Specimens. an organic buffer with a pka of 6.90 at 20 C. Use of. this buffer was first described by Pasculle et al. (12).

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Sept. 1981, p /81/ $02.00/0 Vol. 14, No. 3 Improved Semiselective Medium for Isolation of Legionella pneumophila from Contaminated Clinical and Environmental Specimens PAUL H. EDELSTEIN Division of Infectious Disease, Research and Medical Service, Veterans Administration Wadsworth Medical Center, Los Angeles, California 90073, and Department of Medicine, University of California, Los Angeles, California Received 11 February 1981/Accepted 23 April 1981 Eighteen types of media, all of which used charcoal yeast extract medium as a base, were tested for their ability to support the growth of a stock strain of Legionella pneumophila. Fifteen of these contained antimicrobial agents. Five antibiotic-containing media which performed best in growing the stock strain were then tested for their ability to grow L. pneumophila from clinical respiratory tract specimens. One of the media, BMPAa, which contained cefamandole, polymyxin B, anisomycin, an organic buffer, and a-ketoglutarate, performed better than the others. This medium greatly enhanced recovery of L. pneumophila from contaminated clinical specimens and, in conjunction with an acid wash treatment, facilitated recovery of the bacterium from potable water. One of the major difficulties in diagnosing Legionnaires disease is the inability to grow Legionella pneumophila from contaminated respiratory tract specimens. With few exceptions, this has limited the types of specimens yielding L. pneumophila to normally sterile fluids and tissues, such as transtracheal aspirates (TTA) and lung biopsies (7). Since L. pneumophila is often present in the sputum of patients with Legionnaires disease, as detected by direct fluorescent-antibody testing, a medium which would inhibit selectively the upper respiratory flora would permit the routine isolation of this organism from such sputum specimens (7). Also much needed is a medium which would facilitate the growth of L. pneumophila from environmental specimens, which often contain bacteria. Our laboratory has reported previously on the use of a semiselective medium containing vancomycin and polymyxin B in the isolation of L. pneumophila from three contaminated lung specimens (6). I now report on improvements to this medium which frequently enable the isolation of L. pneumophila from sputum specimens of patients with Legionnaires disease and also facilitate the recovery of the organism from environmental specimens. 298 MATERIALS AND METHODS Media. Eighteen types of media, all based on charcoal-yeast extract (CYE) medium, were tested for their ability to support the growth of a stock strain of L. pneumophila (Table 1) (8). Most of the media contained ACES buffer (Sigma Chemical, St. Louis, Mo.), an organic buffer with a pka of 6.90 at 20 C. Use of this buffer was first described by Pasculle et al. (12). Fifteen of the media were supplemented with antimicrobial agents; these were chosen on the basis of in vitro agar dilution susceptibility testing of L. pneumophila (16) and known activities against other microorganisms. Synergy testing, using CYE medium, was also used for testing some of the antimicrobial combinations against a stock strain of L. pneumophila (Wadsworth isolate 242D) (1). Use of a-ketoglutarate was suggested by James C. Feeley. Antimicrobial agents were filter sterilized with a 0.22-jim filter (Millipore Corp., Bedford, Mass.) and added to the basal CYE medium after it had been cooled to 40 C and after L-cysteine and ferric pyrophosphate had been added; the ph was then adjusted. The ACES buffer and the monopotassium salt of a-ketoglutarate (Sigma) were added to the basal CYE medium before steam sterilization at 121 C. The selenium, in the form of sodium selenate (ICN, Plainview, N.Y.), was also added to the basal medium before autoclaving. Anisomycin and polymyxin B sulfate were gifts from Pfizer Pharmaceuticals (New York, N.Y.); vancomycin hydrochloride, cefazolin sodium, and cefamandole lithium were gifts of Eli Lilly & Co. (Indianapolis, Ind.). The abbreviations used to describe the media were derived as follows: ACES buffer is designated by B, cefazolin by C, anisomycin by A, a-ketoglutarate by a, cefamandole by M, polymyxin B by P, vancomycin by V, and sodium selenate by S. In addition, since two different concentrations of cefamandole and three different concentrations of polymyxin B were tested in various combinations, subscripts have been added to the M and P abbreviations. No subscript for cefamandole means 4 jg/ml, and a subscript of 2 means 8,ig/

2 VOL. 14, 1981 TABLE 1. SELECTIVE MEDIUM FOR L. PNEUMOPHILA 299 Formulations of media testeda Component VP VAP BCYE BCYEa BCYEaS CAP BCAP BMPA BMPAa BMPAaS BMP2A BMP2Aa BMP4A BMP4Aa BM2PA BM2PAa BM2P2A BM2P2Aa a All media used a CYE medium base. ml. Similarly, P represents 80 U/ml, and P2 represents 160 U/ml of polymyxin B. For example, BMP2Aa medium is buffered CYE medium supplemented with 4,sg of cefamandole per ml, 160 U of polymyxin B per ml, anisomycin, and a-ketoglutarate. This information is summarized in Table 1. Five of the 15 types of antimicrobial agent-containing media were used for primary isolation of L. pneumophila from clinical and environmental specimens. These media were selected on the basis of superior stock strain growth or "sensitivity" (see below). These were VP.5, VAP.5, CAP, BCAP, and BMPAa media. Not all media were used simultaneously; in fact, they were developed and studied in the order described. For example, VAP.5 was compared with CAP medium, and BCAP was compared with BMPAa medium. Determination of sensitivity ofmedia. A clinical isolate (242D) of L. pneumophila, serogroup I, frozen at -70 C in 1-ml portions in skim milk, was used to test relative sensitivities of the media. Serial 10-fold dilutions of the thawed milk suspension were made by using Trypticase soy broth (BBL Microbiology Systems, Cockeysville, Md.). These were plated in 0.1-ml volumes on the media to be tested, as well as onto non-antimicrobial agent-containing media. Dilutions were always carried to the point at which fewer than 10 colonies were present on the antimicrobial agentfree plates. For all stages of the study, plates were incubated at 35 C and 90% relative humidity in 2.5% CO2 before the use of buffer in CYE medium. The incubation conditions were changed to humidified air at 35 C after ACES buffer was incorporated into CYE medium. This was because better growth of L. pneumophila was observed in air than in CO2 with the buffered medium (P. H. Edelstein and A. Saito, unpublished data), as is the case with growth of L. micdadei on BCYE medium (12). Length of time for growth to appear and number and size of colonies were used for comparison. Comparison of media by using environmental and clinical specimens. Environmental specimens, all of which were potable water, were either plated directly or concentrated 10-fold by centrifugation and then inoculated in 0.1-ml volumes onto BCYEa and BMPAa media simultaneously. These specimens were also treated with an acid wash, as described by Bopp and colleagues (3). This consists of diluting the specimen 1:10 in 0.2 M HCl-KCl buffer, ph 2.2. This is then mixed in a Vortex-type mixer and incubated at room temperature (21 C) for 4 min. A portion of 0.1 ml is then plated directly onto the culture plates. Clinical specimens were plated simultaneously onto antimicrobial agent-free and semiselective media as described previously (7). Organisms other than L. pneumophila were identified by conventional methods, and L. pneumophila was identified on the basis of growth, morphological, biochemical, and immunological characteristics (4, 17). Determination of stability of cefamandole in BMPAa medium. Freshly made plates of BMPAa medium made at the same time were either refrigerated at 5 C or placed in a humidified 35 C incubator. Three strains of Morganella morganii and one of Providencia rettgeri, all of which were highly resistant to polymyxin B, were used to assess the stability of the cefamandole component. Cefamandole minimal inhibitory concentrations of the four strains, as tested by agar dilution, were >2.0, >2.0, 2.0, and 0.5 ug/ml, respectively. The organisms were incubated overnight in Mueller-Hinton broth (MHB) at 37 C. Broths were diluted 1:2,000 with Mueller-Hinton broth. The diluted broth suspensions of organisms were then spot inoculated onto the test media, as well as onto blood agar, using a 10-pl loop. The inoculated media were incu- Medium ACES a-ketoglu- Vancomycm (g/ Anisomy- C. (jig,/ Beazl Polynun (Umi a Cefazol.Qtg/2 Cefaman- buffer (%) tarate (%o) gl cm ug/ ()g/ml) B (U/ml) ml) (MM) g

3 300 EDELSTEIN bated in a humidified incubator at 35 C and observed daily for growth for 5 days. Media held at 35 C were compared with media held at 5 C, using as controls growth on blood agar and that on the freshly made BMPA medium. Experiments were performed in duplicate, and BMPA was tested with and without 0.1% a-ketoglutarate. Statistical analysis. Proportions were compared by using chi-square analysis with Yates' correction. Student's t test was used to compare means and paired samples (5). RESULTS Sensitivity of media. Simultaneous comparison of CYE, VAP.5, and CAP media revealed that there was a -2-log1o decrease in number as well as a 1- to 2-day delay in appearance of colonies of L. pneumophila on the selective media as compared with growth on CYE medium. The colonies growing on CAP medium were larger than those on VAP.5 medium after identical periods; there were two to three times more colony-forming units (CFU) on CAP medium than on VAP.5 medium for identical inoculum sizes. Simultaneous comparison of BCYE, BCYEa, BMPA, and BCAP media revealed that growth on BCYEa and BMPA media was equivalent. Growth occurred on these two media after a 50- h incubation; growth appeared 1 to 2 days later on BCYE and BCAP media, depending upon inoculum size. At equivalent incubation periods there were 50 to 100 times more CFU on BCYEa and BMPA media than on BCAP medium, and three to five times more CFU on BCYEa and BMPA media than on BCYE medium. On plates containing the smallest inoculum, there was no growth on BCAP medium after 146 h of incubation, whereas there were 1 and 3 CFU on BMPA and BCYEa media, respectively, after 50 h; growth did not occur on BCYE medium until after 98 h of incubation, and yield on this plate was considerably lower than on the other two plates yielding growth. Yield was next compared on BMPA, BMPAa, BMP2A, BMP4A, BM2PA, and BM2P2A media. For the highest inoculum (_800 CFU/plate), growth occurred on all plates after 3 days of incubation, but was not numerically equivalent. The greatest number of CFU occurred on BMPAa and BMP2A media, with fewer on BMPA, BMP4A, BM2PA, and BM2P2A, media in descending order; there was a fourfold difference in CFU between BMPAa and BM2P2A media. This difference held true for the lower inoculum sizes. Overall, the best growth was on BMPAa medium. This study was repeated with the addition of BCYEa medium to the group of media being tested. Again, although all plates supported growth after 3 days of incubation, the J. CLIN. MICROBIOL. plates with best growth were BMPAa and BMP2A media. Growth on BCYEa medium was equivalent to that on BMPA medium, both of which contained 20% fewer CFU than did BMPAa and BMP2A media. Again, BMP4A, BM2PA, and BM2P2A media had about 30% fewer CFU than did the media which supported growth the best. Next, all media which contained 0.1% a-ketoglutarate were studied. These were BMPAa, BMP2Aa, BMP4Aa, BM2PAa, BM2P2Aa, BCYEa, and BCYEaS media. Earliest (48 h) growth was observed on BMPAa, BMP2Aa, BCYEa, and BCYEaS media. The greatest yield of growth was also observed on these plates; there was a twofold difference between the plates with highest and lowest yields. All media tested supported growth, even at low inoculum sizes (5 CFU/plate). Growth on the seleniumcontaining medium was no better than on the same medium without selenium. Environmental specimens. Ninety-three potable water specimens were plated on both BMPAa and BCYEa media. Nineteen grew L. pneumophila, with 14 growing on both BMPAa and BCYEa media, 4 on BMPAa medium alone, and 1 on BCYEa medium alone. Five of 15 positive BCYEa plates versus 12 of 18 positive BMPAa plates had pure growth of L. pneumophila (P < 0.025). There was no significant difference in the average number of CFU per plate, which was 17.6 for BMPAa medium and 16.4 for BCYEa medium, with standard deviations of 32.5 and 28.1, respectively (P > 0.10). The mean incubation times required for first growth of L. pneumophila on BCYEa and BMPAa media, respectively, were 3.0 and 3.4 days, with modes of 3 days for both and ranges of 2 to 4 and 3 to 7 days, respectively (P > 0.10). Three specimens which were culture positive for L. pneumophila with acid treatment were negative for Legionella because of overgrowth of contaminants with the same specimens not treated with acid. On the other hand, three specimens which were culture positive for L. pneumophila with non-acid-treated specimens were culture negative when the same specimens were acid treated. All three of these false-negative specimens could be accounted for on the basis of low numbers of L. pneumophila organisms in the specimens and the 1:10 dilution of the acid-treated sample before plating. This dilution factor also was reflected in the average number of colonies per plate for the acid-treated samples; the average plate counts for acidtreated specimens plated onto BMPAa and BCYEa media were 2.20 and 3.07 CFU, respectively, with respective standard deviations of

4 VOL. 14, and 4.70 (P < 0.05 for acid treated versus non-acid treated). The mean incubation times for these acid-treated specimens were 3.2 and 4.0 days for BYCEa and BMPAa media, respectively, with respective ranges of 3 to 4 and 3 to 6 days and respective modes of 3 and 4 days. These incubation times were not significantly different than those for the non-acid-treated specimens (P > 0.10) or between the two types of media (P > 0.10). Sixty-four percent of specimens positive by any method were positive on BCYEa medium plated with acid-treated material, and 80% were positive on BMPAa medium plated with the same specimens. Sixty-seven percent of positive BCYEa and 92% of positive BMPAa plates yielded pure growth of L. pneumophila when plated with acid-treated samples. This difference was not significant, nor was the difference in plate purity for acid treatment versus non-acid treatment (P > 0.10 for both). The acid-treated samples which were overgrown by contaminants were overgrown chiefly by Pseudomonas sp. Clinical specimens. Use of VP.5 medium was abandoned when it became apparent that yeast rapidly overgrew the medium and that use of anisomycin was feasible since it had no inhibiting effects on L. pneumophila at levels of up to 160,tg/ml. Use of VP.5 medium did enable growth of L. pneumophila from two contaminated lung specimens when use of CYE medium did not yield the organism because of contaminants (6). Use of VAP.5 medium limited fungal overgrowth, and its use led to isolation of L. pneumophila from a lung heavily contaminated with Staphylococcus aureus; inoculation of the same specimen on CYE medium failed because of staphylococcal overgrowth (6). However, some S. aureus and S. epidermidis strains overgrew other VAP.5 plates. Cefazolin was substituted for vancomycin because of the greater sensitivity of CAP medium compared with VAP.5 medium and because no synergism or additive effect against L. pneumophila was noted in vitro between cefazolin and polymyxin B. L. pneumophila grew from a sputum specimen plated on CAP, but not from simultaneously inoculated CYE and VAP.5 media (patient 3, reference 11). Fifteen specimens (sputum-14, TTA-1) grew L. pneumophila when inoculated on CAP medium, but failed to on CYE medium. Two TTA specimens grew L. pneumophila when inoculated on CYE medium, but did not on CAP medium. Since BMPAa medium appeared to be more sensitive than BCAP medium, and because of the broader antibacterial range of activity of cefamandole versus cefazolin, BMPAa medium SELECTIVE MEDIUM FOR L. PNEUMOPHILA 301 was next evaluated. Twelve specimens (sputum- 10, TTA-1, lung-1) grew L. pneumophila when inoculated on BMPAa medium, but not on BCYEa medium. None of 10 specimens which grew L. pneumophila on BCYEa medium failed to grow the bacterium on BMPAa medium. The difference in yield on BMPAa medium versus BCYEa medium was highly significant statistically (P < 0.001). All 13 specimens which yielded L. pneumophila when inoculated on CAP medium also yielded it on simultaneously inoculated BMPA medium, but one specimen yielded L. pneumophila on BMPAa medium and not on BCAP medium. Colony size was always larger on BMPAa medium than on CAP or BCAP medium, and numbers of CFU were always greater on BMPAa medium. Incubation times for first growth of L. pneumophila were almost identical for paired positive platings on BCYEa and BMPAa media, with means of 3.0 and 3.1 days, respectively, and identical modes and ranges of 3 and 2 to 4 days, respectively (P > 0.10). Seventeen of 22 positive BMPAa plates versus 6 of 10 positive BCYEa plates had pure growths of L. pneumophila (P > 0.10). There was a significant difference in the average number of colonies per plate (77.8 for BMPAa medium and 6.2 for BCYEa medium, with standard deviations of and 12.2, respectively [P < 0.005]). Table 2 shows the selectivity of BMPAa medium for both clinical and environmental specimens. Stability of the cefamandole component of BMPA medium. Freshly made BMPA TABLE 2. Inhibitionoforganisms bybmpaamedium plated with clinical and environmental specimens % of strains com- Organism pletely inhibiteda Yeasts (11) Proteus mirabilis (10) Klebsiella sp (18) Escherichia coli (18) Morganella morganii.80 (5) Staphylococcus aureus.80 (10) Viridans streptococci (16) Diphtheroids (9) Coagulase-negative staphylococci (12) Pseudomonas sp. 15 (27) Legionellapneumophila... 2 (52) Group D streptococci... (35) Serratia sp... 0 (8) a Number of strains studied is shown in parentheses. Total (100%) inhibition represents the number of distinct strains which grew only on nonselective media. Values of less than 100% represent the percentage of strains which grew either on BMPAa medium alone or on both BMPAa medium and a nonselective medium.

5 302 EDELSTEIN plates, with and without 0.1% a-ketoglutarate, completely inhibited all three strains of M. morganii and the P. rettgeri strain. Plates held for 6 or more days at 35 C before inoculation had no inhibitory effect whatsoever, whereas plates held at 5 C for up to 15 days before inoculation completely retained their inhibitory characteristics. Plates held at 5 C for 20 or more days before inoculation allowed one of the M. morganii strains (cefamandole agar MIC >2 tig/ml) to grow heavily after 48 h of incubation; all three other organisms were completely inhibited on plates held at SoC for up to 28 days before inoculation, even after 5 days of incubation at 370C. DISCUSSION These studies support the use of an antibioticcontaining CYE medium to increase the yield of L. pneumophila from clinical and environmental specimens. It is apparent that the addition of both ACES buffer and a-ketoglutarate to CYE medium increases its sensitivity. The addition of selenium, as proposed by Smalley and colleagues (13), was of no benefit. It is probable that the benefits noted originally for selenium were influenced by use of a nonoptimal basal medium. Data obtained from primary plating of clinical and environmental specimens support the use of BMPAa medium as a means of increasing both selectivity and sensitivity. The overall result is a dramatic increase in yield of positive cultures in clinical specimens and a not so dramatic but significant yield increase in environmental specimens. The shelf life of BMPAa medium, based on data reported here as well as stability data for anisomycin and polymyxin B, should be about 2 to 3 weeks (2, 9, 10, 14). We have used BMPAa after 6 weeks of storage at 50C without apparent loss of specificity. Acid treatment of environmental specimens appears to increase the yield from contaminated specimens; when this is combined with the use of BMPAa medium, there is a further improvement in yield over that seen when BCYEa medium is used with acid-treated samples. However, for some samples, use of acid washing decreases yield. Thus, optimal yield will be obtained for environmental samples when they are plated onto both BMPAa and BCYEa media, in both acid-treated and non-acid-treated form. Whether acid treatment can increase the yield of L. pneumophila from clinical specimens remains to be determined. Use of the negative enrichment procedure proposed by Thorpe and Miller (15) is a different approach which may also be helpful in increasing the yield of L. pneumophila from environ- J. CLIN. MICROBIOL. mental specimens. However, the low sensitivity of their method may preclude its use except in the face of high concentrations of L. pneumophila. Whether increasing the concentrations of polymyxin B and cefamandole in BMPAa medium would further increase the specificity of the medium without altering sensitivity is conjectural. The sensitivity of BMP2Aa medium was equivalent to that of BMPAa medium when tested with the stock strain, making it likely that it could be used in place of BMPAa medium. This may depend upon the presence of antibacterial materials such as antibody, complement, and antimicrobial agents in specimens to be tested. Whether the twofold difference in yield between BMPAa and BM2P2Aa media would influence yield in primary plating also needs investigation. It is apparent that BMPAa medium, although considerably more sensitive and specific than VP.5 medium, is not ideal. The organisms causing major problems, Pseudomonas sp. and group D streptococci, unfortunately are commonly found in the oral flora of patients at this hospital. Different means of inhibiting these organisms must be found to produce a truly selective medium. Since the sensitivity of BMPAa medium is almost identical to that of BCYEa medium, a laboratory with limited funds could use BMPAa medium with little risk of decreased yield. Whether this would hold true for the primary isolation of other Legionella species needs to be determined. For this reason, I recommend that both BCYEa and BMPAa be used for the isolation of Legionella sp. from clinical specimens. Both types of media should also be used for environmental specimens, which should be plated in acid-treated and non-acid-treated form. ACKNOWLEDGMENTS This investigation was supported by the Medical Research Service of the Veterans Administration. I thank Sydney Finegold for support and critiques, William Martin for M. morganii and P. rettgeri strains, as well as susceptibility testing of these strains, and Janis Bridge, Judy Snitzer, and Julie Stohler for excellent technical assistance. LITERATURE CITED 1. Anhalt, J. P., L. D. Sabath, and A. L. Barry Special tests: bactericidal activity, activity of antimicrobics in combination, and detection of 6l-lactamase production, p In E. Lennette (ed.), Manual of clinical microbiology, 3rd ed. American Society for Microbiology, Washington, D.C. 2. Anonymous Polymyxin-B sulfate, p. 8: In M. J. Reilly (ed.), American hospital formulary service. American Society of Hospital Pharmacists, Washington, D.C.

6 VOL. 14, 1981 SELECTIVE MEDIUM FOR L. PNEUMOPHILA Bopp, C. A., J. W. Sumner, G. K. Morris, and J. G. Wells Isolation of Legionella spp. from environmental water samples by low-ph treatment and use of a selective medium. J. Clin. Microbiol. 13: Cherry, W. B., B. Pittman, P. P. Harris, G. A. Hebert, B. M. Thomason, L. Thacker, and R. E. Weaver Detection of Legionnaries disease bacteria by direct immunofluorescence staining. J. Clin. Microbiol. 8: Dixon, W. J., and F. J. Massey Introduction to statistical analysis, 3rd ed. McGraw-Hill Book Co., New York. 6. Edelstein, P. H., and S. M. Finegold Use of a semiselective medium to culture Legionella pneumophila from contaminated lung specimens. J. Clin. Microbiol. 10: Edelstein, P. H., R. D. Meyer, and S. M. Finegold Laboratory diagnosis of Legionnaires' disease. Am. Rev. Respir. Dis. 121: Feeley, J. C., R. J. Gibson, G. W. Gorman, N. C. Langford, J. K. Rasheed, D. C. Mackel, and W. B. Baine Charcoal-yeast extract agar: primary isolation medium for Legionella pneumophila. J. Clin. Microbiol. 10: Korzybski, T., Z. Kowszyk-Gindifer, and W. Kurylowicz Antibiotics: origin, nature, and properties, vol. 3, p American Society for Microbiology, Washington, D.C. 10. Martin, J. E., and J. S. Lewis Anisomycin: improved antimycotic activity in modified Thayer-Martin medium. Public Health Lab. 35: Meyer, R. D., P. H. Edelstein, B. D. Kirby, M. H. Louie, M. E. Mulligan, A. A. Morgenstein, and S. M. Finegold Legionnaires' disease: unusual clinical and laboratory features. Ann. Intern. Med. 93: Pasculle, A. W., J. C. Feeley, R. J. Gibson, L. G. Cordes, R. L. Myerowitz, C. M. Patton, G. W. Gorman, L. L. Carmack, J. W. Ezzell, and J. N. Dowling Pittsburgh pneumonia agent: direct isolation from human lung tissue. J. Infect. Dis. 141: Smalley, D. L., P. A. Jaquess, and J. S. Layne Selenium-enriched medium for Legionella pneumophila. J. Clin. Microbiol. 12: Tanner, F. W., B. A. Sobin, and J. F. Gardocki Some chemical and biological properties of anisomycin, p In H. Welch and F. Marti-Ibaiiez (ed.), Antibiotics annual Medical Encyclopedia, Inc., New York. 15. Thorpe, T. C., and R. D. Miller Negative enrichment procedures for isolation of Legionella pneumophila from seeded cooling tower water. Appl. Environ. Microbiol. 40: Washington, J. A., and V. L. Sutter Dilution susceptibility test: agar and macro-broth dilution procedures, p In E. Lennette (ed.), Manual of clinical microbiology, 3rd ed. American Society for Microbiology, Washington, D.C. 17. Weaver, R. E., and J. C. Feeley Cultural and biochemical characterization of the Legionnaires' disease bacterium, p In G. L. Jones and G. A. Hebert (ed.), "Legionnaires": the disease, bacterium, and methodology. Centers for Disease Control, Atlanta, Ga. Downloaded from on December 8, 2018 by guest