ABSTRACT
Coinfections with more than one human immunodeficiency virus type 1 (HIV-1) subtype appear to be the source of new recombinant strains and may be commonplace in high-risk cohorts exposed to multiple subtypes. Many potential dual infections have been identified during the HIV Superinfection Study in Mbeya, Tanzania, where 600 female bar workers who are highly exposed to subtypes A, C, and D have been evaluated every 3 months for over 3 years by use of the MHAacd HIV-1 genotyping assay. Here we describe an in-depth, longitudinal analysis of the viral quasispecies in a woman who was triply infected with HIV-1 and who developed AIDS and passed away 15 months after enrollment. The MHA results obtained at 0, 3, 6, 9, and 12 months revealed dual-probe reactivities and shifts in subtype over time, indicating a potential dual infection and prompting further investigation. The multiple infection was confirmed by PCR amplification of three genome regions by a multiple primer approach, followed by molecular cloning and sequencing. A highly complex viral quasispecies was found, including several recombinant forms, with vpu/gp120 being the most diverse region. A significant fluctuation in molecular forms over time was observed, showing that the serial sample format is highly desirable, if not essential, for the identification of multiple infections. In a separate experiment, we confirmed that the detection of coinfections is more efficient with the use of multiple amplification primers to overcome the primer bias that results from the enormous diversity in the HIV-1 genome.
The high level of diversity of the human immunodeficiency virus type 1 (HIV-1) genome is caused by the ability of the virus to mutate, recombine, and replicate with a short generation time. The high mutation rate of 0.3 nucleotide changes per genome per replication cycle (16) is the result of inaccurate viral DNA synthesis by the viral enzyme reverse transcriptase, which is error-prone partly due to its lack of 3′→5′ exonuclease proofreading activity (4, 6, 14). The enzyme also switches between the two RNA templates that are packaged in each virion approximately three times per genome per replication cycle (1, 16). When a cell is simultaneously infected with two or more HIV-1 strains, recombination during replication leads to a novel viral genome which contains genetic information from both proviruses. Some mosaic HIV-1 genomes circulate broadly in the human population and are now designated “circulating recombinant forms” (CRFs), 16 of which have been described to date (2, 28, 34). The number of CRFs will continue to increase as HIV-1-infected populations are evaluated completely at the molecular level.
HIV-1 exists in vivo as a quasispecies, a population of highly related but genetically distinct viruses (8, 33). This rapid variability provides the virus with the capability to adapt its genome to escape selective pressures from the immune system and antiretroviral therapy. Several reports have shown correlations between HIV-1 evolution in an infected individual and the progression to AIDS (7, 32), the development of drug resistance (5, 17, 24), and vertical transmission (23).
Adequate sampling of the viral quasispecies is challenging, yet necessary, for the detection of multiple, molecularly distinct infections. The full recovery and characterization of strains from a multiply infected individual over the course of HIV-1 infection require new approaches and sensitive methods.
HIV-1 subtypes A, C, D, and CRF10_CD and many unique recombinant forms are cocirculating in Tanzania, a setting that facilitates the evaluation of multiple infections with different subtypes and the recombination between them (12, 15, 18, 19, 26). In this report, we describe an in-depth, longitudinal analysis of the viral quasispecies in a woman with late-stage HIV-1 infection who was 1 of 600 female bar workers constituting a high-risk cohort in a longitudinal HIV Superinfection Study (HISIS) in Mbeya Region, Tanzania, that is described elsewhere (13, 27). The study participants were evaluated every 3 months for up to 4 years. HIV-1-positive samples were screened by a multiregion hybridization assay (MHAacd) (11). The objective of this work was to characterize the viral quasispecies in a triply infected HIV-1-positive individual as accurately as possible by using a multiple primer approach for nested PCR amplification of three 1.4-kb parts of the genome. Two primer sets used in four pair-wise combinations for each PCR were used to minimize bias in the recovery of strains. This approach was employed to give us a more complete picture of the viral population within the patient, and based on our results, this approach does increase the likelihood of detecting dual infections.
MATERIALS AND METHODS
Study population.Six hundred women from a high-risk population in Mbeya, Tanzania, were enrolled between September and November 2000 for a prospective HIV Superinfection Study (HISIS) within the Mbeya Medical Research Programme. The women have been working in bars along the Trans-African highway and are highly exposed to subtypes A, C, and D, the three predominant HIV-1 subtypes circulating in East Africa. After giving informed consent, each participant provided blood samples at enrollment and every 3 months thereafter for a period of up to 4 years. In 2001, the HISIS project provided health care to all participants, which included the treatment of all acute infectious diseases, screening for and treatment of sexually transmitted diseases, and since 2003, cotrimoxazole prophylaxis for women with CD4 counts below 200 (27). Since September 2004, all women with low CD4 counts have been referred for antiretroviral treatment within the Southern Highland Care and Treatment Programme.
Sample processing and MHAacd.HIV-1 antibody testing and the isolation of peripheral blood mononuclear cells (PBMCs) were done in Mbeya, Tanzania. The HIV-1 status of each patient was determined by using two diagnostic HIV enzyme-linked immunoassay tests (Enzygnost Anti HIV1/2 Plus [Dade Behring, Liederbach, Germany] and Determine HIV 1/2 [Abbott, Wiesbaden, Germany]). Discordant results were resolved by Western blotting (Genelabs Diagnostics, Geneva, Switzerland). PBMCs were isolated by Histopaque-1077-1 (Sigma Diagnostics, Taufkirchen, Germany) density gradient centrifugation and shipped in liquid nitrogen to the Tropical Institute in Munich, Germany, where DNA extraction, MHAacd, PCR amplification, and cloning were performed. For HIV-positive samples, DNAs were typically extracted from 200,000 PBMCs (High Pure viral nucleic acid kit; Roche Diagnostics, Mannheim, Germany), and screening for the HIV-1 subtype was done by a multiregion hybridization assay (MHAacd) (11) on an ABI 7700 sequence detection system (Applied Biosystems, Darmstadt, Germany). Plasma viral loads were determined with RNAs extracted from plasma by use of an Amplicor HIV-1 MONITOR test, version 1.5 (Roche Molecular Systems, Branchburg, N.J.).
In-depth analysis of participant 507.One participant who appeared to be multiply infected, as assessed by MHAacd, was studied at 0, 3, 6, 9, and 12 months.
To prove that the serial samples came from the same individual, we used an AmpFeSTR Profiler kit (Applied Biosystems, Foster City, Calif.), based on a method using length variations of short tandem repeat loci in the human genome for unique human identification. With this PCR amplification kit, the repeat regions of nine short tandem repeat loci and a segment of the X-Y homologous gene amelogenin were coamplified, with one primer of each locus-specific primer pair being labeled with a fluorescent dye (35). Electrophoresis of the PCR products to separate the alleles according to size and fluorescence detection were done on an ABI PRISM 3100 genetic analyzer (Applied Biosystems, Foster City, Calif.). The multiple infection was confirmed by sequencing multiple clones at each of three genome regions for each time point.
PCR amplification.Using nested PCR, we generated amplicons of 1.1 to 1.4 kb from regions 1 (gag/pol), 2 (vpu/gp120), and 3 (gp41/nef). Region 1 stretches from the p17 protein gene in gag to the beginning of the pol gene, region 2 starts in the middle of vpu and covers all of the gp120 gene, and genome region 3 contains the gp41 gene and half of the nef coding region. To increase the sensitivity for recovering all viral quasispecies, we used a multiple primer approach. Two outer primer pairs and two inner primer pairs were chosen and used in four different combinations in four separate PCRs. The PCR mixtures contained a 200 μM concentration of each deoxynucleoside triphosphate, 1.5 mM MgCl2, 15 mM Tris-HCl, pH 8.0, 50 mM KCl, 0.2 μM (each) primers, and 3.75 U of AmpliTaq Gold (Applied Biosystems, Darmstadt, Germany). Five microliters (for genome regions 1 and 2) or 10 μl (for genome region 3) of extracted DNA was added to the first-round PCRs, while 1 μl (for regions 1 and 3) or 3 μl (for region 2) of the first-round products was enough for the second round of amplification. The thermocycler protocol was as follows: 95°C for 10 min; 35 cycles of 95°C for 10 s, annealing for 30 s, and 72°C for 2 min; and one cycle of 72°C for 10 min. Figure 1 shows the locations of the second-round amplicons, and the primers and annealing temperatures are described in Table 1. Five microliters of each second-round PCR product was subjected to agarose gel electrophoresis, and the products were visualized by ethidium bromide staining.
Locations of second-round PCR amplicons in the HIV-1 genome. Region 1 refers to gag/pol, region 2 refers to vpu/gp120, and region 3 refers to gp41/nef. Each fragment was generated by a multiple-primer PCR approach and was between 1.1 and 1.4 kb long. Numbers below the fragments indicate HXB2 locations.
PCR primer sequences with HXB2 locations and annealing temperatures for regions 1, 2, and 3
Cloning and sequencing.The four PCR products for each region were pooled, and 4 μl of the mix was cloned into the pCR4-TOPO vector according to the instructions of a TOPO TA cloning kit for sequencing (Invitrogen, Karlsruhe, Germany). At least 24 clones were picked and grown overnight in 3.3 ml of Luria-Bertani broth (Invitrogen, Karlsruhe, Germany) plus kanamycin at 37°C. After purification of the plasmid DNA (QIAprep spin miniprep kit; QIAGEN, Hilden, Germany), 3 μl was used for enzymatic digestion with 8 U of EcoRI (New England Biolabs, Frankfurt am Main, Germany) and 10× NEB (5 mM NaCl, 10 mM Tris-HCl, 1 mM MgCl2, 0.0025% Triton X-100, pH 7.5). Incubation was performed at 37°C for 2 h, and agarose gel electrophoresis was performed with 10 μl of the digestion products. Clones with inserts of the expected size were shipped to the Henry M. Jackson Foundation, Rockville, Md., where sequencing and phylogenetic analysis took place. Between 17 and 25 positive clones for each region and time point were sequenced by using fluorescent dye terminators (Applied Biosystems, Foster City, Calif.), and electrophoresis and data collection were done on an ABI Prism 3100 genetic analyzer (Applied Biosystems, Foster City, Calif.).
Phylogenetic analysis.The sequences were assembled with Sequencher, aligned with reference sequences of all HIV-1 subtypes, and phylogenetically analyzed with the MEGA, version 2.1, program (20) and the SEQBOOT, DNAPARS, DNADIST, NEIGHBOR, and CONSENSE modules of the PHYLIP package incorporated in the GDE interface (30). Distance analysis was performed with the Kimura two-parameter model and a transition/transversion ratio of 2.0. Bootscanning (25, 29) and informative site analysis software, which tabulates subtype associations in 10-bp increments, were used to identify recombinant breakpoints, which were confirmed by visual inspections of the sequence alignment and/or by subregion trees.
Evaluation of multiple-primer PCR approach.The last available sample from the patient (at 12 months) and the vpu/gp120 genome region were chosen for a separate experiment to determine whether the multiple-primer PCR approach was more sensitive than a single nested PCR for the recovery of the viral quasispecies. Extracted DNAs were separately amplified with the four different primer combinations by nested PCR as described for the first set of experiments, but the four second-round amplicons were screened by MHA and cloned individually. At least 20 clones from each amplification were sequenced and phylogenetically analyzed by the same procedures as those used previously.
RESULTS
Within the HIV Superinfection Study (HISIS) in Mbeya, Tanzania, 407 of 600 (67.8%) female bar workers tested HIV-1 positive at baseline and were screened for their HIV-1 subtype every 3 months by use of a multiregion hybridization assay (MHAacd). This genotyping assay is based on subtype-specific fluorescent probes binding to HIV DNA in five genome regions (11). Among the many potential dual infections that have been identified during this study, one participant with late-stage HIV-1 infection was selected for analysis. Serological assays at enrollment indicated that the woman had a cured syphilis infection and was, as were 86% of all bar workers, infected with herpes simplex virus 2. She was healthy at enrollment, reported short episodes of fever and diarrhea at the 6- and 9-month follow-up visits, and presented with a very reduced general condition and AIDS-defining symptoms such as weight loss, chronic fever, and cough (with a duration of longer than 1 month) at the 12-month follow-up. At all time points, her viral load was above 500,000 copies/ml; her CD4 count only became available later in the study. The woman missed the subsequent follow-up at month 15. A verbal autopsy conducted among relatives revealed an AIDS-related opportunistic infection as the cause of death.
The initial MHAacd screening of the five available time points showed a dual infection with subtypes A and C. It was interesting that the recognition of the different subtypes was inconsistent in the longitudinal follow-up (data not shown).
For confirmation of the dual infection, three fragments from each time point, each of which was approximately 1.4 kb, were amplified, cloned, and subsequently sequenced. The three fragments, for the gag/pol (1), vpu/gp120 (2), and gp41/nef (3) regions, covered all genome regions probed by the MHA. In total, 298 sequences were derived, with between 17 and 25 sequences from each time point and genome region. A phylogenetic analysis identified a highly complex viral quasispecies in all three regions, including several different recombinant forms. The details of the phylogenetic analysis are described below.
Region 1 (gag/pol).Ninety-eight sequences of the gag/pol genome region were drawn in equal numbers from the 0-, 3-, 6-, 9-, and 12-month samples. As shown in Fig. 2, three molecular forms were identified, namely, a pure subtype A form (form I) and two AC recombinant forms (forms II and III). Supported by a significant bootstrap value, the A portions of the three forms were similar, indicating that the A strain was one of the parental strains of the two AC recombinants. Both recombinants had similar structures but differed in the position of the first breakpoint. A subregion tree (bootstrap 100) (Fig. 3) indicated that the C portions originated from the same parental C strain, although it was not detected as a pure subtype C strain among the analyzed sequences. Further confirmation of these findings was done by a distance analysis of each subregion. The mean distances between the A portions of each form and between the C portions of forms II and III did not differ significantly from the heterogeneity within the sequence clusters of each form and were all below 3.0%. Form II was the predominant strain, representing 60% of all sequences, followed by form I, the pure A strain, at 34%. Form III was observed more rarely than the other forms (6% of all sequences), and only at the 6- and 9-month follow-ups. No obvious trend in the proportions of the three forms over time could be observed, although their proportions did fluctuate between time points (Fig. 2C).
(A) Neighbor-joining (NJ) phylogenetic tree of HIV-1 region 1 (gag/pol) nucleotide sequences taken 0 (circles), 3 (diamonds), 6 (squares), 9 (triangles), and 12 (pentagons) months after enrollment. Numbers at branch nodes refer to bootstrap values; only values of>70% are shown. (B) Structures of the three molecular forms confirmed by bootscanning and subregion trees as described in the legend to Fig. 3. (C) Histogram representing the proportion of each molecular form in all sequences of each time point.
(A) Bootscans of the three molecular forms in region 1 (gag/pol), with the y axis referring to the bootstrap values and with recombinant breakpoints being marked on the x axis. (B) NJ phylogenetic subtrees of the genome regions between the breakpoints with a subset of analyzed sequences. Forms I (circles), II (triangles), and III (squares) cluster together differently according to their subtypes in each subregion. Letters next to the sequence clusters refer to HIV-1 subtypes.
Region 2 (vpu/gp120).Figure 4 shows the viral diversity in vpu/gp120. Among the 104 sequences, at least 12 different molecular forms were identified, making vpu/gp120 the most diverse region studied for this individual. The patient was apparently triply infected, with one strain being subtype C (form I) and with two divergent subtype A strains, A1 (form II) and A2 (form III). The mean distance between the two A clusters was 9.2% higher than the heterogeneity within the sequence of each A variant (about 2%), confirming the existence of two distinct A strains. However, intrasubtype recombination events following infection probably led to less divergence between the two A forms compared to the distance between A reference strains and the two A clusters (about 15%). The three pure subtypes were found to be the dominating strains at all time points, with A2 representing 43%, C representing 28%, and A1 representing 20% of all sequences. Furthermore, four intersubtype recombinants (forms IV to VII) and five different A1A2 recombinants (combined within form VIII) were identified. Only one sequence could be identified for each of these recombinants. The construction of subregion trees (Fig. 5) with high bootstrap values and visual inspection confirmed that all recombinants emerged through a single crossover event between two of the three parental strains (forms I to III). Fluctuations in the frequencies of the molecular forms were high. Forms I to III were present at visit 0, 1, and 3, but at visit 2 only form III (subtype A2) persisted and represented almost all sequences. At visit 4, only forms I and III could be detected. The different A1C and A2C recombinant forms (IV to VII) emerged and disappeared at various visits, suggesting that none had a distinct selective advantage. The A1A2 recombinant strains could be detected at all visits except for the last one.
(A) NJ phylogenetic tree of HIV-1 region 2 (vpu/gp120) nucleotide sequences taken 0 (circles), 3 (diamonds), 6 (squares), 9 (triangles), and 12 (pentagons) months after enrollment. Numbers at branch nodes refer to bootstrap values; only values of >70% are shown. (B) Structures of the molecular forms confirmed by bootscanning and subregion trees (see Fig. 5). (C) Histogram representing the proportion of each molecular form in all sequences of each time point and its fluctuation with time.
Phylogenetic subtrees for recombinant fragments IV, V, VI, and VII (see Fig. 4) of the molecular forms found in region 2 (vpu/gp120). Only one or two sequences of each form, the outgroup, and A, C, and D reference sequences are included in the trees. Numbers at branch nodes refer to bootstrap values. The HXB2 location is given for each fragment.
Region 3 (gp41/nef).The analysis of the fragment spanning gp41 and the beginning of nef included 96 sequences. Figure 6 illustrates the structures and temporal fluctuations of the four molecular forms detected for this genome region. The most abundant was form I, a subtype A strain, but three different AC recombinant forms (forms II, III, and IV) were also found. The A and C portions of all of these molecular forms were from the same A and C parental strains, as confirmed by using subregion trees (Fig. 7) and distance analysis. The sequence heterogeneity within each form, as well as the distance between the A portions of the viral variants and between the C portions of forms II, III, and IV, was below 3.0%. In contrast to the other genome regions studied, form I, which was the A subtype, dominated at all time points and represented 84% of all sequences. Form II could represent the parental form for forms III and IV, which would require one and two crossovers, respectively. These were the most rare forms in the quasispecies. The hatched fragment in form IV could not be classified, probably because it was quite short.
(A) NJ phylogenetic tree of HIV-1 gp41/nef nucleotide sequences taken 0 (circles), 3 (diamonds), 6 (squares), 9 (triangles), and 12 (pentagons) months after enrollment. Numbers at branch nodes refer to bootstrap values; only values of >70% are shown. Sequence “h” was hypermutated. (B) Structures of the four molecular forms confirmed by bootscanning and subregion trees (see Fig. 7). Numbers below the breakpoints and the edges of fragments indicate HXB2 locations. (C) Histogram representing the proportion of each molecular form in all sequences of each time point and its fluctuation with time.
Phylogenetic subtrees for the recombinant fragments (see Fig. 6) of the molecular forms in region 3 (gp41/nef). Only one or two sequences of each form, the outgroup, and A, C, and D reference sequences are included in the trees. Numbers at branch nodes refer to bootstrap values. The HXB2 location is given for each fragment.
Evaluation of multiple PCR primer approach.A separate experiment was set up to explore whether the multiple primer approach had actually augmented the recovery of the diverse viral quasispecies from this multiply infected individual. After amplification of the vpu/gp120 fragment from the sample taken at 12 months with the four different primer combinations (a, b, c, and d), the PCR products were cloned separately to identify the strains that were amplified by each of the four reactions. At least 20 clones from each reaction were sequenced. As shown in Fig. 8, the analysis of 103 sequences revealed that the sample contained at least three different molecular forms. Subtype A contributed 81% and subtype C contributed 18% of all analyzed sequences. An AC recombinant of the parental strains was found only once.
(A) NJ phylogenetic tree of HIV-1 vpu/gp120 nucleotide sequences obtained from visit 4 only. A (diamonds), B (squares), C (triangles), and D (pentagons) refer to the sequences obtained from the corresponding PCR products generated by the four different primer combinations. For the fifth group, all PCR products were pooled (circles) before cloning and sequencing. (B) Structures of the three molecular forms and primer combinations that recovered them.
The separate analysis of each of the four primer combinations revealed a significant amplification bias. With setups b and d, subtype C could not be detected at all, and with setups a and c and the original experiment, in which all PCR products were pooled before cloning, subtype C was recovered and represented at least 20% of the analyzed sequences.
Since the number of sequences that can be analyzed by cloning and sequencing is limited, the more sensitive MHAacd assay, which can detect unequal mixtures of 1 in 3,000 (11), was used to prove the presence or absence of the two molecular forms in the four setups. Similar to the sequencing results, the MHA confirmed the presence of strains A and C in setups a and c and the absence of strain C in setups b and d (data not shown).
DISCUSSION
An evaluation of blood samples collected every 3 months during an HISIS participant's last year of life revealed multiple concurrent infections with at least three different HIV-1 strains. MHAacd indicated an AC dual infection. Cloning and sequencing confirmed this finding and, furthermore, distinguished two distinct A strains in addition to the subtype C strain and several recombinant forms. However, the three parental strains could only be recovered by using genome region 2. In this and other multiple infection cases from HISIS studied to date, it has often been observed that the complexity of molecular forms varies by the genome region examined (data not shown), and the fact that more viral diversity was recovered for vpu/gp120 than for the other genome regions studied is not unusual: it may be that the recombination and selection of strains “purify” some genome regions more than others in the setting of dual infection. In any case, a highly complex viral quasispecies did emerge during this late-stage infection, including inter- and intrasubtype recombination, with vpu/gp120 being the most diverse genomic region. The tendency towards more homogeneity among strains in the late stage of AIDS, as reported earlier (7), could not be observed in this case.
Whether the patient was simultaneously or sequentially infected with the three HIV-1 strains cannot be determined. However, the woman was highly exposed to multiple partners for several years, which makes a superinfection a distinct possibility. Another case of superinfection in this cohort, in a participant who seroconverted during the study, has already been confirmed (McCutchan et al., submitted for publication).
The high viral load of >500,000 copies/ml in all follow-ups was most likely the result of the late-stage HIV-1 infection but may also have been caused by the concurrent infection with multiple HIV-1 strains, as shown earlier (10).
This report confirms that multiple HIV-1 infections can occur and shows that several HIV-1 strains can persist over time, without any single strain gaining predominance. This phenomenon may be more likely when immune responses are waning and there is minimal immune pressure. However, we do not claim to have determined the exact proportions of the circulating forms, and the existence of undetected viral strains is even within the bounds of possibility. The exact amplification of all viral quasispecies can be hampered by several factors, as follows. (i) Every PCR utilizes a limited number of proviral copies. In our case, 27 to 66 proviral DNA copies extracted from 400,000 to 1,000,000 PBMCs were used for the primary PCR (9). (ii) Primer bias exists, which we have tried to minimize but have definitely not eliminated. (iii) Finally, the number of clones analyzed can have an effect on the amplification of strains.
The choice of primer sequences can bias the recovery of strains and may preclude the detection of multiple infections altogether. To our knowledge, this is the first systematic evaluation of single versus multiple primer approaches for the recovery of diverse quasispecies. We have demonstrated that two of the primer combinations used for the vpu/gp120 genome region only amplified the subtype A strain, whereas the other two were able to detect a dual infection by recovering subtypes A and C. Due to the rapid HIV-1 evolution in individuals and the community enhancing viral diversity, mismatches between sample and primer sequences may occur frequently, and thus the multiple primer approach is an efficient method for increasing the likelihood of detecting multiple infections compared to the conventional, single nested PCR.
To ensure a more accurate reproduction of the circulating quasispecies, even more than two primer combinations per round could be employed in future PCR approaches.
The serial sampling format may also provide better detection of coinfections than a cross-sectional sampling frame, taking into account the high fluctuation and recombination of the molecular forms over time that we described in this case.
In addition, our data show once more that HIV-1 recombination is a common phenomenon in multiply infected individuals. Within individuals, recombination can lead to radically different genomic combinations and will have a much more dramatic impact on viral evolution than do nucleotide substitutions.
We believe that the characterization of the viral quasispecies emerging within an HIV-1-infected individual during the course of infection and their correlation with host immune responses will be informative in the quest for an HIV-1 vaccine that protects against multiple HIV-1 subtypes.
ACKNOWLEDGMENTS
This work was supported by the European Commission (DG XII, INCO-DC) and by a cooperative agreement between the Henry M. Jackson Foundation for the Advancement of Military Medicine and the U.S. Department of Defense.
The views and opinions expressed herein do not necessarily reflect those of the U.S. Army or the Department of Defense.
We thank the excellent staff at the Mbeya Medical Research Programme who conducted the HISIS study, especially Vera Kleinfeldt, Frowin Nichombe, Weston Assisya, and Clemence Konkamkula, and all participants in the study.
FOOTNOTES
- Received 11 November 2004.
- Accepted 15 February 2005.
- ↵*Corresponding author. Mailing address: Department of Infectious Diseases and Tropical Medicine, Ludwig-Maximilians-University, Leopoldstr. 5, 80802 Munich, Germany. Phone: 49 89 2180-3925. Fax: 49 89 33 60 38. E-mail: martina.gerhardt{at}lrz.uni-muenchen.de.
REFERENCES
- American Society for Microbiology
