OK, I wrote this many moons ago – in 1978, for my first degree in Genetics. At the time, I felt that the issue was pretty much “case proved” and I moved on to do other things in my life…….Ten years ago I sent it to a Soil Association researcher who was impressed with its having been even discussed back then. But, hey, life goes on and we all grow wiser. One day this issue might even be taken seriously……..
Do drug resistance (R) factors present a threat to the therapeutic use of antibiotics in modern medicine?
Antibiotic resistance in pathogenic bacteria is of concern to clinicians; the aim of this essay is to examine the causes of such resistance and the optimum use of antibiotics in the face of this threat.
The discovery of antibiotics revolutionised medicine. However the optimism engendered by the use of sulphonamide and penicillin in the 1930s was muted somewhat when resistant pathogens emerged. After the Second World War several semi-synthetic antibiotics were evolved (!) by the biochemists/ pharmacists/ drug companies and were quickly put to widespread use. It seemed that bacterial infection could be conquered for, surely, if a pathogenic strain became, presumably by mutation, resistant to one antibiotic it would succumb to a second.
All went to plan at first. Then disturbing reports emanated from Japan of Shigella strains concomitantly resistant to several antibiotics, namely Sulphonamide (Su), Tetracycline (Tet) and Chloramphenicol (Chl). The rise in frequency of resistance was acute and the epidemiology such that a thesis involving mutation, as for penicillin resistance, could not hold. Typically a patient could be infected with a drug sensitive Shigella strain and, after a couple of days treatment with a single antibiotic, excrete multiply resistant strains, insensitive to drugs the patient had not been exposed to, as well as the drug used in the therapy.
The theory was put forward that multiple resistance must be present as a factor in the host’s gut and be transferred to the ingested pathogen. Akiba et al (1960) demonstrated that multiple resistance in E. coli could be passed as a unit to a drug sensitive Shigella strain in vitro. This was subsequently confirmed in vivo. The phenomenon, termed “transferable drug resistance”, seemed a major threat to the use of antibiotics.
Research showed the “transferable drug resistance factor” to be extra-chromosomally encoded and capable of mediating its own transfer to a recipient cell. The factor, christened an R-factor, was subjected to various in vitro studies (see Appendix 1). Its resistances fall broadly into two categories. Firstly there are those producing degredative enzymes e.g. Chl, Ampicillin (Amp) resistance genes, and secondly there are those which prevent the cell from concentrating the antibiotic, as sensitive strains do, e.g. Tet resistance. Typically R-factors carry 3-4 resistance determinants tho some carry 6 or more, in a variety of compositions – see Table 1.
||Tc, Str, Su
||Tc, Str, Su, Chl
||Tc, Str, Su, Chl
||Tc, Str, Su, Chl, Ap
||Tc, Str, Su
||Tc, Str, Su, Ap
||Str, Su, Ap
In vitro crosses were carried out between sensitive and non-sensitive bacterial strains to determine whether the data of Akiba were more general. Anderson (1968) summed up the results. Normally in mixed culture R-factor transfer occurs at a low level, 10-5/donor cell/hour, but under optimum conditions of culture and physiology of the donor and recipients transfer could be epidemic, yielding 50% transfer after an initial 12 hour lag phase. Datta (1972) carried out many such crosses using a wide variety of donor and recipient types and demonstrated that the nature of donor and recipient strain is not restricted to a few organisms but can occur between most gram negative enteric bacteria and also many other gram negative organisms. The R-factor seemed to be a “promiscuous” plasmid and in vitro studies did nothing to soothe the clinician’s nerves.
Or, at least, promiscuous in vitro under optimal conditions. What was the potential for in vivo transfer? There was circumstantial evidence that this did indeed occur. However all this data was derived from clinical observation and was not quantitative. Obviously these clinical observations required explanation but simpler model systems were examined first.
Salzman and Klemm (1968) reared “germ free” mice, fed on antibiotic free food. They subsequently permitted colonisation of the mouse guts with a single drug sensitive bacterial strain, as prospective recipient strain. Donor strains were fed and faecal bacteria examined over the subsequent month for resistant bacteria derived from the sensitive strain. They were found after 18 hours when (5×10-3)% of recipients were resistant. This figure rose to 1% after two weeks and 21% after four weeks.
Following this in 1969 Reed et al used “microbially defined” mice to show more or less the same, although they achieved less transfer.
In 1970 Guinée took the work a stage further. He conducted similar experiments to the above, using rats, but added the parameter of feeding tetracycline to the rats, using a donor R-factor conferring tetracycline resistance. His results are in Table 2.
|Ppm tetracycline in drinking water
= R-factor transfer
||Rats fed R+ donor
||Rats not fed donor
Clearly there was greater selection in favour of R+ recombinants in the presence of tetracycline at a higher dose.
H.W.Smith (1969a) carried out a fairly large scale investigation into the in vivo transfer of R-factors between strains of enterobacteria in chickens, calves and pigs. The bulk of his survey was carried out on chickens whose guts he colonised with a Naladixic acid resistant but otherwise antibiotic sensitive strain. To these otherwise “normal” chickens he fed donor R+ strains and screened for feacal NalR, drug resistant recombinants. His overriding conclusion was that “in vitro transfer is not necessarily an indication of in vivo transfer”. Both strains must be reasonably good colonisers of the gut, so they may reach the caecumwhere transfer takes place. Even so strains differ greatly in their recipient abilities. Salmonella typhimurium phage type 29 he found to be far and away the best recipient. However all R+ recombinants would be eliminated from the animal hosts very quickly in the absence of fed antibiotics, selecting for a resistance coded for by the donor R-factor.
Before we can look the clinician in the eye and comment upon his data we must experimentally examine the human system as well. Can in vivo transfer be demonstrated in the human gut? Simply the answer is “yes” – results are, unsurprisingly, similar in nature to those found using other mammalian species. However very few studies have been performed.
H.W.Smith (1969b) used one volunteer who carried no faecal R-factors and was not fed with antibiotics. From his gut flora the predominant strain of E.coli was isolated and a NalR mutant prepared. This strain colonised the gut very well, persisting for two years.
18 strains of E.coli from a wide variety of sources were used as R-factor donors. They were taken in test doses ranging from 104 organisms once to 109 for 7 consecutive days. Feacal NalR R+ organisms were counted.
One strain, B119 from ox, taken as 109 organisms for seven days, persisted for eleven days following cessation of the dose. Half its resistance determinants were transferred (Str, Neo from Str, Neo, Su, Tet) at high levels and the transformants persisted in the gut for 18 days. Otherwise, except for two human derived strains, colonisation by the donor was non-existent or very poor (Maximum was 11 days, mean 2 or 3). Resistance transfer was even less frequent, no transformed strain persisted longer than 7 days and the percentages of transfer were low.
Anderson et al (1973) demonstrated in vivo transfer to marked recipients in the human gut but only when selected for by feeding antibiotics concomitantly with the cross. Again they used strains derived from the volunteers’ faecal flora so there was no problem of colonisation of their guts with donors or recipients. However it was found that cessation of antibiotic selection caused the R+ recombinants to rapidly disappear.
So, what of the clinical data? Well Lebek (1963) reported on the “in vivo transfer of an R-factor in the gut of two infants” – only it was in German.
Later Farrar et al (1972) reported a similar case. Here two children, who shared an incubator, developed a Shigellosis a few days after birth. The first was treated with Ampicillin until it was demonstrated that the infecting strain was Str, Tet, Amp resistant. Subsequently the infant was treated with Kanamycin. The other child developed the illness after discharge from hospital, and was readmitted and treated with Kanamycin. However there was no improvement. Its faecal Shigella were shown to be now Kanamycin resistant as well as Str, Tet, Amp. In vitro studies showed this new resistance to segregate from Str, Amp, Tet resistance and it was concluded that resistance to Kanamycin was borne on a separate plasmid. This plasmid must have been acquired by the Shigella strain after infecting the infant, presumably from some organism already resident in the infant’s gut.
These cases, tho’, surely only parallel those of Salzman and Klemon (1968). More relevant are the observations made in the introduction that sensitive infecting Shigella sonnei could emerge resistant. However here again there was intense antibiotic selection for resistant strains. In the main, it seems that drug resistant strains only persist in animals or humans to whom antibiotics are fed.
But, then, where do R-factors come from? Where is the reservoir?
Anderson proposes a solution. He studied an outbreak of Salmonella typhimurium in English calves in the 1960s. It seemed that antibiotics could be used as prophylactics to prevent spread of infections through intensively reared herds of calves. Also, in chickens, antibiotics were shown to increase meat yield. Initially the method was effective but, surprise, surprise, soon the pathogen S. typhimurium was shown to be resistant to drug therapy in these calf herds, and there was epidemic spread of the infection. The outbreak was centred on one dealer and contained to “factory farmed” herds. As the epidemic spread and different antibiotics were used so patterns of resistance changed eventually incorporating 5 or 6 resistances from the original 3. Tetracycline resistance neared 100%.
The predominant phage type involved in the infection was 29 (Smith’s “best recipient”). Anderson and Datta (unpublished obs, as far as I can tell) Examined strains of Salmonella typhimurium isolated from human patients. They found a similar pattern to that obseved in calves. The predominant phage type was 29 and it showed a similar increase in multiple resistance, as did other infecting S.typhimurium. (Anderson 1968, see Appendix 3).
The message is clear – animals fed antibiotics, if handled unhygenically in the slaughterhouse and subsequently, provide a reservoir of drug resistant pathogens. This is obviously true. However it seems to be of limited significance as S.typhimurium is a relatively mild pathogen and is normally eliminated from its human host in a matter of days – together with its R-factor. Further the illness is not treated with antibiotics and so there is no selection for any recombinants, say E.coli, in the host gut.
The second source is the hospital itself. Over the years there has been intense selective pressure placed on the hospital bacterial population to become drug resistant. Furthermore the strains populating a hospital are human derived and more likely to colonise the human gut. Thus even tho’ the S.typhimurium of Anderson’s calves remained resistant for two years after cessation of antibiotic prophylaxis this was not significant to human medicine, except in that it showed that long term selection could streamline an R-factor so it was beneficial to its host in the absence of antibiotic selection, and could compete with other enteric bacteria. Similarly one might expect human derived pathogens to have evolved in the hospitals which had R-factors conferring other advantages to its host. A possible example of this is a lac R-factor (OK, so lac is transposable) isolated by Smith and Parnell (1975). So what evidence is there for a hospital reservoir of R-factors?
A survey by Salzmann and Klemm (1967) in the Bronx Municipal Hospital showed that more than 20% of the hospital’s personnel carried drug resistant bacteria on their hands! Whether this is a reflection of that hospital’s standard of hygiene or is a general phenomenon is not known. However studies have been performed on the frequencies of resistance of patients on admission and discharge and there is an increase observed from the level of 10% or so observed in the general population to about 20% (Salzmann and Klemm, 1966)(with apologies my gross over simplification). The case reported by Farrar et al (1972) demonstrated that a Kanomycin resistant organism was picked up by the infant – presumably in the hospital. Lebek’s (1963) data also have the same interpretation. Further there are cases reported of pathogens causing secondary (e.g. post operative) infections in hospitals being drug resistant (Gardner and Smith, 1969). Clearly there is then a hospital reservoir of R-factors, presumably this leaks back into the general population thro’ discharged patients, hospital personnel and other hospital visitors.
So what of the general population? It seems from very variable studies that there is a 10-20% incidence of R-factors in healthy non-antibiotic treated people. However no fluctuation can be shown except in the incidence of multiple drug resistance which appears to have increased. The existence of R-factors in pre-antibiotic societies has been demonstrated. Davis and Anandan showed Chl Str Tet Su and other R-factors in a tribe in Borneo and Gardner et al (1969) found R-factors in Solomon Islanders gut flora. A lyophylised bacterial sample from 1945 has similarly shown multiple drug resistance. In principle this is not surprising since antibiotics were invented by fungi not mankind and had been used for, er, a long time giving rise to resistant bacteria. What is surprising is that the R-factors coded for up to four resistances. It suggests that R-factors must have been fairly widespread prior to large scale antibiotic use.
However some resistances have probably appeared subsequently to their introduction (tho’ how can one demonstrate this?) It is interesting to note the discovery that many resistances are carried on elements much smaller than plasmids (transposons) capable of dissociating themselves from their host and reinserting themselves at a different (specific) D.N.A. site. This phenomenon accounts for the increase in numbers of resistances determined by R-factors, tho’ whether this is by a transdution or by hopping of transposons from one plasmid to another in an intact cell is open to question (it’s probably both anyway).
So what conclusions can we draw from this rather conflicting or, at least, opposingly slanted information?
The primary ones are obvious. It was a mistake to use antibiotics so liberally, tho’ an understandable mistake. Consequently prophylactic use of antibiotics has been generally outlawed. The threat to human health now seems to reside in the continued position of hospitals as reservoirs of R-factors. From the data reported above one would predict that this relatively high frequency should fall (be falling) concomitant with the reduction in antibiotic use. However this fall may not be rapid as “tailor made” R-factors may well exist, conferring advantages other than drug resistance to their hosts. Positive efforts should thus be made to eliminate such R-factors. In this respect transfer of drug resistance by transposons could lead to complications, but, I feel, should be of little general significance.
It seems we cannot look forward to the complete elimination of drug resistance, however we should try to restore the situation to as near as possible to the “pre-antibiotic age”, when only fungi knew of bacteriocides and human beings relied on their immune systems.
Chris Hemmings Dec 12th, 1977. Essay written in study for BSc, Genetics, awarded 1978.
Anderson et al (1975) have carried out physical studies on R -factors derived from a wide variety of sources. They divided into five plasmid compatibility or exclusion groups. Each such group had its members alike in molecular weight/contour length and resistances encoded. Furthermore they demonstrated that within a group members showed high homology in hybridisation studies, whereas intergroup studies showed low homology. Presumably each group arose from a specific ancestor, of which there were, obviously, five.
The molecular weight of an R-factor is such that about fifty functions may be encoded. An explanation is offered for these genes in the text.
The question of colonisability of a given gut by a bacterium seems crucial to this argument. Apart from the case quoted there is little data available. Working without the complication of R-plasmids, Cooke et al (1972) showed a varied ability of E.coli strains to colonise three volunteers’ guts. Using a 1011 cells dose one strain colonised for 120 days. A 108 dose of any particular strain survived for different periods of time in each of the three volunteers. In general, though, survival was for a short duration, using either human or animal derived strains – although a couple of strains persisted for a month.
Several studies of the patterns of resistance in clinical isolates have been made. The most comprehensive is that of Manten, Guinee et al (1971). Their survey is of the 11 years 1959-1966 of virtually every Salmonella isolate obtained in the Netherlands over that period, so amounting to an epidemiology study.
S.typhimurium took a similar course to that of the British isolates. Tetracycline resistance rose sharply, peaking in 1965-1966. Chloramphenicol resistance remained very low throughout the survey, peaking in 1963 at 1.6%. Ampicillin resistance followed that of Tetracycline – although it was not used in animal husbandry – presumably it was a passenger on the original R-factor(s) selected. The human and animal patterns were very similar – presumably because Salmonellas are transmitted by unhygienic handling of animal carcasses. Luckily chloramphenicol resistance was not one of the passenger resistances for it might then have posed a threat to the treatment of S.typhi. Of the 610 S.typhi isolates, though, only three were Chloramphenicol resistant.
The authors noted that the death rate from Salmonellosis had fallen to less than half its original level by the end of the survey, but could not say whether the incidence of R-factors had increased or declined.
Akiba et al. Jap J Microb 4, 219-227, 1960.
Anderson, ES. Am Rev Microb 22,1968.
Anderson et al. J Med Microb 6, 461-473, 1973.
Anderson et al. J Gen Microb 91, 376-382, 1975.
Cooke et al. J Med Microb 5, 361-369, 1972.
Datta, N. J Gen Microb, 70, 453-460, 1972.
Davis and Anandan. New Eng J Med 282, 117-122, 1970.
Farrar et al. J Infect Dis 126, 27-33, 1972.
Gardener and Smith, DH. Ann Int Med, 71, 1-9, 1969.
Guinee PM. J Bacteriol 102, 291-292, 1970.
Gardener et al. Lancet, 2, 774-776, 1969.
Manten et al. Bulletin WHO 45, 85-93, 1971.
Reed et al. J Bacteriol, 100, 22- , 1969.
Saltzman and Kleman. Antimicrobial Agents and Chemoth, 7, 97-100, 1967.
Saltzman and Klemm. Anti microb Ag and Chemoth, 212-220, 1966.
Saltzman and Klemm. Proc Soc Exp Biol, 128, 392-394, 1968.
Smith, HW. J Med Microb, 3, 165-180, 1969a.
Smith, HW Lancet, 1174, 1969b.
Smith, HW and Parsell. J Hyg, 75, 275-292, 1975.