OUP user menu

Iron and infection: the heart of the matter

John J. Bullen, Henry J. Rogers, Paul B. Spalding, Charles G. Ward
DOI: http://dx.doi.org/10.1016/j.femsim.2004.11.010 325-330 First published online: 1 March 2005


Bacterial resistance to antibiotics is a major threat to clinical medicine. However, natural resistance to bacterial infection, which does not depend on antibiotics, is a powerful protective mechanism common to all mankind. The availability of iron is the heart of the matter and the successful functioning of these antibacterial systems depends entirely upon an extremely low level of free ionic iron (10−18 M) in normal tissue fluids. This in turn depends on well-oxygenated tissues where the oxidation–reduction potential (Eh) and pH control the binding of iron by unsaturated transferrin and lactoferrin.

Bacterial virulence is greatly enhanced by freely available iron, such as that in fully-saturated transferrin or free haemoglobin. Following trauma a fall in tissue Eh and pH due to ischaemia, plus the reducing powers of bacteria, can make iron in transferrin freely available and abolish the bactericidal properties of tissue fluids with disastrous results for the host. Hyperbaric oxygen is a possible therapeutic measure that could restore normal bactericidal systems in infected tissues by raising the Eh and pH.

  • Iron
  • Infection
  • Sepsis
  • Trauma
  • Oxygen

1 Introduction

The rapid increase in bacterial resistance to antibiotics is a threat to clinical medicine. A recent fact sheet states that nearly two million patients in the United States acquire an infection in hospital each year and of these about 90,000 die from their infection, compared with about 13,000 in 1992. More than 70% of the bacteria that cause these infections are resistant to at least one of the drugs most commonly used for treatment [1]. An illustration of the pattern of increased resistance is shown with Enterococci where the percentage of isolates resistant to vancomycin rose from zero around 1980 to 25% in 1990 [2]. Similar problems exist with methicillin resistant Staphylococci, where the additional resistance to vancomycin can render infection with this organism particularly difficult to deal with [3].

It is frequently pointed out that the continued use of antibiotics provides the selective pressure for bacteria to develop resistance, but repeated passage of these organisms from patient to patient may also have a major effect on bacterial virulence. This cannot be tested in man, but an experimental example shows its significance. A strain of Pseudomonas aeruginosa from a hospital patient was relatively avirulent for mice and had an LD50 of 2.6 ×106. After only 16 unselected passages in mice, the LD50 fell to 4.09 ×104 nearly a 100-fold increase in virulence [4]. This shows that the factors responsible for growth in vivo with this particular organism are extremely labile and easily enhanced by passage. Similar conditions may arise in man. This raises the question of natural resistance to infection, which is directly influenced by bacterial virulence.

2 Natural resistance to infection

Mankind is endowed with very effective mechanisms of natural resistance that are responsible for our survival as a species during countless millennia in the past, in spite of repeated attacks by bacteria. Although, there have been numerous casualties from infections like bubonic plague, tuberculosis, syphilis, cholera and many others, there has always been a proportion of the population that survived these attacks and passed on these protective mechanisms to future generations.

Natural resistance in man is not easy to measure, but one example gives an idea of its effectiveness. Attempts made to infect human volunteers with hospital strains of Staphylococci were largely unsuccessful. At least seven million cocci injected intradermally were required to produce a small pustule. However, when sutures contaminated with about 30,000 cocci were inserted very severe stitch abscesses developed within 24 h, reaching the size of an orange despite treatment with penicillin, to which the organism was sensitive [5]. These experiments showed that normal human tissues are highly resistant to infection with Staphylococci, unless there is some local injury to initiate infection.

Normal physiological mechanisms of resistance can sometimes be enhanced. The incidence of infection in patients undergoing colorectal resection was halved by increasing the concentration of inspired oxygen to 80% during the operation and for 2 h thereafter compared with patients receiving 30% oxygen. This has two main advantages since infection is prevented in its very early stage, and oxygen is relatively inexpensive and easy to administer [6]. Natural resistance of this kind needs to be exploited.

3 The heart of the matter: iron availability

3.1 The normal low-iron environment in vivo

Many different factors contribute to natural resistance to infection but it is now clear that these protective systems can only function successfully in an environment where the normal concentration of free ionic iron is about 10−18 M [7], which can be regarded as virtually zero. This low-iron environment is due to the iron binding proteins transferrin and lactoferrin, which are normally only 30–40% saturated with iron and have very high association constants for ferric iron of about 1036[8]. The low-iron environment can be reinforced by interleukin-1 released during an infection in response to fever [9]. Thus, in experimental infections with Francisella tularensis in human volunteers, the transferrin saturation fell to 10% or less in typical clinical cases [10]. This hypoferraemia is a normal response to infection and appears to be a part of natural mechanisms of resistance [11].

3.2 Freely available iron and bacterial virulence

The ability of freely available iron to diminish or destroy normal resistance and increase bacterial virulence has been demonstrated repeatedly in experimental infections involving at least 18 different bacterial species [8]. Guinea pigs survived intraperitoneal infection with approximately 50,000 Escherichia coli O111, and the organisms had largely disappeared in 60 h, but died from septicaemia when the same number of bacteria were given with iron (5 mg kg−1), or haemoglobin, (170 mg) or haematin (100 mg) [12].

With Vibrio vulnificus there was a dramatic reduction of the LD50 for mice from 6 ×106 to approximately one organism by the injection of ferric ammonium citrate (4 µg g−1) [13].

The injection of ferric iron (5 mg kg−1) with the original unpassaged strain of P. aeruginosa[4] in mice produced a reduction of the LD50 from 1.26 ×106 to 4.74 ×105, an increase in virulence of approximately 2-fold. After 16 passages, the injection of iron caused a dramatic fall of the LD50 from 4.09 ×104 to less than 10 organisms, an increase in virulence of approximately 4000-fold. Thus, the increased virulence due to passage plus the added effect of freely available iron resulted in an increase of virulence of over 100,000-fold compared with the original strain [4]. This suggests that passaged organisms may not only be more virulent in the normal host but may also be capable of a very large increase in virulence if iron becomes freely available.

The mechanisms whereby potential pathogens acquire the iron necessary for growth in vivo, and the expression of virulence, are extremely complex and diverse. For example, the genera Escherichia, Klebsiella, Salmonella and Shigella produce the iron chelators enterobactin and aerobactin, which have extremely high affinities for ferric iron, while Pseudomonas species possess effective siderophores in pyochelin and pyoverdin. All these organisms produce specific outer membrane proteins involved with iron uptake. A different method of iron acquisition is a direct interaction of the bacterial cell with the iron binding protein such as employed by members of the Pasteurellaceae and Neisseriaceae.

Haem compounds are a source of iron for many bacteria. Yersinia pestis for example can utilize free haem, haemoglobin, haemoglobin bound to haptoglobin, and haem bound to haemopexin. Virulent Y. pestis store haem, giving rise to pigmented colonies of bacteria. A non-pigmented mutant was avirulent, but had its virulence fully restored by added iron. These and other mechanisms are discussed in detail elsewhere [14].

It is significant that natural resistance to infection operates effectively in the normal low-iron environment in spite of bacterial iron acquiring mechanisms. Only when iron is freely available are these protective mechanisms abolished [8,12,13].

4 Clinical aspects of freely available iron

4.1 Iron overload in acute leukaemia

Patients with acute leukaemia frequently suffer from iron overload and are unusually susceptible to Candida albicans infection. Of 34 patients with acute myoblastic leukaemia in relapse 21 had elevated serum iron levels of >120 µg% (normal range 70–120 µg%) and in 17 of these 21 patients the transferrin was 100% saturated. An additional 10 patients had a transferrin saturation greater than 50%. Among all these patients, seven had significant C. albicans agglutination titers (1:320 or greater) and two had systemic candidiasis with positive blood cultures [15]. Normal human serum is strongly inhibitory for C. albicans, whereas the fungus grows profusely in leukaemic sera with 100% saturation of the transferrin, and in normal serum if the saturation is raised to 100% [15]. This suggests that the increased susceptibility of leukaemic patients to C. albicans infection is probably due to the ability of the fungus to multiply rapidly in tissue fluids containing freely available iron.

Iron overload may also be made worse by chemotherapy. Seventy patients with leukaemia had raised levels of transferrin saturation (51–59%) before treatment. All developed granulocytopenia (<1000 cells µl−1) during treatment with cytosine arabinoside and daunorubicin and the transferrin saturation rose to 96% or above. There was also a significant fall in total iron binding capacity. Fungal infections were suspected in 59% of these patients and confirmed in 19% [16]. In addition, it has been found that some patients with myeloid leukaemia have circulating low-molecular-mass iron complexes, measured by the bleomycin assay, of about 1.0 µmol l−1. This was increased to about 9.0 µmol l−1 after chemotherapy [17].

Neutropenia is usually held responsible for an increase in susceptibility to infection, but the fact that an increase in neutrophils produced by treatment with granulocyte colony-stimulating factors failed to demonstrate any significant benefit for overall survival or disease-free survival [18,19] suggests that some other major factor is involved. Increased availability of iron seems to be the most likely candidate.

4.2 Iron availability in blood transfusions and blood substitutes

In normal individuals, the amount of free haemoglobin found in plasma is only about 0.3–0.7 mg dl−1. Even this small amount may be caused by damaging some red cells while taking the sample [20]. Both haemoglobin and haematin can readily provide iron for bacteria and are extremely effective in abolishing natural resistance to infection [12,14].

Blood transfusions appear to increase susceptibility to bacterial infection [21,22]. Promotion of infection is often attributed to immune suppression [22]. This is difficult to define but it is known that free haemoglobin definitely promotes infection [12,14] and that stored blood deteriorates with the red cells becoming more fragile. In addition, there is an increase in low-molecular-weight bleomycin chelatable iron in plasma depleted red blood cell units stored for more than 10 days without haemolysis [23]. It is therefore, quite possible that haemoglobin from lysed red cells, or low-molecular-weight iron complexes in transfused blood are responsible for the increased susceptibility to infection, and it would be interesting to measure the concentration of haemoglobin and bleomycin chelatable iron in the plasma of patients receiving blood transfusions [17,20].

Blood substitutes based on modified haemoglobin have been investigated both experimentally and in clinical trials [24]. The main benefits proposed are freedom from infectious agents and the ability to transport oxygen. Unfortunately, modified haemoglobin also promotes infection. Pyridoxalated glycolaldehyde-polymerised human haemoglobin was equally effective as native human haemoglobin in promoting infection in mice with E. coli O18 K1. Intraperitoneal injection of 103E. coli resulted in a brief invasion of the blood stream but the organisms later disappeared and all the animals survived. Animals given 30 mg of modified haemoglobin intravenously immediately before intraperitoneal infection died from septicaemia with bacterial blood counts reaching 108 ml−1[24].

A clinical trial with patients suffering from severe haemorrhagic shock given 500–1000 ml of 10% diaspirin cross-linked haemoglobin showed a death rate after 28 days of 46%, compared with 17% in those given saline (P= 0.003). The exact reasons for the increased mortality are not given but clearly, the haemoglobin was not a suitable resuscitation fluid [25].

5 Trauma, iron availability and infection

5.1 The role of Eh and pH

Normal well-vascularised tissues are highly resistant to infection, and the key to this is adequate oxygenation [26]. An increase in susceptibility to infection caused by hypoxia is usually attributed to interference with oxidative killing of bacteria by polymorphonuclear leucocytes [26]. This may be true with some organisms such as Staphylococci where anaerobic conditions reduced the rate of killing by polymorphs to 25% compared with 85% with air. However, this did not apply to bacteria like Streptococcus viridans or P. aeruginosa where anaerobic killing was identical to that seen under aerobic conditions, while with others like Clostridium perfringens or Bacteroides fragilis the reduction in killing under anaerobic conditions was only slight (5–10%) [27]. Thus, it seems unlikely that the effects of hypoxia can be attributed entirely to a generalized depression of phagocytic killing against all the different bacteria encountered in clinical sepsis. However, the level of tissue oxygenation does have an indirect but profound influence on freely available iron [28].

Two physical conditions, the oxidation–reduction potential (Eh) and the pH are of fundamental importance since they control the binding of iron to transferrin. The Eh of normal guinea pig muscle is about +300 mV, pH 7.4. At this Eh the iron bound to transferrin is in the ferric form. If the Eh is lowered to −140 mV or below the ferric iron is reduced to the ferrous form and is no longer bound by the protein, and can be available to bacteria [29].

Human plasma at an Eh of about +200 mV, pH 7.5 was strongly bactericidal against E. coli with about 99.9% of approximately 50,000 bacteria/ml−1 being destroyed in 1 h. When the Eh was lowered to about −400 mV pH 7.5 by adding a reducing agent the bactericidal effect was stopped, and after a delay of 2 h the remaining bacteria, about 50/ml, grew rapidly to reach over 106 ml−1 in a further 6 h. Ferrous iron was detected in the plasma. This effect can be reversed. Small numbers of E. coli grew rapidly in normal human plasma at an Eh of about −400 mV, pH 7.5. When the Eh was raised to about +200 mV, pH 7.5 by introducing air into the system rapid killing started within 2 h. The ability to restore the bactericidal effect is significant because it shows that a low Eh does not irreversibly damage the bactericidal system [30]. The pH of tissue fluids is equally important because the binding of iron by transferrin is progressively lost at a pH of 6.7 or below, and this is accompanied by the loss of bactericidal effects against many different organisms [28,30].

5.2 Hypoxia and bacterial virulence

Trauma frequently results in local ischaemia and a reduction or interruption of the blood supply can have a marked effect on the Eh of tissues. Thus, injection of 10 µg of adrenaline into guinea pig muscle caused a fall in Eh from +300 mV to under +200 mV, which lasted for 2 h [31]. In experimental skin infections in guinea pigs injection of 2 µg of adrenaline produced local hypoxia and a fall in Eh. This greatly promoted infection with some bacteria. There was a 100,000-fold reduction in the minimum infecting dose with Cl. perfringens, 10,000-fold with E. coli, 60–80-fold with Staphylococci, 10-fold with Strep. pneumoniae, but none with Vibrio cholerae[32].

One may ask why the injection of adrenaline should produce such different results with different bacteria. The answer almost certainly depends on the reducing power of the individual organisms. Bacteria effect reduction as a natural result of metabolic activity. Substrates like glucose must be metabolized to obtain the energy necessary for proliferation and in this process O2 may be reduced by the bacterial electron transport system. However, bacteria differ greatly in their reducing power. Organisms like Staph. aureus can initiate growth at a relatively high Eh of +300 mV but only reach a final Eh in cultures of between −100 and −150 mV [33]. C. perfringens requires an Eh of +60 mV or lower to initiate growth but can produce intensely reducing conditions of −400 mV [31]. E. coli on the other hand, can initiate growth at +300 mV but can reduce the medium to −400 mV or below. The intense reducing activity of E. coli is shown by the fact that fifteen million molecules of oxidant may be reduced by one cell in 1 min [33].

These data may help to explain some of the results obtained with adrenaline. Thus, organisms like Cl. perfringens and E. coli, which possess very powerful reducing systems appear to be able to take advantage of a reduction of the skin Eh to −28 to −58 mV so that the minimum infecting dose was reduced by 10,000–100,000-fold. With other bacteria like Strep. pneumoniae with far less powerful reducing systems the minimum infecting dose fell by only 10-fold [32]. This has important implications for trauma. A fall in tissue Eh and pH due to injury could provide just the right conditions for promoting infection with those bacteria that can take advantage of a fall in Eh. Once the Eh falls below −140 mV iron from transferrin would be freely available for rapid bacterial growth, and the bactericidal properties of the tissue fluids would be abolished [28,29].

5.3 Effects of highly reducing conditions in vivo

Studies on intraperitoneal infections with Cl. perfringens in passively immunised guinea pigs showed that very large numbers of bacteria (109–1010) produced highly reducing conditions (Eh −420 mV) and a low pH (6.4) in the peritoneal cavity which led to a large increase in vascular permeability, shock and death. The exudates were completely non-toxic for normal animals but rich in protein [31]. These data raise the possibility that any massive infection with bacteria capable of producing highly reducing conditions may lead to loss of normal vascular permeability and result in haemoconcentration and the development of shock. In man, a fall in oxygen tension below 15–20 mm Hg in the small vessels of the arm even for a short time causes an increase in vascular permeability [31].

In septic peritonitis in man it has been shown that the mean pH in drainage fluid from 59 patients was 6.75 compared with 7.49 in 105 uninfected patients. The pO2 in 17 infected patients was zero or near zero [34]. Unfortunately, a zero pO2 does not indicate whether reducing conditions exist and to explore this it would be essential to measure the Eh and pH. It is also important to assess the total viable bacterial count in infected tissues. Only very large numbers of bacteria can produce highly reducing conditions, and much more data on this aspect is required.

5.4 Hyperbaric oxygen

Treatment of infected tissues with hyperbaric oxygen to raise the Eh might be very useful since it would restore normal bactericidal systems nullified by highly reducing conditions. At the same time, it would be essential that the pH was restored to normal values. In practice the problem of free iron being provided by haem compounds would remain, but exactly what happens in these circumstances needs to be explored [31].

6 Conclusions

The heart of the matter in bacterial infection is a competition for iron between the host and the invading organism. In normal individuals, two factors control the availability of iron for bacteria. Haem compounds are unavailable since they are enclosed in erythrocytes or other tissue cells. In tissue fluids such as plasma, severe iron restriction due to the iron binding proteins is controlled by the Eh and pH, which in turn is directly maintained by the level of oxygenation.

Freely available iron can have devastating effects on natural resistance to infection, and result in the abolition of protective bactericidal mechanisms in tissue fluids, which can lead to rapid bacterial growth. This can overwhelm phagocytic systems with disastrous results for the host.

Hyperbaric oxygen may be useful for the treatment of infected tissue by restoring tissue Eh and pH, together with normal bactericidal systems. This requires further investigation.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
View Abstract