ANALYTICAL BIOCHEMISTRY 126, 13 1- 138 ( 1982)

Analysis of Nitrate, Nitrite, and [15N]Nitrate in Biological Fluids

LAURA C. GREEN, DAVID A. WAGNER, JOSEPH GLOGOWSKI, PAUL L. SKIPPER,

JOHN S. WISHNOK, AND STEVEN R. TANNENBAUM
Department of Nutrition and Food Science, Massachusetts Institute of Technology,
Cambridge, Massachusetts 02139
Received February 22, 1982

A new automated system for the analysis of nitrate via reduction with a high-pressure
cadmium column is described. Samples of urine, saliva, deproteinized plasma, gastric juice,
and milk can be analyzed for nitrate, nitrite, or both with a lower limit of detection of 1.0
nmol NOT or NOg/ml. The system allows quantitative reduction of nitrate and automatically
eliminates interference from other compounds normally present in urine and other biological
fluids. Analysis rate is 30 samples per hour, with preparation for most samples limited to
simple dilution with distilled water. The application of gas chromatography/mass spectrometry
for the analysis of “NO; in urine after derivatization to ‘5N02-benzene is also described.

To study the metabolism of nitrate in hu- removing the small amounts of nitrate as
mans (1) and in rats (2), we required an an- well. Other investigators have also reported
alytical system to determine nitrate (NO;) that samples as complex as urine were not
and nitrite (NO;) in biological media. Our amenable to nitrate analysis by cadmium re-
requirements for such a system included duction (5,6).
short analysis times and high selectivity, i.e., Our successful attempts at analyzing
the absence of both false positive responses NO; in urine utilized the advantages of high-
and false negatives due to interfering sub- pressure liquid chromatography (HPLC)’
stances. Automation of as much of the anal- technology, particularly the large sample di-
ysis as possible was desired. lution and high surface-to-volume ratio of
Initial attempts were made to analyze column pathway to sample. The efficiency
NO; by the method of Stainton (3) using of nitrate reduction at high pressures through
cadmium wire to reduce NO; to NO; in a a column packed with fine particles of cop-
Technicon autoanalyzer system. We found per-plated cadmium metal was 99 + 5% of
that this method worked well for nitrate in nitrate added to urine. Interferences were
water but poorly for nitrate in urine. In urine, thus effectively eliminated in this system.
substantial interferences were present, even The automated system that we have devel-
at 1: 100 dilutions of urine in water, that in- oped is described below.
hibited the reduction of NO; to NO;. Phys- In metabolic tracer and isotope dilution
iological concentrations of phosphate and studies of NO; (1,2) we used the stable iso-
citrate, as have been systematically investi- tope 15N. To quantify excreted 15N0:, we
gated by other investigators (4), were found adapted the method of Tesch and co-workers
to interfere. Other interfering substances are (7) for the nitration of benzene by nitrate in
apparently also present in urine, and we urine and analyzed the nitrobenzene by GC-
could find no combination of cation ex-
change, anion exchange, and/or treatment
’ Abbreviations used: HPLC, high-pressure liquid
with alum that would allow sufficient re- chromatography; GC-MS, gas chromatography-mass
moval of the interfering substances without spectrometry; SIM, selected ion monitoring.

131 0003-2697/82/ 150 13 l-08$02.00/0

Copyright 0 1982 by Academic Press, Inc.
All rights of reproduction in any form reserved.
132 GREEN ET AL.

MS using selected ion monitoring (SIM). equipped with a 30-~1 sample loop. The so-
These adaptations are also described below. lenoid valves are controlled by signals from
the sample carousel, with delays which allow
MATERIALS AND METHODS the sample injector loop to load for 25 s dur-
ing the middle of the 40 s sample uptake
Automated NO; and NO? Analysis time. If only a small volume of sample is
Chemicals. Sodium nitrate, sodium ni- available, the sample loop may instead be
trite, sulfanilamide, and naphthylethylene- loaded manually with a syringe with as little
diamine dihydrochloride were analytical re- as 10 ~1. There is a 2-min lag time before a
agent grade and were obtained from Mal- sample leaving the cup enters the injector.
linckrodt (St. Louis, MO.). Cadmium filings The Griess reagent consists of 1 part 0.1%
(-100 mesh) were obtained from Ventron naphthylethylenediamine dihydrochloride in
Speciality Products, Alfa Chemical (Danvers, distilled water plus 1 part 1% sulfanilamide
Mass.). (or sulphanilic acid) in 5% concentrated
Systemdescription. A schematic represen- H3P04, the 2 parts being mixed together
tation of the system for the automated anal- within 12 h of use and kept chilled. Each part
ysis of nitrite and nitrate appears in Fig. 1. may be stored refrigerated for up to 2 months.
The principles of operation are as follows. The mobile phase buffer is 5% aqueous
Nitrate in a sample is reduced by the cad- NH&l adjusted to pH 9.0 with sodium bor-
mium column to NO;, which reacts with a ate, and may be stored refrigerated for up to
Griess reagent to form a purple azo dye. several months. Both the Griess reagent and
Nitrite in a sample needs no reduction, and buffer are degassed before use by stirring un-
so reacts with the Griess reagent when the der a partial vacuum at room temperature
cadmium column is bypassed, as well as for several minutes.
when it is utilized. The color of the product A double-headed reciprocating pump
dye is developed by a 60°C water bath, (Milton Roy Co., Riviera Beach, Fla., mini-
cooled by a 0°C water bath, and its absor- pump) delivers the buffered mobile phase at
bance at 546 nm is detected by a flow- 1.O ml/min and Griess reagent at 0.5 ml/min.
through uv/visible spectrophotometer (Model The Griess reagent stream enters the buffered
440, Waters Associates, Milford, Mass.). The sample stream via a mixing tee, after the lat-
detector is connected both to a strip chart ter has passed through the cadmium reduc-
recorder (Linear Model 252, Arthur H. Tho- tion column.
mas Co., Philadelphia, Pa.) and to an inte- The reduction column is high-pressure
grator (Supergrator II, Columbia Scientific Teflon (TFE) tubing (Supelco, Bellefonte,
Industries, Austin, Tex.). Pa.), 10 in. in length, l/S in. o.d., 2.7 mm
Samples (minimum volume for auto- i.d., packed with copper-plated cadmium fil-
mated analysis = 0.3 ml) are loaded onto a ings. Stainless-steel tubing may also be used.
Technicon (Tarrytown, N. Y.) sample car- The Cd filings are plated with copper by stir-
ousel; the sampling needle alternates between ring in a 5% aqueous solution of CuSO,.
each sample cup and a refilling distilled water Excess metallic Cu is removed by copious
reservoir, sampling for 40 s and washing for washing with distilled water. Further washing
80 s, at a flow rate of 0.32 ml/min. Thus 30 with 0.1 N HCl solubilizes and removes
samples are analyzed per hour. This timing Cd(OH)*. The Cd is finally washed in dis-
sequence allows baseline resolution between tilled water, then stored in NH&l buffer for
successive sample peaks. up to 2 months awaiting packing.
Samples are injected onto the column via The column is packed by swirling the Cd
an air-actuated, solenoid valve-controlled in- in buffer and quickly pouring it into the tub-
jection valve (Rheodyne, Model No. 7010) ing, the far end of which has been capped
 Pump : Griess Reagent

P Sample Injector
Waste (to sink)
Integrator Pump : NH4CI Buffer
f
I I

Peristaltic Pump
Strip Chart Visible Light 0°C 60°C
Recorder Absorbance Water Water
Detector Bath Bath
Supply and Timer

Sample Carousel-to sink)

FIG. 1. Schematic representation of the system for the automated analysis of nitrite and nitrate.
134 GREEN ET AL,

with a fritted l/8- to I/ 16-m. reducing union. stainless-steel tubing may be used as well.
Gravity and vigorous tapping or vibrating Pieces are joined with Swagelok ferrules,
pack the Cd into the column. The freshly nuts, unions, or tees.
packed column is capped with a fritted l/8- Sample preparation. Samples are generally
to l/16-in. reducing union, and buffer is diluted to give NaN03 or NaNO* concentra-
pumped at 2 ml/min through the column for tions of between 5.0 and 20.0 pM. Animal
30 to 60 min. The small void which may or human urine or human saliva is typically
form at the leading end of the column is filled diluted I:20 in distilled water and can be
with more wet cadmium. The packed col- analyzed as is. Gastric juice samples are neu-
umn is positioned in the system and condi- tralized with 5% aqueous NH&l buffer and
tioned by repeated sampling of a 20 PM are diluted 1: 10 for NO; analysis, or 1:3 for
NaN03 standard until a maximal response NO; analysis.
is obtained. The average lifetime of a well- Serum, plasma, or whole milk is depro-
packed cadmium column is 800-1000 sam- teinized before analysis as follows. To a 500-
ples. Erratic responses and/or decreasing col- ~1 sample is added 100 ~135% sulfosalicyclic
umn efficiency (> 10% drop in conversion of acid. Treated samples are mixed by vortexing
NO; to NO;) generally signal the end of a every 5 min and allowed to react for 30 min
given column’s reliability. at room temperature. They are then centri-
It should be noted that not only NO; but fuged at 10,OOOg for 15 min. Two hundred
also any preformed NO; in a sample will microliters of supematant, 300 ~15% aqueous
contribute to the Griess-positive response NH&l buffer, and 60 ~15% NaOH are com-
after passage through the cadmium column. bined for analysis.
If the column is bypassed, however, only Aqueous standards of NaN03 and NaNO*,
NO; and not NO; will yield a response. ranging from 5.0 to 50 pM, are analyzed daily
Therefore, in the analysis of samples that to check column efficiency. Stock urine sam-
contain both species, one portion passes ples and 10 pM NO; standards are also an-
through the cadmium column, yielding a alyzed periodically to check for system vari-
composite response for NO; and NO;, an- ability. Samples are analyzed in duplicate.
other portion bypasses the cadmium column,
yielding a response for NO, alone, and the
“NO; Analysis in Urine
sample’s NO; content is calculated by dif-
ference. Sample preparation. The acid-catalyzed
The column positioned upstream to the nitration of benzene by nitrate was adapted
cadmium column is high-pressure Teflon for the analysis of 15NO; in urine from the
(TFE) tubing (Supelco), 9 in. long, l/8 in. method of Tesch et al. (7). One milliliter of
o.d., 2.7 mm i.d., packed with Dowex AG urine is mixed with 5 ml benzene, then 5 ml
SOW X 12, 200-400 mesh, cation-exchange concentrated H2S04 is added slowly. The
resin. This column does not retain NO; or reactants are mixed in a shaker bath at room
NO;. It serves as a “clean-up” column, trap- temperature for 5- 10 min. The benzene layer
ping interfering substances from biological is removed, passed through a silica cartridge
samples, thus protecting the cadmium col- Sep-Pak (Waters Associates) and concen-
umn and the rest of the system. It is also the trated slowly under a N2 stream to 200 ~1.
only column through which samples pass Analysis. The ratio of [“N]- to
when the cadmium column is switched out [ 14N]nitrobenzene is determined by GC-MS
of this system and analysis is for NO; alone. on a Hewlett-Packard 5992 C&/MS System
All of the system tubing, downstream of (HP, Palo Alto, Calif.), with selected ion
the reciprocating pump, is l/16 in. o.d., 1.58 monitoring at m/z = 123 (M) and m/z 124
mm i.d., Teflon tubing. Low dead-volume (A4 + 1). The gas chromatography separation
 ANALYSIS OF NITRATE AND NITRITE 135

is on a 3-ft X l/4-in. glass column packed 15N is calculated as follows:

with 3% OV225 on 100/120 Supelcoport
atom percent excess “N =
(Supelco) using a carrier flow of helium of
25 ml/min and an oven-temperature pro- ‘24NB - 0.070 ‘23NB
gram from 90 to 120°C, at 16”C/min. Under 2
‘23NB + lz4NB
these conditions, the retention time of nitro- - (O.O70)‘*‘NB + 0.006 ‘23NB
benzene is 1.8 min. Complete mass spectra
of urine samples and nitrobenzene standards
where ‘23NB and ‘24NB are the peak areas of
are run to confirm identity. the m/z 123 and m/z 124 nitrobenzene ions.
Calculations. The 15N isotope enrichment This simplifies to
of nitrobenzene was calculated according to
Biemann (8). The nitroaromatics are partic- R - 0.070
ularly susceptible to A4 - 1 fragmentation atom percent excess j5N =
R + 0.936 ’
due to the loss of a hydrogen atom from the
aromatic ring. A nitrobenzene standard (un- where R = ‘24NB/‘23NB. To correct for the
labeled) is used to quantify the extent of the M - 1 fragment, we assumed that a fraction,
M- 1 peakbySIMofm/z 122(M- l),m/ f, of ‘23NB loses a hydrogen and that the
z 123 (M), and m/z 124 (M + 1). The per- same fraction f of the ‘24NB loses a hydrogen.
centage of the (M - 1) ion varies daily de- so,
pending on ion source condition, flow rate,
and column pressure. Normalized peak the true peak area of 124NB: A = a
heights of an unlabeled nitrobenzene stan- l-f’
dard are: where a = 124NB observed,
Ion mon- b - fA
itored (m/z) 122 123 124 125 the true peak area of 123NB: B = ___
Mass M-l M it+1 Iv+2 1-f’
Intensity 0.05” 1.00 0.070 0.006b where b = ‘23NB observed, and, B = c/f,
’ Changes daily. where c = 122NB observed. Solving this gives
’ Obtained from Beynon’s table (Ref. (9)).
A (1 -f)O
Using these values the atom percent excess ii= 1 - f(1 + a/b)’
where

f = (I +$)-[(I +2$-4;(1 +;+;)1’”

2(1+:+x) *

The corrected formula therefore is

nitrate, or both nitrate and nitrite, are routed
A/B - 0.070
atom percent excess 15N = through the cadmium reduction column: the
A/B + 0.936 ’ standard lines for the column are shown in
Fig. 2. The equations of these standard lines
RESULTS AND DISCUSSION are
The automated analytical system de-
scribed here responds in a quantitative linear A 546nm = O.O023[NO;] - 0.0038 [l]
fashion to nitrite, nitrate, or both within the
range of 1 to 300 pM. Standards containing and
136 GREEN ET AL.

0.72

0.66
/ l

0.60
t

0.48 I-

: 0.36
,’
a
L
,” 0.30
2
0.24

0 18
t

006

I I I I I I I I I I
0 30 60 90 120 150 180 210 240 270 300

[NO; ]or[NO;],pM
FIG. 2. Standard curves for nitrate and nitrite: cadmium column switched into the system. Absorbance
at 546 nm is the absorbance reading on the flow-through spectrophotometer (Waters Model No. 440)
at the apex of the peak. The flow cell path length is 1 cm and its volume is 12.5 ~1.

A 546nm= O.O023[NO,-]- 0.0020. [Z] mium column and are quantified as follows.
First, the sample is analyzed with the cad-
The slopes of Eqs. [l] and [2] are the same, mium column switched out of the system,
indicating that the efficiency with which the and the resultant absorbance at 546 nm is
cadmium column reduces NO; to NO; is entered into Eq. [3] to calculate the nitrite
essentially 100%. In fact, column efficiencies concentration. Next, the sample is analyzed
averaged 99 f 5%. with the cadmium column in the system, and
Aqueous standards containing nitrite but the resultant absorbance is entered into the
no nitrate are routed only through the clean- rearranged combination of Eqs. [l] and [2]
up column, bypassing the cadmium column. to yield the nitrate concentration.
The linear response of absorbance at 546 nm
of the Griess chromophore as a function of
nitrite concentration is governed by the equa- [NO;] = A54a~o;;;0058 - [NO;]. t41
tion
For samples containing NO; but no NOS,
A546nm= O.O036[NO;] - 0.0041. [3]
such as normal urine, the nitrate content is
Samples containing NO; and NO; are simply calculated from Eq. [ 11.
analyzed both with and without the cad- Standard curves generated by additions of
 ANALYSIS OF NITRATE AND NITRITE 137

TABLE 1
5Or
RANGE OF NITRATE AND NITRITE CONCENTRATIONS
IN VARIOIJSBIOLOGICALFLUIDS

Biological [NO?1 [NOrI

fluid (m) (NM)

Urine 250-2000 ND”

Saliva 200-600 30-210
Plasma 15-60 ND
Gastric juice 50-85 0.40-60b
Milk 20-30 ND

’ Not determined.
b Nitrite concentrations at the high end of this range
are from people with gastric pathologies.

terfering substances which cochromatograph

with nitrobenzene. The yield of nitrobenzene
from nitrate is about 50%, but, since only the
“N-to-14N ratio is required, this is accept-
01
able. A previously reported analysis of
[NO;;O 2o 3o 15NO; requires conversion to Nr gas (1 l),
I” Ur~ne.~M, Added
with at least 1 mg N needed. Our method
FIG. 3. Analytical recovery of nitrate added to urine. requires 10 pg N.
The basic components of our automated
nitrate to urine, diluted 1:40 in distilled wa- analytical NO;/NO; system are not new.
ter, are linear and show recoveries of added Griess described the diazotiation of an aryl
nitrate of 95 rC_5% (Fig. 3). This result sug- amine by nitrite and coupling of the product
gests that substances in urine neither inter- to form an azochromophore in 1879 (12).
fere with nor enhance the quantitative re- Cadmium metal, with or without copper
duction of nitrate and development of the
Griess chromophore. Neither day-to-day nor
column-to-column responses, as monitored
by sampling a stock solution of urine, stored
frozen, varied by more than 3%.
Thousands of samples of human and rat
urine, human saliva, and plasma have been
analyzed for nitrate and nitrite using this sys-
tem (1,2,10). Typical ranges of nitrate and
nitrite levels in biological samples are shown
in Table 1.
A standard curve for 15NO; analysis is
shown in Fig. 4. This curve was generated by
analyzing aqueous standards to which
“NO; and 14NO; had been added in known
ratios. Analysis of 15NOS in urine gives a sim- 0 20 40 60 80

ilar curve (data not shown). Observed excess Atom Percent Ex~ess’~N,Added
atom percentage of 15N correspond exactly FIG. 4. Observed rates of ‘5NOT:‘4N0; as a function
to those added, and we have found no in- added isotopes in known ratios.
138 GREEN ET AL.

treatment, has been used to reduce nitrate 2. Green, L. C., Tannenbaum, S. R., and Goldman,
P. (1981) Science 212, 56-58.
to nitrite since the 1950s and 1960s (13- 16). 3. Stainton, M. P. (1974) Anal. Chem. 46, 1616.
The novel aspect of our system is the use of 4. Davison, W., and Woof, C. (1979) Analyst 104,
technology designed for high-pressure liquid 385-390.
chromatography to solve the problems that 5. Radomski, J. L., Palmiri, C., and Heam, W. L.
otherwise arise due to interfering substances (1978) Toxicol. Appl. Pharmacol. 45, 63-68.
6. Witter, J. P., Gatley, S. J., and Balish, E. (1979)
in biological samples. Science 205, 1335-1337.
Davison and Woof (4) have noted that in- 7. Tesch, J. W., Rehg, W. R., and Sievers, R. E. (1976)
terferences in cadmium reduction of nitrate J. Chromatogr. 126, 743-755.
might be reduced by increasing the amount 8. Biemann, K. ( 1962) Mass spectrometry, in Organic
of cadmium used. We have shown that this Chemical Applications (Biemann, K., ed.), pp.
223-225, McGraw-Hill, New York.
improvement can in fact be achieved with 9. Beynon, J. H., and Williams, A. E. (1963) Mass and
the greater availability of cadmium provided Abundance Tables for Use in Mass Spectrometry,
by small particles of metal packed in a high p. 18, Elsevier, Amsterdam.
pressure column. Workers with many sam- IO. Wagner, D. A., Schultz, D. S., Deen, W. M., Young,
ples to analyze for nitrite, nitrate, or both V. R., and Tannenbaum, S. R. (1982) in prep
aration.
might also consider an analysis time of 2 min Il. Burris, R. H. and Wilson, P. W. (1957) in Methods
per sample to be advantageous. of Enzymology (Colowick, S. P., and Kaplan,
N. O., eds.), Vol. 4, pp. 355-365, Academic Press,
ACKNOWLEDGMENTS New York.
12. Griess, J. P. (I 879) Ber. Deutsch. Chem. Ges. 12,
This work was supported by NC1 Grant I-PO l- 426.
CA26731-02. The authors wish to thank Dale Schultz 13. Grau, R., and Mima, A. (1957) 2. Anal. Chem. 158,
for his assistance on the 15N calculations. 182-189.
14. Follett, M. J., and Ratcliff, D. W. ( 1963) J. Sci. Food
REFERENCES Agr. 14, 138-144.
15. Morris, A. W., and Riley, J. P. (1963) Anal Chim.
1. Green, L. C., Ruiz de Luzuriaga, K., Wagner, Acta 29, 272-279.
D. A., Rand, W., Istfan, N., Young, V. R., and 16. Wood, E. D., Armstrong, F. A. J., and Richards,
Tannenbaum, S. R. (198 1) Proc. Natl. Acad. Sci. F. A. (1967) J. Mar. Biol. Assoc. U. K. 47, 23-
USA 78,17&l-7768. 31.