"Long before it's in the papers"
January 28, 2015


Computers help churn out new cancer remedies

Sept. 29, 2006
Special to World Science  

In what they’re call­ing one of the most ex­cit­ing can­cer re­search de­vel­op­ments in years, sci­en­tists are de­vel­op­ing ways to make com­put­ers churn out new can­cer re­me­dies—with no need for an­y­one to even know how they work.

An atomic-level mod­el of part of the hu­man an­dro­gen re­cep­tor, a mol­e­cule in the bo­dy linked to pros­tate can­cer.  Many pros­tate can­cer drugs work by di­rect­ly or indi­rect­ly block­ing the ac­tiv­i­ty of this mol­e­cule, which al­lows hor­mones called an­dro­gens to cir­cu­late. (Cour­te­sy Uni­ver­si­ty of California-San Fran­cis­co)

Called chem­i­cal ge­no­mic screen­ing, the tech­nique is de­signed to by­pass the hard, some­times fu­tile work of try­ing to learn pre­cise­ly what goes wrong in a spe­cif­ic can­cer in or­der to fix it. 

The tech­nique ex­ploits the fact that an or­gan­ism’s state at any time de­pends not on­ly on its genes, but on which genes are ac­tive, since a gene can al­so lie dor­mant, unused. Par­tial ac­ti­va­tion is al­so pos­si­ble. 

With­in a cell, the ac­ti­va­tion sit­u­a­tion at a giv­en time re­sults in a dis­tinct pro­file, which ex­ist­ing tech­nolo­gies can re­c­ord.

In the new tech­nique, re­search­ers feed in­to a com­put­er an ac­ti­va­tion pro­file linked to a par­ti­c­u­lar form of can­cer. The ma­chine then checks this against a da­ta­base of known drugs, which con­tains pre­vi­ous­ly known in­for­ma­tion on how each drug changes gene ac­ti­va­tion pat­terns. 

Fi­nal­ly, the com­put­er lists which of these com­pounds tend most strongly to con­vert the “sick” pro­file, which had been fed in­to it, in­to a pro­file known to be as­so­ci­at­ed with a health­i­er state. By fix­ing the pro­file, sci­en­tists rea­son, the drug may help rem­e­dy the un­der­ly­ing prob­lem. All this can oc­cur with lit­tle or no know­l­edge of the ma­la­dy’s causes. 

Re­search­ers stress that they’re not giv­ing up on learn­ing caus­es—in­deed, this could en­hance the re­sults—but in the mean­time, short­cuts to new treat­ments could br­ing des­per­ately need­ed re­lief to mil­lions.

The tech­nique “promises to sig­nif­i­cantly en­hance the drug dis­cov­ery pro­cess,” wrote Har­vard Med­i­cal School’s Scott A. Arm­strong and col­leagues in a pa­per de­scrib­ing some of the new find­ings, in the Sept. 28 on­line is­sue of the re­search jour­nal Can­cer Cell.

But re­search­ers al­so cau­tioned that the tech­nol­o­gy, still at an ear­ly stage, is­n’t clear­ly ca­pa­ble of pro­vid­ing cures. For now, it’s geared to­ward help­ing to con­vert par­tic­u­larly vir­u­lent forms of can­cer in­to more man­age­a­ble ones, mak­ing them bet­ter treat­able by ex­ist­ing reme­dies. These could be ad­min­is­tered along­side the new­ly found treat­ment.

In their pa­per, Arm­strong and col­leagues de­scribed work with vic­tims of child­hood acute lym­pho­blas­tic leu­ke­mi­a, a can­cer of the blood and bone mar­row. Sci­en­tists had pre­vi­ously found that a sub­set of these chil­dren have a par­tic­u­larly poor prog­no­sis. This is as­so­ci­at­ed with a weak re­sponse to a com­mon first-line treat­ment, the hor­mone glu­co­cor­ti­coid.

Arm­strong’s team found that in this “glu­co­cor­ti­coid-resistant” group, can­cerous cells tend to have a spe­cif­ic pro­file of gene ac­ti­va­tion, or gene ex­pres­sion, as it’s tech­ni­cal­ly called. In the oth­er vic­tims, the pro­file is dif­fer­ent.

The re­search­ers ap­plied the new tech­nol­o­gy to­ward con­verting the glu­co­cor­ti­coid-resistant cells in­to non-resistant cells. They fed the ex­pres­sion pro­files of both types in­to a database of 453 known, genome-wide ex­pres­sion pro­files re­sult­ing from treat­ment of var­i­ous cell types with dif­fer­ent drugs.

The database, called The Con­nec­tiv­i­ty Map, re­vealed that an ex­ist­ing drug called ra­pamycin should re­verse the bad pro­file, re­search­ers said. 

In­ves­ti­gat­ing fur­ther, they found that ra­pamycin af­fects mol­e­cules linked to a pro­cess that leads to a form of cel­lu­lar “sui­cide.” This would be a log­i­cal point for a can­cer drug to work at, be­cause this “sui­cide” is im­pli­cat­ed in the ail­ment. Nat­u­ral­ly, healthy cells kill them­selves when they start to be­come ma­lig­nant, pro­tect­ing the body against can­cer. Full-blown dis­ease oc­curs when this su­i­cide sys­tem, called apop­to­sis, fails.

In sum, ra­pamycin and glu­co­cor­ti­coid to­geth­er may be a use­ful treat­ment, the re­search­ers said. Ra­pamycin is cur­rent­ly used to pre­vent the body from re­ject­ing or­gan trans­plants.

In a sep­a­rate pa­per pub­lished in the same is­sue of the jour­nal, the med­i­cal school’s Todd R. Golub and col­leagues took an anal­o­gous ap­proach to pros­tate can­cer. 

Hor­mone treat­ments of­ten pro­vide in­i­tial suc­cess in fight­ing this ill­ness. The me­di­cines work by block­ing a mol­e­cule, called a re­cep­tor, that al­lows hor­mones called an­dro­gens to cir­cu­late. But the drugs tend to stop work­ing eventually be­cause tu­mor cells evolve a re­sist­ance to them, and an­dro­gen cir­cu­la­tion re­vives. 

How this hap­pens is un­clear; but one help­ful fact is that the high- and low-an­dro­gen states have dif­fer­ent ex­pres­sion pro­files, the re­search­ers said. Again us­ing the Con­nec­tiv­i­ty Map, they iden­ti­fied two plant-derived prod­uct­s—ce­las­t­rol and gedun­in—as pow­er­ful an­dro­gen “in­hibitors” that can switch the pro­file.

The pro­files are ob­tained us­ing mi­croar­rays, ti­ny ar­rays of DNA se­quences cor­re­spond­ing to dif­fer­ent genes. 

When a gene in the body is ac­tive, it pro­duces a mol­e­cule called RNA that is chem­i­cally re­lat­ed to the gene’s own DNA. Great­er ac­ti­va­tion means more RNAs. These can al­so be con­verted in­to a form that will link chem­i­cally with the DNA for that gene. To ob­tain the pro­file, re­search­ers ex­tract a cell’s RNA, con­vert it and dump it on­to the mi­croar­ray. Then each RNA mol­e­cule sticks to a chem­i­cal “part­ner” on the ar­ray. The re­sult is a pat­tern of at­tach­ments that re­flects the gene ac­ti­va­tion pro­file.

* * *

Send us a comment on this story, or send it to a friend


Sign up for

On Home Page         


  • St­ar found to have lit­tle plan­ets over twice as old as our own

  • “Kind­ness curricu­lum” may bo­ost suc­cess in pre­schoolers


  • Smart­er mice with a “hum­anized” gene?

  • Was black­mail essen­tial for marr­iage to evolve?

  • Plu­to has even cold­er “twin” of sim­ilar size, studies find

  • Could simple an­ger have taught people to coop­erate?


  • F­rog said to de­scribe its home through song

  • Even r­ats will lend a help­ing paw: study

  • D­rug may undo aging-assoc­iated brain changes in ani­mals

Cancer is something that affects us all. One of the more serious cancers out there today is mesothelioma lung cancer because of how incurable it is. Thankfully, stories like this give hope to cancer patients everywhere that a cure is just around the corner. If you or someone you know is suffering from mesothelioma cancer, get the latest news and information about the steps that are being taken towards a cure. Find out if hiring a mesothelioma lawyer is something you should consider if you feel you were wrongly exposed to asbestos which has been linked to mesothelioma.

Apparel and Accessories at National Geographic National Geographic Books

In what they’re calling one of the most exciting developments in cancer research in years, scientists are developing ways to get computers to churn out new cancer treatments—with no need for anyone to even know how they work. Called chemical genomic screening, the technique is designed to skirt the hard, sometimes futile work of trying to learn precisely what goes wrong in a specific cancer and tailor drugs to fix it. The technique exploits the fact that an organism’s state at any time depends not only on its genes, but on which genes are active or inactive at a given time, since a gene can lie dormant and have no effect. Partial activation is also possible. Within a cell, the activation situation at a given time results in a distinct profile that existing technologies can read. In the new technique, researchers feed into a computer an activation profile linked to a particular form of cancer. The machine then checks this against a database of known drugs, which contains previously known information on how each drug changes gene activation patterns. Finally, the computer lists which of these compounds tend most strongly to convert the “diseased” profile, which had been fed into the machine, into a profile associated with a healthier state. By fixing the profile, scientists reason, the drug may help remedy the underlying problem, all of which can occur with little or no knowledge of its causes. Researchers stress that they’re not giving up on learning causes—indeed, this could enhance the results—but in the meantime, the shortcuts to new treatments could bring relief to millions. The technique “promises to significantly enhance the drug discovery process,” wrote Harvard Medical School’s Scott A. Armstrong and colleagues in a paper describing some of the new findings, in the Sept. 28 online issue of the research journal Cancer Cell. But researchers also cautioned that the technology, still at an early stage, isn’t clearly capable of providing cures. For now, it’s geared toward helping to convert particularly virulent forms of cancer into more manageable ones, making them better treatable by existing remedies. These could be administered alongside the newly found treatment. In their paper, Armstrong and colleagues described work with victims of childhood acute lymphoblastic leukemia, a cancer of the blood and bone marrow. Scientists have previously found that a subset of these children have a particularly poor prognosis. This is associated with a weak response to a common first-line treatment, the hormone glucocorticoid. Armstrong’s team found that in this “glucocorticoid-resistant” group, cancerous cells tend to have a specific profile of gene activation, technically called gene expression. In the other victims, the profile is different. The researchers applied the new technology toward converting the glucocorticoid-resistant cells into non-resistant cells. They fed the expression profiles of both types into a database of 453 known, genome-wide expression profiles resulting from treatments of various cell types with different drugs. The database, called The Connectivity Map, revealed that an existing drug known as rapamycin could potentially reverse the bad expression profile, researchers said. Investigating further, they found that rapamycin affects molecules linked to a process that leads to a form of cellular “suicide.” This would be a logical point for a cancer drug to work at, because this “suicide” is implicated in cancer. Naturally, healthy cells kill themselves when they start to become malignant, protecting the body against cancer. Full-blown disease occurs when this suicide system, called apoptosis, fails. In sum, rapamycin and glucocorticoid together may be a useful treatment, the researchers said. Rapamycin is currently used to prevent the body from rejecting organ and bone marrow transplants. In a separate paper published in the same issue of the journal, the medical school’s Todd R. Golub and colleagues took an analogous approach to prostate cancer. Hormone treatments often provide initial success in battling this illness. But they tend to stop working eventually, when the tumors evolve a resistance to the drugs. The drugs work by blocking a molecule, called a receptor, that allows hormones called androgens to circulate. Failure of the therapy is associated in part with a revival of the androgen transmission in tumors, which occurs despite the drugs. How this happens is unclear; but one helpful fact is that the high-androgen and low-androgen states have different gene expression profiles, the researchers said. Again using the Connectivity Map, they identified two natural, plant-derived products—celastrol and gedunin—as powerful androgen “inhibitors” that can switch the profile. The profiles are obtained using microarrays, tiny arrays of DNA sequences corresponding to different genes. When a gene in the body is active, it produces a molecule called RNA that is chemically related to the gene’s own DNA. Greater activation means more of these molecules. An RNA molecule can also be converted into a form that will link chemically with the DNA for that gene. To obtain the profile, researchers extract a cell’s RNA and dump it onto the microarray. Then each RNA molecule attaches itself to a chemical “partner” on the microarray. The result is a pattern of attachments that reflects the gene activation profile.