiography and electrodermal activity), and motor systems (electromyography) within a single task. The experiments are well designed, and the multichannel data collection approach is a smart move, although the sheer number of dependent variables is perhaps too ambitious. The project has the potential to advance understanding of the dynamic propagation of activation across multiple psychophysical systems and to develop more robust, stable, and multipatterned signals that may reveal new mechanisms underlying complex behavior, as well as the variability in these signals across individuals. The links to computer science and engineering provide further opportunities to find psychophysical signatures of behavior that can be used to predict behavior. The co-registration of psychophysiological measures to study behavior as well as the use of those markers to predict future behavior are contemporary and important avenues of basic research.
Despite its opportunities for discovery, the project faces significant challenges, which were generally acknowledged by the PI. Some of these challenges are logistical. The PI reported that the recruitment of participants is slow and difficult. Several goals of this project, particularly those related to building and validating classifiers and assessing individual differences among psychophysiological response patterns, will require a much larger number of participants. Recruitment criteria could be relaxed, especially given the artificial nature of the friend and foe stimuli.
Aside from logistics, two additional major scientific challenges need to be overcome. The first concerns the separation of signal and noise. In isolation, psychophysiological measures need to be collected over many trials in order to augment signal and reduce noise. The interrelations among measures will undoubtedly introduce other sources of noise because the signals are not perfectly correlated. This casts doubt on the investigators’ ability to predict behavior in a single event, no matter how successful the modeling of past behaviors may be. The second major challenge concerns the processing of data and the development of classifiers. A variety of assumptions and analytic choices need to be made within an iterative set of processes, including the following: How could data be aggregated? How will physiological responses that unfold over different time scales be integrated? How will missing data be handled? The degree to which these questions can be fully explored and refined within the short duration of the project is not clear.
The basic goals of the project (co-registration of psychophysiological data and the use of such data to predict behavior) lie within a scientifically crowded space. The research group would probably benefit from input from outside researchers. In perhaps an effort to cleave a distinctive area of work, the focus of this project has become too narrow, with the ultimate stated goal being the prediction of trigger pulls
at better than chance levels. The logical extension of this goal, however, is unclear. The current project will likely not scale to behavior in real-world scenarios. The project cannot set aside, as it seems to do at present, basic questions regarding the cognitive mechanisms underlying decision making—the locus of greatest gains within the decision-making process that could be leveraged from the psychophysiological markers identified and the efficacy/use of prediction to modify or enhance action. While the project constitutes basic research, its opportunities for basic research advancement are perhaps artificially limited by an intent to aim the final goals of the project toward application.
Project: Effect of Quorum Sensing Molecules on the Production of Bacterial Nanocellulose Materials
This is an ongoing 3-year project that was started in FY18 with the overall objective of manipulating and controlling the production of bacterial nanocellulose through a combination of chemical and/or genetic interventions.
Overall, this was an interesting project that has good potential for broad impact. Nanocellulose, in particular bacterial nanocellulose (BNC), has a number of material properties that make it a very attractive target for research and development. Methods to produce purified BNC and/or to scale up the production of BNC are both active areas of research in many laboratories around the world. The crowded competition space in this research area underlines the potential utility and interest in BNC as a functional material.
The progress made on this project in its first year seems to be good, and it has largely followed the timeline set at the project outset. Optimization of growth conditions for the bacteria G. xylinus has been performed, characterization of the bacterial films and identification of various quorum sensing molecules has been accomplished, and the manipulation of quorum sensing (QS) molecule levels and characterization of BNC growth and structure has started to be performed. The QS molecules that were identified all fall into the class of homoserine lactones (HSL). Interestingly, the addition of different HSL molecules to the bacterial growth media produces BNC biofilms with different morphologies and different fiber thicknesses; however, the overall thickness and growth kinetics of the biofilm itself were not significantly altered. There is still much work to be done here, and it remains to be seen if these different BNC morphologies have unique underlying crystal structures that might make them useful in different applications.
Even with the progress made in the first year, the next steps of this project are not clear. There is a stated objective to characterize the genes LuxI-LuxR that control HSL production, but because BNC production can be directly controlled by the addition of exogenous HSL molecules, the reasons for this effort are not entirely clear. It would seem that the goal needs to be to subvert quorum sensing and simply control it. Similarly, the use of genetic or genomic engineering was cited at several different points in the presentation, but the methods and objectives of performing this work were not obvious. Other academic research laboratories are following this procedure, and it would seem more beneficial to make use of the advances made by others rather than trying to reinvent things.
Project: Effect of Toxicants on the Regulation of Endothelial Barrier Function
This basic science project is developing an in vitro assay system to assess how different chemical warfare agents alter the endothelial-cell barrier of the vasculature. A screen of the growth properties and
responses of several different endothelial cell lines to four potential chemical warfare agents (two organophosphates, malathion and malaoxon; a microbial toxin, lipopolysaccharide; and a chemotherapeutic agent, bis(2-cholroethyl)amine hydrochloride) was undertaken. Two assays were developed to assess barrier function, including (1) a clever impedance assay that measures how effectively cultured cells hinder electron flow and (2) a macromolecular tracer assay that measures the ability of a cell monolayer to slow the transfer between two compartments of a reagent like fluorescein albumin (FITC-albumin) or fluorescein isothiocyanate–dextran (FITC-dextran). The screen and assays revealed differences in the growth rates of the various endothelial cell lines, altered susceptibilities to and effects of the toxicants, and nonlinear responses to varying concentrations of the agents. For reasons relating to reproducibility and cell growth properties, future experiments will focus on one cell line termed human microvascular endothelial cells (HMEC-1), which are immortalized endothelial cells originally derived from a neonatal skin biopsy. These future experiments will assess how chemical agents alter the structure of a cell monolayer, the cytoskeleton of the cell, and intracellular signaling through receptor tyrosine kinase pathways.
Opportunities associated with this project include the potential to develop a rapid in vitro assay to assess effects of chemical warfare agents on blood vessels. Low-level exposure to agents such as organophosphates are known to cause edema and other symptoms associated with a loss of patency in the endothelial cells lining arteries, veins, and capillaries. An ability to reproducibly determine whether a known or unknown agent compromises the vasculature would provide insight into mechanisms of toxicity and perhaps point the way to more effective therapies to treat the wounded.
Challenges of this project include the large amount of prior work on the toxic effects of agents on endothelial cells, which reduces the potential discovery impact of the work. Challenges also include how accurately the chosen cell culture model represents the effects of an agent on the human body in the field and the narrow focus of the future mechanistic studies on tyrosine kinase signaling, which may be affected by some but not all agents and which may not reveal actionable therapeutic targets.
How chemical warfare agents affect the vasculature is an important area of study with direct relevance to the Army. Given this importance, a larger, more focused, and longer-term effort than that represented by a single project carried out on a part-time basis is warranted. Such an effort would involve the use of in vivo as well as in vitro models of vascular injury, larger numbers of chemical agents and toxins, and more personnel.
Project: Epigenetic “Memory” During Bacterial Adaptation to Environmental Changes
The hypothesis to be tested exploits the well-studied bacterium Salmonella typhimurium to determine if the organism has an epigenetic memory of physiological states that persist after the environment responsible for a certain state changes. Heritable differences have been widely described in bacteria, with the traditional cause being conventional genetic change. Other heritable differences have capitalized on stochastic variation of different molecular mechanisms, which gives some members of a bacterial population a growth advantage, which is then selected on. This study is unique in seeking adaptations common to most cells in a population grown under one condition that persists for up to 40 rounds of cell division in a new condition. The principal tool is quantitative mass spectrometry, though deoxyribonucleic acid (DNA) and ribonucleic acid (RNA) samples are archived at various time points to allow the mechanism behind any individual heritable memory to be pursued. Bacterial infectious disease is a persistent problem with deployment into foreign environments and close-quarters living. A better understanding of bacterial biology is relevant to controlling it.
As conceived, this project is extremely demanding technically in that it requires multiple cell cultures to be propagated in parallel without any experimentally induced difference between replicate cell cultures.
Moreover, the data have been simply measurements of protein abundance, and these measurements miss about half the proteins in the cell. They also miss any post-translational modifications, which are common mechanisms by which cells adapt to changes in their environment. As conceived, the project is also conceptually demanding in that the bet hedging mechanisms typically used by bacteria to hedge against changes in their environment occur in one-in-a-thousand or so cells and hence are not well suited to bulk biochemical approaches. Finally, the physiological adaptation sought by the researchers would only be detectable if it were to survive not just cell division but also multiple rounds of the growth cycle involving stationary phase, lag phase, and log phase.
At a minimum, if one is going to persist with the biochemical approach to this question a big change is needed to increase the odds of success. The cells need to be grown in a chemostat so that whatever state they may have achieved has a chance of being maintained long enough to be detected. The expectation that any heritable state will be stable to multiple excursions of a growth cycle could probably only be supported by a prion mechanism, which would probably be missed by the proposed mechanism.
Far better would be to develop a strategy that exploits genetics—the very reason that salmonella has provided so much useful information to the world. The proposed study will look at mutants affecting DNA methylation once a memory is detected, but such candidate gene approaches fail more often than they succeed. There are probably a half-dozen or more genetically based methods of approaching this question, which would all be workable.
One conservative alternative (if one believes that heritable changes in the synthesis of a protein is the hallmark of memory) would be to use reporter gene libraries. Reporter genes that continue to be expressed at the same level they were expressed at in lysogeny broth (LB) once diluted into the low magnesium medium would be a hallmark of a physiological memory. Those reporters would also be the tools that would allow the isolation of mutants, which would reveal the mechanism behind the memory. It would be optimal to use reporters that have good detectability in a colony but relatively short half-lives in other circumstances to best detect changes in the rate of synthesis. A good reporter would allow the study to be combined with microfluidics to gain single cell resolution.
There is the sense that the availability of high-end mass spectrometry equipment in ECBC was a driver for developing projects that could use it. The bottom line is that the research question within this project is an interesting one; however, the approach needs serious modification, and the better approaches are substantially easier to execute.
Project: Structure, Modeling, and Prediction of Cystine Knot Miniproteins
This project encompasses basic research on functional protein design, specifically, a class of thermodynamically robust cystine knot motif sequences. Rather than a completely de novo design, the project was based on bioinformatic analysis followed by a combination of in silico design and rational sequence variation. This approach is a common and sound one for protein engineering and has been successfully demonstrated in many cases. Cystine knot proteins, and miniproteins in general, have been identified as useful targets for protein engineering, due to the fact that they can be synthetically accessed in high throughput by either solid-phase chemical synthesis or recombinant DNA methods. Both have advantages and both were explored by the researchers as avenues for the preparation of these materials.
The motivation for this research was the identification of robust scaffolds that could be employed as high-affinity binders against selectable targets, although no specifics were provided on the latter. The protein engineering community has much interest in the development of such reagents as replacements for antibodies in applications such as biosensors and therapeutics, among others. In many cases antibodies are laborious to develop and produce and display pronounced sensitivity to aggregation and activity
loss during storage or transport. Alternative scaffolds are available that are based on synthetic tandem repeat proteins (e.g., DARPins), but a technological need exists for more structurally diverse classes of binders that could potentially sample a wider range of functional binding space. The cystine knot miniproteins have an advantage in that many such proteins have been identified and a number of them have been structurally characterized at atomic resolution. In addition, the native role of these proteins involves binding to target proteins, so it seems logical that they might serve as substrates for the engineering of novel binders. A similar logic was employed in the development of synthetic tandem repeat binders; however, these proteins form extended, rather than compact, structures and would probably display significantly different binding characteristics from compact protein folds such as the cystine knots.
The objectives of the research were clearly stated and make sense within the context of conventional protein engineering approaches. The rationale was sound in terms of the preliminary analysis, and the hypothesis was clearly stated and verifiable. Nevertheless, the problem is a challenging one to address in a limited time frame with limited research staff.
It seemed that protein engineering and synthetic biology were an evolving research and technology thrust of the ECBC program; nonetheless, it was not made clear whether the in-house resources or scientific expertise of the investigators were sufficient to effectively address the problem, especially in such a short time frame. The investigators did respond to critiques and modify the scientific approach and methods after the first year of funding, but the intrinsic challenges could not be effectively overcome in the subsequent year of funded research.
The development of novel classes of binding agents represents a useful goal, as stated above, for the development of reagents for use in pharmaceuticals and biosensors. It seems that these reagents have utility to the Army, even though the development time may be extensive; specific targets were not discussed. This protein engineering program fits well within the emerging synthetic biology thrust of the ECBC program. The development of high-affinity binding agents through protein engineering is a topic of significant research effort in the broader community of researchers in synthetic biology. High-affinity nucleotide-based binding agents (e.g., aptamers) have been known for quite some time and robust synthetic methods, such as SELEX, are available for their preparation. For many reasons, particularly scale and chemical stability, aptamers may not be the optimal choice. Most biological binding interactions involve proteins; therefore, de novo design and protein engineering represents a logical solution to this problem. The investigators focused on a class of protein motifs that display such a functional role.
Protein engineering is not necessarily straightforward experimentally, especially in terms of predictable success within a short and definable time frame. Because most proteins are highly evolved for a particular functional role and set of conditions, even structurally informed mutagenesis approaches, similar to those employed in this research project, usually result in a limited degree of success even with bioinformatic analysis and high-resolution structural data. The investigators found that synthetic peptides, prepared either using ribosomal deoxyribonucleic acid (rDNA) or chemical synthesis, often did not fold correctly or were insoluble. Cystine knot proteins can have complex post-translational processing that enables them to fold correctly despite having a limited hydrophobic core and, therefore, a thermodynamic driving force for folding. While the investigators did observe folding in a few cases, it was overall problematic. Often, as in the case of SELEX for aptamers and conventional protein engineering, creation and functional screening of large libraries are employed to identify suitable candidates with the desired activities that are capable of folding into the correct structure. Given that, even with preliminary high-end in silico screening (e.g., Rosetta), success rates are within the range of 5-10 percent (often far less), it may not be surprising that significant problems were encountered in making progress toward the specific aims of the research. The aforementioned process of developing a protein screen and generating a library is very labor- and resource-intensive and not necessarily a guarantee of success.
The development of tailorable and designable protein-based ligands based on robust structural motifs, such as the cystine knots described in this project, would offer the opportunity to integrate into other synthetic biology projects, especially in the nanocrystalline cellulose-based biomaterials that were described during the review. It also seems likely that these protein-based ligands could be integrated with other more conventional materials-based platforms, such as metal-organic frameworks (MOFs), which are under development at ECBC—especially in consideration of the structural and thermodynamic robustness of the cystine knot miniproteins. These considerations were not described in detail during the review, but they appear to be a significant motivation for the research, based on the associated commentary.
ECBC Crosscutting Findings
More interaction with the scientific community could improve some of these ECBC projects. In the case of the project Effect of Quorum Sensing Molecules on the Production of Bacterial Nanocellulose the work would benefit greatly from increased interactions outside of ECBC because this type of work is an active area of research in many biotechnology, engineering, and materials science laboratories. Much could be gained from scientific discussions and collaborations with a broader range of scientists. The project Effect of Toxicants on the Regulation of Endothelial Barrier Function could benefit from more interactions with scientists working in the general area of researching chemical warfare agents.
The project Epigenetic “Memory” During Bacterial Adaptation to Environmental Changes would also benefit from creating a context in which the PI can get an outside critique from microbial biologists actively studying heritable variation in bacteria. As conceived, this project has a narrow technical base and fails to embrace the power of alternative, less technically demanding strategies. Funds spent on external consultants would be money well spent. Additionally, this project has all the hallmarks of a project that was developed in isolation, without the benefit of the critical interactive discussion and planning among people skilled in this line of research. The PIs of all proposals need to seek an external vetting of their proposals. The outside review of the ideas and contemporary meetings they attend would help to place the work in a broader context.
Recommendation: ECBC should seek external vetting for its scientific proposals.
Recommendation: ECBC should promote interactions of its researchers with the broader scientific community in order to advance greater scientific understandings and enhance its projects.
For both ARDEC and ECBC input from the broader scientific community through discussions and collaborations would serve to progress their projects.
Recommendation: ARDEC and ECBC should promote interactions of their researchers with the broader scientific community in order to advance greater scientific understandings and enhance their projects.