Sharing malaria ideas: panning for vaccine leads and non-invasive diagnostics

I forgot to share these sooner.  These are the grant applications I submitted to the Gates Foundation’s Grand Challenges most recently in early 2009.  I suppose it was wishful thinking that an unaffiliated, dreamy scientific thinker like myself would have a shot at winning these grants, but I tried.  As I heard Vinod Khosla say once, probably quoting someone else, “try and fail, but don’t fail to try.”  (Here btw is an interesting and inspiring list of quotes about failure).  I definitely learned a lot and the experience helped me think about how innovations are “sourced” from independent, academic, business, nonprofit and public sectors in the US.  I share these pasted below (links to original files at bottom of post) in the hopes that others may find these ideas stimulating or perhaps useful to ultimately help those who suffer from malaria.  I also want to thank my friends and colleagues who gave me feedback on these ideas.  Please rely on the linked files as the best sources, as the formatting is a bit janky below.  I’m mostly including the text below to help with search engine results if anyone happens to be looking for this stuff!

Idea 1:  Identifying malaria vaccine leads via a screen for “immunologically relevant” antigens.

File: GCE_novel antigen screen – v7

Abbreviations: MHC, major histocompatibility class. DC, dendritic cell.

Section I. A novel liver-stage malaria vaccine

The problem

Why is there no 90%+ clinically effective p. falciparum malaria vaccine after decades of research1,2,21? There are 3 explanations: 1) the vaccine must be practical to manufacture, store and administer; 2) the vaccine needs to target the parasite during a prolonged stage of infection to give T cells time to respond; 3) the immune system needs priming in order to sensitively detect and eliminate otherwise difficult-to-detect malaria-infected cells.

To date, the only vaccine approach showing 90%+ efficacy in a human study was derived from whole parasites3,4. Unfortunately this raises multiple practical problems. The vaccine requires large doses of parasites to be reared within mosquitoes and surgically isolated by skilled technicians. The vaccine is administered monthly for up to 10 months. Because it involves preserving whole cell living organisms, the vaccine must be stored at -20°C or colder3,4. Instead of a whole parasite approach, it would be more practical to use established peptide antigen-based approaches that are used for most common vaccines. These are easy to administer once, manufacture at scale, and store in the field at 4°C. We decided that a vaccine should start from the premise of being peptide-based until we are confident that possible antigens have been exhausted. Fortunately there is plenty to explore: less than 5% of malaria proteins have been studied in any way; far fewer have been investigated as vaccine antigens.

Malaria is an intracellular infection and therefore needs to be cleared by the cytotoxic T cell aspect of immunity. But after the body is exposed to an infective antigen, it takes 2-3 days to select and proliferate enough T cells to mount an effective response6-10. At the same time, p. falciparum malaria migrates through different tissues and organs during its complex life cycle within the human host. The most prolonged phase of this life cycle is the liver stage, lasting 7-10 days. This is also the stage where malaria begins to proliferate after entering the host from a mosquito bite. Each sporozoite that survives from the mosquito bite results in ~105 merozoite offspring. Most of p. falciparum malaria symptoms are caused by the extremely high burden of parasites that clog micro vessels. Thus the liver stage is a most desirable stage for attacking malaria.

The third challenge results from the fact that the cytotoxic T cell response needs lymph node priming in order to effectively identify and purge infected liver cells. If there is no such priming, as the case for an unvaccinated patient, malaria can evade immune detection in the liver. However, when dendritic cells in lymph nodes are exposed to malaria antigens, they can have a potent effect on purging infected liver cells. While these approaches have met with some success in animal models, none have yet successfully translated into effective human vaccines. A new liver stage vaccine would need to specifically focus on its ability to prime skin-based dendritic cells after the mosquito bite. This in turn would provide the immune system enough time to create a cytotoxic T cell response targeted towards malaria after it had moved on to the liver.


The solution

We propose an antigen screen that takes the above three challenges into account. The approach aims to identify p. falciparum peptides that can be presented as antigens by dendritic cells after infection, and that are also detectable as presented by infected liver cells. Finding such peptides would allow creation of a vaccine that can prime a T cell response in the skin lymph nodes, then permit the activated T cells to recognize and purge infected liver cells. The approach is a novel application, based on existing biochemical and cell biological methods.

During the first phase of the screen, human DCs will be incubated with p. falciparum sporozoites in vitro. This step will mimic the interaction of sporozoites with skin lymph node DCs that occurs within hours of the mosquito bite6-7. These antigen presenting cells digest invading parasites and present peptide fragments on their surface via MHC proteins. In the lymph node, many T cells then evaluate these antigens and proliferate if there is a strong match for the antigens. Thus we first aim to characterize the most common antigens that DCs present via MHC. Established methods exist to isolate these peptides biochemically and then sequence them with mass spectrometry11,12.

In the second phase of the screen, cultured human hepatocytes will be incubated with malaria sporozoites to mimic liver infection. A proportion of these cells will digest the malaria parasites and present malaria peptides on their cell surface via their MHC proteins. Similar to the DC protocol, these peptides will be isolated and sequenced. In the final phase of the screen, peptide sequences will be compared between DC-presented peptides and hepatocyte-presented peptides. Any peptide sequences that are similar in the two groups will be become leads for further characterization as potential vaccine antigens.

The approach is unlike all other attempts to identify vaccine antigens. Several labs have used functional genetic screens to identify proteins to target with vaccines14-17. However, the limitation of these approaches is that they identify proteins important to p. falciparum’s function, but not specifically proteins that are recognized and attacked by the immune system after the parasites have infected hepatocytes. In other words, these approaches risk creating a vaccine towards an antigen that is invisible to the immune system. Other vaccine approaches have targeted malaria surface proteins. This is a logical place to start, but does not ensure that infected hepatocytes will be targeted by T cells for destruction. Instead, our approach will screen for the most immunologically relevant antigens and therefore aims to maximize the potency of a potential vaccine.

Potential issues with the solution

It is possible our approach may not identify any common antigens between lymph node DC and hepatocyte antigen presentation. We think this unlikely because in vivo only about 3 hours transpire between mosquito bite and liver infection5,8. In this brief window, protein expression is unlikely to change dramatically. Since hepatocytes express MHC1 and sometimes MHC2 receptors7, we would expect the pathogen protein presentation to be similar for MHC1-bound antigens in DCs and hepatocytes. Furthermore, several antigens are known to be processed by DCs and hepatocytes6,7.

Our approach may also benefit from focusing on different antigen-presenting cells. We think this unlikely based on evidence to date that indicates that DCs in skin lymph nodes and hepatocytes are the main APCs for pre-erythrocytic stages of malaria7. Cells that do not have a strong role include liver-based DCs and Kupffer cells7. However, an antigen-presenting role is still unknown for liver cells such as stellate cells or liver sinusoidal endothelial cells7. In the event our initial experiments do not identify any antigens, we would modify the protocol to study these other possible antigen-presenting cells.

A common concern with malaria vaccine design is antigenic variability of malaria strains. Our approach would address this in three ways. First the screen relies on antigens presented by the host, so a vaccine targeting these antigens will destroy infected host cells and not directly kill the parasites. Thus there will not be a direct selection pressure stimulating the parasites to mutate. Second, if multiple candidate antigens are identified from the screen, next steps would include creating a multivalent vaccine from these antigens. This could further reduce the probability of parasites developing mutation resistance to the vaccine. Last, we can filter out any identified antigen candidates with genes that encode known variability regions such as the var region.


Section II. Experimental design and methods

Aim 1 ($40K; 5 months): Characterize peptides presented by hepatocytes. In our screen, hepatocytes are more likely to present difficulties in identifying strong signals of MHC-presented malaria peptides because these cells are not “professional” antigen-presenting cells (as DCs are known to be). So we will focus first on hepatocytes. An IRB protocol would be drafted and approval obtained to collect otherwise discarded surplus cells from liver biopsies and bone marrow biopsies. Primary human hepatocytes would be isolated from liver biopsies and cultured per established methods. Cryopreserved p. falciparum sporozoites (MR4, Manassas, Virginia) would be thawed and then incubated with monolayers of hepatocytes for 16-24 h20. Hepatocytes would be harvested at 3, 6, and 9 days to capture the duration of the liver stage. Peptides bound to hepatocyte MHC proteins would be isolated using the direct biochemical method11,12. Positive controls would include measuring expression of known sporozoite antigens such as CSP and TRAP in the MHC protein-isolated fractions. Negative controls would include measuring expression of intracellular proteins bound to MHC proteins. Peptides would be sequenced with mass spectrometry. Measurable outcome: sequences of hepatocyte MHC-bound malaria peptides.


Aim 2 ($40K; 5 months): Characterize peptides presented by dendritic cells. This section would be approached similarly to the hepatocyte Aim 1 with the following exceptions. Bone marrow cells would be cultured from discarded bone marrow biopsies. Bone marrow progenitors would be cultured into DCs per Plebanski et al (2005). Cryopreserved p. falciparum sporozoites (MR4, Manassas, Virginia) would be thawed and then incubated with monolayers of DCs for 16-24 h20. Positive controls would include similar sporozoite antigens to Aim 1, while negative controls would include intracellular DC proteins that would not be expressed by MHC proteins. Measurable outcome: sequences of dendritic cell MHC-bound malaria peptides.


Aim 3 ($20K; 2 months): Analyze overlapping peptide sequences to identify candidate antigens. With peptides identified , we would next begin analyzing these peptides to identify overlaps. Pep-Miner software would be used to conduct the overlap analysis12. Overlapping peptide fragments would be compared to protein databases to identify the relevant malaria proteins. Measurable outcomes: identification of shared peptides. List of likely antigenic proteins for further characterization.


Translational value and next steps: Peptides identified from the screen would be evaluated for their ability to stimulate a T cell immune response in animal models, and thus become potential future vaccine candidates.



  1. Moran M et al. The Malaria Product Pipeline: Planning for the Future (2007)
  2. Richie TL et al, Nature 415:694 (2002)
  3. Hoffman et al., US Patents US2005/0208078, US2005/0220822
  4. Hoffman SL et al., J Inf Diseases, 185:1155 (2002)
  5. Yamauchi LM et al., Cellular Microbiology, 9:1215 (2007)
  6. Prudencio M et al., Nature Reviews Microbiology, 4:849 (2006)
  7. Frevert U et al. Cellular Microbiology 10:1956 (2008)
  8. Amino R et al., Nature Medicine 12:220 (2006)
  9. Frevert U et al., PLoS Biology, 3:e192 (2005)
  10. Amino R et al., Cell & Host Microbe, 3:88 (2008)
  11. Cox AL et al., Science 264:716 (1994)
  12. Shoshan SH et al., Pharmacogenomics 5:845 (2004)
  13. Aregawi M et al. The World Malaria Report, 2008 (WHO).
  14. Florens L et al. Nature 419: 520 (2002)
  15. Sam-Yellowe TY et al. Journal of Proteome Research 3: 995 (2004)
  16. LaCount DJ et al. Nature 438: 103 (2005)
  17. Wuchty S et al. Journal of Proteome Research 6: 1461 (2007)
  18. Mueller A-K et al., Nature, 433:164 (2005)
  19. van Dijk MR et al., PNAS, 102:12194 (2005)
  20. Plebanski M et al., Immunol Cell Biol, 83:307 (2005)
  21. Wykes M et al., International Journal for Parasitology, 27:705 (2007)
  22. Graves P et al, Vaccines for preventing malaria (pre-erythrocytic) (Cochrane Collection, 2008)


Idea 2: Identifying malarious scents for a future surveillance device.

File: GCE_Scent detector – v4 doc

Section I. What is your idea? Malarious humans release odors that attract mosquitoes. We propose identifying these molecules, as a first step towards designing an environmental malaria detector.

As malaria control efforts gain momentum and continue to be successful in some areas, there is a growing need for devices that can help detect outbreaks of malaria, thus helping governments strategically prioritize their finite resources for malaria control.

A recent study demonstrated that malarious patients harboring the transmissive form of the parasite release odors that can attract mosquitoes two times more often than other malarious patients or uninfected patients (1). A device that could “sniff” these molecules would permit surveillance of a village for malaria outbreaks. This would be particularly useful in areas with insufficient coverage of community health workers (CHWs).

Here we propose to identify the odor molecules, using an established clinical study method and gas chromatography (GC). This proof-of-principle work will be a springboard for designing inexpensive, point-of-use electronic “sniffer” devices to detect malaria outbreaks in resource-poor areas. The impact of such a surveillance device would be enormous. Several countries have reduced malaria cases significantly, including Rwanda, Eritrea, and Uganda. If this progress were to be reversed by a resurgence, several million additional children would get malaria each year (2).

The first non-invasive malaria detector, and the first detector for strategic surveillance.

Our approach would be innovative for several reasons. First, no research group is currently working on identifying these molecules. Koella and colleagues have moved on to other topics (3). Second, there is no precedent for a non-invasive malaria detector, which this project would ultimately aim to accomplish. The only other options for positive identification of malaria are blood smears and rapid diagnostic tests (4,5). Finally, no technology exists that would help a country maintain the necessary vigilance after malaria prevalence has declined.

In concept the putative detector would resemble a fire alarm — highly sensitive, always sampling the environment, and only emitting a warning signal when it detects a positive. In practice, the “alarm” may not be a sound, but rather an electronic signal to an epidemiology center. The detector would require minimal maintenance and staff attention. Thus a developing country could take an evidence-based approach in deciding where to allocate limited resources to control a new outbreak of malaria.

Possible issues with our approach

Mosquitoes are attracted to bite malarious humans by both intrinsic factors (e.g., carbon dioxide, sweat, warmth, skin moisture, and body odor) (6-8) and malaria-specific factors (1). One concern is that our proposed studies could mistakenly identify intrinsic factors. However, Koella et al demonstrated that mosquitoes are attracted two-fold more to gametocytemic patients than controls. This attraction returned to baseline intrinsic attraction after treatment with anti-malarial drugs (1). Thus we plan to replicate these methods, and survey gases from patients before and after treatment with antimalarial drugs, and focus only on molecules with higher concentrations before versus after antimalarial treatment.

The technology would only identify patients with gametocytemia. These represent approximately 52% to 66% of patients, based on a 19-month longitudinal study of children in Kenya (9). One concern could be that a future technology would not be detecting all parasitemic patients (gametocytemic or not). However, our detector would not be used at the point of care, the way a rapid diagnostic test would. Instead, it would be designed to identify patients harboring the transmissible form of the parasite. Thus only identifying the gametocytemic patients is desirable, because only these patients represent transmission, and would be epidemiologically significant harbingers of an outbreak. Of course, any symptomatic patient would be treated by local health workers, regardless of the results of our scent detector.

Another concern is with detection sensitivity. GC can detect samples at a parts per billion concentration. While it may sound difficult to isolate these molecules, precedent exists even in commercial devices, for detecting and characterizing odors at the ppb range in air. More broadly, there is a precedent for commercially available detectors for gaseous molecules. These rely on a range of technologies, from relatively inexpensive detectors based on chemical reactions with substrates (e.g., Breathalyzer tests), to more expensive MS-based units (e.g., GE’s EntryScan for airport security). Our focus for this grant is identifying molecules for detection. After doing so, we would then further optimize a device to affordably detect these molecules. Other options could include electronic micro-sniffers currently in development (10).

Section II. How will you test it?

The experimental plan will use established methods: patient “olfactometer” to collect gaseous samples, GC-MS, and leverage previous clinical protocols and relationships with the Kenyan government.

Aim 1 ($20K). Establish collaboration and clinical protocols. Establish collaboration with Koella and Mukabana to extend findings from Kenya study, using olfactometers to capture gas eluants from asymptomatic patients with confirmed parasitemia or gametocytemia (1,6,11). Draft clinical research and ethics protocols, for review by collaborators and approval from Kenyan National Ethic Review Committee. Design study to recruit 10 groups of 3 children each, to identify asymptomatic patients who fall into one of 3 groups: 1) asymptomatic, 2) parasitemic but not gametocytemic, and 3) gametocytemic. Methods would follow those of LaCroix et al (2005), including immediately taking any symptomatic patients to the local health clinic for treatment. Parasitemia would be verified via thick blood films, prepared by and analyzed by our group.

Methods from LaCroix et al would be modified as follows: at sundown, when patients are monitored while sleeping in tents connected to the olfactometer, effluent gas from each tent would be collected hourly into vacuum tubes, for later analysis by GC-MS. In the beginning of the study, pilot GC-MS experiments will be conducted to verify that some molecules can be detected. Positive controls be tested by exposing a small vial of nail polish remover (acetone) for 10 seconds at sundown, and verifying identification via GC-MS. Asymptomatic patients would be analyzed before any treatment, and then 2 weeks following treatment with sulfadoxine-pyrimethamine, also per methods of LaCroix et al (2005).

In the event collaboration with the Kenyan government fails, we will leverage relationships in Ghana or Rwanda, as backup locations.

*Measurable outcomes: Completed clinical research protocol. Collaboration action plan for study activities and endpoints. Approved clinical research protocol. Kenyan site identified and relationships established with health workers and regional authorities.

*Expected time: 4-6 months.

Aim 2 ($80K). Conduct clinical study and identify odor molecules. The study would be conducted as described above. The focus of Aim 2 would be on conducting the clinical trial, analyzing samples with GC-MS, and identifying molecules based on MS profiles.

In the event that GC-MS samples are not stable during the transport each morning from research site to urban research lab, we will explore alternatives, such as transporting the asymptomatic malarious patients to an urban hospital, housing them in vacant beds, and conducting the GC-MS in real time by the patient’s bedsides. In the event the GC samples are too impure, at this stage we would experiment with column modifications, to selectively look at different aspects of the gaseous mixture. For instance, we could use a cationic filter to remove all anionic materials, then analyze the remaining cationic and neutral compounds. In the event the odors are not concentrated enough for GC detection, we would generate respiratory and other body odor concentrates (12).

*Measurable outcomes: Completed study with balance of patients in each study group. Identified molecules based on GC-MS.

*Expected time: 6-9 months.

Translational value and next steps: Identified scent compounds. Next steps would be engineering an inexpensive, sensitive detector device and conducting field tests.

Abbreviations: GC, gas chromatography. GC-MS, gas chromatography and mass spectrometry. CHW, community health worker.


  1. Lacroix R et al., PLoS Biology, 3:e298 (2005)
  2. Aregawi M et al., The World Malaria Report, 2008 (WHO)
  3. Jacob Koella, personal communication, 10/14/08.
  4. Moody A, Clinical Microbiology Reviews, 15:66 (2002)
  6. Mukabana WR et al., Malaria Journal, 3:1 (2004)
  7. Mboera LEG et al., Medical and Veterinary Entymology, 14:257 (2000)
  8. Takken W et al., Annu. Rev. Entomol. 44:131 (1999)
  9. Jones TR et al., Am. J. Trop. Med. Hyg., 56: 133 (1997)
  10. Raman B et al., Anal. Chem, ePub 10/15/08
  11. Mukabana WR et al., Malaria Journal, 1:17 (2002)
  12. Ghaninia M et al., J Exp Biology 211:3020 (2008)




  1. […] laboratory researcher, clinical researcher, freelance research scientist (see grant attempts here),  multi-industry consultant (with McKinsey and freelance), and a few […]


  2. […] laboratory researcher, clinical researcher, freelance research scientist (see grant attempts here),  multi-industry consultant (with McKinsey and freelance), and a few […]


  3. […] of thinking I’d become a full-time, independent research scientist (e.g., examples of grants for diagnosis and treatment of malaria I drafted here), I later found a job at a health care company that was full time, but less intense, and so I had […]


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: