Interlaboratory Variation in a Diaminopimelic Acid Assay: Influence on Estimated Duodenal Bacterial Nitrogen Flow P. H. ROBINSON1 Department of Animal SCience University of Alberta Edmonton. Alberta, Canada TaG 2P5 G. COTO Institute of Animal Science Calle 30. 768-1 Nuevo Vedado. La Habana, Cuba M. D. STERN Department of Animal SCience University of Minnesota
St Paul 55108 D. M. VEIRA Animal Research Centre2 Agriculture Canada Research Branch Neatby Building Ottawa, Ontario, Canada K1 A oes ABSTRACT
(Key words: bacterial nitrogen flow, diaminopimelic acid, rumen bacteria)
Samples of ruminal bacteria and duodenal digesta were collected from two dairy cows fed a 65% forage diet. Samples were sent blind to four laboratories for diaminopimelic acid analysis. Analyzed values differed among laboratories within sample type, and concentrations ranked as follows: laboratory D greater than laboratory A greater than laboratory B greater than laboratory C. Consideration of differences in actual procedures used among laboratories resulted in several hypotheses to explain some of the interlaboratory variation. Using diaminopimelic acid values from each laboratory to estimate duodenal bacterial nitrogen flow showed that laboratory D estimated a 17% higher flow than the average for laboratories A, B, and C, which were similar.
INTRODUCTION
Received February 2, 1990. Accepted May 14, 1990. lCurrent address: Agriculture Canada Research Station, PO Box 20280, Fredericton, New Brunswick, Canada E3B 4Z7. 2contribution Number 1652.
1990 J Dairy Sci 73:2929-2933
Modeling animal metabolism to understand and predict a perfonnance response of animals to dietary or environmental challenges has become widely established in recent years. Efforts such as those of France et al. (6) and Fox et al. (5) have led to integration of published and unpublished results of animal experimentation into multifaceted computer models. Models often develop as conceptual approaches to a specific problem but rapidly develop into computer simulations that require numbers. This has led modelers to review scientific papers with the purpose of identifying numerical values to describe parameters influenced by various dietary or environmental challenges. But is it valid to combine values from studies at various locations, utilizing different procedures, to fulfill modeling needs for data? A recent NRC publication (10) used this approach to predict rumen microbial yield from various measurable animal and diet characteristics. Yield was predicted separately for dairy cattle, beef cattle, and sheep, but microbial yield values within each class of animal were collected from scientific papers in which the
2929
2930
ROBINSON BT AL.
method utilized differed markedly. Several laboratories have examined different microbial markers, and although some have shown no difference (4, 8, 16, 18), most have reported differences in duodenal microbiallbaeterial N flow due to the marker selected (1, 4, 7, 8, 9, 11, 12, IS, 18). Potential differences due to marker is recognized in NRC (10); as the marker used in each study is listed. However, differences in estimates of microbial flow among studies that used the same marker may reflect different marker analysis procedures or, within a procedure, among laboratory modifications that influence results. The objectives of this study were 1) to measure interlaboratory variation in diaminopimelic acid (DAPA) assay results where all laboratories cite the same procedure and 2) to evaluate the influence of interlaboratory variation on estimated duodenal bacterial N flow. MATERIALS AND METHODS
samples
Two lactating dairy cows fitted with a large diameter ruminal cannula and a duodenal cannula in the proximal duodenum [see Robinson and Sniffen (14) for details of diets, cows, and cannulae] were fed diets of 6S% forage and 3S% concentrate on a DM basis. Ruminal bacteria were isolated 6 times per cow, and du0denal digesta were collected 8 times per cow as previously described (14). Samples were pooled within cow to yield one sample of ruminal bacteria and one sample of duodenal digesta per cow. Samples were analyzed for DM, organic matter (OM), and N as previously described (13). laboratories
Four laboratories, those of the authors, were identified that used the DAPA assay procedure of Czerkawski (2) and have published microbial flow data with cattle or sheep. Subsamples of pooled samples were sent to each laboratory to be assayed for DAPA content using a modified procedure of Czerkawski (2). Each sample was sent blind, and cooperators were requested to: 1) replicate sample analysis two to four times, 2) incorporate DAPA recovery correction in the assay, 3) report individual replicate Journal of Dairy Science Vol. 73,
No. 10, 1990
assays, and 4) submit a detailed copy of the analysis procedure to the coordinator. Dlamlnoplmellc Acid Analysis Methods
All cooperating laboratories claimed that their procedure was that of Czerkawski (2) or a modification. Specifics of the actual DAPA procedures used in each laboratory are summarized in Table 1. Laboratories differed in specifics of the procedures used. All laboratories used similar sample weights for assay and used 6N HQ for hydrolysis, although this varied from 1 to lOX the volume of the original procedure. Cooperators maintained hydrolysis conditions similar to that of the original procedure, with the exception of laboratory D, which hydrolyzed for only 4 h but at a temperature of 14S"C rather than 10S"C. Laboratories A and C cleaned the hydrolysate similar to the original procedure, whereas B and D dispensed with activated charcoal in favor of simple filtration. The original procedure called for cleaned hydrolysate to be reduced in volume by drying at 10S"C. All modified this method: A used a hot water bath. B used a steam bath, D used a rotovap at 70"C, and C did not reduce the volume. Laboratories A and B used the same column resin as in the original procedure, whereas C and D used an alternate resin with similar properties. Only D maintained the same volume of resin per column. Others reduced it by about 60%. Laboratories B, C, and D maintained column conditions and buffer sequences of the original procedure, but A increased the temperature to 55"C and the buffer sequence was changed due to use of an automated column system. Eluant collected by all laboratories exceeded that of the original procedure by 33 to 100%. Eluant was reduced in volume similar to the original procedure by A and B. However, C dried it at 80'C, and D did not reduce the volume. All laboratories used the same ratio of volume-reduce eluant to ninhydrin reagent, except A, which increased it IO-fold from .S to 5. For color development, all boiled samples for S min as specified by the original procedure and read the color at 425 om, except for B which used 430 om. Only A and B used background correction as stipulated by El-Shazley and Hungate (3) in the original procedure of Czerkawski (2). All laboratories corrected results for DAPA
2931
DIAMINOPlMBllC ACID ASSAY TABlE 1. Details of the diaminopimelic acid (DAPA) assays used by tbe laboratories. laboratory
OrigiDal
procedure l Sample weight, DIg Digesta Bacteria
6N HCl added Hydrolysis Time, h Temperature, Vessel Hydrolysate
cleanup
D
ISO
ISO
ISO
200
100
100
75
100
3
30ml
16 lOS Tubes
24 lOS Ampules
Cha1'coaJl
Charcoal/
chromosorb W
m1/SO
16 lOS Ampules
Micropore chromosorb W filter
Hot water
CG120 6 x I 3.1 55 3.25 rising 1
Eluant volume reduction
CG120 5 x 1.4 7.7 Ambient 3.414.2 1 100 30 Rota-vap or steam bath
Quantification Sample ninhydrin, ml Boiling time, min Wavelength. nm Background comction DAPA recovery correction DAPA standard source
2/4 5 425 Yes Yes Sigma
Column Resin Dimensions, em Volume, em3 Temperature, O'C Buffer sequence Volume to column, ml Buffer 3.4 discarded, ml Buffer 4.2 collecled, ml
C
SO-ISO
Dry at lOS'C
Hydrolysate Volume reduction
B
SO-ISO 3ml
'c
A
bath
DIg
20 lOS Tubes
4 145 Tubes
Cha1'coall
Celite
celite
Steam bath
None
Roto-vap @ 70'C
CGl20 4..5 X 1 3..5 Ambient 3.4/4.2 2 100
IRP~9
IRP-69 9.6 x 1 7..5 Ambient 3.4142 5 100 50 None
bath
Steam bath
6 x 1 3.1 Ambieot 3.4142 .5 100 60 Dry at 80'C
5/1 5 425 Yes Yes Flub
2/4 .5 430 Yes Yes Sigma
2/4 7 425 No Yes Sigma
92-96
40
None HoI water
3 m1/S0 DIg
IOml
2/4 .5 425 No Yes Sigma
lCzerkawski (2).
RESULTS
recovery, which exceeded 90% in all cases. The DAPA standard used was Sigma (Sigma Chemical Co., St. Louis, MO) for all laboratories, except A, which used Flub (Buch S.G., Switzerland). Statistical Analysis
The study was designed as a 4 x 4 factorial arrangement with four laboratories and four samples. Differences among laboratories were determined by Student Neuman Keuls multiple range test using version 5.16 of the statistical package of SAS (17). All significance was determined at the P A > B = C. DISCUSSION
Because laboratories claim to be using the same analytical procedure, the procedures are often assumed to be identical. On the basis of the actual procedures used by the four laboratories in this study, differences in available equipment, reagents, and facilities, as well as efforts at simplification, lead to significant changes in the original procedure. These resulting differences among procedures probably are responsible for observed differences in DAPA concentration among laboratories. However, this is not
TABLE 3. Analytical results of diaminopimelic acid (DAPA) assays. 1 Sample type Cow 1 Sample 1
LaboratmrA Replicate 1 2 3
4 Mean
B
C
Cow 2 Sample 2
0
2.10 1.79 1.76 2.29 2.00 1.84 1.71 2.21 1.75 1.85 1.75 2.05 1.79 1.98 1.82 1.74 2.25
ABC 2.90 3.48 3.25 3.25 3.27
Cow 1 Sample 3
0
ABC
Cow 2 Sample 4
0
A
B
C
o
(mg DAPAlg of organic matter) - - - - - - - - - - 2.86 2.64 3.15 .81 .70 .63 .98 .80.69 .78 .97 2.86 2.65 3.11 .81 .68 .60 .95 .83 .71 .67 .98 2.79 2.65 .74.68 .70 .82 .71 .69 2.83 2.65 .87.66 .62 .89 .71 .62 2.84 2.65 3.13 .81 .68 .64.96 .84 .71 .69 .98
IStatistically significant laboratory, sample, and laboratory x sample effects (P 1 > 3 = 4 (P A > B > C (P
St Paul 55108 D. M. VEIRA Animal Research Centre2 Agriculture Canada Research Branch Neatby Building Ottawa, Ontario, Canada K1 A oes ABSTRACT
(Key words: bacterial nitrogen flow, diaminopimelic acid, rumen bacteria)
Samples of ruminal bacteria and duodenal digesta were collected from two dairy cows fed a 65% forage diet. Samples were sent blind to four laboratories for diaminopimelic acid analysis. Analyzed values differed among laboratories within sample type, and concentrations ranked as follows: laboratory D greater than laboratory A greater than laboratory B greater than laboratory C. Consideration of differences in actual procedures used among laboratories resulted in several hypotheses to explain some of the interlaboratory variation. Using diaminopimelic acid values from each laboratory to estimate duodenal bacterial nitrogen flow showed that laboratory D estimated a 17% higher flow than the average for laboratories A, B, and C, which were similar.
INTRODUCTION
Received February 2, 1990. Accepted May 14, 1990. lCurrent address: Agriculture Canada Research Station, PO Box 20280, Fredericton, New Brunswick, Canada E3B 4Z7. 2contribution Number 1652.
1990 J Dairy Sci 73:2929-2933
Modeling animal metabolism to understand and predict a perfonnance response of animals to dietary or environmental challenges has become widely established in recent years. Efforts such as those of France et al. (6) and Fox et al. (5) have led to integration of published and unpublished results of animal experimentation into multifaceted computer models. Models often develop as conceptual approaches to a specific problem but rapidly develop into computer simulations that require numbers. This has led modelers to review scientific papers with the purpose of identifying numerical values to describe parameters influenced by various dietary or environmental challenges. But is it valid to combine values from studies at various locations, utilizing different procedures, to fulfill modeling needs for data? A recent NRC publication (10) used this approach to predict rumen microbial yield from various measurable animal and diet characteristics. Yield was predicted separately for dairy cattle, beef cattle, and sheep, but microbial yield values within each class of animal were collected from scientific papers in which the
2929
2930
ROBINSON BT AL.
method utilized differed markedly. Several laboratories have examined different microbial markers, and although some have shown no difference (4, 8, 16, 18), most have reported differences in duodenal microbiallbaeterial N flow due to the marker selected (1, 4, 7, 8, 9, 11, 12, IS, 18). Potential differences due to marker is recognized in NRC (10); as the marker used in each study is listed. However, differences in estimates of microbial flow among studies that used the same marker may reflect different marker analysis procedures or, within a procedure, among laboratory modifications that influence results. The objectives of this study were 1) to measure interlaboratory variation in diaminopimelic acid (DAPA) assay results where all laboratories cite the same procedure and 2) to evaluate the influence of interlaboratory variation on estimated duodenal bacterial N flow. MATERIALS AND METHODS
samples
Two lactating dairy cows fitted with a large diameter ruminal cannula and a duodenal cannula in the proximal duodenum [see Robinson and Sniffen (14) for details of diets, cows, and cannulae] were fed diets of 6S% forage and 3S% concentrate on a DM basis. Ruminal bacteria were isolated 6 times per cow, and du0denal digesta were collected 8 times per cow as previously described (14). Samples were pooled within cow to yield one sample of ruminal bacteria and one sample of duodenal digesta per cow. Samples were analyzed for DM, organic matter (OM), and N as previously described (13). laboratories
Four laboratories, those of the authors, were identified that used the DAPA assay procedure of Czerkawski (2) and have published microbial flow data with cattle or sheep. Subsamples of pooled samples were sent to each laboratory to be assayed for DAPA content using a modified procedure of Czerkawski (2). Each sample was sent blind, and cooperators were requested to: 1) replicate sample analysis two to four times, 2) incorporate DAPA recovery correction in the assay, 3) report individual replicate Journal of Dairy Science Vol. 73,
No. 10, 1990
assays, and 4) submit a detailed copy of the analysis procedure to the coordinator. Dlamlnoplmellc Acid Analysis Methods
All cooperating laboratories claimed that their procedure was that of Czerkawski (2) or a modification. Specifics of the actual DAPA procedures used in each laboratory are summarized in Table 1. Laboratories differed in specifics of the procedures used. All laboratories used similar sample weights for assay and used 6N HQ for hydrolysis, although this varied from 1 to lOX the volume of the original procedure. Cooperators maintained hydrolysis conditions similar to that of the original procedure, with the exception of laboratory D, which hydrolyzed for only 4 h but at a temperature of 14S"C rather than 10S"C. Laboratories A and C cleaned the hydrolysate similar to the original procedure, whereas B and D dispensed with activated charcoal in favor of simple filtration. The original procedure called for cleaned hydrolysate to be reduced in volume by drying at 10S"C. All modified this method: A used a hot water bath. B used a steam bath, D used a rotovap at 70"C, and C did not reduce the volume. Laboratories A and B used the same column resin as in the original procedure, whereas C and D used an alternate resin with similar properties. Only D maintained the same volume of resin per column. Others reduced it by about 60%. Laboratories B, C, and D maintained column conditions and buffer sequences of the original procedure, but A increased the temperature to 55"C and the buffer sequence was changed due to use of an automated column system. Eluant collected by all laboratories exceeded that of the original procedure by 33 to 100%. Eluant was reduced in volume similar to the original procedure by A and B. However, C dried it at 80'C, and D did not reduce the volume. All laboratories used the same ratio of volume-reduce eluant to ninhydrin reagent, except A, which increased it IO-fold from .S to 5. For color development, all boiled samples for S min as specified by the original procedure and read the color at 425 om, except for B which used 430 om. Only A and B used background correction as stipulated by El-Shazley and Hungate (3) in the original procedure of Czerkawski (2). All laboratories corrected results for DAPA
2931
DIAMINOPlMBllC ACID ASSAY TABlE 1. Details of the diaminopimelic acid (DAPA) assays used by tbe laboratories. laboratory
OrigiDal
procedure l Sample weight, DIg Digesta Bacteria
6N HCl added Hydrolysis Time, h Temperature, Vessel Hydrolysate
cleanup
D
ISO
ISO
ISO
200
100
100
75
100
3
30ml
16 lOS Tubes
24 lOS Ampules
Cha1'coaJl
Charcoal/
chromosorb W
m1/SO
16 lOS Ampules
Micropore chromosorb W filter
Hot water
CG120 6 x I 3.1 55 3.25 rising 1
Eluant volume reduction
CG120 5 x 1.4 7.7 Ambient 3.414.2 1 100 30 Rota-vap or steam bath
Quantification Sample ninhydrin, ml Boiling time, min Wavelength. nm Background comction DAPA recovery correction DAPA standard source
2/4 5 425 Yes Yes Sigma
Column Resin Dimensions, em Volume, em3 Temperature, O'C Buffer sequence Volume to column, ml Buffer 3.4 discarded, ml Buffer 4.2 collecled, ml
C
SO-ISO
Dry at lOS'C
Hydrolysate Volume reduction
B
SO-ISO 3ml
'c
A
bath
DIg
20 lOS Tubes
4 145 Tubes
Cha1'coall
Celite
celite
Steam bath
None
Roto-vap @ 70'C
CGl20 4..5 X 1 3..5 Ambient 3.4/4.2 2 100
IRP~9
IRP-69 9.6 x 1 7..5 Ambient 3.4142 5 100 50 None
bath
Steam bath
6 x 1 3.1 Ambieot 3.4142 .5 100 60 Dry at 80'C
5/1 5 425 Yes Yes Flub
2/4 .5 430 Yes Yes Sigma
2/4 7 425 No Yes Sigma
92-96
40
None HoI water
3 m1/S0 DIg
IOml
2/4 .5 425 No Yes Sigma
lCzerkawski (2).
RESULTS
recovery, which exceeded 90% in all cases. The DAPA standard used was Sigma (Sigma Chemical Co., St. Louis, MO) for all laboratories, except A, which used Flub (Buch S.G., Switzerland). Statistical Analysis
The study was designed as a 4 x 4 factorial arrangement with four laboratories and four samples. Differences among laboratories were determined by Student Neuman Keuls multiple range test using version 5.16 of the statistical package of SAS (17). All significance was determined at the P A > B = C. DISCUSSION
Because laboratories claim to be using the same analytical procedure, the procedures are often assumed to be identical. On the basis of the actual procedures used by the four laboratories in this study, differences in available equipment, reagents, and facilities, as well as efforts at simplification, lead to significant changes in the original procedure. These resulting differences among procedures probably are responsible for observed differences in DAPA concentration among laboratories. However, this is not
TABLE 3. Analytical results of diaminopimelic acid (DAPA) assays. 1 Sample type Cow 1 Sample 1
LaboratmrA Replicate 1 2 3
4 Mean
B
C
Cow 2 Sample 2
0
2.10 1.79 1.76 2.29 2.00 1.84 1.71 2.21 1.75 1.85 1.75 2.05 1.79 1.98 1.82 1.74 2.25
ABC 2.90 3.48 3.25 3.25 3.27
Cow 1 Sample 3
0
ABC
Cow 2 Sample 4
0
A
B
C
o
(mg DAPAlg of organic matter) - - - - - - - - - - 2.86 2.64 3.15 .81 .70 .63 .98 .80.69 .78 .97 2.86 2.65 3.11 .81 .68 .60 .95 .83 .71 .67 .98 2.79 2.65 .74.68 .70 .82 .71 .69 2.83 2.65 .87.66 .62 .89 .71 .62 2.84 2.65 3.13 .81 .68 .64.96 .84 .71 .69 .98
IStatistically significant laboratory, sample, and laboratory x sample effects (P 1 > 3 = 4 (P A > B > C (P
