Psychoneuroendocrinology 143 (2022) 105827

Contents lists available at ScienceDirect

Psychoneuroendocrinology
journal homepage: www.elsevier.com/locate/psyneuen

What are oxytocin assays measuring? Epitope mapping, metabolites, and

comparisons of wildtype & knockout mouse urine
Gitanjali E. Gnanadesikan a, b, *, Elizabeth A.D. Hammock c, Stacey R. Tecot a, d,
Rebecca J. Lewis e, Russ Hart f, g, C. Sue Carter h, i, Evan L. MacLean a, b, j, k
a
School of Anthropology, University of Arizona, Tucson, AZ 85721, USA
b
Cognitive Science Program, University of Arizona, Tucson, AZ 85721, USA
c
Department of Psychology and Program in Neuroscience, The Florida State University, Tallahassee, FL 32306, USA
d
Laboratory for the Evolutionary Endocrinology of Primates, University of Arizona, Tucson, AZ 85721, USA
e
Department of Anthropology, University of Texas at Austin, Austin, TX 78712, USA
f
Arbor Assays, Ann Arbor, MI 48108, USA
g
21 Grams Assays Inc, Chelsea, MI 48118, USA
h
Kinsey Institute, Indiana University, Bloomington, IN 47405, USA
i
Department of Psychology, University of Virginia, Charlottesville, VA 22903, USA
j
Psychology Department, University of Arizona, Tucson, AZ 85721, USA
k
College of Veterinary Medicine, University of Arizona, Tucson, AZ 85721, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Oxytocin has become a popular analyte in behavioral endocrinology in recent years, due in part to its roles in
Oxytocin social behavior, stress physiology, and cognition. Urine samples have the advantage of being non-invasive and
Solid-phase extraction minimally disruptive to collect, allowing for oxytocin measurements even in some wild populations. However,
ELISA
methods for urinary oxytocin immunoassay have not been sufficiently optimized and rigorously assessed for their
Urine
potential limitations. Using samples from oxytocin knockout (KO) and wildtype (WT) mice, we find evidence of
Matrix interference
Knockout mice considerable interference in unextracted urine samples, with similar distributions of measured oxytocin in both
genotypes. Importantly, although this interference can be reduced by a reversed-phase solid-phase extraction
(SPE), this common approach is not sufficient for eliminating false-positive signal on three immunoassay kits. To
better understand the source of the observed interference, we conducted epitope mapping of the Arbor Assays
antibody and assessed its cross-reactivity with known, biologically active fragments of oxytocin. We found
considerable cross-reactivity (0.5–52% by-molarity) for three fragments of oxytocin that share the core epitope,
with more cross-reactivity for longer fragments. Given the presence of some cross-reactivity for even the tri­
peptide MIF-1, it is likely that many small protein metabolites might be sufficiently similar to the epitope that at
high concentrations they interfere with immunoassays. We present a new mixed-mode cation-exchange SPE
method that minimizes interference—with knockout samples measuring below the assay’s limit of detec­
tion—while effectively retaining oxytocin from the urine of wildtype mice. This method demonstrates good
parallelism and spike recovery across multiple species (mice, dogs, sifakas, humans). Our results suggest that
immunoassays of urine samples may be particularly susceptible to interference, even when using common
extraction protocols, but that this interference can be successfully managed using a novel mixed-mode cation
exchange extraction. These findings imply that previous conclusions based on urinary oxytocin measur­
ements—especially those involving unextracted samples—may need to be reassessed.

1. Introduction pharmacology. Early work demonstrated its importance for maternal

behavior (Pederson and Prange, 1979; Van Leengoed et al., 1987) and
Oxytocin has been the focus of considerable research effort over the pair-bonding (reviewed in Carter et al., 1995). Recently, evidence has
last several decades, especially in behavioral endocrinology and been mounting for oxytocin’s role in a myriad of social behaviors and

* Correspondence to: School of Anthropology, University of Arizona, P.O. Box 210030, Tucson, AZ 85721-0030.
E-mail address: [email protected] (G.E. Gnanadesikan).

https://doi.org/10.1016/j.psyneuen.2022.105827
Received 4 March 2022; Received in revised form 2 June 2022; Accepted 3 June 2022
Available online 8 June 2022
0306-4530/© 2022 Elsevier Ltd. All rights reserved.

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

cognitive processes, as well as stress physiology (reviewed in Feldman, emotion, finding that dogs’ uOT increased after feeding, exercising, and
2012; Carter, 2014; Froemke and Young, 2021). petting, but not drinking water. In contrast, Marshall-Pescini et al.
Making sense of the oxytocin literature can be challenging, partially (2019) found that affiliative physical contact between dogs and people
due to the variety of experimental methods used. For example, admin­ did not produce consistent changes in uOT levels measured in either the
istration of exogenous oxytocin can produce dosage-dependent results dogs or their owners. Wirobski, Range et al. (2021) found that in pet
(Bales et al., 2007; Cardoso et al., 2013; Peters et al., 2014), and the dogs, uOT was associated with physical contact with their owner;
effects of administration are not necessarily the same as the effects of however, in hand-raised but pack-living dogs and wolves, there was no
endogenous oxytocin (Crockford et al., 2014). While administration association between measured uOT and physical contact with a bonded
studies may be fruitful for pharmacology, studies of endogenous human, suggesting the importance of the animal’s life experience and
oxytocin are crucial for understanding the biology of the oxytocin sys­ the strength of the human-animal bond.
tem, its function, and its evolution. However, studies of endogenous Primatologists have also made use of uOT measurements in a variety
oxytocin concentrations have often struggled to reliably measure of lab and field studies. Using SPE and high-performance liquid chro­
oxytocin, and there are disagreements regarding the best approaches for matography (HPLC), Seltzer and Ziegler (2007) found that a captive
measuring oxytocin, especially by immunoassay (e.g. McCullough et al., male common marmoset’s uOT was lower during a 48-hour social
2013; Lefevre et al., 2017; MacLean et al., 2019). Furthermore, endog­ isolation period than during the subsequent visual reintroduction to his
enous oxytocin measurements and patterns of change are often uncor­ mate. In captive tamarins, Snowdon et al. (2010) found high levels of
related across biological matrices, especially urine (Amico et al., 1987; individual uOT variation, with pair-bonded mates exhibiting similar
Feldman et al., 2011; Weber et al., 2017). While there are good reasons concentrations that were partially explained by their affiliative behav­
that measurements might differ across matrices— specifically different iors. Similarly, in captive common marmosets, strongly bonded pairs
rates of clearance and windows of integration (e.g. urine accumulates exhibited synchronous fluctuations in uOT concentrations (Finkenwirth
oxytocin over the entire time since the last void) (Amico et al., 1987; et al., 2015), and uOT was associated with infant care behaviors such as
Mitsui et al., 2011)—it is unclear to what extent measurements across infant-licking and proactive food sharing in all group members (Fin­
matrices are functionally similar. Complicating these comparisons, kenwirth et al., 2016). In field studies of baboons, higher uOT concen­
sample preparation techniques—such as extraction—that have been trations have been associated with a female’s estrous stage and their
developed and validated in one matrix are often assumed to work proximity maintenance with consort partners (Moscovice and Ziegler,
equally well in other matrices, despite their different compositions. 2012). In field studies of chimpanzees, uOT has been associated with
Using blood plasma samples from oxytocin knockout (KO) and social bonding (Crockford et al., 2013), food sharing (Wittig et al.,
wildtype (WT) mice, we previously demonstrated that while some 2014), and intergroup conflict (Samuni et al., 2017). Lastly, in field
commercially available immunoassay kits (e.g. Enzo Life Sciences) are studies of bonobos, female-female sexual behavior was associated with
susceptible to interference, others (e.g. Arbor Assays) are not, even with increased oxytocin levels, compared with both a baseline condition and
unextracted (diluted 1:8) mouse plasma samples (Gnanadesikan et al., inter-sexual copulation (Moscovice et al., 2019).
2021). Here, we turn this approach towards improving and validating Thus, uOT studies have already documented a variety of intriguing
urinary oxytocin (uOT) measurements. While plasma samples are often associations with social behavior. However, there are differences across
available in laboratory research settings, blood collection is often studies and species regarding the specific behavioral associations found,
impossible or disruptive. In studies of wild populations, blood collection and generalizations remain elusive. Furthermore, some studies have
usually involves trapping animals, and even in captive settings or human failed to find the expected effects (e.g. Marshall-Pescini et al., 2019), and
studies, it may affect behaviors of interest. Thus, the advantages of there is likely publication bias in this regard.
non-invasive hormone measures for behavioral endocrinology have long
been appreciated both in general (Strier and Ziegler, 2005) and for 1.2. Challenges for immunoassay of urinary neuropeptides
oxytocin research specifically (Crockford et al., 2014). However,
developing reliable methods has been challenging. Compounding these challenges in interpreting the urinary oxytocin
Urine samples provide a promising avenue for non-invasive behav­ literature, the specificity and reliability of uOT techniques has not al­
ioral endocrinology for several reasons: 1) urine can be caught in a ways been tested rigorously. Even when extensive validation work has
container, on an absorbent material, or pipetted up off surfaces without been conducted (e.g. Seltzer et al., 2010; Wirobski, Schaebs et al., 2021),
handling the animal; 2) urination may occur throughout the day, assessment of potential false-positive signal has not been possible,
enabling collection at multiple time points; 3) many species produce lacking a “ground truth” (i.e. known concentrations). Additionally,
volumes of urine that are sufficient to assay multiple analytes in the although many studies have employed some form of extraction, it re­
same sample. mains unknown to what extent these methods are sufficiently selective
to retain oxytocin while eliminating potentially interfering molecules.
1.1. Urinary oxytocin and behavior across species Urine contains a large quantity and variety of metabolites (Bouatra
et al., 2013) that we expect might interfere with immunoassays, and
Many studies have already utilized uOT measurements and found thus extraction and immunoassay specificity may be particularly
interesting associations with social behavior. In humans, uOT has been important for this matrix. Are inconsistencies across studies due to
associated with perceptions of a romantic partner’s bonding behavior, meaningful differences in species, behaviors, and sampling windows, or
regardless of gender (Algoe et al., 2017), and with performance on a are they methodological artifacts due to differences in extraction and
facial visual search task in men (Saito et al., 2014). In children, a immunoassay? To shed light on these methodological questions, in this
mother’s touch and speech after a stressful event increased uOT while study we tested multiple methods on both oxytocin KO and WT mice,
decreasing salivary cortisol (Seltzer et al., 2010). Interestingly, in searching for techniques that minimized interference in the KO samples
mothers, uOT was higher after interacting with other people’s children while maximizing signal in the WT samples.
compared to their own children, perhaps due to oxytocin’s role in the The specificity of immunoassays can be affected by many aspects of
onset—rather than maintenance—of maternal behavior, or perhaps their chemistry. The antibody is particularly important, as different
simply due to a novelty effect (Bick and Dozier, 2010). antibodies may bind different portions of a molecule. This region of an
Increasingly, uOT has been used in studies of human-animal in­ antigen that is bound by the antibody is known as the epitope. Aiming to
teractions. Nagasawa et al., (2009, 2015) demonstrated that uOT in dogs shed light on potential sources of interference, we worked with Pepscan
and their owners was associated with mutual gazing. Mitsui et al. (2011) Presto BV (Lelystad, The Netherlands) to perform epitope mapping of
suggested that oxytocin could be used as a biomarker of positive the Arbor Assays antibody. Motivated by these results, we also tested the

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

cross-reactivity of the Arbor Assays antibody with three known, bio­ across 2–3 individuals within the same—blinded—genotype code, due
logically active fragments of oxytocin (Uvnäs Moberg et al., 2019), as to limited sample volume. All samples were from male mice. Six indi­
well as the effects of two extraction methods on retention of these vidual Oxt+/+ wildtype samples were used; 10 individual Oxt− /−
fragments. knockout samples were pooled as necessary to create 6 pooled samples
Lastly, using a mixed-cation exchange method that minimized of 1–3 individuals each.
interference in knockout mouse samples, we tested its analytical validity
on the Arbor Assays kit across a variety of species (mice, sifakas, dogs, 2.2. Sifaka samples
and humans). We focused on the Arbor Assays kit because it is the
predominate platform that we use in our work (Gnanadesikan et al., Verreaux’s sifaka (Propithecus verreauxi) samples were collected from
2021), although we ran a small number of samples on the Cayman a habituated wild population at Ankoatsifaka Research Station in Kir­
Chemical and Enzo Life Sciences kits for comparison. We used samples indy Mitea National Park, Madagascar, under IACUC Protocol #13–470
from species that the investigators work with regularly in their research from the University of Arizona and IACUC Protocol #2017–152 from the
and that represent different orders of mammals with potentially inter­ University of Texas at Austin. Fresh urine samples were collected from
esting biological differences (i.e. size, diet, urine concentration). known, identified individuals by catching the stream directly with foil,
catching the stream with foil off of tree trunks, or pipetting urine off of
2. Methods foliage with disposable pipettes. Samples were pipetted into cryovials,
which were immediately placed into a thermos with either instant ice
2.1. Mouse subjects and sample collection packs or tubes of frozen water until transferred into a liquid nitrogen
tank within 4 h. Samples were kept frozen in a liquid nitrogen dewar
As in our previous work (Gnanadesikan et al., 2021), Oxttm1Zuk mice during transport from the field site to Antananarivo and while awaiting
(Nishimori et al., 1996) were maintained on a C57BL/6J background shipment to the US. They were shipped to the University of Arizona in a
and bred at Florida State University; all breeding and handling pro­ charged liquid nitrogen shipper and frozen at − 80 ◦ C until analysis at
cedures were approved by the Institutional Animal Care and Use Com­ the Laboratory of the Evolutionary Endocrinology of Primates (LEEP).
mittee (IACUC) following the Guide for the Care and Use of Laboratory We created one pool (of two samples) for assessment of parallelism
Animals (Protocols 2020–21 and 202 000 042). Oxt+/− breeder pairs and a second pool (adding seven additional samples) to assess spike
were continuously housed. The first morning of the appearance of a litter recovery. Sample information is provided in the appendix (table A5).
was noted as postnatal day 0 (P0). Mice were weaned, tagged, and tailed
for genotyping on P21 and group housed by sex. After genotyping, 2.3. Dog and human samples
same-sex mice were re-housed by genotype, so that only mice of the
same genotype were housed together. Mice were housed on a 12:12 L:D Dog (Canis lupus familiaris) and human (Homo sapiens) urine pools
cycle in open wire-top caging with wood chip bedding and provided ad were created through volunteer collections among laboratory staff/
libitum food (LabDiet Rodent 5001) and water. In total, samples from students and their pet dogs, as well as using extra volume collected for
159 mice of this strain (WT: 73, KO: 86) were used in this study. ongoing studies, under University of Arizona IACUC Protocol #16–175
Additionally, adult C57BL/6J (B6) wildtype mice were purchased from and IRB Protocol #1 808 883 345.
The Jackson Laboratory (total n = 20).
Urine samples were passively collected from adult male and female 2.4. Solid-phase extractions
mice during routine handling on the day of planned euthanasia for other
purposes. During live handling to confirm the mouse identification (ID 2.4.1. Hydrophilic-Lipophilic Balanced (HLB)
number) by ear tag prior to euthanasia, the mouse was held over a Hydrophilic-Lipophilic Balanced reversed-phase extractions were
microcentrifuge tube to collect urine. After confirming the mouse ID conducted using the Oasis PRiME HLB 1 cc cartridges (Waters Corpo­
number, the tube was closed, labeled with the ID number, and frozen. ration, Milford, MA, USA, Part Number: 186 008 055) and a positive
Samples were shipped frozen to the University of Arizona on dry ice and pressure manifold (Biotage PRESSURE+48). All samples were diluted
stored at − 80 ◦ C until time of assay. Each urine sample, identifiable by into an equal volume of 0.1% trifluoroacetic acid (TFA) in water, vor­
ID number, was assigned a genotype code by E.A.D.H. to enable texed for at least 30 s, and then centrifuged. Exact speeds and times
experimental design, assay layout planning, and sample pooling (within varied, in part due to sample volumes; all samples were centrifuged at
genotype when necessary), while retaining sample blinding for G.E.G. either 10 000 rpm (RCF = 9632 x g) for 5 min or 5000 rpm (RCF = 4696
for assays directly comparing knockout and wildtype individuals. x g) for 15 min.
SPE cartridges were conditioned first with 1 ml acetonitrile (ACN)
2.1.1. Genotyping with 0.1% TFA, and then with 1 ml 0.1% TFA in water. The entire
DNA from tail samples was genotyped for Oxt alleles using a common sample supernatant was loaded onto the cartridge. Each cartridge was
forward primer (5’-TCAGAGATTGAACAAGACGCC) and specific reverse then washed with 1 ml 10% ACN, 0.1% TFA. Finally, samples were
primers for WT (5’-TCAGAGCCAGTAAGCCAAGC) and KO (5’- eluted with 2 ml 30% ACN, 0.1% TFA. This procedure is identical to that
ACTTGTGTAGCGCCAAGTGC). Using a hot start and 40 cycles of 30 s reported in Gnanadesikan et al. (2021), although 2 ml of eluant was used
each of 94 ◦ C, 57 ◦ C, and 72 ◦ C, the primers generated a wildtype allele in all cases, to maximize recovery. It is worth noting that the ACN
of approximately 500 bp and a knockout allele of approximately 180 bp, content of the elution (30%) is lower than is commonly used in other
resolved by gel electrophoresis. extraction protocols (e.g. 60% or 85%). In reversed-phased SPE, higher
organic content is expected to elute more molecules—especially more
2.1.2. Samples in each analysis hydrophobic molecules—which could include additional sources of
For a summary of mouse samples used in this study, see table A3. The interference. Thus, elution with lower ACN content should theoretically
first experiment was conducted on unextracted urine samples: 10 result in a more selective extraction.
wildtype Oxt+/+ samples (7 female, 3 male) and 14 knockout Oxt− /−
samples (8 female, 6 male). The second experiment, as well as the 2.4.2. Mixed-mode Cation Exchange (MCX)
methods development work reported in the appendix used a series of Mixed-mode Cation Exchange extractions were performed using
sample pools. In total, these pools used samples from 76 wildtype in­ Oasis PRiME MCX 1 cc cartridges (Waters Corporation, Milford, MA,
dividuals (mix of C57BL/6J and Oxt+/+) and 62 knockout individuals USA, Part Number: 186 008 917) and a positive pressure manifold
(Oxt− /− ). For experiment three, some individual samples were pooled (Biotage PRESSURE+48). All samples were diluted into an equal volume

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

of loading buffer (200 mM ammonium formate with phosphoric acid to vasopressin.

achieve pH 5). Samples were vortexed for at least 30 s and then
centrifuged, as above, at either 10 000 rpm (9632 × g) for 5 min or 5000 2.5.1. Coefficients of variation
rpm (4696 × g) for 15 min For unextracted samples of both genotypes (experiment 1), the mean
The MCX cartridges were first conditioned with 1 ml methanol intra-assay coefficient of variation (CV) between duplicates was 7.1%.
(MeOH) and 1 ml of deionized water (Thermo Scientific #751 628). The For MCX-extracted samples (experiment 3), the mean intra-assay CV for
entire supernatant of the sample was then loaded onto each cartridge. samples that measured above the assay’s limit of detection (n = 6) was
Cartridges were washed with successive 1 ml washes: 1 ml deionized 7.4%. Results were not compared across plates (except for the multi-kit
water, 4 ml wash buffer (60% loading buffer, 40% MeOH), 1 ml comparison).
deionized water. Finally, each sample was eluted in 2 ml of elution
buffer (50% MeOH, 50% ammonium hydroxide solution). The pH of the 2.6. Epitope mapping
elution buffer was 12 in the three-way kit comparison; the subsequently
optimized protocol—used for the individual-level analyses—uses pH 11. Epitope mapping of the Arbor Assays antibody was conducted by
We experimented with various combinations of loading, wash, and Pepscan Presto BV (Lelystad, The Netherlands). The basic approach in­
elution buffers and volumes to optimize this protocol, as reported in the volves synthesizing peptide fragments and testing an antibody’s affinity
appendix. for each sequence (Geysen et al., 1984). First, Pepscan constructed ar­
rays of peptide fragments to assess the linear and conformational
2.4.3. All extracted samples epitope binding; these arrays consisted of all possible 10-amino-acid
The eluted samples were centrifuged at 3000 rpm (1690 × g) for 5 fragments of oxytocin/neurophysin I prepropeptide, the larger precur­
min to ensure that liquid on the side of the tubes settled at the bottom. sor protein from which oxytocin is derived. Next, they performed
All samples were stored at − 80 ◦ C overnight and lyophilized the next fine-mapping of the epitope region by using single amino acid sub­
day using a CentriVap (Labconco model #7 810 016) with the following stitutions for each of the 9 amino acids in oxytocin to determine how
settings: CentriVap unheated, cold trap − 80–85 ◦ C, vacuum 0.3–0.4 each substitution affected antibody binding. The binding of the antibody
mbar. When lyophilization was complete, and immediately before assay, to each of the synthesized peptides was tested in a Pepscan-based
samples were reconstituted in assay buffer. Individual samples were immunoassay. This approach enables a comparison of the antibody’s
reconstituted in 250 μl assay buffer; certain pools for multi-kit com­ affinity for different peptide sequences, which can shed light on the
parisons or validations used larger volumes, with concentration factors antibody’s specificity; sequences that do not demonstrate high levels of
specified below. Reported concentrations are corrected for the concen­ antibody binding are unlikely to interfere with immunoassay measure­
tration or dilution factors inherent in the extraction. ment, while sequences that show binding levels similar to the identified
epitope may cause interference.
2.4.4. Extraction efficiency
Extraction efficiencies were determined by extracting spiked buffer. 2.7. Oxytocin fragments
We used the stock oxytocin solution from the Arbor Assays kit and
diluted it to 1000 pg/ml in Arbor Assay’s oxytocin assay buffer. Ex­ Given that fragments of oxytocin are known to be formed through
tractions were performed on 250 μl aliquots, while an additional aliquot multiple metabolic processes (Uvnäs Moberg et al., 2019), we hypoth­
was frozen for assay unextracted. Reported extraction efficiencies were esized that these fragments might cross-react with immunoassay anti­
calculated as (extracted measurement / unextracted measurement) × bodies. We focused on three biologically active fragments of oxytocin
100. Extraction efficiency was also assessed in dog urine by adding that have been studied previously: OT 2–9, OT 4–9, and OT 7–9, where
spiked buffer (10%, ~1000 pg/ml) to a dog urine sample (90%), and the numbers indicate the residues of oxytocin (a nonapeptide) contained
measuring the sample with and without the spike, as well as the unex­ in the fragment. OT 7–9 is more commonly known as
tracted spiked buffer. Reported extraction efficiencies were calculated as melanocyte-inhibiting factor 1 (MIF-1). In order to assess the
(spiked urine measurement)/ (0.1 × unextracted spike + 0.9 × urine cross-reactivities of these fragments on the Arbor Assays kit, we pur­
measurement) × 100. Extraction efficiency was not assessed in other chased MIF-1 from Cayman Chemical (Catalog #24 476) and had OT
species at this time; however we encourage future studies to assess 2–9 and OT 4–9 synthesized by Biomatik (Ontario, Canada). Since
extraction efficiency in the relevant species and matrix. oxytocin is cyclic, with residues 1 and 6 connected by a disulfide bridge,
OT 2–9 and OT 4–9 were synthesized with a cystine at position 6 (cys­
2.5. Oxytocin assays teine-cysteine, connected by a disulfide bridge), which is consistent with
known metabolic pathways (Uvnäs Moberg et al., 2019). For more in­
Except where otherwise noted, oxytocin was measured using the formation on these fragments, see Fig. 4.
Arbor Assays Oxytocin Enyzme Immunoassay (EIA) kit (Catalog #K048- We reconstituted a single 10 mg aliquot of MIF-1 in 10 ml of
H5), following the recommended assay protocol (although not using the deionized water, freezing reconstituted aliquots for future assays. The
extraction solution provided). The reported sensitivity for this kit is 17.0 synthesized fragments came in 1 mg aliquots, so we used a new aliquot
pg/ml and the lower limit of detection is 22.9 pg/ml; the limit of for each assay; we reconstituted each aliquot in 1 ml deionized water to
detection of a given experiment is dependent on (multiplied or divided create a 1 mg/ml solution, thus variation in these aliquots could
by) the dilution or concentration factor inherent in the extraction, contribute to variation in measured cross-reactivities across assays. On
specified below. Arbor Assays reports that cross-reactivity is 94.3% for the day of each assay, the 1 mg/ml solution was diluted in assay buffer to
isotocin, 88.4% for mesotocin, and less than 0.15% for vasotocin and the starting concentration of the assay. In a first experiment, we created
arginine vasopressin. 1 μg/ml starting solutions and then conducted five 1:10 part-to-whole
In the three-way kit comparison, we also used the Enzo Life Sciences (p:w) serial dilutions for each fragment to identify what orders of
(Catalog #ADI-900–153A0001) and Cayman Chemical (Catalog #500 magnitude would result in measurements within the standard curve.
440) kits. The manufacturer-reported sensitivities are 15 pg/ml for Enzo Informed by these results, we conducted a second assay with serial di­
and 20 pg/ml for Cayman. The reported cross-reactivities for the Enzo lutions targeting the middle of the standard curve (see table A6 for input
kit are: mesotocin 7.0%, Arg8-vasotocin 7.5%, and < 0.02% for all other concentrations). Reported mean cross-reactivities were calculated based
reported compounds including Arg8-vasopressin. Reported cross- on those dilutions that measured within 30–70% of the assay’s
reactivities for the Cayman kit are 100% for mesotocin and isotocin maximum binding (B0) as the percentage of the fragment’s known
and < 0.01% for all other reported compounds including Arg8- concentration that was measured by the assay. Although immunoassay

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

results are usually reported by weight (e.g. pg/ml), since the fragments (Mean ± SD: WT = 930.8 ± 512.7 pg/ml, KO = 885.7 ± 295.6 pg/ml;
are very different sizes, we report cross-reactivities by both molarity and Welch’s t(13.5) = 0.248, p = 0.808; Fig. 1). We interpreted this as
weight. The cross-reactivities-by-molarity represent a direct comparison indicating a high level of interference in the immunoassay and pro­
based on the number of peptides present, while cross-reactivities-by- ceeded to explore various extraction options to eliminate or minimize
weight may be helpful in considering what fraction of a measurement this interference.
might be due to each of these fragments. In a third assay, we explored
whether the HLB and MCX extractions removed these fragments. Each 3.2. Experiment 2: basic HLB extraction
fragment was prepared in Arbor Assay’s oxytocin assay buffer to a
concentration that we expected would result in approximately 30% of Since solid-phase extraction is commonly used to remove interfering
the maximum binding, based on the previous assays. An aliquot was molecules from biological matrices, we evaluated the performance of a
frozen overnight to measure unextracted, and two aliquots were relatively simple hydrophilic-lipophilic balanced (HLB) extraction.
extracted using the HLB and MCX solid phase extraction methods. The Similar reversed-phase extraction techniques using different cartridges
extractions were performed on 250 μl of sample, the eluate was lyoph­ and buffers have frequently been used in oxytocin measurements across
ilized, and the sample was reconstituted in 250 μl of assay buffer at the matrices and species (e.g. Szeto et al., 2012; Crockford et al., 2013;
time of assay. As a baseline for comparison, the same procedure was Bienboire-Frosini et al., 2017), and we had previously validated this
applied to a spiked buffer sample composed of Arbor Assay’s oxytocin specific HLB method in mouse plasma (Gnanadesikan et al., 2021). With
assay buffer spiked with ~1000 pg/ml of the kit standard. urine samples, this extraction resulted in a clear genotype contrast, but
there was still considerable interference in the knockout samples (WTpool
2.8. Analytical validations = 611.9 pg/ml; KOpool = 188.03 pg/ml). Subsequent replications found
similar results (not reported). We assessed whether an increased wash
Analytical validations were conducted with wildtype mouse, sifaka, volume would eliminate the observed interference, but instead found
dog, and human sample pools. Each species differed in the amount of that we lost signal in both genotype pools, and the genotype contrast was
sample extracted (and thus starting concentration/dilution factor) but not improved (WTpool = 116.49 pg/ml; KOpool = 72.81 pg/ml).
were otherwise treated identically. Parallelism was assessed via serial
dilution (2:3, p:w) of urine pools and calculation of the coefficient of 3.3. Experiment 3: improved MCX extraction method
variation on corrected concentrations at each dilution (Andreasson
et al., 2015). Starting concentration factors for each parallelism were In order to improve our extraction selectivity, we developed and
1.33x for mouse and sifaka, 5.3x for dog, and 16x for human. optimized a new method using a mixed-mode cation exchange (MCX)
Spike recovery was assessed using samples at a single dilution (ratio cartridge. This enabled us to manipulate the retention using both pH and
of volume applied to the cartridge for SPE to the volume reconstituted): organic content of the loading, wash, and elution buffers. For details of
2:5 for mouse, 1:2 for sifaka, 4x for dogs, and 8x for humans. All spike the methods development, see the appendix (Section 3, figures A1–A5).
recovery extractions were reconstituted in 250 μl assay buffer; sample Using this new method, we then compared the MCX and HLB
volumes were 100 μl for mouse, 125 μl for sifaka, 1 ml for dog, and 2 ml extraction methods across three different oxytocin immunoassay kits,
for humans. Spiked samples consisted of 90% sample matrix at the hypothesizing that differences in the kit chemistries—especially anti­
specified dilution and 10% synthetic oxytocin in assay buffer (kit stan­ bodies—might lead to different susceptibilities to interference. We
dards 1–5). Percent recovery was calculated as (observed / expected) x found that the HLB extraction was susceptible to varying levels of
100, where the expected values were measured independently by adding interference across kits, but the MCX extraction minimized this inter­
each spike to assay buffer. ference on all three kits (KO measurements 13− 39 pg/ml, corrected
concentrations 21− 66 pg/ml; WT measurements 124–171 pg/ml, cor­
2.9. Statistical software rected concentrations 206–286 pg/ml; Fig. 2; table A2).
While these results were much improved, we wanted to further
Statistical analyses were conducted in R version 4.1.1 (R Core Team, minimize the KO interference, with the aim of increasing the WT:KO
2021). For the unextracted urine comparison in experiment 1, distri­ ratio. We achieved this on the MCX cartridge by decreasing the pH of the
butions were not significantly different from a normal distribution elution buffer from 12.0 to 11.0 for a more selective elution. We then
(Shapiro-Wilk’s WWT = 0.948, pWT = 0.644; WKO = 0.955, pKO = 0.677), tested individual samples (some pooled) to examine the genotype
so Welch’s t-test was used to compare the means. For the HLB extrac­ contrast again on the Arbor Assays kit. All knockout samples measured
tions in experiment 2 and the three-way kit comparison in experiment 3, below the assay’s limit of detection. As such, these values should not be
single measurements were made on pooled samples and no statistical interpreted (and in some cases had high CVs due to low measurement);
tests were used. For the individual-level results in experiment 3, sample nevertheless, the difference between groups is statistically significant
sizes were small (n = 6) for each genotype, so we used a Mann-Whitney (Mann-Whitney W = 0, p = 0.002; Fig. 3). We therefore used a pH 11.0
test to avoid normality assumptions. See the appendix for a discussion of elution buffer for all subsequent assays.
power analyses. Using spiked buffer samples, we also assessed the extraction effi­
ciency of the MCX method; in one assay across three samples, the
3. Results extraction efficiency was 83%, while in another assay across two sam­
ples, the extraction efficiency was 92%. Measured in spiked dog urine,
3.1. Experiment 1: unextracted urine extraction efficiency was 121%.

Given our past success assaying unextracted mouse plasma samples 3.4. Epitope mapping
(Gnanadesikan et al., 2021), we began by assessing the genotype
contrast in unextracted diluted mouse urine samples on the Arbor Assays Given the considerable interference that we observed in unextracted
EIA kit. Due to limited sample volume for some individuals, some and HLB-extracted urinary samples, we decided to further probe the
samples were assayed at a 1:10 (p:w) dilution into Arbor Assay buffer, properties of the anti-oxytocin polyclonal rabbit antibody used in the
while others were measured at a 1:25 dilution. Based on a serial dilution Arbor Assays EIA. We therefore worked with Arbor Assays and Pepscan
of a single wildtype C57BL/6J urine sample, we expected both of these to perform both linear and conformational epitope mapping of this
dilutions to be within range of the assay (see appendix, table A1). antibody. Epitope mapping tests an antibody’s affinity to a library of
Knockout samples were indistinguishable from wildtype samples synthesized peptide fragments, thereby determining its epitope(s) and

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

Fig. 1. Boxplot comparing the dilution-corrected measured

oxytocin concentrations in unextracted urine samples
across genotypes on the Arbor Assays kit. Depending on
available sample volume, samples were assayed at a 1:10
(circle) or 1:25 (triangle) dilution into assay buffer. The
mean measured oxytocin concentrations were not signifi­
cantly different across genotypes (Mean ± SD: WT =
930.8 ± 512.7 pg/ml, KO = 885.7 ± 295.6 pg/ml; Welch’s
t(13.5) = 0.248, p = 0.808). Note that this pattern was true
even with the presence of a single high WT sample;
excluding this point, Mean WT = 819.6 ± 395.5, which is
below the KO mean. The limits of detection are 229 pg/ml
for samples assayed at a 1:10 dilution and 573 pg/ml for
those assayed at a 1:25 dilution.

Fig. 2. One pooled sample for each genotype was

measured on three commercial kits after either an HLB or
MCX solid phase extraction (1.67x concentration through
SPE in both cases). HLB samples demonstrated varying
levels of interference across kits. The MCX extraction
minimized this interference and resulted in relatively
concordant results across kits. The gray rectangles span the
range of the kits’ sensitivities and limits of detection
(adjusted for the extraction concentration factor): Arbor
lower limit of detection = 38 pg/ml, Cayman sensitivity
= 33 pg/ml, Enzo sensitivity = 25 pg/ml.

assessing its specificity. 3.5. Oxytocin fragment cross-reactivities and extractions

The epitope mapping results identified a single epitope within
oxytocin/neurophysin I prepropeptide, corresponding to residues 2–9 of Given what we learned about the epitope of the Arbor Assays anti­
oxytocin (YIQNCPLG). Further testing of this region— by assessing body, and especially that the core epitope consists of residues 7–9, we
single amino acid substitutions—revealed that the core epitope is PLG investigated the cross-reactivity of this antibody with known, biologi­
(Fig. 4, A6). Almost any change to these residues resulted in highly cally active fragments of oxytocin (Uvnäs Moberg et al., 2019). Specif­
disrupted binding, although the magnitude of this change was depen­ ically, we purchased OT 2–9, OT 4–9, and OT 7–9, the latter more
dent on the exact substitution; substitutions outside of this core region commonly known as melanocyte-inhibiting factor 1 (MIF-1). Across
had less impact on antibody binding (see supplementary data file for raw multiple dilutions, the average cross-reactivities-by-molarity measured
epitope mapping results). within the range of 30–70% of the assay’s maximum binding (B0) were
The linear and loop arrays produced similar results, as did versions 52% for OT 2–9 (also 52% by-weight), 22% for OT 4–9 (32% by-weight),
allowing the free cysteines to bridge, suggesting that the conformational and 0.5% for MIF-1 (2% by-weight; Figs. 4 and 5; table A6).
structure of oxytocin—including the disulfide bridge between the Finally, we tested both the HLB and refined MCX extractions on each
cysteine residues—is not important for epitope recognition, consistent of the three oxytocin fragments, as well as oxytocin itself. While both
with earlier studies reporting immunoreactivity with the linear form of extractions retained > 90% of oxytocin itself, extraction considerably
oxytocin after a reduction/alkylation procedure (Brandtzaeg et al., reduced the measured oxytocin concentrations for all three fragments.
2016). Both extractions removed most of the fragments, reducing the assay’s

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

musculus), Verreaux’s sifaka (Propithecus verreauxi), domestic dogs

(Canis lupus familiaris), and humans (Homo sapiens).
The wildtype mouse pool diluted parallel to the standard curve (CV
of corrected concentrations = 10.1%; figure A8) and exhibited good
spike recovery for samples measured at a 2:5 dilution factor (mean ± SD
= 114 ± 7%). Similarly, pooled sifaka urine samples demonstrated good
parallelism (CV of corrected concentrations = 17.5% over full range;
5.11% for lower 5 dilutions; figure A9) and spike recovery measured at a
1:2 dilution (mean ± SD = 114 ± 6%). A dog urine pool diluted parallel
to the standard curve (CV of corrected concentrations = 6.0%; figure
A10) and exhibited acceptable spike recovery when assessed at a 4x
concentration (mean ± SD = 119 ± 9%).
Lastly, a human urine pool diluted relatively parallel to the standard
curve, especially for higher concentrations (figure A11). The CV of
corrected concentrations over the entire range (16x–2.1x) was 18%;
excluding the fourth dilution, which had a duplicate CV of 25%, this rose
to 19%. However, the last two dilutions measured somewhat higher than
Fig. 3. Genotype contrast using a more selective mixed-mode cation exchange expected, thus for the first three dilutions alone, the CV of corrected
SPE chemistry and the Arbor Assays kit. Each point represents an individual concentrations was 6.6%. Human urine also exhibited acceptable spike
sample, some of which were pooled across 2–3 samples within the same recovery when measured at an 8x concentration (mean ± SD = 123
(blinded) genotype code, due to limited sample volume, resulting in six samples ± 8%). It should be noted, however, that this pool had a specific gravity
per genotype. A clear genotype distinction can be seen between knockout (KO) of 1.023 and a corrected concentration of approximately 7 pg/ml.
and wildtype (WT) samples (Mann-Whitney W = 0, p = 0.002). Human population means for specific gravity are generally 1.01–1.02
(Goldberger et al., 1995; Miller et al., 2004). Thus, even with an 8x
measurement by 1–2 orders of magnitude. The extractions were more concentration, some samples will be near or below the limit of detection
effective at eliminating OT 4–9 and MIF-1; for OT 2–9—the molecule of this assay, and higher concentration factors may be necessary for
most similar to oxytocin—the MCX extraction removed considerably more dilute samples. More experimentation is needed to determine
more of the fragment (measured 117.15 pg/ml) than did the HLB whether an ideal protocol exists for samples of all concentrations, or
extraction (measured 283.05 pg/ml; see Fig. 6). what the best methods are for compensating for the wide variation in
specific gravity (and consequently oxytocin concentrations) in human
urine. Relatedly, in some subsequent experiments, spike recovery for
3.6. Analytical validations: mouse, sifaka, dog, and human
certain human urine pools was above acceptable thresholds, so further
optimization may be necessary for human samples.
Using the optimized MCX method, we conducted analytical valida­
tions on pooled samples from four species/strains: wildtype mice (Mus

Fig. 4. Oxytocin is a nonapeptide formed from the cleavage of oxytocin/neurophysin I prepropeptide. Our epitope mapping of the Arbor Assays anti-oxytocin
antibody demonstrates that the epitope spans oxytocin residues 2–9, while the core epitope consists of residues 7–9. We purchased three biologically active frag­
ments of oxytocin to test their cross-reactivity with the Arbor Assays antibody, finding high cross-reactivity for 2–9, intermediate cross-reactivity for 4–9, and low
cross-reactivity for 7–9 (also known as MIF-1). Cross-reactivities reported here are by-molarity (see text for cross-reactivities by-weight). Since oxytocin is cyclic, with
residues 1 and 6 connected by a disulfide bridge, OT 2–9 and OT 4–9 were synthesized with a cystine at position 6 (cysteine-cysteine), consistent with known
metabolic pathways.

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

Fig. 5. Measured oxytocin concentrations on the Arbor

Assays EIA kit vs. known concentrations of three biologi­
cally active fragments of oxytocin (log scale). Only mea­
surements within the range of 30–70% of the assay’s
maximum binding are shown and used to calculate cross-
reactivity, the percent of the known fragment concentra­
tion that is measured as oxytocin. OT 2–9, the fragment
most similar to oxytocin, had the highest cross-reactivity
(52% by-molarity and by-weight), followed by OT 4–9
(22% by-molarity, 32% by-weight). MIF-1, the smallest
fragment tested, displayed the lowest cross-reactivity
(0.5% by-molarity, 2% by-weight). Input and measured
concentrations, percent binding, and calculated cross-
reactivities are presented in table A6.

Fig. 6. Oxytocin and each of the three oxytocin fragments

(in Arbor Assays’ assay buffer) were measured with no
extraction and after one of two solid-phase extractions:
HLB or MCX. Both extractions retain most of the oxytocin
present in spiked buffer (HLB: 100%, MCX 93%). However,
the extractions eliminate most of the signal measured for
each fragment of oxytocin, with the MCX extraction
removing more of oxytocin 2–9 than the HLB extraction.
Note that the actual starting concentrations for each frag­
ment are higher than their measured concentrations (OT
2–9: 2500 pg/ml; OT 4–9: 4000 pg/ml; MIF-1: 100 000 pg/
ml).

4. Discussion and 130 – 950 pg/ml (Mean ± SD = 430 ± 190) after doing so. The dog
pool that we used for both parallelism and spike recovery analyses had
Our results suggest that many existing urinary oxytocin methods may concentrations of 40 – 50 pg/ml (no SG correction), and in another study
be suboptimal. We found that unextracted urine samples—measured at a in progress, pet dog samples have generally measured < 100 pg/ml
1:10 or 1:25 dilution in assay buffer—were indistinguishable on the before SG-correction, and typically < 250 pg/ml after SG-correction.
Arbor Assays kit across oxytocin knockout and wildtype mice. A basic In methods development work, higher concentrations are often
reversed-phase SPE on the Waters Oasis PRiME HLB cartridge improved assumed to be better (e.g. higher extraction efficiency or improved
the genotype contrast, but still exhibited considerable interference, storage). This assumption is tempting, in part because many studies
which was observed at varying levels across three different commer­ suffer from an inability to detect analytes at low concentrations. How­
cially available immunoassay kits. This interference was eventually ever, this assumption may lead us astray without a careful consideration
minimized with the development of a new SPE method using the Waters of potential sources of interference, which may be particularly abundant
Oasis PRiME MCX cartridge, which utilizes a mixed-mode cation ex­ in urine. On the other hand, it has also been assumed—without sufficient
change chemistry. Although we have not tested all of the cartridge- empirical evidence—that high oxytocin concentrations are biologically
extraction-kit combinations that have previously been used in the implausible. This and other recent work (Gnanadesikan et al., 2021)
literature, our findings across three different commercially available kits suggest that some methods, at least, are not susceptible to interference in
suggest that there is strong potential for interference in immunoassays of mouse samples.
urinary oxytocin, even following sample extraction. It is notable that the Feldman et al. (2011) found that in human samples, while plasma
concentrations we report using the MCX method are lower than those and salivary measures of oxytocin were correlated, extracted urinary
reported with some other methods. For example, Wirobski, Range et al. measurements were not. Weber et al. (2017) also found no correlation
(2021) reported pet dog concentrations in the range of 77 – 1500 pg/ml between unextracted plasma and unextracted urinary oxytocin mea­
(Mean ± SD = 440 ± 290) before correcting for specific gravity (SG), surements in premature human infants. While there are many possible

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

reasons for these findings, including that different biological matrices challenges for urinary oxytocin work, the method we report here min­
are thought to integrate over different periods of time, another plausible imizes measurement of other small peptide fragments, including
interpretation in light of our results is that the urine measurements in oxytocin metabolites and other interfering molecules.
these studies were susceptible to interference. This interpretation is It should also be noted that while other immunoassay kits use
supported by studies that have found correlations between plasma and different antibodies, given that the oxytocin molecule is so small, other
extracted urinary oxytocin in certain conditions and time windows antibodies likely have a similar susceptibility to interference, even if
(Amico et al., 1987; Francis et al., 2016). their epitope and/or core epitope is slightly different. For example, we
It should be noted that many of the previously demonstrated asso­ previously demonstrated that the Enzo Life Sciences kit is susceptible to
ciations between urinary oxytocin measurements and social behavior interference with plasma samples in situations where the Arbor Assays
may be robust, despite these potential limitations, especially if kit is not, presumably reflecting differences in their epitopes or speci­
employing a change-from-baseline approach. Nevertheless, many ficity (Gnanadesikan et al., 2021). Similarly, Wirobski, Schaebs et al.
studies may have inflated concentrations due to the type of interference (2021) found that the Arbor Assays kit performed better than the Enzo
we demonstrated here. This interference may also contribute to negative kit for human and canine urinary oxytocin, with better spike recoveries,
findings if the noise introduced by interference obscures an underlying lower intra-assay CVs, and increased sensitivity. Nevertheless, given the
effect, which may be especially important for studies of individual fundamental constraint posed by the small size of the oxytocin molecule,
differences. improved immunoassay of urine samples will most likely rely on
While the success of our method across a wide variety of mammalian improved extraction techniques, such as the solid-phase extraction
species—including members of the orders Primates, Rodentia, and method that we report here.
Carnivora—is promising, the extraction method as reported here may Whether immunoreactivity from oxytocin’s metabolites should be
not be optimal for all species. It should be noted, however, that this considered interference depends on the perspective of the end user.
method worked well even in wild sifaka samples, which are highly Indeed, analytes measured in excretory products such as urine and feces
concentrated due to their extremely limited water intake; therefore we often involve binding to metabolites of the target analyte rather than the
do not expect highly concentrated urine to be problematic. Nevertheless, parent molecule, and in some cases this may be desirable (Palme et al.,
we strongly encourage anyone who aims to extend this work to other 2013). It has recently been proposed that many of oxytocin’s functions
species and populations to perform a careful analytical validation with may derive from active fragments of the nonapeptide (Uvnäs Moberg
their own samples, particularly parallelism and spike recovery, although et al., 2019), suggesting that oxytocin metabolites may be relevant
these tests alone may not be sufficient to determine the absence of biomarkers for some aspects of the oxytocinergic system. Our results
interference. indicate that several bioactive fragments of oxytocin are nearly entirely
Our epitope mapping results illuminate one of the core components eliminated using solid-phase extraction, making this an important
of any immunoassay: the antibody. The importance of the core epito­ consideration for the protocols described here. Given the abundance of
pe—spanning residues 7–9—is consistent with the manufacturer’s re­ interfering molecules in urine, it will be challenging for immunoassays
ports of low cross-reactivity with mesotocin and vasopressin, since both to measure oxytocin metabolites accurately in urinary samples; how­
of these related nonapeptides differ from oxytocin at the eighth residue. ever, this may be possible in other matrices (e.g. plasma, CSF) and may
It is worth noting that a novel form of oxytocin has been documented in contribute to divergent results with extracted and unextracted samples
some species of New World monkeys; this form of oxytocin has a sub­ in some studies.
stitution at residue 8, with proline replacing the ancestral leucine (Lee Some oddities of urinary oxytocin measurement have been reported
et al., 2011). Given our epitope mapping results, we would expect previously. Schaebs et al. (2019) performed a validation of dog and wolf
binding for this form of oxytocin to be significantly reduced. Interest­ samples on the Enzo kit, but found that despite good parallelism, the
ingly, although the identity of each residue in the rest of oxytocin (1− 6) spike recoveries and extraction efficiencies were concerningly high.
is less important for antibody binding, cross-reactivity drops rapidly for Reyes et al. (2014) reported that 18% of their samples produced
smaller fragments. out-of-range high measurements on the Enzo kit even after solid-phase
This demonstrated cross-reactivity is important for several reasons. extraction. Interestingly, when assayed at a two-fold dilution, not only
First, measurements of oxytocin by immunoassay likely include some were the samples within range, but the corrected oxytocin concentra­
amount of signal from these bioactive fragments; this interpretation is tions were also within range, suggesting some form of interference when
supported by the observation that injection of radiolabelled oxytocin (in assayed at higher concentrations. Reyes et al. (2014) hypothesize that
marmosets) results in urinary samples that produce multiple radioactive this may be due to dehydration and overly concentrated samples, sup­
fractions after HPLC, including ones other than the standard (Seltzer and ported by the finding that out-of-range samples had higher creatinine
Ziegler, 2007). Second, although the cross-reactivity for MIF-1 seems values. Higher creatinine was also associated with higher vasopressin
low, it is a very small fragment, with a sequence that is common in a measurements, higher specific gravity, and more acidic samples. Our
variety of peptides. This cross-reactivity may partially explain why work suggests that either high concentrations of metabolites or the
immunoassay of urinary samples is more difficult than for plasma altered pH of dehydrated samples could explain this result, and the MCX
samples, given that urine is a waste product, full of thousands of known method—with the pH of buffers optimized for both retaining oxytocin
and unknown metabolites (Bouatra et al., 2013). A BLASTP (Altschul and minimizing interference—improves assay performance
et al., 1990) search of the NCBI Protein Reference Sequences database considerably.
(O’Leary et al., 2016), restricted to Homo sapiens, on oxytocin 4–9 re­ Lastly, we would like to caution that this and our previous work
veals 678 unique sequences with 100% identity (i.e. containing (Gnanadesikan et al., 2021) emphasize the need for detailed reporting of
QNCPLG), and many more with partial identity. This result suggests that immunoassay methods, not only in validation papers, but also accom­
a large number of metabolites might have sufficient sequence similarity panying any empirical result. This reporting should include the details of
to cross-react with the Arbor Assays anti-oxytocin antibody, although extraction methods and immunoassay kits, sample and buffer volumes,
we do not know which specific fragments are present in urine or what dilution factors, and other aspects of the sample treatment and pro­
metabolic processes produce them. Importantly, oxytocin has been cessing protocols. Attention to these details will aid in both interpreta­
shown to be one of the least abundant—while still detecta­ tion and replication of reported results.
ble—metabolites in human urine, while other metabolites can be more
abundant by up to 10 orders of magnitude (Bouatra et al., 2013). Thus, 5. Conclusions
even peptide fragments with low cross-reactivities could interfere when
present in high enough concentrations. Although this presents Our immunoassay results indicate that many existing methods of

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

measuring urinary oxytocin may be susceptible to interference. Our Health under award number R21HD095217, by the National Institute of
epitope mapping and fragment cross-reactivity results suggest this may Mental Health under award number MH114994, and by the National
be attributable to the abundance of metabolites in urine. Nevertheless, Science Foundation Graduate Research Fellowship Program under grant
we show that a more selective extraction method, utilizing a mixed- number DGE-1 746 060. Any opinions, findings, and conclusions or
cation exchange (MCX) chemistry, is successful at minimizing this recommendations expressed in this material are those of the authors and
interference. This method has the potential to improve urinary oxytocin do not necessarily reflect the views of the National Institutes of Health or
studies in a wide variety of species and to open doors for reconsidering the National Science Foundation.
the relationships between urinary oxytocin and other physiological and
behavioral measures. Appendix A. Supporting information

Funding Supplementary data associated with this article can be found in the
online version at doi:10.1016/j.psyneuen.2022.105827.
This study was funded in part by the National Institutes of Health
(R21HD095217 to E.L.M. and S.R.T. and MH114994 to E.A.D.H.). Sifaka References
sample collection was funded by a University of Texas at Austin
Research Grant and the National Science Foundation (BES-1 719 654) to Algoe, S.B., Kurtz, L.E., Grewen, K., 2017. Oxytocin and social bonds: the role of oxytocin
in perceptions of romantic partners’ bonding behavior. Psychol. Sci. 28 (12),
R.J.L. Sifaka sample analysis was funded by a National Science Foun­ 1763–1772. https://doi.org/10.1177/0956797617716922.
dation grant (BES-1 719 655) to S.R.T. G.E.G. was additionally funded Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J., 1990. Basic local
by the National Science Foundation Graduate Research Fellowship alignment search tool. J. Mol. Biol. 215 (3), 403–410. https://doi.org/10.1016/
S0022-2836(05)80360-2.
Program (DGE-1 746 060), the P.E.O. Scholar Award, and FRISCO. The Amico, J.A., Ulbrecht, J.S., Robinson, A.G., 1987. Clearance studies of oxytocin in
Laboratory for the Evolutionary Endocrinology of Primates (LEEP) was humans using radioimmunoassay measurements of the hormone in plasma and
funded by the University of Arizona College of Social and Behavioral urine. J. Clin. Endocrinol. Metab. 64 (2), 340–345. https://doi.org/10.1210/jcem-
64-2-340.
Sciences, School of Anthropology, Institute for the Environment, Pro­
Andreasson, U., Perret-Liaudet, A., van Waalwijk van Doorn, L.J., Blennow, K.,
vost’s Office, and Bio-5 Institute. Arbor Assays partially funded the Chiasserini, D., Engelborghs, S., Teunissen, C.E., 2015. A practical guide to
epitope mapping analyses. immunoassay method validation. Front. Neurol. 6 (Aug), 1–8. https://doi.org/
10.3389/fneur.2015.00179.
Bales, K.L., van Westerhuyzen, J.A., Lewis-Reese, A.D., Grotte, N.D., Lanter, J.A.,
CRediT authorship contribution statement Carter, C.S., 2007. Oxytocin has dose-dependent developmental effects on pair-
bonding and alloparental care in female prairie voles. Horm. Behav. 52 (2),
Gitanjali E. Gnanadesikan: Conceptualization, Data curation, 274–279. https://doi.org/10.1016/j.yhbeh.2007.05.004.
Bick, J., Dozier, M., 2010. Mothers’ concentrations of oxytocin following close, physical
Formal analysis, Funding acquisition, Investigation, Methodology, interactions with biological and nonbiological children. Dev. Psychobiol. 52 (1),
Visualization, Writing – original draft. Elizabeth A.D. Hammock: 100–107. https://doi.org/10.1002/dev.20411.
Conceptualization, Resources (mouse samples), Writing – original draft. Bienboire-Frosini, C., Chabaud, C., Cozzi, A., Codecasa, E., Pageat, P., 2017. Validation
of a commercially available enzyme immunoassay for the determination of oxytocin
Stacey R. Tecot: Conceptualization, Funding acquisition, Resources in plasma samples from seven domestic animal species. Front. Neurosci. 11 (SEP)
(laboratory, sifaka samples), Project administration, Writing – original https://doi.org/10.3389/fnins.2017.00524.
draft. Rebecca J. Lewis: Resources (laboratory, sifaka samples), Writing Bouatra, S., Aziat, F., Mandal, R., Guo, A.C., Wilson, M.R., Knox, C., Wishart, D.S., 2013.
The human urine metabolome (sep). PLoS ONE 8 (9), e73076. https://doi.org/
– review & editing. Russ Hart: Conceptualization, Funding acquisition, 10.1371/journal.pone.0073076.
Methodology, Writing – review & editing. C. Sue Carter: Conceptuali­ Brandtzaeg, O.K., Johnsen, E., Roberg-Larsen, H., Seip, K.F., MacLean, E.L., Gesquiere, L.
zation, Writing – review & editing. Evan L. MacLean: Conceptualiza­ R., Wilson, S.R., 2016. Proteomics tools reveal startlingly high amounts of oxytocin
in plasma and serum. Sci. Rep. 6 (0317), 1–7. https://doi.org/10.1038/srep31693.
tion, Funding acquisition, Methodology, Project administration, Cardoso, C., Ellenbogen, M.A., Orlando, M.A., Bacon, S.L., Joober, R., 2013. Intranasal
Resources (laboratory equipment), Writing – original draft. oxytocin attenuates the cortisol response to physical stress: a dose-response study.
Psychoneuroendocrinology 38 (3), 399–407. https://doi.org/10.1016/j.
psyneuen.2012.07.013.
Declaration of Competing Interest
Carter, C.S., 2014. Oxytocin pathways and the evolution of human behavior. Annu. Rev.
Psychol. 65 (1), 17–39. https://doi.org/10.1146/annurev-psych-010213-115110.
R.H. founded Arbor Assays and is a board member. The authors Carter, C.S., Courtney Devries, A., Getz, L.L., 1995. Physiological substrates of
mammalian monogamy: The prairie vole model. Neurosci. Biobehav. Rev. 19 (2),
declare no other competing interests.
303–314. https://doi.org/10.1016/0149-7634(94)00070-H.
Crockford, C., Wittig, R.M., Langergraber, K., Ziegler, T.E., Zuberbuhler, K., Deschner, T.,
Acknowledgements 2013. Urinary oxytocin and social bonding in related and unrelated wild
chimpanzees. Proc. R. Soc. B: Biol. Sci. 280 (1755), 1–8. https://doi.org/10.1098/
rspb.2012.2765.
We would like to acknowledge the Pepscan team for their work on Crockford, C., Deschner, T., Ziegler, T.E., Wittig, R.M., 2014. Endogenous peripheral
the epitope mapping, as well as Arbor Assays for their contributions. In oxytocin measures can give insight into the dynamics of social relationships: a
particular, Bobbi O’Hara at Arbor Assays was extremely helpful at review. Front. Behav. Neurosci. 8 (March), 1–14. https://doi.org/10.3389/
fnbeh.2014.00068.
multiple stages of this work. We also benefited significantly from con­ Feldman, R., 2012. Oxytocin and social affiliation in humans. Horm. Behav. 61 (3),
versations with Terence Gorman at Waters, who advised us throughout 380–391. https://doi.org/10.1016/j.yhbeh.2012.01.008.
the process of developing and optimizing the MCX method. Sifaka Feldman, R., Gordon, I., Zagoory-Sharon, O., 2011. Maternal and paternal plasma,
salivary, and urinary oxytocin and parent-infant synchrony: considering stress and
samples were collected and transported with funding by the National affiliation components of human bonding. Dev. Sci. 14 (4), 752–761. https://doi.
Science Foundation under award number BES-1 719 654, with the org/10.1111/j.1467-7687.2010.01021.x.
permission of CAFF/CORE, the Ministry of Water and Forests, and Finkenwirth, C., van Schaik, C., Ziegler, T.E., Burkart, J.M., 2015. Strongly bonded
family members in common marmosets show synchronized fluctuations in oxytocin.
Madagascar National Parks, and the assistance of the University of Physiol. Behav. 151, 246–251. https://doi.org/10.1016/j.physbeh.2015.07.034.
Antananarivo, MICET, the Ankoatsifaka Research Station, the Sifaka Finkenwirth, C., Martins, E., Deschner, T., Burkart, J.M., 2016. Oxytocin is associated
Research Project staff, and Meredith Lutz, Diary Razafimandimby, with infant-care behavior and motivation in cooperatively breeding marmoset
monkeys. Horm. Behav. 80, 10–18. https://doi.org/10.1016/j.yhbeh.2016.01.008.
Fanomezantsoa Razafimalala, Mc Antonin Andriamahaihavana, Sylvia
Francis, S.M., Kirkpatrick, M.G., de Wit, H., Jacob, S., 2016. Urinary and plasma oxytocin
Rahobilalaina, Hannah Carbonneau, and Allison Hays. We would also changes in response to MDMA or intranasal oxytocin administration.
like to acknowledge the positive and constructive comments of two Psychoneuroendocrinology 74, 92–100. https://doi.org/10.1016/j.
anonymous reviewers in improving this paper. This material is based psyneuen.2016.08.011.
Froemke, R.C., Young, L.J., 2021. Oxytocin, neural plasticity, and social behavior. Annu.
upon work supported by the Eunice Kennedy Shriver National Institute Rev. Neurosci. 44, 359–381. https://doi.org/10.1146/annurev-neuro-102320-
of Child Health & Human Development of the National Institutes of 102847.

10

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
G.E. Gnanadesikan et al. Psychoneuroendocrinology 143 (2022) 105827

Geysen, H.M., Meloen, R.H., Barteling, S.J., 1984. Use of peptide synthesis to probe viral Peters, S., Slattery, D.A., Uschold-Schmidt, N., Reber, S.O., Neumann, I.D., 2014. Dose-
antigens for epitopes to a resolution of a single amino acid. Proc. Natl. Acad. Sci. dependent effects of chronic central infusion of oxytocin on anxiety, oxytocin
USA 81 (13 I), 3998–4002. https://doi.org/10.1073/pnas.81.13.3998. receptor binding and stress-related parameters in mice. Psychoneuroendocrinology
Gnanadesikan, G.E., Hammock, E.A., Tecot, S.R., Carter, C.S., MacLean, E.L., 2021. 42, 225–236. https://doi.org/10.1016/j.psyneuen.2014.01.021.
Specificity of plasma oxytocin immunoassays: a comparison of commercial assays R Core Team. (2021). R: A Language and Environment for Statistical Computing
and sample preparation techniques using oxytocin knockout and wildtype mice. [Computer software manual]. Vienna, Austria. Retrieved from https://www.r-pro
Psychoneuroendocrinology 132 (July), 105368. https://doi.org/10.1016/j. ject.org/.
psyneuen.2021.105368. Reyes, T.L., Galinsky, A.M., Hoffmann, J.N., You, H.M., Ziegler, T.E., McClintock, M.K.,
Goldberger, B.A., Loewenthal, B., Darwin, W.D., Cone, E.J., 1995. Intrasubject variation 2014. Social peptides: Measuring urinary oxytocin and vasopressin in a home field
of creatinine and specific gravity measurements in consecutive urine specimens of study of older adults at risk for dehydration. J. Gerontol. - Ser. B Psychol. Sci. Soc.
heroin users. Clin. Chem. 41 (1), 116–117. https://doi.org/10.1093/clinchem/ Sci. 69, S229–S237. https://doi.org/10.1093/geronb/gbu104.
41.1.116. Saito, A., Hamada, H., Kikusui, T., Mogi, K., Nagasawa, M., Mitsui, S., Hiraki, K., 2014.
Lee, A.G., Cool, D.R., Grunwald, W.C., Neal, D.E., Buckmaster, C.L., Cheng, M.Y., Urinary oxytocin positively correlates with performance in facial visual search in
Parker, K.J., 2011. A novel form of oxytocin in new world monkeys. Biol. Lett. 7 (4), unmarried males, without specific reaction to infant face. Front. Neurosci. 8 (8 JUL),
584–587. https://doi.org/10.1098/rsbl.2011.0107. 1–7. https://doi.org/10.3389/fnins.2014.00217.
Lefevre, A., Mottolese, R., Dirheimer, M., Mottolese, C., Duhamel, J.R., Sirigu, A., 2017. Samuni, L., Preis, A., Mundry, R., Deschner, T., Crockford, C., Wittig, R.M., 2017.
A comparison of methods to measure central and peripheral oxytocin concentrations Oxytocin reactivity during intergroup conflict in wild chimpanzees. Proc. Natl. Acad.
in human and non-human primates. Sci. Rep. 7 (1), 1–10. https://doi.org/10.1038/ Sci. 114 (2), 268–273. https://doi.org/10.1073/pnas.1616812114.
s41598-017-17674-7. Schaebs, F.S., Marshall-Pescini, S., Range, F., Deschner, T., 2019. Analytical validation of
MacLean, E.L., Wilson, S.R., Martin, W.L., Davis, J.M., Nazarloo, H.P., Carter, C.S., 2019. an Enzyme Immunoassay for the measurement of urinary oxytocin in dogs and
Challenges for measuring oxytocin: the blind men and the elephant? wolves. Gen. Comp. Endocrinol. 281 (May), 73–82. https://doi.org/10.1016/j.
Psychoneuroendocrinology 107 (May), 225–231. https://doi.org/10.1016/j. ygcen.2019.05.015.
psyneuen.2019.05.018. Seltzer, L.J., Ziegler, T.E., 2007. Non-invasive measurement of small peptides in the
Marshall-Pescini, S., Schaebs, F.S., Gaugg, A., Meinert, A., Deschner, T., Range, F., 2019. common marmoset (Callithrix jacchus): a radiolabeled clearance study and
The role of oxytocin in the dog–owner relationship. Animals 9 (10), 1–19. https:// endogenous excretion under varying social conditions. Horm. Behav. 51 (3),
doi.org/10.3390/ani9100792. 436–442. https://doi.org/10.1016/j.yhbeh.2006.12.012.
McCullough, M.E., Churchland, P.S., Mendez, A.J., 2013. Problems with measuring Seltzer, L.J., Ziegler, T.E., Pollak, S.D., 2010. Social vocalizations can release oxytocin in
peripheral oxytocin: Can the data on oxytocin and human behavior be trusted? humans. Proc. R. Soc. B: Biol. Sci. 277 (1694), 2661–2666. https://doi.org/10.1098/
Neurosci. Biobehav. Rev. 37 (8), 1485–1492. https://doi.org/10.1016/j. rspb.2010.0567.
neubiorev.2013.04.018. Snowdon, C.T., Pieper, B.A., Boe, C.Y., Cronin, K.A., Kurian, A.V., Ziegler, T.E., 2010.
Miller, R.C., Brindle, E., Holman, D.J., Shofer, J., Klein, N.A., Soules, M.R., O’Connor, K. Variation in oxytocin is related to variation in affiliative behavior in monogamous,
A., 2004. Comparison of specific gravity and creatinine for normalizing urinary pairbonded tamarins. Horm. Behav. 58 (4), 614–618. https://doi.org/10.1016/j.
reproductive hormone concentrations. Clin. Chem. 50 (5), 924–932. https://doi.org/ yhbeh.2010.06.014.
10.1373/clinchem.2004.032292. Strier, K.B., Ziegler, T.E., 2005. Advances in field-based studies of primate behavioral
Mitsui, S., Yamamoto, M., Nagasawa, M., Mogi, K., Kikusui, T., Ohtani, N., Ohta, M., endocrinology. Am. J. Primatol. 67 (1), 1–4. https://doi.org/10.1002/ajp.20165.
2011. Urinary oxytocin as a noninvasive biomarker of positive emotion in dogs. Szeto, A., Ph, D., Mccabe, P.M., Ph, D., Nation, D.A., Ph, D., Ph, D., 2012. Evaluation of
Horm. Behav. 60 (3), 239–243. https://doi.org/10.1016/j.yhbeh.2011.05.012. enzyme immunoassay and radioimmunoassay methods for the measurement of
Moscovice, L.R., Ziegler, T.E., 2012. Peripheral oxytocin in female baboons relates to plasma oxytocin. Psychosom. Med. 73 (5), 393–400. https://doi.org/10.1097/
estrous state and maintenance of sexual consortships. Horm. Behav. 62 (5), 592–597. PSY.0b013e31821df0c2.
https://doi.org/10.1016/j.yhbeh.2012.08.011. Uvnäs Moberg, K., Handlin, L., Kendall-Tackett, K., Petersson, M., 2019. Oxytocin is a
Moscovice, L.R., Surbeck, M., Fruth, B., Hohmann, G., Jaeggi, A.V., Deschner, T., 2019. principal hormone that exerts part of its effects by active fragments. Med.
The cooperative sex: Sexual interactions among female bonobos are linked to Hypotheses 133 (August), 109394. https://doi.org/10.1016/j.mehy.2019.109394.
increases in oxytocin, proximity and coalitions. Horm. Behav. 116 (August), 104581 Van Leengoed, E., Kerker, E., Swanson, H.H., 1987. Inhibition of post-partum maternal
https://doi.org/10.1016/j.yhbeh.2019.104581. behaviour in the rat by injecting an oxytocin antagonist into the cerebral ventricles.
Nagasawa, M., Kikusui, T., Onaka, T., Ohta, M., 2009. Dog’s gaze at its owner increases J. Endocrinol. 112 (2), 275–282. https://doi.org/10.1677/joe.0.1120275.
owner’s urinary oxytocin during social interaction. Horm. Behav. 55 (3), 434–441. Weber, A., Harrison, T.M., Sinnott, L., Shoben, A., Steward, D., 2017. Plasma and urinary
https://doi.org/10.1016/j.yhbeh.2008.12.002. oxytocin trajectories in extremely premature infants during NICU hospitalization.
Nagasawa, M., Mitsui, S., En, S., Ohtani, N., Ohta, M., Sakuma, Y., Kikusui, T., 2015. Biol. Res. Nurs. 19 (5), 549–558. https://doi.org/10.1177/1099800417718266.
Oxytocin-gaze positive loop and the coevolution of human-dog bonds. Science 348 Wirobski, G., Range, F., Schaebs, F.S., Palme, R., Deschner, T., Marshall-Pescini, S., 2021.
(6232), 333–336. https://doi.org/10.1126/science.1261022. Life experience rather than domestication accounts for dogs’ increased oxytocin
Nishimori, K., Young, L.J., Guo, Q., Wang, Z., Insel, T.R., Matzuk, M.M., 1996. Oxytocin release during social contact with humans (dec). Sci. Rep. 11 (1), 14423. https://doi.
is required for nursing but is not essential for parturition or reproductive behavior. org/10.1038/s41598-021-93922-1.
Proc. Natl. Acad. Sci. USA 93 (21), 11699–11704. https://doi.org/10.1073/ Wirobski, G., Schaebs, F.S., Range, F., Marshall-Pescini, S., Deschner, T., 2021. Analytical
pnas.93.21.11699. and physiological validation of an enzyme immunoassay to measure oxytocin in dog,
O’Leary, N.A., Wright, M.W., Brister, J.R., Ciufo, S., Haddad, D., McVeigh, R., Pruitt, K. wolf, and human urine samples. Sci. Rep. 11 (1), 1–11. https://doi.org/10.1038/
D., 2016. Reference sequence (RefSeq) database at NCBI: current status, taxonomic s41598-021-92356-z.
expansion, and functional annotation. Nucleic Acids Res. 44 (D1), D733–D745. Wittig, R.M., Crockford, C., Deschner, T., Langergraber, K.E., Ziegler, T.E.,
https://doi.org/10.1093/nar/ gkv1189. Zuberbu¨hler, K., 2014. Food sharing is linked to urinary oxytocin levels and bonding
Palme, R., Touma, C., Arias, N., Dominchin, M.F., Lepschy, M., 2013. Steroid extraction: in related and unrelated wild chimpanzees. Proc. R. Soc. B Biol. Sci. 281 (1778),
Get the best out of faecal samples. Wien. Tiera¨rztl. Mon. 100 (9–10), 238–246. 1–10. https://doi.org/10.1098/rspb.2013.3096.
Pederson, C.A., Prange, A.J., 1979. Induction of maternal behavior in virgin rats after
intracerebroventricular administration of oxytocin. Proc. Natl. Acad. Sci. USA 76
(12), 6661–6665. https://doi.org/10.1073/pnas.76.12.6661.

11

Downloaded for Fakultas Kedokteran Universitas Hasanuddin ([email protected]) at Hasanuddin University from ClinicalKey.com
by Elsevier on February 20, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.