Skip to main page content

• HOME
• Current Issue
• Archive
• Alerts
• About ASM
• Contact us
• Tech Support
• Journals.ASM.org

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gentry, D. R. Articles by Holmes, D. J. Search for Related Content PubMed PubMed Citation Articles by Gentry, D. R. Articles by Holmes, D. J.

 Previous Article  |  Next Article 

Antimicrobial Agents and Chemotherapy, June 2007, p. 2048-2052, Vol. 51, No. 6
0066-4804/07/$08.00+0     doi:10.1128/AAC.01066-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Stepwise Exposure of Staphylococcus aureus to Pleuromutilins Is Associated with Stepwise Acquisition of Mutations in rplC and Minimally Affects Susceptibility to Retapamulin

Daniel R. Gentry,^* Stephen F. Rittenhouse, Lynn McCloskey, and David J. Holmes

Department of Microbiology, MMPD-CEDD, GlaxoSmithKline, Collegeville, Pennsylvania 19426

Received 23 August 2006/ Returned for modification 4 October 2006/ Accepted 20 March 2007

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
To assess their effects on susceptibility to retapamulin in Staphylococcus aureus, first-, second-, and third-step mutants with elevated MICs to tiamulin and other investigational pleuromutilin compounds were isolated and characterized through exposure to high drug concentrations. All first- and second-step mutations were in rplC, encoding ribosomal protein L3. Most third-step mutants acquired a third mutation in rplC. While first- and second-step mutations did cause an elevation in tiamulin and retapamulin MICs, a significant decrease in activity was not seen until a third mutation was acquired. All third-step mutants exhibited severe growth defects, and faster-growing variants arose at a high frequency from most isolates. These faster-growing variants were found to be more susceptible to pleuromutilins. In the case of a mutant with three alterations in rplC, the fast-growing variants acquired an additional mutation in rplC. In the case of fast-growing variants of isolates with two mutations in rplC and at least one mutation at an unmapped locus, one of the two rplC mutations reverted to wild type. These data indicate that mutations in rplC that lead to pleuromutilin resistance have a direct, negative effect on fitness. While reduction in activity of retapamulin against S. aureus can be seen through mutations in rplC, it is likely that target-specific resistance to retapamulin will be slow to emerge due to the need for three mutations for a significant effect on activity and the fitness cost of each mutational step.

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The pleuromutilin class of antibiotics are protein synthesis inhibitors that target the large subunit of bacterial ribosomes (4). Biochemical analysis (6) and subsequent cocrystallization studies (9) have shown that pleuromutilins inhibit the peptide bond formation reaction and bind the peptidyl transferase region of ribosomes. Resistance to pleuromutilins due to ribosome mutation has been demonstrated in a limited number of organisms. Early studies with Escherichia coli mapped resistance to tiamulin, a semisynthetic pleuromutilin, near the closely linked rplC (encoding ribosomal protein L3) and rplD (encoding ribosomal protein L4) genes (2), which Bosling et al. (3) later showed was due to mutations in rplC. Pringle et al. (7) demonstrated that resistance to tiamulin in Brachyspira spp. is due to mutations in rplC, rrl (encoding 23S rRNA), or both. Isolates with the highest level of resistance possessed three mutations, two in rrl and one in rplC. More recently, Kosowska-Shick et al. (5) reported that multiple, serial passage resistance selection of Staphylococcus aureus in the presence of retapamulin (see below) selected for single and double mutations in rplC.

Although used for veterinary applications, no pleuromutilin has been developed for human use. Due to the emergence of resistance to existing antibiotics, a new generation of pleuromutilins is being developed for treatment of human disease, the first of which is retapamulin for topical use. S. aureus is an important skin pathogen against which retapamulin is particularly effective in vitro. The purpose of the study reported here was to determine if target-based resistance to pleuromutilins can be generated in S. aureus and, if so generated, what the in vitro activity of retapamulin against the isolated mutants is. We found that, while a reduction in susceptibility can be generated in the laboratory at a low mutation rate, the magnitude of this effect on MICs remains fairly small and is associated with a reduction in cell fitness.

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Strains and culture conditions. Resistant mutants were generated in the common laboratory Staphylococcus aureus strain RN4220. Cultures were grown in brain heart infusion, when high cell density was desired, or LB medium. MICs were determined by following CLSI methodology (8).

Drug exposure and isolation of first, second, and third mutants. To generate and isolate spontaneous S. aureus mutants with reduced susceptibility to pleruromutilins, an overnight culture of S. aureus RN4220 (in brain heart infusion medium) was plated onto an LB agar plate containing 1.0 µg/ml tiamulin. Colonies that arose were then patched onto an LB plate (no drug) and, after 16 h at 37°C, replica plated onto LB agar containing 1, 5, 10, and 20 µg/ml of tiamulin.

To isolate second-step mutants, an overnight culture of the first-step G152D L3 mutant (2 x 10⁹ CFU total) was plated on LB agar containing 10 µg/ml of tiamulin. Colonies were recultured on plates containing 10 µg/ml of tiamulin to verify resistance.

Third-step mutants with a further reduction in susceptibility to tiamulin could not be generated when selected at 20 µg/ml. To isolate third-step mutants, the second-step mutants were plated with 5 other pleuromutilin compounds at 2 times their respective MICs. All colonies that arose after extended incubation were replated on media with each compound at the selective concentration to confirm resistance.

PCR and sequencing. Genes targeted for sequencing were amplified by PCR using PCR Superscript High Fidelity (Invitrogen). Primers used are described in Table 1. PCR was carried out by adding a single colony to the enzyme-primer mix. After a 5-min denaturation at 94°C, 30 cycles of 30 s at 94°C, 30 s at 48°C, and 2 min at 72°C were carried out. PCR fragments were purified on QIAGEN spin columns and sequenced using the same primers used for the amplification.

View this table: [in this window] [in a new window]

TABLE 1. Primers used for PCR amplification of target genes

Marker rescue of G144R G152D D159Y triple mutation. The primers L3regupsma and L3regdownsma were used to amplify the region around rplC from a mutant encoding the G144R G152D D159Y triple mutation. The resulting amplicon was inserted into the S. aureus-E. coli shuttle vector pCU1 (1) and ultimately electroporated into S. aureus RN4220, selecting for the plasmid-encoded chloramphenicol resistance. The resulting plasmid carrying strain was grown on LB agar containing 1, 5, or 10 µg/ml tiamulin alongside RN4220 (pCU1) as a control.

Reversion analysis of third-step mutations. All of the third-step, most highly pleuromutilin-resistant mutants grew very slowly on all media tested. To isolate faster-growing variants for analysis, single colonies of each mutant grown on LB agar plates supplemented with 20 mg/ml tiamulin were streaked on LB agar plates without supplementation. After greater than 3 days at 37°C, fast-growing variants arose for most mutants tested. At least 2 fast-growing colonies from each mutant were purified on LB agar and subjected to MIC and sequence analysis.

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Isolation of resistant mutants. S. aureus mutants spontaneously resistant to tiamulin were readily isolated. From approximately 5 x 10⁹ CFU plated, about 100 CFU formed on LB agar supplemented with 1 µg/ml tiamulin. From this batch, 50 colonies were purified and further characterized. The tiamulin MICs of these mutants ranged from 2 to 4 µg/ml. rplC was amplified via PCR and sequenced. Three different mutations in rplC were identified which led to G155R, D159Y, or S158L changes in the L3 protein (Table 2). Previous work in-house and elsewhere (5) identified S158L, G152D, and an in-frame deletion of residues 151 to 153. These mutations, therefore, define a region of L3 capable of causing decreased tiamulin susceptibility in S. aureus.

View this table: [in this window] [in a new window]

TABLE 2. First-step S. aureus mutants in ribosomal protein L3 and associated tiamulin MICs

As mentioned above, a previous study (unpublished) identified a strain carrying the G152D mutation (see also reference 5). Concurrent with the first-step mutant isolation experiments, second-step mutants using this mutant as the parent were selected. After 3 days on LB agar with 10 µg/ml tiamulin, 14 colonies appeared. Upon regrowth, all colonies grew on plates containing 10 µg/ml of tiamulin, while none grew on plates containing 20 µg/ml of tiamulin. MICs for tiamulin against the second-step mutants (tiamulin MICs, 16 µg/ml) were 64-fold higher than that obtained for the wild-type strain RN4220. The rplC gene of each isolate was amplified and the sequence determined. All strains possessed one of two additional mutations in L3: D159Y (n = 12) or G155R (n = 2).

Repeated attempts to isolate third-step mutants by plating the two second-step mutants on higher concentrations of tiamulin failed. From these experiments, it was estimated that the frequency of third-step mutants to tiamulin is less than 1 x 10^–10. To see if further resistance to any pleuromutilin compound could be generated, representative isolates carrying the two second-step mutations were plated with 5 different pleuromutilin compounds at 2x their MIC. After 5 days, resistant colonies arose on one plate. Given that all second-step mutations arose in L3, the obvious first characterization of the third-step mutants was to sequence the rplC. Eight of the 11 strains possessed a third mutation in L3. Seven of the eight possessed the same three mutations, G152D, G155R, and D159Y. One strain acquired a new mutation, G144R, while the remaining three did not have an additional mutation in L3.

MIC analysis of mutants. Tiamulin MICs against the third-step mutants were >256-fold higher than those determined for the wild-type strain (Table 3). To investigate the effect of L3 mutations on S. aureus susceptibility to retapamulin, the antibacterial activity of retapamulin was determined against the mutants isolated in this study (Table 3). Retapamulin inhibited first-step mutants at concentrations of 0.25 to 0.5 µg/ml, demonstrating a four- to eightfold decrease in susceptibility against these mutants in comparison with the MIC obtained for the wild-type strain. Against the second- and third-step mutants, retapamulin inhibited these organisms at concentrations of 1 to 4 µg/ml, representing an MIC increase of 16- and 64-fold, respectively. The increases in retapamulin MICs for each acquired mutation (4- to 8-, 16-, and 64-fold) were lower than those seen for tiamulin (8- to 16-, 64-, and >256-fold against first-, second-, and third-step mutants, respectively). In addition, the absolute concentrations of retapamulin required to inhibit each acquired mutant were lower than those seen for tiamulin (0.25 to 0.5 versus 2 to 4 µg/ml, 1 versus 16 µg/ml, and 4 versus 32 to >64 µg/ml against first-, second-, and third-step mutants, respectively).

View this table: [in this window] [in a new window]

TABLE 3. MICs of various compounds against S. aureus rplC mutants

In addition to determining MICs for the pleuromutilins tiamulin and retapamulin, MICs of various antibiotics with diverse modes of action were also determined. MICs for chloramphenicol, tetracycline, levofloxacin, and mupirocin were unchanged in the mutants tested relative to their wild-type parent.

As mentioned above, 3 third-step mutants did not acquire a third mutation in L3. These isolates, designated 2A1-1, 2A1-7, and 2A1-8, are presumed to have an unmapped mutation or mutations at a locus other than rplC. MIC analysis of these mutants (Table 4) reveals no changes in MICs for compounds other than pleuromutilins, indicating that the unmapped mutation(s) are specific for pleuromutilins.

View this table: [in this window] [in a new window]

TABLE 4. MICs of various compounds against S. aureus rplC/unmapped mutants and their fast-growing variants

Marker rescue. As noted above, some third-step mutants lacked a third alteration in L3. This raised the question of whether the resistance exhibited by rplC triple mutants was solely due to the observed changes in rplC and not due to mutations elsewhere on the chromosome. To test this, the mutant rplC encoding the G144R G152D D159Y change was chosen for a "marker rescue" experiment. The gene with 1.0 kb of flanking DNA was amplified and cloned into the S. aureus-E. coli shuttle vector pCU1 (to make the plasmid pCU1-G144R G152D D159Y) and introduced into S. aureus RN4220 by electroporation. The resulting strain was then streaked on LB agar containing tiamulin. While no resistant colonies arose from RN4220 carrying pCU1 alone, several colonies appeared from the strain carrying pCU1-G144R G152D D159Y. These colonies were purified on 1 µg/ml tiamulin, and MICs were measured from the resulting isolates. Of 8 isolates tested, all were resistant to tiamulin (MICs, 32 to 64 µg/ml). Also, all were susceptible to chloramphenicol, indicating that the plasmid was lost and that the mutations were transferred to the chromosome from the plasmid. Sequencing of rplC from the resulting strains confirmed the presence of the G144R G152D D159Y-encoding mutations. From these results, we conclude that the increased resistance is solely due to the changes in rplC.

Attempts to locate unmapped mutations. An attempt was made to identify the locus or loci responsible for the resistance exhibited by the three isolates that lacked a third mutation in L3. The genes encoding ribosomal proteins L2, L4, L10, L22, and L27 were sequenced due to the predicted proximity of these proteins to L3 or the peptidyl transferase center. None of the isolates were found to possess mutations in those proteins. Finally, domains V and VI of the 23S rRNA genes were sequenced and no mutations were found.

Phenotypic and reversion analysis of third-step mutants. For most of the generated mutants, significant effects on growth were noted. As judged by colony formation on rich media, isolates bearing the D159Y mutation exhibited a slight defect in growth. All other single mutants appeared to grow similar to the wild-type strain. A definite defect in growth was noted in isolates carrying the G152D G155R double mutation, while those carrying the D159Y G152D double mutations suffered a severe growth defect, as demonstrated by colonies of approximately 1 mm forming only after incubation on blood agar for 36 h or more. Among the slower-growing colonies, faster-growing colonies often appeared. The third-step mutants exhibited an extremely severe growth defect. Measurable colonies did not form on blood agar until incubation for an excess of 72 h. Over time, as with the D159Y G152D double mutant, faster-growing variants arose randomly. Additionally, the slow growth and instability of the mutants made it difficult to carry out detailed growth measurements.

For most of the slow-growing third-step mutants, faster-growing variants arose at a high frequency. In the case of strains carrying the triple mutation G152D D159Y G155R, three such faster-growing variants were purified, rplC was sequenced, and it was found that each of the three mutants acquired an additional mutation in rplC. Two of the isolates carried the alteration H134N, while the other carried the alteration A150T. MIC analysis revealed that pleuromutilin MICs decreased (Table 3). The triple mutant G152D D159Y G144R also grows slowly but is more stable, and fast-growing variants were not isolated.

Just as for the rplC triple mutants above, reversion analysis was carried out on the other type of third-step mutants, those that have 2 mutations in rplC (coding for changes G152D D159Y in L3) and an unmapped mutation or mutations. Based on growth on solid media, these mutants have the most severe growth defect. The isolates 2A1-1 and 2A1-7 yielded a high frequency of fast-growing variants, while none were isolated from 2A1-8. rplC from two variants each of 2A1-1 and 2A1-7 were sequenced, and it was found that the mutation causing the G152D alteration in L3 reverted to wild type. MIC analysis indicated that pleuromutilin MICs decreased to levels similar to the isolates that carry just the D159Y alteration (Table 4).

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
This paper describes the isolation of spontaneously arising pleuromutilin-resistant mutants of S. aureus. As with other bacteria, resistance was mapped to the gene encoding ribosomal protein L3, rplC. Results similar to ours were reported for S. aureus from multistep passage experiments using retapamulin (5). Kosowska-Shick et al. (5) found that prolonged passage led to an increase in retapamulin MICs and that, in some cases, the increased MICs correlated with two mutations in rplC. Here we report direct observation that a stepwise reduction in pleuromutilin susceptibility occurs concurrently with stepwise acquisition of mutations in rplC. Further, we have found that the accumulation of three mutations in rplC leads to even higher levels of reduced pleuromutilin susceptibility, albeit at a significant cost to the fitness of the organism. We found that the slow-growth phenotype often reverted to faster growth. Analysis of some of the faster growing isolates proved telling in the relationship between the growth defect and pleuromutilin resistance. Fast-growing variants of an isolate with the L3 alteration G152D D159Y G155R acquired a fourth alteration in L3, either H134N or A150T, and also exhibited a decrease in pleuromutilin MICs (Table 3). Likewise, fast-growing variants of isolates that have two mutations in rplC and an unmapped mutation or mutations reverted both genotypically, having lost one of the two rplC mutations, and phenotypically, exhibiting a decrease in pleuromutilin MICs. In both these classes of mutants, alterations that lead to pleuromutilin resistance cause a decrease in fitness and that any subsequent increase in fitness leads to further alteration in L3 with a concomitant decrease in pleuromutilin resistance.

In addition to the method of isolating mutants being fundamentally different (we isolated spontaneous mutants by direct plating while Kosowska-Shick et al. isolated resistant mutants by passage on sub-MIC concentration of compounds), our studies reported here complement and add to those of Kosowska-Shick et al. For example, prolonged passage did not lead to and likely would not have led to the isolation of the third-step mutations that we report here given the extremely slow growth rate of the mutants isolated. Without the third-step mutants, the clear correlation of decreased fitness and pleuromutilin resistance would not have been uncovered. Further, the previous work did not formally dismiss the possibility that mutations in genes other than rplC contribute to pleuromutilin resistance. The marker rescue experiments described here show that reduced pleuromutilin susceptibility is directly due to the three mutations in rplC.

We uncovered the existence of an unmapped locus or loci that add to pleuromutilin resistance in the presence of two mutations in rplC (Table 4). The contribution of this locus or loci is mysterious given that reversion of one rplC mutation results in a faster growing and pleuromutilin-sensitive phenotype, likely suggesting that the unmapped locus or loci cannot affect pleuromutilin susceptibility in the absence of two mutations in rplC. A clue to the nature of this mutation is the increase in the tetracycline sensitivity of the triple mutant (Table 4). The triple mutants exhibit increased susceptibility to tetracycline which reverts with the reversion of one rplC mutation. Because tetracycline and pleuromutilins act at different stages of translation and even bind different subunits, a complete analysis of these mutants is out of the scope of this report.

Clearly, the most common laboratory-generated mutations leading to reduced pleuromutilin susceptibility occur in ribosomal protein L3. While first and second mutations do cause an elevation in pleuromutilin MICs, a significant decrease in activity is not seen until a third mutation in L3 is acquired. This was especially apparent for retapamulin, where activity against first- and second-step mutants was only 4- to 16-fold less than that for the wild-type strain and all retapamulin MICs against first- or second-step mutants were 1 µg/ml. Retapamulin inhibited all third-step mutants at a concentration of 4 µg/ml. Given that the frequency of third-step mutants to pleuromutilins is <1 x 10^–10, it is highly unlikely that three successive mutational events would occur in ribosomal protein L3 in S. aureus during therapeutic exposure to retapamulin. While elevated pleuromutilin MICs can be generated through mutations in L3, the magnitude of this effect on retapamulin activity against S. aureus remains fairly low and only occurs at a great cost to cell fitness. Therefore, based on the need for a third mutation for a significant effect on activity, a low mutation rate for third-step mutations in L3, and the fitness cost of each mutational step, it is likely that target-specific resistance to retapamulin will be slow to emerge.

 
* Corresponding author. Mailing address: UP1345, GlaxoSmithKline, 1250 S. Collegeville Road, Collegeville, PA 19426. Phone: (610) 917-7504. Fax: (610) 917-7901. E-mail: dan.r.gentry{at}gsk.com

Published ahead of print on 2 April 2007.

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1
1. Augustin, J., R. Rosentstein, B. Wieland, U. Schneider, N. Schnell, G. Engelke, K. Entian, and F. Gotz. 1992. Genetic analysis of epidermin biosynthetic genes and epidermin-negative mutants of Staphylococcus epidermidis. Eur. J. Biochem. 204:1149-1154.[Medline]
2
2. Bock, A., F. Turnowsky, and G. Hogenauer. 1982. Tiamulin resistance mutations in Escherichia coli. J. Bacteriol. 151:1253-1260.[Abstract/Free Full Text]
3
3. Bosling, J., S. M. Poulsen, B. Vester, and K. S. Long. 2003. Resistance to the peptidyl transferase inhibitor tiamulin caused by mutation of ribosomal protein L3. Antimicrob. Agents Chemother. 47:2892-2896.[Abstract/Free Full Text]
4
4. Hunt, E. 2000. Pleuromutilin antibiotics. Drugs Future 25:1163-1168.
5
5. Kosowska-Shick, K., C. Clark, K. Credito, P. McGhee, B. Dewasse, T. Bogdanovich, and P. C. Appelbaum. 2006. Single and multistep resistance selection studies on the activity of retapamulin compared to other agents against Staphylococcus aureus and Streptococcus pyogenes. Antimicrob. Agents Chemother. 50:765-769.[Abstract/Free Full Text]
6
6. Poulson, S. M., M. Karlsson, L. B. Johansson, and B. Vester. 2001. The pleuromutilin drugs tiamulin and valnemulin bind to the RNA at the peptidyl transferase centre on the ribosome. Mol. Microbiol. 42:1091-1099.
7
7. Pringle, M., J. Poehlsgaard, B. Vester, and K. S. Long. 2004. Mutations in ribosomal protein L3 and 23S ribosomal RNA at the peptidyl transferase centre are associated with reduced susceptibility to tiamulin in Brachyspira spp. isolates. Mol. Microbiol. 54:1295-1306.[CrossRef][Medline]
8
8. NCCLS. 2003. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; approved standard, 6th ed. (M7-A6). National Committee for Clinical Laboratory Standards, Wayne, PA.
9
9. Schlunzen, F., E. Pyetan, P. Fucini, A. Yonath, and J. M. Harms. 2004. Inhibition of peptide bond formation by pleuromutilins: the structure of the 50S ribosomal subunit from Deinococcus radiodurans in complex with tiamulin. Mol. Microbiol. 545:1287-1294.

---

Antimicrobial Agents and Chemotherapy, June 2007, p. 2048-2052, Vol. 51, No. 6
0066-4804/07/$08.00+0     doi:10.1128/AAC.01066-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

This article has been cited by other articles:

• Kilaru, S., Collins, C. M., Hartley, A. J., Bailey, A. M., Foster, G. D. (2009). Establishing Molecular Tools for Genetic Manipulation of the Pleuromutilin-Producing Fungus Clitopilus passeckerianus. Appl. Environ. Microbiol. 75: 7196-7204 [Abstract] [Full Text]  
• Traczewski, M. M., Brown, S. D. (2008). Proposed MIC and Disk Diffusion Microbiological Cutoffs and Spectrum of Activity of Retapamulin, a Novel Topical Antimicrobial Agent. Antimicrob. Agents Chemother. 52: 3863-3867 [Abstract] [Full Text]  
• Woodford, N., Afzal-Shah, M., Warner, M., Livermore, D. M. (2008). In vitro activity of retapamulin against Staphylococcus aureus isolates resistant to fusidic acid and mupirocin. J Antimicrob Chemother 62: 766-768 [Abstract] [Full Text]  
• Miller, K., Dunsmore, C. J., Fishwick, C. W. G., Chopra, I. (2008). Linezolid and Tiamulin Cross-Resistance in Staphylococcus aureus Mediated by Point Mutations in the Peptidyl Transferase Center. Antimicrob. Agents Chemother. 52: 1737-1742 [Abstract] [Full Text]  
• Daum, R. S. (2007). Skin and Soft-Tissue Infections Caused by Methicillin-Resistant Staphylococcus aureus. NEJM 357: 380-390 [Full Text]  

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Gentry, D. R. Articles by Holmes, D. J. Search for Related Content PubMed PubMed Citation Articles by Gentry, D. R. Articles by Holmes, D. J.