点击显示 收起
Swedish Institute for Infectious Disease Control, Solna
Microbiology and Tumor Biology Center, Stockholm, Sweden
St. Jude Children's Research Hospital, Memphis, Tennessee
Streptococcus pneumoniae isolates of serotypes 1, 4, 6B, 7F, 14, and 19F belonging to clonal types with known invasive disease potential in humans were used to infect C57BL/6 and BALB/c mice. Most isolates were able to colonize the nasopharynx for 7 days. One serotype 19F isolate of the clonal type ST162 had higher bacterial numbers than other isolates and clonal types of the same serotype. Serotype 4 clones caused the most-severe invasive disease, whereas serotype 1 clones caused low-level bacteremia without disease symptoms. BALB/c mice were more likely than C57BL/6 mice to develop meningitis. Disease kinetics varied significantly between clonal types. Although most induced a robust tumor necrosis factor response, some isolates of serotype 1 and 7F did not, suggesting that invasive disease caused by different clonal types may result in different degrees of host response. Capsular serotype, other clonal properties, and host factors are important for the development of pneumococcal disease.
Some capsular serotypes of Streptococcus pneumoniaesuch as 1, 4, and 7Fare primarily found in patients with invasive disease and are rarely found in healthy carriers, whereas other serotypessuch as 6B, 14, 19F, and 23Fare frequent colonizers in the nasopharynx of healthy preschool children but also may cause invasive disease. Also, particular serotypessuch as 3, 6B, and 19Fare more often associated with severe invasive disease and mortality in humans than are other serotypessuch as 1 and 7F [14]. The pneumococcal capsule provides an antiphagocytic property to the bacterium, and it is likely that different capsular polysaccharides induce different degrees of antiphagocytic responses. Furthermore, capsular polysaccharides may be recognized by C-type lectins in the spleen, and this recognition promotes clearance of the bacteria from the bloodstream [5]. It is possible that different capsular chemistries affect the interaction with these lectins. Therefore, the capsular polysaccharides act as dominant virulence factors in pneumococci, and epidemiological studies have suggested that the capsular serotype, rather than other clonal properties, determines the potential attack rate of the baterium [6, 7]. However, in a study comparing carrier and invasive isolates from the Stockholm area during a 1-year period, we identified clones of the same serotype that differed in their occurrence in carriers, compared with their occurrence in patients with invasive disease, and their ability to cause invasive disease, and this finding suggests that clonal properties other than capsular serotype also affect the ability of the bacterium to cause invasive disease in humans. These clonal differences may be diverse, such as variation in the distribution and composition of horizontally acquired pathogenicity islands [8].
Because variation in host susceptibility is not taken into account when the attack rate of pneumococci is assessed in an epidemiological study, such studies need to be complemented by infection studies in inbred mice. In the present study, we examined inbred mice after intranasal (inl) or intraperitoneal (ip) challenge with 13 isolates belonging to 10 clonal types of serotypes 1, 4, 6B, 7F, 14, and 19F with known odds ratios (ORs) for invasiveness [4]. Our results verify that the capsular serotype is important for disease outcome but that other clonal properties influence colonization and the invasive disease potential to a significant extent. Furthermore, disease severity in humans was related to the degree of proinflammatory response in mice. Finally, our data suggest that the ability of the pneumococci to cause meningitis, the most severe form of invasive pneumococcal disease, depends to a significant degree on host factors.
MATERIALS AND METHODS
Bacterial strains used.
The pneumococcal strains used were the sequenced S. pneumoniae invasive TIGR4 strain and 13 clinical pneumococcal isolates selected to provide a wide range of clones with different abilities to cause invasive disease in humans (table 1). These clones were identified in 2 previous molecular epidemiological studies in which we compared nasopharyngeal isolates from healthy children with invasive isolates from the Stockholm area obtained during the same year and thereby were able to obtain estimates on ORs for invasiveness for a number of clones [4, 9].
Creation of bioluminescent derivatives of the clinical isolates.
Stable bioluminescent derivatives of the clinical isolates were created using chromosomal DNA from the highly bioluminescent S. pneumoniae strain D39 Xen 7. D39 Xen 7 carries the Tn4001 luxABCDE Kmr cassette, as described elsewhere [10, 11]. This results in production of luciferase and its substrate and emission of light, which is monitored in mice with a charge-coupled device camera (Xenogen). Transformants were selected on blood plates containing kanamycin (400 g/mL) and were grown at 37°C in 5% CO2.
Experimental pneumonia and sepsis model in mice.
Experiments performed in Memphis, Tennessee, were approved by the Animal Care and Use Committee at the St. Jude Children's Research Hospital. Female BALB/c and C57BL/6 mice, 45 weeks old, were obtained from Jackson Laboratory. Experiments performed in Stockholm, Sweden, were approved by the Ethical Committee for Animal Experiments in Stockholm and were conducted in accordance with the European Communities Council Directive 86/609/EEC. Female C57BL/6 mice, 46 weeks old, were bred and housed in the animal facility of the Microbiology and Tumor Biology Center at the Karolinska Institute (Stockholm, Sweden). All mice were kept on a 12-h light-dark cycle and had access to standard food and tap water as desired.
Mice were anesthetized by light inhalation of isoflurane (Forene; Abbott) and were inoculated either ip or inl with live S. pneumoniae ( 5 × 106 cfu), grown in C+Y medium [12], in a volume of 200 L or 20 L, respectively. For ip experiments, the original pneumococcal strains were used, and for inl experiments, the created bioluminescent strains were used. Most strains without bioluminescence were also tested in the inl model (data not shown). We found no evidence that the lux gene has an effect on virulence. Mice were monitored for 7 days to assess their health status, and each day they were given a clinical score for disease severity. Disease severity was graded on a scale of 05, with 0 meaning the mouse was healthy and exhibited normal motor activity, 1 meaning the mouse showed decreased spontaneous activity and a hunched posture, 2 meaning the mouse had tremor and piloerection and loss of vigilance, 3 meaning the mouse turned upright in >5 s after being inverted, 4 meaning the mouse had positional passivity and did not turn upright after being inverted, and 5 meaning the mouse did not move. A mouse was considered to have meningitis if it presented with clinical symptoms, including ataxia and imbalance, combined with bioluminescence from the skull. Mice were killed at 72 h or 168 h after challenge or when they had a disease severity score >3. In the experiments on short-time kinetics after ip challenge, mice were killed randomly at 4 or 6 h after challenge.
Distribution and quantification of bacteria in mice and determination of tumor necrosis factor (TNF) in blood.
To monitor bioluminescent S. pneumoniae infections, mice were anesthetized with isoflurane and imaged at different time points after challenge, by use of the IVIS Imaging System 100 Series (Xenogen). Blood was collected, serially diluted in PBS, and plated on blood agar plates. Bacteremia was estimated from the number of colony-forming units found per milliliter of blood. Bacterial burden of the lungs was estimated from the number of colony-forming units found per gram of homogenized lung tissue. Nasopharyngeal-tracheal lavage samples were obtained by flushing 0.2 mL of PBS through the tracheas of dead mice and collecting the first 1020 L of liquid that could be extracted from the nares. Colonization in the nasopharynx was estimated from the number of colony-forming units found per milliliter of nasopharyngeal-tracheal lavage samples. TNF was measured in mouse serum using commercial ELISA kits (R&D Systems) at 4 and 6 h after ip challenge.
Data analysis.
For survival studies, differences were analyzed by the Kaplan-Meier log-rank test. Data on colony-forming units in blood after ip challenge and TNF responses were analyzed using Student's t test. For measurements of bioluminescence, Living Image software (version 2.50; Xenogen) was used, and comparisons were made with Student's t test. For comparisons of frequencies, Fisher's exact test was used. P < .05 was considered to be statistically significant.
RESULTS
Clonal and serotype-specific properties in colonization and frequency of invasive disease.
Representatives of pneumococcal clones belonging to 6 serotypes (table 1) that have known ORs for invasiveness [4] were genetically tagged with the lux gene and inoculated into C57BL/6 and BALB/c mice.
Colonization after inl challenge.
All but 3 of the pneumococcal isolates colonized 100% of the C57BL/6 and BALB/c mice at 168 h after challenge (data not shown), as monitored by measurement of colony-forming units in nasopharyngeal-tracheal lavage samples. The exceptions were ST2281, ST3061, and ST12414, which had a weak ability to colonize.
Identification of a hypercolonizing 19F clone.
Three isolates of serotype 19F belonging to 2 different clonal types were tested for colonization in the inl mouse model. Tracking of bacteria using bioluminescence showed that the invasive and carrier isolates of ST16219F had a remarkably strong ability to colonize both the C57BL/6 and the BALB/c mice, a trait not found for the other clone (ST42519F) (figure 1A and 1B). In mice infected with ST16219F, the intensity of bioluminescence from the nose increased during the first 7 days after challenge. The intensity remained unchanged through day 9 and disappeared by day 16 after challenge. In BALB/c mice, at day 7 after challenge, the mean radiance was 1291 photons/s/cm2/steradian (photons per second leaving a square centimeter radiating into an angle of 1 steradian) for those infected with ST16219F and was -361 photons/s/cm2/steradian (i.e., no detectable emission of light) for those infected with ST42519F (P < .005). In C57BL/6 mice, the mean radiance was 497 and 70 photons/s/cm2/steradian, respectively (P = .053). Calculation of radiance was normalized to the background radiance of the mice, so negative values were possible. We confirmed the bioluminescence data by obtaining nasopharyngeal-tracheal lavage samples from mice at day 7 or day 28 after challenge, and bacteria were found in the nasopharynx at both time points (data not shown). A mean of 1 × 106 cfu/mL was found in the C57BL/6 mice challenged with ST16219F, compared with a mean of 5 × 104 cfu/mL in the C57BL/6 mice challenged with ST42519F. Similar differences were found in the BALB/c mice (data not shown).
Pneumonia, bacteremia, and survival after inl challenge.
In C57BL/6 mice, the 2 ST2054 isolates were the most able to cause lethal systemic disease after inl challenge, and ST1386B was the next most able (figure 2A). Members of these 2 clonal types were the only isolates that caused high levels of bacteremia and death (figure 2A and 2B). Once bacteremia was established, the mice were unable to clear ST2054 infection and died of septicemia, whereas clearance of bacteria was evident in infection with ST1386B. Thus, whereas most mice infected with ST1386B were bacteremic at 24 h after challenge, no mouse was bacteremic at 168 h after challenge, and no mouse died >48 h after challenge (figure 2A and 2B).
Clones causing systemic disease (ST2054 and ST1386B) also caused pneumonia, as was revealed by the observation of colony-forming units in lung tissue (figure 2C). However, at 168 h after challenge, colony-forming units of some other clones (ST1917F, ST30714, ST16219F, and ST42519F) could also be recovered from lung tissue of mice that did not have clinical symptoms (figure 2C). The invasive ST30714 isolate caused pneumonia in 20% of mice, whereas the carrier ST55514 isolate did not cause pneumonia in any mouse. Although the majority of mice infected with ST16219F developed pneumonia (figure 2C), few developed bacteremia (figure 2B), which is in accordance with the results of previous studies [1315].
In contrast, BALB/c mice showed an increased susceptibility to pneumonia after inl challenge with ST55514, compared with C57BL/6 mice. Of the BALB/c mice, 50% (5/10) developed pneumonia, whereas none (0/15) of the C57BL/6 mice had developed pneumonia by 168 h after challenge (P < .05). Furthermore, BALB/c mice showed a tendency to be more susceptible to ST1386B infection than C57BL/6 mice: only 70% (7/10) of the BALB/c mice were alive at 168 h after challenge, compared with 93% (14/15) of the C57BL/6 mice (2 = 2.1; P = .15).
Different TNF responses in C57BL/6 mice after ip challenge.
There were large differences between serotypes and between clones in their abilities to grow in blood, to mount an inflammatory response, and to cause death in C57BL/6 mice after ip challenge (figure 4AC). As was expected, the most virulent isolates after inl challenge (those of ST2054) were also highly virulent after ip challenge. Members of this clonal type induced high levels of bacteremia and high levels of TNF in plasma (figure 4A and 4B). Signs of disease were seen at 6 h after challenge. Fatal sepsis led to a median survival time of 14 h (figure 4C). ST1386B and ST1766B showed a similar pattern of rapid, fatal septicemia that was associated with high levels of TNF. Both ST42519F and ST16219F induced an increased TNF response, but the TNF response in mice at 6 h after challenge was much stronger with the hypercolonizing ST16219F isolate than with ST42519F (P < .0001) (figure 4B). The level of bacteremia was only slightly lower after challenge with ST42519F than with ST16219F (median, 3.60 × 106 vs. 9.40 × 106 cfu/mL of blood at 4 h after challenge [P = .41] and 2.00 × 107 vs. 5.00 × 107 cfu/mL of blood at 6 h after challenge [P = .79]), and the growth rate was the same for both clones (figure 4A). The median survival times were 19 h after challenge with ST16219F and 26 h after challenge with ST42519F (median survival ratio, 1.37 [95% confidence interval, 1.0061.731]; 2 = 2.15; P = .14) (figure 4C).
Significantly, isolates belonging to clonal types of serotypes 7F (ST1917F) and 1 (ST2281 and ST3061) induced a very low TNF response (figure 4B), even though the number of bacteria in the bloodstream was relatively high. The number of colony-forming units in blood remained unchanged between 4 and 6 h after challenge with the ST1917F isolates, which suggests that there was an effective clearing of bacteria from the host even in the absence of a TNF response. However, eventually all mice developed fatal sepsis within 40 h after challenge with these isolates, with a median survival time of 17 h.
The median survival times for the C57BL/6 mice infected with ST2281 and ST3061 were 44 h and 168 h, respectively (2 = 34.21; P < .05) (figure 4C). There were also significant differences in the overall survival rate after 168 h: 0% after challenge with the ST2281 clone and 32% after challenge with the ST3061 clone. This finding exemplifies that isolates of the same serotype but of a different clonal type may differ in their ability to escape the process of host clearance during systemic infection.
ST12414 and ST55514 were the least virulent isolates after ip challenge (figure 4C). ST12414 induced a low level of bacteremia that correlated to no TNF response at 4 and 6 h after challenge (figure 4B). The survival rate was 50%, with a median survival time of 143 h. ST55514 was less virulent, with a 90% survival rate and a median survival time of >216 h (2 = 3.9; P < .05) (figure 4C).
DISCUSSION
We have recently shown that pneumococci belonging to different serotypes and clones can be divided into 3 major groups on the basis of their ability to cause invasive disease in humans [4]. The first group includes serotypes most commonly found in invasive disease and rarely found in carrierssuch as serotypes 1, 4, and 7Fthat have a high invasive disease potential. Although mortality in humans has been shown to be very low (0%) for infections with serotype 1 and serotype 7F isolates, it has been demonstrated to be intermediate to high (10%) for infections with serotype 4 isolates [1, 18]. The second group includes serotypes commonly found both in invasive disease and in carrierssuch as serotypes 14 and 6Bthat have an intermediate invasive disease potential. Invasive disease attributed to infection with serotypes 14 and 6B isolates has been associated with intermediate (7%) and high (25%) mortality, respectively [1]. The third group includes serotypes primarily found in carrierssuch as serotype 19Fthat have a low invasive disease potential. Isolates belonging to this serotype were commonly found in carriers at Stockholm day care centers. Invasive 19F isolates from the same region were found relatively infrequently. Nevertheless, invasive disease mediated by serotype 19F has been associated with high APACHE scores and high mortality (25%) [1]. It is possible that pneumococcal variants that exhibit a high attack rate (first group) are more likely than opportunists (third group) to affect healthy individuals, and this explains why, for example, septicemia caused by a serotype 1 isolate usually is milder than that caused by a serotype 19F isolate. Alternativelyor in additionhost responses to different pneumococcal serotypes may differ significantly, and this difference in response can affect both the likelihood of developing invasive disease and disease severity.
Even though capsular polysaccharides represent a major pneumococcal virulence factor, we do not know to what extent the significant differences in invasive disease potential and severity that we and others [4, 6, 19] found for different serotypes are due to the capsule itself and/or to the closer genetic relationship that likely prevails for isolates of the same serotype. In our previous epidemiological study, we found several isolates of the same serotype but of different clonal types that differed in their ability to cause invasive disease. Also, isolates belonging to the same clonal type but of different serotypes, due to serotype switches, were found to have similar ORs for invasiveness [4].
In the present study, we infected C57BL/6 and BALB/c mice with a collection of isolates belonging to pneumococcal clonal types and serotypes with known ORs for invasiveness, as determined in our previous study [4]. Our data demonstrate significant differences in virulence not only between different serotypes but also between different clonal types of the same serotype. In addition, our data highlight the importance of host factors in disease outcome. The ST2054 isolates were the only ones to cause significant mortality after inl challenge. ST205 is a serotype 4 clone that is a common cause of invasive disease in humans in several countries and is rarely seen in healthy carriers. TIGR4, isolated in Norway, seemed to be more likely to cause pneumonia in mice than the Stockholm isolate belonging to the same clonal type (figure 2C). Whether this difference is due to the presence of the virulence-associated rlrA locus in TIGR4 [20] that is not present in the Stockholm isolate remains to be determined. Nearly all other clonal representatives used in the present study were unable to cause lethal invasive disease in mice, and only a limited number caused bacteremia after inl challenge. All mice died after ST2054 reached the bloodstream, whereas in infection with other clonal types, mice were able to clear the bacteria. One notable example was ST2281. This serotype 1 clone caused low-level bacteremia over a long period but did not cause serious disease.
Our findings that BALB/c mice were less susceptible to the most virulent isolates (those of ST2054) than were C57BL/6 mice but were more susceptible to some of the less virulent clones (those of serotypes 6B and 14) reveal the central importance of host responses in the likelihood of developing disease. It is likely that human polymorphisms in host defense functions affect the susceptibility to disease caused by different clonal types. Not only the likelihood of developing disease but also the type of disease might be host dependent. Particularly interesting in this respect was our finding that BALB/c mice were more likely to develop meningitis after challenge with the ST2054 and ST1386B isolates than were C57BL/6 mice. The slower progression of invasive disease and the prolonged period of bacteremia in BALB/c mice might be contributing factors in increasing the likelihood of developing meningitis. A slow progression of systemic disease might, for example, allow time for a more direct transmission of bacteria from the nasopharynx to the meninges [21].
Invasive pneumococcal disease in humans is likely preceded by a colonization phase in the nasopharynx. Little is known about bacterial and host factors that affect colonization. Our mice experiments show that there are significant differences between different clonal types and their ability to colonize. Most notable was our finding of a hypercolonizing 19F isolate belonging to clonal type ST162, a clone frequently found in carriers in the Stockholm area. This isolate appeared to be a hypercolonizer in both C57BL/6 and BALB/c mice, and the phenotype expressed seemed independent of the 19F capsule, because a 19F isolate of a different clonal type did not exhibit this phenotype. Of the 19F isolates, only the hypercolonizing ST16219F isolate was able to reach the bloodstream from the nasopharynx, albeit it did so transiently and at low numbers. Hence, the number of bacteria in the nasopharynx is likely to be an important parameter that affects the likelihood of developing invasive disease.
Interestingly, ip challenge with the serotype 1 and 7F isolates caused virtually no TNF response in mice. These serotypes, which are not included in the 7-valent conjugated pneumococcal vaccine, have been associated with low mortality and low APACHE scores in human systemic pneumococcal disease [1], and preliminary data suggest that they might act as primary pathogens in previously healthy individuals (K. Sjstrm, C. Spindler, . rtqvist, M. Kalin, A. Sandgren, and B. Henriques-Normark, unpublished data). In contrast, clonal types of serotypes 6B and 19F, which are associated with significant mortality in humans, caused a significant TNF response. Our data presented here, together with clinical data, therefore suggest that pneumococci of different clonal types might cause systemic disease via different mechanisms.
Acknowledgments
We thank Ingrid Andersson, Christina Johansson, Gunnel Mllerberg, and Karin Sjstrm, for technical assistance, and the clinical microbiological laboratories and the Department of Communicable Disease Control and Prevention in Stockholm, Sweden, for the invasive pneumococcal isolates.
References
1. Henriques B, Kalin M, Ortqvist A, et al. Molecular epidemiology of Streptococcus pneumoniae causing invasive disease in 5 countries. J Infect Dis 2000; 182:8339. First citation in article
2. Kalin M. Pneumococcal serotypes and their clinical relevance. Thorax 1998; 53:15962. First citation in article
3. Robbins JB, Austrian R, Lee CJ, et al. Considerations for formulating the second-generation pneumococcal capsular polysaccharide vaccine with emphasis on the cross-reactive types within groups. J Infect Dis 1983; 148:113659. First citation in article
4. Sandgren A, Sjostrom K, Olsson-Liljequist B, et al. Effect of clonal and serotype-specific properties on the invasive capacity of Streptococcus pneumoniae. J Infect Dis 2004; 189:78596. First citation in article
5. Kang YS, Kim JY, Bruening SA, et al. The C-type lectin SIGN-R1 mediates uptake of the capsular polysaccharide of Streptococcus pneumoniae in the marginal zone of mouse spleen. Proc Natl Acad Sci USA 2004; 101:21520. First citation in article
6. Brueggemann AB, Griffiths DT, Meats E, Peto T, Crook DW, Spratt BG. Clonal relationships between invasive and carriage Streptococcus pneumoniae and serotype- and clone-specific differences in invasive disease potential. J Infect Dis 2003; 187:142432. First citation in article
7. Brueggemann AB, Spratt BG. Geographic distribution and clonal diversity of Streptococcus pneumoniae serotype 1 isolates. J Clin Microbiol 2003; 41:496670. First citation in article
8. Brown JS, Gilliland SM, Holden DW. A Streptococcus pneumoniae pathogenicity island encoding an ABC transporter involved in iron uptake and virulence. Mol Microbiol 2001; 40:57285. First citation in article
9. Henriques Normark B, Christensson B, Sandgren A, et al. Clonal analysis of Streptococcus pneumoniae nonsusceptible to penicillin at day-care centers with index cases, in a region with low incidence of resistance: emergence of an invasive type 35B clone among carriers. Microb Drug Resist 2003; 9:33744. First citation in article
10. Bricker AL, Camilli A. Transformation of a type 4 encapsulated strain of Streptococcus pneumoniae. FEMS Microbiol Lett 1999; 172:1315. First citation in article
11. Orihuela CJ, Gao G, Francis KP, Yu J, Tuomanen EI. Tissue-specific contributions of pneumococcal virulence factors to pathogenesis. J Infect Dis 2004; 190:16619. First citation in article
12. Lacks S, Hotchkiss RD. A study of the genetic material determining an enzyme in Pneumococcus. Biochim Biophys Acta 1960; 39:50818. First citation in article
13. Takashima K, Tateda K, Matsumoto T, Iizawa Y, Nakao M, Yamaguchi K. Role of tumor necrosis factor alpha in pathogenesis of pneumococcal pneumonia in mice. Infect Immun 1997; 65:25760. First citation in article
14. Tateda K, Takashima K, Miyazaki H, Matsumoto T, Hatori T, Yamaguchi K. Noncompromised penicillin-resistant pneumococcal pneumonia CBA/J mouse model and comparative efficacies of antibiotics in this model. Antimicrob Agents Chemother 1996; 40:15205. First citation in article
15. Briles DE, Hollingshead SK, Paton JC, et al. Immunizations with pneumococcal surface protein A and pneumolysin are protective against pneumonia in a murine model of pulmonary infection with Streptococcus pneumoniae. J Infect Dis 2003; 188:33948. First citation in article
16. McCool TL, Weiser JN. Limited role of antibody in clearance of Streptococcus pneumoniae in a murine model of colonization. Infect Immun 2004; 72:580713. First citation in article
17. Autenrieth IB, Beer M, Bohn E, Kaufmann SH, Heesemann J. Immune responses to Yersinia enterocolitica in susceptible BALB/c and resistant C57BL/6 mice: an essential role for gamma interferon. Infect Immun 1994; 62:25909. First citation in article
18. Henriques Normark B, Kalin M, Ortqvist A, et al. Dynamics of penicillin-susceptible clones in invasive pneumococcal disease. J Infect Dis 2001; 184:8619. First citation in article
19. Mizrachi Nebenzahl Y, Porat N, Lifshitz S, et al. Virulence of Streptococcus pneumoniae may be determined independently of capsular polysaccharide. FEMS Microbiol Lett 2004; 233:14752. First citation in article
20. Hava DL, Camilli A. Large-scale identification of serotype 4 Streptococcus pneumoniae virulence factors. Mol Microbiol 2002; 45:1389406. First citation in article
21. van Ginkel FW, McGhee JR, Watt JM, Campos-Torres A, Parish LA, Briles DE. Pneumococcal carriage results in ganglioside-mediated olfactory tissue infection. Proc Natl Acad Sci USA 2003; 100:143637 [erratum: Proc Natl Acad Sci USA 2004; 101:6834]. First citation in article