Skip to main content Skip to secondary navigation


Main content start


Targeted Drug and Nucleic Acid Delivery

Introduction. Targeting therapies only to specific cells in our bodies -- and thereby increasing efficacy and avoiding side effects – is an exciting idea first proposed by Erlich in 1900. Strikingly, a 2016 review revealed that the average delivery efficiency for more than 200 attempts was less than 1%. The over-riding reason for these failures is that many distinct challenges (Figure 1) must be simultaneously addressed. The Swartz lab is now building upon more than ten years of their own design and testing experience to produce a “smart” virus-like particle that answers all of these challenges.

Our approach. The initial disease focus is end stage prostate cancer using the heavily over-expressed PSMA cell surface marker as our target. The vehicle is produced as shown in Figure 2. The VLP (virus-like particle) is composed of 240 copies of a single multiply-mutated protein. The protein immediately forms the dimers shown, and 120 of these are triggered, simply by adding salt, to spontane-ously assemble into hollow spherical particles filled with toxic cargo (panel A). After assembly, the nanoparticles are stabilized by stimulating disulfide bond formation between the VLP subunits. The shell then becomes a single covalently stabilized molecule. These bonds will dissolve inside the targeted cells to release the toxin and kill the cancer cells.

The VLPs display 120 protruding surface spikes; each decorated with uniquely reactive non-natural amino acids that facilitate precise surface modification (panel B). An anti-PSMA scFv antibody fragment is attached so the particles will bind to the cancer cells with high avidity and will also trigger internalization into the same cells. However, a major reason for failure of the hundreds of attempts described above is that foreign nanoparticles are recognized as invaders (or as circulating garbage) and are efficiently removed by immune system cells in the liver, spleen and lymph nodes. To study this, we load the VLPs with near infrared fluorophores which avoid interference by hemoglobin light absorbance.

Whole body imaging of mice (Figure 3) can then visualize the VLPs showing that VLPs without surface modification are rapidly sequestered in the liver and spleen. To avoid this, we are pioneering the production and attachment of the extracellular domain (ECD) of the CD47 receptor that tells immune cells “don’t eat me”. Preliminary experiments are promising

Progress and plans. Feasibility has now been demonstrated for the major design features, and we are working to improve production methods before carefully studying how the type and number of each VLP surface modification affects nanoparticle pharmacokinetics. We will then optimize the delivery vehicle design with an objective of delivering at least 90% of the injected (or infused) toxin (mertansine) to the tumors for effective cancer destruction.

Precisely Designed Vaccines for Stopping Pandemics and General Applications

Introduction. The COVID-19 global pandemic has been extremely costly in lives lost, physical suffering, and economic disruption. This is truly a global challenge, and we are learning that the most effective deterrent is almost certainly a widely distributed, efficacious vaccine. While anti-COVID progress is now encouraging in the US, much of the world is still suffering; and, unfortunately, a new pandemic threat could emerge at any time. We are therefore working toward a new vaccine technology to address this critical dilemma.

This is a significant challenge because of the speed and magnitude of the required response. Our best chance for avoiding global pandemic emergence is rapid global vaccination. We have therefore designed a program with the objective of providing six billion doses of a safe and effective vaccine within two months of discovering the new pandemic pathogen. Such a response requires highly productive as well as inexpensive production technologies. Acceptable costs can be achieved only with a combination of low recurring expenses as well as low capital costs. In other words, inexpensive raw materials must be efficiently converted into vaccines by very highly productive facilities. To put this in perspective, our objective is to increase productivity and to reduce costs by at least 1oo-fold relative to current technologies. We believe this can only be accomplished by synergistically combining novel vaccine designs with novel production technologies.

Vaccine Design. Because our immune systems have evolved to fight viral invaders, our new vaccine design mimics a natural virus. It is composed a virus-like particle (VLP) scaffolding composed of 240 copies of a single protein. On its surface will be attached a protein antigen capable of eliciting protective responses and one or more adjuvant molecules to both stimulate and direct a potent response. The general design is indicated by the cutaway views shown in Figure 4.

Also indicated is a major advantage of this design. Most of the vaccine nanoparticle can be produced in advance of the pandemic threat and stockpiled. This will help tremendously in rapidly producing the vaccine when needed.

Proposed Pandemic Suppression. When a potential pandemic-causing pathogen emerges, the probability is high that enough will be known about the virus to identify a surface antigen suitable for an effective vaccine. To produce the vaccine, we will simply need to produce that antigen and attach it to the stockpiled vaccine precursor that is stored away. Cell-free protein synthesis (CFPS) methods allow us to test thousands of different antigen production protocols in parallel to identify an optimal production process in only a few days. Using dried or frozen cell extracts that have also been stored, commercial scale production would then begin only about 10 days after antigen identification. Five hundred liter CFPS bioreactors could then produce about 500 million doses a week at very low cost.

Progress and plans. Initial feasibility for such a vaccine has been demonstrated in mice using a lymphoma cancer model. High levels of neutralizing antibodies were elicited, and the antibodies protected against an otherwise lethal cancer challenge. Going forward, we plan to further optimize the vaccine design in mice while testing vaccine candidates for at least three different diseases. We will also demonstrate efficient protein production, vaccine assembly, and process scale-up. The program is further described in this article: .

Developing “Liquid Biopsies” to enumerate and obtain Circulating Tumor Cells (CTCs)

Introduction. Metastatic cancers are extremely difficult to defeat because they continually send out new “seeds” in the blood stream. Although concrete correlations are still lacking, evidence suggests that the number of tumor cells in circulation (CTCs) can indicate the severity of the disease, and, importantly, the effectiveness of the patient’s current treatment regimen. However, oncologists still lack an effective assay that is convenient and accurate. 

Assay Design and Demonstration. We have been developing a simple and fast CTC assay that builds on our VLP technology. Figure 5 shows how such an assay could be conducted by an automated device in an oncologist’s office using a freshly obtained blood sample. The assay uses the multi-functional nanoparticles shown in Figure 6. The same VLPs described above (Figure 2) will be surface modified with: 1) a scFv antibody fragment that recognizes an over-expressed surface marker on the cancer cells, and 2) a luciferase that emits a bright light after a chemical called coelenterazine is added. Figure 7 shows a portion of an image obtained from an assay prototype using an off-the-shelf camera. One hundred prostate cancer cells were mixed with the nanoparticles, applied to the filter, washed, and activated with coelenterazine. All 100 cells were detected as bright blue spots in the image. Early experiments in which cancer cells were mixed with blood samples were also promising.

Plans. Next steps for the project are to optimize the degree of VLP surface modification, design and test device prototypes, prepare image analysis software, and seek a licensee for the technology (already protected by an issued patent). In parallel, methods will be developed for isolating viable tumor cells so they can be analyzed for genetic features, cell surface marker profiles, and sensitivity to a variety of treatment approaches.



Production and Evolution of Improved [FeFe]-Hydrogenases

Introduction. Essentially all of our efforts toward improving the health of our planet have been dependent upon the spectacular enzyme depicted in Figure 8. It turns out that all biological organisms must controll the metabolic exchange of subatomic particles; i.e., electrons and protons. For example, removing accumulated carbon dioxide from our atmosphere and converting it into useful molecules requires multiple electrons and protons. The metabolic processes in our bodies also require the precisely controlled transfer of electrons and protons. It is therefore of fundamental importance that the [FeFe]-hydrogenase (H2ase) shown in Figure 8 in an example of the only class of enzyme that converts these subatomic particles into a molecule, hydrogen (H2), that can be used to transfer and to store metabolic energy. Of  course, it is also highly relevant that hydrogen can be stored and used as a non-metabolic fuel when needed; for example to generate electricity or propel automobiles as well as large trucks and buses.

The enzyme depicted in Figure 8 derives from the anerobic bacterium, Clostridium pasteurianum, and is consequently named CpI. To be functional, it must be loaded with a series of Fe-S centers that are shown on the left for improved clarity as well as in the protein structure diagram. The lower four centers accept electrons from a small protein called ferredoxin and conduct them toward the H-cluster active site (the uppermost two centers). PDC stands for proximal delivery center, and PCC indicates the proximal catalytic center. The top center is where the reaction occurs. It consists of an assembly involving two Fe atoms, 2 cyanides, 3 carbon monoxides, and a unique molecule (S-C-N-C-S) that bridges between the two iron atoms. Protons diffuse into the enzyme through a channel near the top of the enzyme (as depicted in Figure 8) where they are reduced by the electrons to form hydrogen (H2) molecules.

Developing knowledge and methods for [FeFe]-hydrogenase production. We began our H2ase work in 2003. Fortunately, a 2004 publication revealed that CpI activation depends on the HydG, HydE, and HydF maturase proteins that work together to assemble and install the active site 2Fe center (the top cluster in Figure 8). Using a cell extract in which the three maturases had been overexpressed, cell-free synthesis and activation of CpI was then demonstrated. This became an important tool for producing and screening thousands of hydrogenase mutants in searching for improved functional properties.

To gain more knowledge about the critical functions of the muturases and to discover small molecules required for maturation, we also developed an in vitro CpI activation system. First we obtained an E.coli strain engineered to overexpress the isc operon that builds Fe-S clusters. This was then used to develop a process for producing the CpI apoenzyme, an inactive form of the enzyme that has most of the Fe-S clusters but not the active site 2Fe cluster. Next three cell extracts, each containing one of the maturase proteins, were mixed with the apoenzyme and various small molecule substates for in vitro CpI activation. We discovered that pyridoxal phosphate was required for activation in addition to the more obvious Fe and S sources. This system and FTIR spectroscopy conducted at UC Davis enabled the confirmation that all of the the cyanides and carbon dioxides were derived from tyrosine. When this new knowledge was coupled with the new protocols, tyrosines with mass labels at specific sites on the molecule were used to discover previously unknown biochemical reactions that build the active site 2Fe cluster. This work resulted in two Science papers and several reports in PNAS and JACS. The new knowledge also improved cell-free CpI synthesis and activation.

Importantly, we also developed and published a new E.coli process for producing larger quantities of [FeFe]-hydrogenases. This new procedure increased active yields by about 30-fold relative to previous procedures and has become the standard production protocol for [FeFe]-hydrogenase production. There are two main types of [FeFe]-hydrogenases, the one shown in Figure 8 and a smaller enzyme that lacks the electron conducting Fe-S centers. Both types are efficiently produced by our published process.

Evaluating and Improving H2ase Functionality. Even though the [FeFe]-hydrogenases offer extremely fast H2 production rates, they are very rapidly inactivated by oxygen (O2). This an especially serious concern for photosynthetic H2 production since photosynthetic water splitting (a process that provides the required electrons and protons) releases O2 as a required side product. O2 sensitivity is also a concern if H2 is to be used for delivering electrons to supercharge the production of commodity biochemicals from atomospheric CO2 (see below). We therefore set out to search for O2 tolerance [FeFe]-hydrogenases.

We began our search with the smaller H2ase, producing and evaluating about 10,000 candidates using our cell-free synthesis production and evaluation platform. The search turned up enzymes with higher activity, but none of the mutants had improved O2 tolerance. We concluded that the larger, more complex enzyme (Figure 8) was required. Using intensive mutageneis, we then identified isolates that appeared much improved. However, for convenience, we had searched for retention of H2 utilization activity after O2 exposure. Since H2ase are bidirectional catalysts, we assumed that this tolerance would translate into tolerance also for enzymes exposed to O2 when working to produce H2. This turned out to incorrect. The “improved” mutants were just as sensitive as the wild type enzyme under the conditions that really mattered.

Screening for the required CpI functionality required the development of a new assay that measures O2 tolerance during hydrogen production.

Figure 9 shows the electron supply pathway used to fuel hydrogen production. It was designed to provide a constant supply of NADPH and reduced Ferredoxin. Figure 10A shows that the hydrogen production rate was first measured without O2 to assess initial CpI activity, and 5% O2 by volume was then added to the reaction headspace to measure the degree of O2 sensitivity. Semilog plots of the slopes of H2 concentration in the headspace over intervals B to E indicated the first order rate constant for activity loss. Panel B shows good agreement over a 20-fold CpI concentration range. This new assay was critical for discovering CpI mutants with reduced O2 sensitivity.

This assay was further streamlined and used to screen over 300 site directed CpI single and double mutants. Even though the previous mutants were more oxygen tolerant only for hydrogen utilization, the locations of those mutations provided useful guidance. A double mutant of CpI was discovered that decreased the first order oxygen-induced inactivation rate constant to about 30% of its native value. While more complete oxygen tolerance would still be desirable, this mutant still appears suitable for use in photobiological hydrogen production. Importantly, these mutations appear to provide further guidance for improving oxygen tolerance. The mutations are located between the PDC Fe-S cluster and the electron delivery chain of Fe-S clusters and confer increased local hydrophobicity. In future work, we plan to explore the resultant structure-function hypothesis by futher increasing hydrophobicity in that region and also between other Fe-S clusters in the electron conductance pathway.


Hydrogen Production from Biomass-Derived Sugars

One of the early projects for biological hydrogen production focused on the use of biomass-derived sugars such as glucose, xylose, and arabinose. The general metabolic scheme is shown in Figure 11 for glucose utilization and suggests that 11 H2’s could be produced from each glucose. If feasible, crop wastes could provide the key raw material (H2) needed to remove N2 from the atmosphere and convert it into the nitrogen fertilizer needed for the following year’s crops.

While general feasibility for this scheme was demonstrated, conversion yields and produc-tivities were much lower than expected. In a recent opinion article, Professor Swartz suggests that this was caused by a rate limitation imposed by CpI. Future work will search for CpI mutants that support much higher productivities and conversion rates.


Photosynthetic Hydrogen Production

Introduction. Cost effective photosyn-thetic hydrogen production is admittedly a very ambitious objective. Like many of the projects described here, success will require many coordinated advances. It is therefore essential that all of the limitations and opportunities be identified in advance so the “deal breaker” obstacles can be overcome to provide overall project feasibility.

Bioreactor and organism design. The basic bioreactor and cellular metabolism schemes are diagrammed in Figure 12. The bioreactor must contain the photosynthetic cell suspension and absorb solar irradiation while removing the product gases, O2 and H2, and avoiding temperature rises that could damage the culture. The photosynthetic bacterium must be engineered for broad spectrum light absorbance and for efficient delivery of the electrons and protons (produced from water splitting) to the hydrogenase.

The contrast between normal photosynthetic metabolism and our target is illustrated in Figure 13. Normally, 9 ATP’s and 6 NADPH’s (at a minimum) are needed to convert 3 CO2’s into one 3-carbon biomolecule. In addition, the NADPH/NADP+ ratio, i.e., the thermodynamic potential must be kept high. To satisfy these need, at least 4 photons are required to split water and several more are required to provide high electrochemical potentials.

These thermodynamic driving forces are needed to pump protons out of the cytoplasm to form the proton gradient that drives the aggressive ATP production needed for carbon fixation and protein synthesis. Additional photons (captured by PSI) are also needed to provide the required NADPH electrochemical potential. These energy intensive metabolic processes would be used to grow the cells. However, during photosynthetic H2 production the cells would be deprived of CO2 to avoid carbon fixation and growth. Photons will then only be needed for water splitting so that hydrogen can be produced with much higher solar efficiencies (assuming the hydrogen partial pressure is kept low by active harvesting). If necessary, an “uncoupling protein” will be synthesized to leak the protons derived from water splitting into the cytoplasm. If structures called phycobillisomes are also produced so that the entire visible spectrum is absorbed for water splitting, unprecedented photosynthetic efficiency is expected.

Overall system design. For this project, biotechnology alone will not be sufficient. The tremendous advantage offered by hydrogen production is the ability to store energy. Hydrogen could be produced along with photovoltaic electricity on sunny days and stored. When solar irradiation is not sufficient for photovoltaic electricity production, the stored H2 (and O2) would be used to satisfy electricity demands.H2 is also being adopted as a potential fuel for buses and trucks. Figure 14 illustrates how the photosynthetic H2 reactor could be integrated with a gas harvesting and storage system. 

As suggested above, it will be important to maintain a low H2 partial pressure in the bioreactor. This will minimize the thermodynamic electron potential required for H2 synthesis. The same measure will maintain a low O2 partial pressure to minimize damage to the hydrogenase and photosystems. This low pressure can be provided by cooling the exhaust gases in conjunction with a compressor that removes the harvested gases from the bioreactor. The heat will be transferred to large underground thermal reservoirs. The gases will then be further cooled until the O2 is liquified and removed to storage. Then the H­2 will also be liquified and stored.

When solar irradiation is low, the stored gases can then be used in fuel cells to generate electricity. Vaporization and expansion of the stored H2 and O2 will absorb the heat that was removed previously to restore the low temperatures of the thermal reservoirs. Because of thermodynamic inefficiencies and heat absorbed by the solar bioreactor, some excess heat will need to be removed by other means. To minimize the need for refrigeration, we propose that the solar collector can be used at night to radiate heat into outer space.

Initial demonstration of H2 production. As explained above, one of the major “deal breakers” is finding a prolific H2ase that is not inactivated by O2. The system described would minimize inactivation by lowering pO2 in the bioreactor, but H2ase durability is still a concern. To test performance of our improved CpI mutant, we established the in vitro H2 photosynthesis system diagramed in Figure 15A.

The PSI photosystem and plastocyanin were purified from a photosynthetic bacterium. Using electrons donated by ascorbate, light irradiation drove hydrogen production catalyzed by either the native CpI (WT=wild type) or the CpI mutated to reduce O2 sensitivity. The decrease in H2 production rate was measured and used to calculate the first order inactivation rate constant. Previous experiments had indicated that the anaerobic H2 production rate was stable over 80 minutes.

The results confirmed that the mutated H2ase was much more stable in the presence of O2. The measured inactivation rate constants are shown along with what percent of the anaerobic hydrogen accumulation was observed in the presence of either 2.5 or 5% O2. These results validated that the assay that we had developed to measure O2 sensitivity was predictive for photosynthetic hydrogen production. The observations also suggest that the mutated CpI would retain about 30% of its activity after a 10 hr solar exposure (with 2.5% O2) while the native enzyme would retain only 1.5% activity. This new H2ase stability is a tremendous advantage. Of course, this is only a first step, but we suggest that it is a very important one.

Future work will focus on further mutating CpI for even better oxygen tolerance and also to better facilitate acceptance of electrons with lower thermodynamic potential. In parallel, it will be important to screen and reengineer photosynthetic bacteria for efficient photosynthetic growth and effective transition to photosynthetic hydrogen production. Additionally, we believe that we can develop nocturnal procedures that will repair the H2ase that was damaged during solar irradiation.


Carbon Negative Production of Commodity Chemicals

Introduction. The modern chemical industry does an amazing job of converting fossilized vegetative matter into products that have become essential for modern lifestyles. Currently, that industry releases significant quantities of CO2 into the atmosphere. We propose to establish innovative technology to reverse this detrimental consequence. The new biotechnology will both directly and indirectly pull CO2 from the atmosphere to convert it into new materials for producing durable goods. Overall, this can contribute significantly toward sustainable mitigation of global warming.

The transformational advance is efficient transfer of electrons from renewable electricity into biomolecular synthesis using cell-free bioprocessing. However, as with the other projects, economically attractive bioprocess technology requires the coordinated development of several synergistic advances. Recently, ARPA-E (Department of Energy) has agreed to fund a demonstration project designed to establish the feasibility for establishing multiple biochemical production plants throughout the midwestern corn belt. CO2 will be drawn from the exhaust gases of bioethanol facilities. Initially, glucose will also be used as a raw material since it is available from the corn starch produced locally and contains carbon fixed from the atmosphere by the corn plants. However, this foundational technology, designed to produce succinic acid, will then allow the transition to technologies using CO2 as the only carbon source. We envision a new chemical industry that will contribute significantly toward stopping global warming while also providing thousands of good jobs in the corn belt.

Process Concepts. The overall process is diagramed in Figure 16. The heart of the system is a cell-free reactor with catalysis provided by a crude cell extract. The extract will be produced from an engineered E.coli culture with overexpressed pathway proteins. During preparation, the extract will be triggered to inactivate enzymes catalyzing side reactions that would otherwise lower conversion yields.

The extract will then be augmented by adding separately produced hydrogenase, FNR, and ferredoxin to catalyze the reverse of the pathway reactions shown in Figure 9. These enzymes will harvest electrons from H2 provided by an electrolyzer and transfer them to NADP+ or NAD+ as needed. These extra electrons will allow two CO2 molecules and one glucose molecule to be converted into two molecules of succinate. This provides much higher glucose conversion yields and fixes CO2 from bioethanol offgas into a biochemical that can potentially find its way into a plastic or other useful product.

The cell-free approach allows on-line monitoring of reaction intermediates so that glucose and CO2 feeding and other adjustments can be made to support very high and sustained volumetric productivities. However, this performance will require continuous product removal. This is indicated in Figure 16 by the ultrafiltration (UF) unit that will retain enzymes in the reactor while removing the aqueous continuous phase containing accumulated succinate, enzyme co-factors, and reaction intermediates. Since the succinate concentration will dominate, it can readily and specifically be removed by an anion exchange capture column. The recycle fluid will then be returned to the reactor after flowing through a chamber (pink color) designed to kill possible microbial contamination and to condition the fluid to maintain catalytic reactivity. When the succinate capture column is saturated, the recycle flow will be diverted to a second capture column and the succinic acid product will be harvested from the first column to regenerate it.

A preliminary technoeconomic analysis suggests that this process can provide excellent profitability as well as an attractive return on investment. Professor Swartz and four collaborators will soon begin the project aided by ARPA-E guidance and funding.


Improving productivity and product diversity for CFPS

Increasing productivity. Cell-free protein synthesis is now being used for producing a wide variety of proteins. Platforms are now available based on human cell extracts and plant cell extracts as well as bacterial. Further, significant progress has been reported by the DeLisa and Jewett labs toward producing glycosylated proteins. While some increases in system volumetric productivity have also been reported, we believe that a significant opportunity for major improvements still exists. This belief is supported by several observations from our lab. The most significant resulted from using data obtained by examining the impact of individually overexpressing nearly 40oo E.coli genes on the expression and activation of four different enzymes. A few of the most beneficial proteins were added at higher concentrations and the genes encoding four of the most detrimental proteins were deleted. When the most influential of these changes were combined with increased amino acid concentrations and pH stabilization, approximately 4 fold increases in product accumulation were achieved. Total protein accumulation as high as 5 mg/ml was reached in a 6 hr reaction period. Unfortunately, due to the complexity of this very large data set, these results were not published.

In future work, we plan to confirm and improve upon these observations. Opportunities exist in improving: energy supply, nucleotide triphosphate supply, mRNA stability, translation initiation rate, translation elongation rate, improved protein folding (discussed in more detail below), and in substrate concentration and biochemical environment stabilization.

Improving protein folding and solubility. While CFPS systems are sometimes called transcription/translation systems, the most demanding challenge is in folding the protein properly. This is particularly true for more complicated proteins of value as pharmaceuticals and also for complex enzymes, especially when cofactors must be assembled and installed and when disulfide bonds must be correctly formed. For these targets, a variety of chaperones and other helper proteins may be advantageous. To address these needs, we are developing a Design of Experiment approach to rapidly discover the individual and synergistic contributions of a variety of folding aids. In addition, we have observed significant advantages from adjusting the biophysical environment relative to, for example, the –SH/S-S redox potential and the ionic strength of the CFPS solution. Specific benefits can also be provided by adjusting concentrations of required cofactors that must be installed to activate enzymes. The access provided by the cell-free approach opens many opportunities to efficiently produce proteins that are difficult to activate during in vivo expression.

Increasing production duration. While improvements in volumetric productivity can be advantageous, additional productivity gains and cost reductions can be gained by extending the productive duration of the CFPS reactions. Past experiments have indicated significant gains from the periodic addition of additional amino acids, nucleotides, energy source molecules, and Mg+2. More recent experiments have focused on maintaining the activities of metabolic components required for sustained energy supply.