The “triage theory”: micronutrient deficiencies cause insidious damage that

accelerates age-associated chronic disease

The  triage  theory  (Ames,  BN  (2006)  PNAS  103-­‐17589-­‐94)  posits  that  the  spectrum  of  functions  for  a  particular  
vitamin  or  mineral  (V/M)  are  managed  by  the  organism  such  that,  when  micronutrient  availability  is  limited,  
functions  required  for  short-­‐term  survival  take  precedence  over  functions  whose  loss  can  be  better  tolerated  
(e.g.  by  selection  for  micronutrient  binding  constants  or  targeted  tissue  distribution).  Ames  proposed  that  a  
consequence  of  this  evolutionary  adaptation  is  an  increase  in  the  risk  of  chronic  diseases  of  aging  when  V/M  
availability  is  limited.  That  nature  may  have  developed  such  a  system  is  logically  consistent  with  an  important  
evolutionary  theory  that  natural  selection  favors  short-­‐term  survival  for  reproduction  over  long-­‐term  health  
[e.g,  Kirkwood  (2008)  J  Intern  Med  2:117-­‐27].  During  evolution  micronutrient  shortages  were  likely  to  be  very  
common,  e.g.  the  15  essential  minerals  are  not  distributed  evenly  on  the  earth;  dietary  sources  and  availability  
also  fluctuated  markedly.  
 
If  the  Triage  theory  is  correct,  the  implications  for  pubic  health  are  enormous.  V/Ms  are  remarkably  
inexpensive.  Intakes  of  a  number  of  V/Ms  are  below  levels  considered  adequate.  These  deficiencies  are  
unusually  widespread  in  poor  countries,  but  also  in  the  US  population  in  all  segments  of  society,  especially  the  
poor,  children,  adolescents,  the  obese,  and  the  elderly.  In  
fact,  it  appears  that  only  a  small  fraction  of  the  
population  is  adequate  in  all  micronutrients.  High  
consumption  of  calorie-­‐rich,  micronutrient-­‐poor  
unbalanced  diets  exacerbates  the  problem.  Yet,  there  is  
little  societal  concern  because  no  overt  pathologies  have  
been  associated  with  levels  of  deficiency  that  are  not  
extreme.  The  triage  theory  predicts  that  more  moderate  
levels  of  deficiency  could  be  causing  insidious  changes  
that,  over  time,  culminate  in  age-­‐associated  chronic  
disease.  
 
A  direct  experimental  test  of  the  triage  theory’s  
prediction  of  a  causal  relationship  between  long-­‐term  
micronutrient  deficiencies  and  chronic  disease  is  virtually  impossible  to  obtain  in  randomized  controlled  trials  
conducted  over  many  years,  as  we  have  discussed  (Ames  et  al  (2007)  Am  J  Clin  Nutr  86:522-­‐3).  On  the  other  
hand,  the  central  premise  of  the  triage  theory,  that  limited  V/M  availability  favors  functions  required  for  short-­‐
term  survival  at  the  expense  of  those  required  for  long-­‐term  health,  can  be  tested.    
 

1
 VITAMIN  K  AND  SELENIUM,  TWO  EXAMPLES  OF  TRIAGE  THEORY  
 
In  these  two  studies,  we  applied  a  triage-­‐based  analysis  to  published  evidence.  The  analyses  were  feasible  
because  the  number  of  proteins  that  require  vitamin  K  or  selenium  are  relatively  few  (14  for  vitamin  K  and  24  
for  selenium),  unlike  V/Ms  such  as  iron  or  zinc  that  are  required  for  functions  of  hundreds  of  proteins.  Brief  
summaries  of  these  two  studies  are  below.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Vitamin  K  [McCann  &  Ames  (2009)  Am  J  Clin  Nutr  90:889-­‐907]  
 
Vitamin  K  is  a  cofactor  for  a  single  enzyme  required  for  the  γ-­‐carboxylation  of  14  different  proteins,  10  with  
known  functions.  γ-­‐carboxylation  is  required  for  these  proteins  to  bind  calcium,  which  is  known  to  be  required  
for  protein  function  in  almost  all  cases.  Most  of  these  10  proteins  are  synthesized  only  in  the  liver  and  are  
required  for  coagulation.  Only  3  of  the  10  
proteins  are  not  involved  in  coagulation.  In  
contrast  to  the  coagulation  factors,  these  
proteins  are  γ-­‐carboxylated  in  extra-­‐hepatic  
tissues.      
 
As  shown  in  Figure  1,  Mouse  knockouts  for  
almost  all  of  the  coagulation  factors  are  lethal,  
suggesting  these  functions  are  required  for  
short-­‐term  survival,  in  contrast  to  knockouts  of  
the  3  extra-­‐hepatic  proteins,  none  of  which  is  
embryonic  lethal.    
 
Figure 1. The dichotomy between the embryonic lethality of
coagulation factors, all predominantly γ-carboxylated in the liver,
and the non-embryonic lethality of other vitK dependent proteins
not involved in coagulation, all of which are predominantly γ-
carboxylated in extra-hepatic tissues.

2
Higher  dietary  intakes  of  vitK1  are  required  for  γ-­‐carboxylation  of  the  limited  number  of  extra-­‐hepatic  proteins  
that  have  so  far  been  examined  than  are  required  for  γ-­‐carboxylation  of  the  hepatic  proteins  involved  in  
coagulation.    
 
As  shown  in  Figure  2,  dietary  inadequacy  of  vitK1  is  linked  epidemiologically  in  some  studies  to  several  age-­‐
related  chronic  conditions  (e.g.,  bone  fractures,  atherosclerosis,  insulin  resistance).  And,  phenotypes  of  the  3  
non-­‐embryonic  lethal  knockouts  have  characteristics  of  a  similar  set  of  chronic  conditions  (e.g.,  arterial  
calcification,  skeletal  abnormalities,  glucose  dysregulation).    
 
 
Figure 2. This figure summarizes results discussed in McCann &
Ames (2009) Am J Clin Nutr 90:889-907. The analysis combines
mouse knock-out and human mutant phenotypes, effects of long-
term warfarin (anticoagulant) therapy, and prospective
epidemiological studies examining associations between vitK intake
and disease incidence. The internal consistency of these results from
diverse disciplines

The  major  dietary  source  of  vitamin  K  (vitK1)  is  

preferentially  directed  to  the  liver.  Furthermore,  it  
appears  that  vitK1  is  mainly  active  in  the  liver  and  
another  form  of  vitamin  K  that  can  be  synthesized  
from  vitK1  is  mainly  active  in  the  γ-­‐carboxylation  of  
extra-­‐hepatic  proteins.    
 
Whether  currently  recommended  intakes  for  vitK1  
are  sufficient  for  optimal  carboxylation  of  the  extra-­‐hepatic  proteins  has  been  questioned  (8).  The  average  
intakes  of  vitK1  in  the  United  States  and  the  United  Kingdom  are  less  even  than  currently  recommended  
intakes  (which  are  primarily  based  on  levels  required  for  adequate  coagulation  function).  Thus,  a  large  percent  
of  the  population  may  not  be  receiving  sufficient  vitK1  for  optimal  γ-­‐carboxylation  of  proteins  important  for  
long-­‐term  bone  and  heart  health.    
 
Collectively,  results  form  a  coherent  mechanistic  picture  consistent  with  triage  theory.    The  localization  of  the  
γ-­‐carboxylation  of  coagulation  factors  in  the  liver,  and  of  other  vitamin  K  dependent  proteins  to  extra-­‐hepatic  
tissues,  sets  up  a  dichotomy  that  takes  advantage  of  the  preferential  distribution  of  vitK1  to  the  liver  to  
preserve  coagulation  functions  when  vitamin  K  availability  is  scarce.    
 
Selenium  [McCann  &  Ames  (2012)  FASEB  J  25:1793-­‐1814]  
 
In  mammals,  the  great  majority  of  dietary  selenium  (Se)  is  incorporated  as  selenocysteine  during  translation  
at  the  active  site  of  25  selenoproteins  (SPs)  in  humans  (24  in  rodents).  The  substitution  of  Se  for  sulfur  at  the  
catalytic  site  has  a  major  effect  on  enzyme  potency  (13)  and  can  increase  efficiency  by  _100-­‐fold  (14,  15).  SPs  
are  required  for  life,  as  demonstrated  by  the  lethality  of  mouse  KOs  unable  to  incorporate  Se  into  SPs  (16),  
and  by  severe  impairment  if  SP  synthesis  is  knocked  out  in  specific  tissues  (e.g.,  refs.  17,  18).  About  a  dozen  
SPs  are  well  characterized,  most  of  which  are  enzymes  involved  in  redox  regulatory  pathways  and  
maintenance  of  optimal  redox  status.  This  review  focuses  on  12  SPs  with  known  mouse  or  human  mutants  
with  phenotypes  that  permit  a  classification  of  essential  or  nonessential.  
 
Preferential  protection  against  Se  deficiency  occurs  between  and  within  tissues.  For  example,  Se  is  
preferentially  distributed  to  some  tissues,  such  as  the  brain  and  reproductive  tissues,  at  the  expense  of  others.  
3
A  broad  spectrum  of  both  essential  and  nonessential  SPs  are  present  in  these  favored  tissues  and  also  in  all  
other  tissues,  presenting  a  very  different  picture  from  vitamin  K-­‐dependent  proteins,  where  all  essential  
proteins  are  predominantly  activated  in  the  same  tissue  (liver).    
 
Our  triage  analysis  focused  on  within-­‐tissue  hierarchies  of  SP  activities  or  concentrations  measured  in  vivo  in  
experiments  where  the  sensitivities  of  essential  and  nonessential  SPs  to  Se  deficiency  are  directly  compared.  
While  these  hierarchies  have  been  widely  discussed,  this  is  the  first  systematic  analysis.    
 
Mouse  knockout  phenotypes  were  available  for  13  SPs.  Five  (Gpx4,  Txnrd1,  Txnrd2,  Dio3,  and  Sepp1)  were  
classified  as  essential  and  7  (Gpx1,  Gpx  2,  Gpx  3,  Dio1,  Dio2,  Msrb1,  and  SelN)  nonessential.    
 
On  modest  Se  deficiency,  nonessential  selenoprotein  activities  and  concentrations  are  preferentially  lost,  with  
one  exception  (Dio1  in  the  thyroid,  which  we  predict  is  conditionally  essential).  An  example  is  in  Figure  1.  
 
Figure  1.  Sensitivity  of  Gpx4  and  overall  Gpx  activities  to  
modest  and  severe  Se  deficiency  in  multiple  tissues.  Essential  
SPs  are  represented  by  squares  and  nonessential  SPs  by  
circles:  blue  squares,  Gpx4;  red  circles,  overall  Gpx  activity  
(see  text).  When  _1  experimental  result  was  available  in  a  
given  tissue,  means  _  sd  are  plotted.  Within  a  given  tissue,  
means  that  do  not  share  a  common  letter  (a,  b)  are  significantly  
different  (P_0.05).  Points  plotted  without  sds  represent  
results  of  single  experiments.  
 

 
 
Mechanisms  include  the  requirement  of  a  special  
form  of  tRNA  sensitive  to  Se  deficiency  for  
translation  of  nonessential  selenoprotein  mRNAs   Figure 2. Mammalian Sec tRNA[Ser]Sec Structures of modified
nucleosides are highlighted in boldface. The figure is from
except  Dio1,  shown  in  Figure  2.     Hatfield et al (2006) Prog. Nuc. Acid Res Mol. Biol. 81:97-142.  

The  same  set  of  age-­‐related  diseases  and  conditions,  

including  cancer,  heart  disease,  and  immune  
dysfunction,  are  prospectively  associated  with  modest  Se  deficiency  and  also  with  genetic  dysfunction  of  
nonessential  selenoproteins,  suggesting  that  Se  deficiency  could  be  a  causal  factor,  a  possibility  strengthened  
by  mechanistic  evidence.    
 
Modest  Se  deficiency  is  common  in  many  parts  of  the  world;  optimal  intake  could  prevent  future  disease.