Table 6 Ž
. Cytokine production mean pgrml1 s.d.
Baseline 12 months
a,b,c,d,e
IFN-g Single
18.012.9 28.216.7
Pair 25.213.4
32.519.0 Group
190.67.5 75.329.0
a,b
IL-2 Single
20.08.67 4.746.86
Pair 22.45.78
2.694.82 Group
25.23.86 8.755.37
f
IL-4 Single
0.781.47 1.402.37
Pair 0.420.97
1.592.11 Group
0.230.79 2.242.13
a,b,c,d
IL-10 Single
20.55.68 10.76.02
Pair 19.65.25
12.16.43 Group
40.56.40 10.03.72
a
P F 0.05 for comparison across housing conditions.
b
P F 0.05 for comparison across timepoints.
c
P F 0.05 for housing condition=timepoint interaction.
d
P F 0.05 for planned comparison at baseline timepoint only.
e
P F 0.05 for planned comparison at 12 month timepoint only.
f
Results are reported, but the values are below the accurate detection limits of the test kit.
x w Ž
. x
P ’s F 0.001 and significant interactions for IFN-g F 2,30 s 31.89, P F 0.001 and w Ž
. x Ž
. IL-10 F 2,30 s 22.11, P F 0.001
Table 6 .
4. Discussion
Numerous statistically significant differences in cell-mediated immune responses were identified across the three housing conditions, supporting our initial hypothesis that
housing condition can affect a variety of in vitro assays of cell-mediated immune responses. Specifically, the ratios of CD4
q
to CD8
q
lymphocytes were significantly lower for singly housed than for socially housed subjects after the 4-month timepoint of
the study. Additionally, pathogen-induced lymphocyte proliferation, proliferation re- sponses to some mitogens, NK activity, and production of three of four cytokines, all
differed across housing conditions. Clearly, the cell-mediated immunological responses of singly housed, pair housed, and group housed adult rhesus monkeys are not identical.
Differences in responses existed after 4, 8, and 12 months of the study. The presence of statistically significant differences at the 12-month timepoint suggests that acclimation
to experimental housing conditions may be slow to develop, taking considerably longer than previously suspected.
In general, our lymphocyte subset data are in agreement with previous reports in which lower CD4
q
rCD8
q
were observed for subjects under conditions that involved w
stressful manipulations to their psychosocial environments mother–infant separations Ž
. Coe et al., 1988, 1989; Laudenslager et al., 1990; Terao et al., 1995 , enforced social
Ž .
instability Kaplan et al., 1991; Cohen et al., 1992; Line et al., 1996 , and separation Ž
from andror reunion with their social group Gust et al., 1991, 1992, 1993; Gordon et .
q q
al., 1992 . Additionally, the ratio of CD4 to CD8
lymphocytes has been identified as a potentially ‘‘trait-like’’ characteristic of laboratory monkeys, varying consistently
across a variety of situations and timepoints, and thus may be a useful measure of Ž
. biobehavioral organization Capitanio et al., 1998a; Lilly et al., 1999 . Our data support
these findings; CD4
q
rCD8
q
for individual monkeys were significantly positively Ž
. correlated across housing conditions and timepoints. Like Capitanio et al. 1998a , but
Ž .
unlike Lilly et al. 1999 , we found no consistency across situations or timepoints for CD4
q
counts. Differences observed in pathogen-specific lymphocyte proliferation responses were of
great interest to us because they represent a specific response to an infectious agent and may be more representative of in vivo immune function than responses to nonspecific
Ž .
mitogens LPS, PWM, etc. . We performed T-cell proliferation assays to S. flexneri, C. jejuni, S. typhimurium, and E. coli; pathogens that can cause severe diarrhea and a
Ž variety of related health problems in captive macaques Holmberg et al., 1982; Hird et
. al., 1984; Reinhardt, 1993; Wolfensohn, 1998 . These organisms were chosen for
specific reasons. First, they are common, but treatable, in many captive primate colonies, thereby representing a more explicit ‘‘ecologically’’ and immunologically
relevant challenge than PHA, ConA, or some of the other commonly used mitogens. Also, while debilitating and potentially life threatening, infection with these organisms is
considerably less devastating than infection with experimental strains of SIV, many of
Ž which have been chosen for their virulence characteristics Benveniste et al., 1996; Joag
. et al., 1998 . This again enhances the ecological and immunological suitability of the
gastrointestinal pathogens as a set of appropriate in vitro experimental challenges. Unlike SIV, most monkeys are likely to have been exposed to and survived infectious
episodes of Shigella, Campylobacter, Salmonella, andror E. coli. Thus, given their severe, but not deadly disease course, these organisms could serve as potential surviv-
able in vivo challenges as well.
Ž . In general, higher Stimulation Index SI values on lymphocyte proliferation assays
are indicative of better immune response. In most comparisons, the highest SI values for all four gastrointestinal pathogens were observed among pair housed subjects, although
the difference between pair housed, and single and group housed subjects combined, was only significant at the 12-month timepoint. Since in vitro proliferation to a given
antigen is a measure of a subject’s specific immunological memory, these data suggest that pair housed monkeys may be better equipped to deal with such infectious agents.
This supposition is supported to some extent by previous findings from our laboratory in which pair housed monkeys had significantly less diarrhea that required treatment, than
Ž .
did either single or group housed monkeys Schapiro and Bushong, 1994 . These are
among the first data from nonexperimentally infected primates that may address the Ž
relationship between responses on in vitro immunological assays proliferation re-
. sponses and in vivo measures of health. Definition of this relationship is one of the key
Ž challenges currently facing the field of primate psychoneuroimmunology Lubach et al.,
. 1995; Laudenslager et al., 1996 .
It is our contention that strong social relationships, particularly the affiliative interac- tions that characterize our pair housed monkeys, may diminish the likelihood of severe
infection with potentially diarrhea-inducing agents. The findings of Capitanio et al. Ž
. 1998b are extremely relevant to this contention. Half of their subjects spent regular
periods interacting in stable groups, while the other half of the subjects spent regular periods in unstable groups. Monkeys in stable groups affiliated more, and more
Ž importantly, survived significantly longer an increase of over 40 in median survival
. Ž
. times , after SIV infection than monkeys in unstable groups Capitanio et al., 1998b . If
stable social groups and high levels of affiliative behavior can increase a monkey’s ability to survive after SIV infection, then it is relatively unsurprising that affiliative
interactions may also lessen the severity of the effects of infection with gastrointestinal pathogens. Just as allowing SIV-infected subjects the opportunity to regularly interact in
stable social groups may enhance experimental validity for SIV studies, social housing may enhance validity for other types of biomedical studies. Thus, from applied
management and research perspectives, pair housing may be an extremely effective strategy to increase levels of social affiliation while simultaneously minimizing
diarrhea-related problems.
While we did find significant differences in proliferation responses to StaphA across Ž
housing conditions the LPS and PWM results were statistically, but not biologically .
significant , our proliferation data, in general, are not in agreement with previous reports. Whereas many have found diminished proliferation responses under conditions
Ž that involved stressful manipulations to subjects’ psychosocial environments Kaplan et
. al., 1991; Boccia et al., 1992; Cohen et al., 1992; Lubach et al., 1995 , we have
repeatedly found no decrements in proliferation responses among animals that were Ž
experiencing psychosocial environments that should have been more stressful Schapiro .
et al., 1998 . This includes marginally higher proliferation responses in unenriched than in enriched rhesus monkeys, significantly higher responses in middle-ranking than in
highest-ranking female rhesus, and the present finding of marginally higher responses in singly housed monkeys. As mentioned previously, we have chosen to emphasize
pathogen-specific immune memory responses because we suspect that proliferation responses to actual disease-inducing agents are more sensitive to subtle changes in
psychosocial conditions than are responses to nonspecific mitogens, and thus, are better indicators of an animal’s capacity to immunologically respond to such ‘‘ecologically
relevant’’ changes.
Our natural killer cell activity data do not agree with previous reports. Diminished NK activity usually prevails under conditions that involve stressful manipulations to
Ž subjects’ psychosocial environments Boccia et al., 1992; Lubach et al., 1995; Terao et
. al., 1995 . We found the lowest levels of NK activity among subjects housed in stable
social groups. This might be understandable if all group housed subjects were the lowest ranking in their groups, but group housed subjects included monkeys of all ranks.
Alternatively, some singly caged subjects may have been more comfortable living alone. However, this is counterintuitive and is not supported by the observed decreases in
CD4
q
rCD8
q
. At the moment, we have no good explanation for either the lower NK activity in group housed monkeys or the relatively low levels of NK activity in our
monkeys overall. Ž
It has been suggested that variability in disease progression in HIV and SIV; Clerici .
and Shearer, 1993; Benveniste et al., 1996; Capitanio et al., 1998b is associated with Ž
. Ž
. shifts in T helper cell TH1 and T suppressor cell TH2 responses. The timing of
acceleration of disease progression in SIV and HIV appears to correlate with a shift Ž
. from TH1 to TH2 responses Benveniste et al., 1996 . A primary motivation for us to
collect and analyze cytokine production data was the prospect that these measures would help us address this shift in T cell responses, and thereby clarify the relationship
between functional immune assays and animal health. Production of IL-2 and IFN-g Ž
. among other cytokines is typically associated with TH1 responses and production of
Ž .
IL-4 and IL-10 among other cytokines is typically associated with TH2 responses. Ž
Although our data reveal no evidence of a TH1 to TH2 shift one really would not be .
expected, given that no disease challenge was involved , we still found significant differences across housing conditions. Cytokine production is an important step in the
immunological response to challenge, and the fact that social housing condition can affect the production of these signaling compounds is additional evidence establishing
the relationship between psychosocial factors and immune responses.
During this longitudinal study, repeated immunological samples were collected at baseline and at timepoints 4, 8, and 12 months into the study. A number of the
immunological responses we measured differed significantly across timepoints, includ- ing E. coli, Salmonella, Campylobacter, PHA, ConA, PWM, and StaphA proliferation
responses; NK activity; and IL-2, IL-10, and IFN-g production. While timepoint related effects are mildly interesting, the existence of differences across housing conditions at
late timepoints, particularly at 12 months, would be of even greater significance. This would be especially true for measures on which there were no housing condition effects
at baseline, implying that acclimation to new housing conditions is a process that takes a considerable period of time for laboratory macaques. By acclimation, we are essentially
referring to a return to baseline response levels. In fact, our data reveal that for 6 of 10 measures on which groups did not differ at baseline, they did differ at 12 months.
Interestingly, for three of four measures on which they differed at baseline, significant differences no longer existed at 12 months. Such results suggest that after a year of
Ž . single housing, monkeys are a not identical to group housed or pair housed monkeys,
Ž . and to a lesser extent, they are b not identical to what they were when they were
Ž .
socially normally housed. It is clear that the relatively simple act of changing a monkey’s social housing
condition affects its immunological responses for long periods. This is of great conse- quence when immunological studies, particularly those involving infection andror
vaccination, are begun after short acclimation periods. The combination of the immuno- logical effects of single housing in addition to experimental infection may be more than
the monkey or a vaccine can accommodate. The clearest example in our data set is the CD4
q
rCD8
q
data. Single, pair, and group housed monkeys CD4
q
rCD8
q
did not
Ž .
differ significantly at the start of the study 0.95, 0.94, and 1.13, respectively , when Ž
. most animals 30 of 36 were actually group housed. Twelve months later, when some
monkeys had been singly or pair housed for a year, CD4
q
rCD8
q
for group housed Ž
. animals had not changed 1.15 , but the ratios for pair housed monkeys had decreased to
0.83, and those for singly caged subjects had fallen more precipitously to 0.68. The StaphA data demonstrate a similar effect, with group housed animals maintaining
Ž .
Ž consistent responses at baseline and 12 months 35.8 and 35.4 , while both single 61.3
. Ž
. to 110 and pair housed monkeys 57.4 to 120 showed dramatic increases in prolifera-
Ž .
tion responses. Lilly et al. 1999 have also found no apparent acclimation on a variety of immunological measures up to 28 weeks after a change in housing condition.
Housing captive macaques in different social conditions results in significant differ- ences in many aspects of their immune profiles. Thus, when studying macaque immune
responses, one must be aware of the effects that housing condition, as one aspect of the psychosocial environment, has on the animals, independent of the experimental manipu-
lation. There are similar influences which also should be accounted for, including
Ž rearing and housing history Laudenslager et al., 1985, 1990; Coe et al., 1989, 1992;
. Ž
Lilly et al., 1999 , dominance rank Boccia et al., 1992; Clarke et al., 1996; Schapiro et .
Ž al., 1998 , social instability Kaplan et al., 1991; Cohen et al., 1992; Line et al., 1996;
. Ž
Capitanio et al., 1998b , pre-sampling exposure to technician activity Capitanio et al., .
Ž .
1996 , and even enrichment history Schapiro et al., 1998 . Therefore, there is consider- able impetus to effectively manage and even to attempt to optimize the psychosocial
environment encountered by nonhuman primate subjects in biomedical research. One way to eliminate undesirable effects of uncontrolled psychosocial influences would be to
apply findings such as those cited above to establish specific criteria for the selection, treatment, and management of primate subjects in the research laboratory.
We have demonstrated that immunological data gathered from singly housed adult rhesus monkeys are not identical to data gathered from pair housed or group housed
adult rhesus monkeys. This generalization applies even after a year spent in the target housing condition, suggesting that long acclimation periods may be necessary for
Ž .
monkeys to immunologically adapt to single caging if they ever do; Lilly et al., 1999 . While we were hoping that our data would clearly identify one type of housing
condition that was superior on all immunological indices, our results do not allow such an interpretation at this time. Measuring a large number of immunological responses is
advantageous when attempting to get a complete picture of an animal’s immunological functioning, but may be disadvantageous when attempting to interpret the large number
of results.
5. Conclusions