This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
6 Neuro-Immune AssociativeLearning
M.-B. Niemi . G. Pacheco‐Lopez . H. Engler . C. Riether . R. Doenlen . M. Schedlowski
The first scientifical reports of classically conditioned immune functions arose from Russian scientific
contemporaries of I. P. Pavlov. Lukyanenko (1961) cites an observation of Makukahin (1911) and a report
by Voronov and Riskin (1925) as possibly the first ones to demonstrate ‘‘conditioned leukocytic reactions.’’
However, according to Luk’yanenko they could not be interpreted properly and were not carried forward.
In the 1920s, Metal’nikov and his colleagues at the Pasteur Institute in Paris, France systematically
documented that an immune response could be elicited without the presence of an antigen but just by
evoking a previous experience, that is, CS exposure. After scratching or heating the skin of guinea pigs with
a warmed metallic plate as CS, the animals received intraperitoneal (i.p.) injections of various bacteria‐derived compounds such as small doses of either Bacillus anthrax, a Staphylococcus filtrate, tapioca
emulsion, or Vibrio cholera as US. After association phase and a delay to allow the return to baseline levels
was completed, the CS alone yielded significant and rapid influx of polymorphonuclear leukocytes into the
peritoneal cavity (Metal’nikov and Chorine, 1926, 1928). As a proof of the biological relevance of the
elicited conditioned response on the immune system, it was reported that conditioned animals survived to
lethal doses of V. cholera or Streptococcus if they were previously behaviorally evoked. Follow‐up experi-
ments replicated and extended these findings (Nicolau and Antinescu‐Dimitriu, 1929; Ostrovskaya, 1929).
Additional evidence occurred largely provided by A. O. Dolin in the Soviet Union and his fellows Krylov,
Flerov, and Luk’yanenko. Both specific and nonspecific immune reactions and both immunosuppression
and immunoenhancement due to conditioning were achieved in mice, rats, guinea pigs, rabbits, dogs, oxen,
monkeys and also in humans (Dolin and Krylov, 1952; Doroshkevich, 1954; Vygodchikov, 1955). At the
same time, but less recognized, reports from Romania showed a conditioned increase in phagocytic activity
of blood polymorphonuclear cells in dogs (Benetato, 1955; Baciu et al., 1965).
In addition, a series of reports published in Switzerland documented conditioned asthma‐likeresponse by using an auditory CS in guinea pigs (Noelpp and Noelpp‐Eschenhagen, 1951a–c, 1952a–c).Such findings were replicated by Ottenberg et al. (1958), and extended to humans (Dekker et al., 1957;
Turnbull, 1962).
In 1975, Ader and Cohen published their seminal work in which they constructed the term behaviorally
conditioned immunosuppression, which retrospectively set the stage for the field of psychoneuroimmunology.
During that time, R. Ader was working on extinction of conditioned taste aversion employing saccharin
taste as CS paired with a drug that induced significant visceral malaise. He realized that some rats of the
conditioned group died during the course of a series of extinctions trials (i.e., CS representation). Moreover,
those animals had received the largest of three different amounts of saccharin and displayed the most
Anti‐OVA antibody titers ↑ Ovalbumin electroacupuncture Rat Huang et al.
(2004)
Antibody response ↓ Keyhole limpet
hyacinine
Chocolate milk Mouse Ader et al. (1993)
Secretion of mast‐cellprotease II ↑
Egg albumin
antigen
Flashing light and
noise by ventilation
fans
Rat MacQueen et al.
(1989)
Primary mixed lymphocyte
response ↓
Antilymphocyte
serum
SAC Rat Kusnecov et al.
(1983)
Splenocyte proliferation, IL‐2and IFN‐g production, and mRNA
expression and synthesis ↓
Cyclosporine A SAC Rat Exton et al.
(1998, 2000);
Pacheco‐Lopezet al. (2005)
T helper:T suppressor subset
ratio ↑
Levamisole SAC Rat Husband et al.
(1987)
NK‐cell activity, IL‐2, lymphocyte
proliferation ↓
Morphine sulfate Distinctive
environment
Rat Coussons‐Readet al. (1994)
IL‐2, IFN‐g, and corticosterone
plasma levels ↑
Staphylococcal
enterotoxin B
SAC Rat Pacheco‐Lopezet al. (2005)
Nitric oxide ↓ Heroin Distinctive
environment
Rat Szczytkowski and
Lysle (2007)
Corticosterone production ↑ Interleukin‐1 b SAC/LiCl or
peppermint odor
Mouse Dyck et al.
(1990)
128 6 Neuro-immune associative learning
Neuro-immune associative learning 6 129
like cyclophosphamide (CY) and cyclosporine A (CsA) have been successfully employed as US, resulting in
conditioned immunosuppressive effects when associated with an appropriate CS such as taste/flavor (Ader
and Cohen, 1975; Exton et al., 2001).
Morphine sulfate as US paired with a distinctive environment as CS served to achieve a condi-
tioned suppression of NK‐cell activity, IL‐2 production, and lymphocyte proliferation (Coussons‐Readet al., 1994).
3 Conceptual Framework for Neuroimmune Associative Learning
The following section aims to provide a framework which takes basic assumptions about stimuli and sites of
action within neuroimmune associative learning into account. Within classical conditioning theory, a
prerequisite for its occurrence is that the CNS must sense both the CS (changes in the external environ-
ment) and the US (changes in the internal environment), where the signals are then processed, associated,
consolidated and recalled at evocation time (Eikelboom and Stewart, 1982; Bovbjerg, 2003; Pacheco‐Lopezet al., 2006, 2007b). Accordingly, the CNS must initiate both the UR and the CR. Translated to neuroim-
mune associative learning, this implies that only immune changes that are detected by the brain can serve as
US, in turn, only immune changes that are executed by the brain should then be called CR. However, the
nature of the US and the CR in most of the neuroimmune associative learning protocols so far reported is
not evident (> Figure 6-2).
3.1 Association Phase
During association phase, there are two possible USs that may be detected by the brain. A directly perceived
US, which is defined as a stimulus that itself is recognized by the CNS. The second kind is an indirectly
perceived US, which is defined to be signaled by intermediary molecules that are then indeed detected by the
CNS. In that case those molecules are the genuine US and the applied drug is a sham US, or indirectly
perceived US. This has important implications, for example, pairing of a CS with a compound that uncondi-
tionally decreases body temperature as US could result in a conditioned hypothermic response; however,
conditioned hyperthermia is also possible, that is, ‘‘paradoxically conditioned effects’’ or ‘‘counterconditioning
effects’’ (e.g., Bull et al., 1991). The proposed framework predicts such possibilities also for conditioned immune
responses and provides a tentative explanation. As Eikelboom and Stewart (1982) proposed, it is necessary to
correctly determine the nature of the US and UR to predict the direction of the CR.
There are two possible afferent pathways for any US, regardless whether directly or indirectly perceived –
a humoral and a neural pathway (see > Section 1.1). Within the neural pathway, the US information, on
detection, may be translated into neural activity. This sensing process requires immunoceptive capacities of
the CNS (Goehler et al., 2000; Blalock, 2005).
The humoral afferent pathway is required more often for indirectly perceived USs that induce molecules
that reach the brain via the blood stream as well as for USs that are not detected locally by the immune
system. In addition, if an indirectly perceived US affects several cell types, all of the involved molecules
become candidates for serving as a genuine US to be detected by the CNS. This implies a more complex,
longer, and therefore maybe slower signaling process for indirectly perceived US compared with directly
perceived US. Therefore, it can be assumed that the CNSmay take longer to respond to an indirectly perceived
US than to a directly perceived US.
3.2 Recall Phase
The CR represents the ultimate proof that an association has formerly taken place. There are two possible
pathways by which immune functions can be modulated by the CNS: the humoral efferent pathway and the
neural efferent pathway (see > Section 1.1). During recall phase, humoral efferent pathway may affect
Theoretical framework for neuroimmune associative learning. At association, the conditioned stimulus (CS) can
potentially be associated with two possible unconditioned stimuli (US). The US directly detected by the CNS is
defined as a directly perceived US. The stimulus that needs one or more intermediary molecules released
by another system to be detected by the CNS is termed indirectly perceived US. Any US, indirectly or directly
perceived, has two possible afferent pathways to the CNS: a neural afferent pathway and a humoral efferent
pathway. At recall, the CNS canmodulate immune function via these twopathways. Thehumoral efferent pathway
may imply changes in neuroendocrine mediators that directly or indirectly modify the immune response. The
neural efferent pathway is supported by direct innervations of primary and secondary lymphoid organs
130 6 Neuro-immune associative learning
neurohormones that in turn affect immune responses. These peripheral effects are diffuse and long‐lastingas any neuroendocrine responses. Direct innervation of primary and secondary lymphoid organs (Elenkov
et al., 2000; Mignini et al., 2003; Tracey, 2007) may be part of the neural efferent pathway. Since several
immune parameters such as T‐cell differentiation (Sanders and Kohm, 2002a; Sanders and Straub, 2002b),
hematopoiesis (Miyan et al., 1998; Artico et al., 2002), T‐ and B‐cell activity (Downing and Miyan,
2000), NK‐cell activity (Katafuchi et al., 1993; Hori et al., 1995), and inflammatory responses (Czura and
Tracey, 2005; Pavlov and Tracey, 2005) are affected by neural activity, and they may also be subjected to be
affected by neuroimmune associative learning protocols. It may also be the case that besides the analyzed
and reported parameters there are still others not yet identified that are controlled or initiated by the
conditioning procedure. The picture becomes more complex considering that the immune system under-
takes sensitization (memory) and habituation (tolerance) processes, basically independent from the CNS.
In addition, it is known that several immune functions underlie circadian rhythms (Buijs et al., 2006).
These conditions complicate the prediction of the final nature and magnitude of the CR, often requesting
experimental trials. In this regard, some immune parameters modulated at recall may be the bizarre
reflection of neural activity that cannot be explained by orthodox learning and memory rules. For instance,
one such scenario would be that the delay between two recall trials is not long enough for certain immune
functions to return to baseline levels. Such successive recall trials may yield an additive effect in these and
related immune parameters rather than an extinction, which may occur in the neural correlates that elicited
these conditioned immune effects. In summary, immune responses can be affected recalling neuroimmune
engrams, but this does not necessarily imply that such immune responses were behaviorally conditioned.
The following section introduces findings of neuroimmune associative learning experiments and discusses
them with regard to rules of orthodox learning and memory theory that are commonly accepted to apply
when nonimmune USs are employed. This is insofar of interest as the conditioned taste aversion paradigm,
which is employed in the majority of neuroimmune associative learning experiments, has certain unique
features. It has been reported that odor–immune association can be established under long interstimulus
intervals (e.g., up to 24 h) (Hsueh et al., 1992). Taste–visceral associations are weak with interstimulus
delays longer than 4 h (Hiramoto et al., 1992; Solvason et al., 1992). Taste/odor–immune association could
occur after only a single CS–US association, whereas visual/auditory/touch–immune association requires
further reinforcement trials (Domjan, 2005). Therefore, given the CNS–immune communication complex-
ity, it seems worth to take a closer look at rules applying to a given neuroimmune associative learning.
4.1 General Learning Rules of Neuroimmune Associative Learning
Intensive training by increasing the number of CS–US trials during association strengthens the conditioned
response. This is a commonly agreed principle in diverse conditioning paradigms and was first described by
Pavlov (1927) and also applying to neuroimmune associative learning (Espinosa et al., 2004). Although one
learning trial conditioning (e.g., pairing antigenic challenge with saccharin, Alvarez‐Borda et al., 1995;
Madden et al., 2001) has been reported many times, it is not a phenomenon general to all USs employed in
neuroimmune associative learning protocols. Therefore, a paradigm including several association trials may
produce more reliable CRs (Espinosa et al., 2004). In this regard, it has been reported that the magnitude of
the conditioned effects on the immunity is larger after intensive learning (Niemi et al., 2007).
Extinction is defined as the reduction in magnitude and duration of the conditioned response as a
consequence of unreinforced trials (i.e., CS alone) (Szczytkowski and Lysle, 2007). The extinction rate was
directly related to the volume of CS presented on the association trial (Ader, 1974). This principle has been
proven to apply to neuroimmune associative learning in many other studies (e.g., Bovbjerg et al., 1984;
Lysle et al., 1988). In contrast, a recent finding indicates that evoking a consolidated taste–CsA engram
several times resulted in a stronger conditioned immunosuppression (Niemi et al., 2007). Similar heterodox
findings have been reported on an analog taste–immunosuppression engram resulting from pairing
saccharin and cyclophosphamide. Here, the immunosuppression was also more pronounced after several
CS unreinforced exposures (Ader and Cohen, 1975; Rogers et al., 1976; Wayner et al., 1978). Possible
explanations for such peculiar results are due to an insufficient delay between recall trials (too‐near trials),in that immune functions do not have enough time to return to baseline levels, an additive effect may result
in a cumulative immunosuppression. An alternative hypothesis is that the conditioned immunosuppressive
state occurring at first recall is sensed by the CNS and acts as a reinforcement to induce reconsolidation
(Berman and Dudai, 2001; Eisenberg et al., 2003; Dudai, 2006) of the taste–immunosuppression engram
and hereby working as an additional association trial.
Regarding the phenomenon of passive forgetting (the retention rate after a delay between association
and recall phases), Markovic et al. (1988) found excellent retention of CS–US association 8 weeks after
association. Here, rats were sensitized with ovalbumin injections; afterward, conditioning comprised
saccharin as CS and ovalbumin as US. No extinction occurred during a 6‐day test period in terms of the
CTA behavior.
Contingency is defined as the occurrence of CS and US and is important for achieving associative
learning. Ader and Cohen (1982) demonstrated this in showing that a partial reinforcement protocol
significantly reduced the conditioned immunosuppressive effect. Apparently, contingency in neuroimmune
associative learning adheres also to common learning principles.
Backward association would be the case if the US precedes the occurrence of the CS. Odor–poly I:C
backward conditioning resulted in conditioned increased NK‐cell activity (Solvason et al., 1992). This
should not be possible for directly (fast) perceived US (see > Section 3), whereas an indirectly perceived
US, which takes longer to be sensed by the CNS, is more likely to occur in parallel with sensory and
encoding steps of the CS, and therefore may become associated. In this case, a backward conditioned group
may help to delineate the nature of a given US or to discover the genuine US, respectively. An alternative
approach would be to systematically vary the CS–US interstimulus interval. In addition, if the UR kinetic is
known, the determination of the time point of the most pronounced CR enables extrapolation to the
genuine US that mediates the relevant information.
Latent inhibition, the CS preexposure may retard or diminish CS–US engram, thus reducing the
strength of the CR at recall. Such learning interference phenomena have been documented also in
neuroimmune associative learning (e.g., saccharin–ovalbumin Chen et al., 2004). However, the naturalistic
relation among stimuli should be considered to estimate the relevance of such a phenomenon in a given
conditioning protocol.
4.2 Features of Neuroimmune Associative Learning
Neuroimmune associative learning across life span: The strength of neuroimmune associative learning seems
to be age‐dependent. For instance Spector et al. (1994) report that old mice (24 months old) were able to
display a conditioned enhancement of NK activity in an odor–poly I:C conditioning protocol, but young
(3 months old) mice showed a stronger conditioned response. This reflects very much the unconditioned
response, which is detectable, although weaker in aged mice. This assumption is in line with reports
demonstrating a diminished conditioned immunosuppression in aged animals (Gorczynski, 1987b, 1991).
Two main factors could be the basis of aging effects on neuroimmune associative learning. Learning and
memory deficits are a common characteristic of CNS senescence (Nomura and Hori, 1996). However it
should also be considered that innervations to the immune organs change across life span, reducing
significantly during elderly stages. As has been reviewed (> Section 3.2), such neuroimmune efferent
neural pathways may be essential to induce the conditioned effect on immunity. Thus, it cannot be assumed
that a given conditioning protocol could be extrapolated to any stage of life.
Gender effects: Few systematic attempts have been undertaken to elucidate the role of gender or estruous
cycle on neuroimmune associative learning. Spector et al. (1994) reported conditioned increases in NK‐cellactivity in both male and female mice. However, it should be taken into account that a significant effect of
gonadal hormones on the neural (Kritzer et al., 2007) as well as on immune processes (Cutolo et al., 2006)
has been documented. Thus, it may not be surprising to find gender effects in a given neuroimmune
associative learning protocol.
Compensatory or paradoxical conditioned effects: These effects on immune responses have been reported
after evocation of a given neuroimmune associative learning. For example, Gorczynski and Kennedy (1984)
varied the time of day at which the initial association trial began and found that conditioned suppression
was developed by taste cues paired with CYduring the light portion of the diurnal cycle, whereas association
that began during the dark portion of the diurnal cycle resulted in either no CR or a conditioned
immunoenhancement. The authors hypothesized that the background level of neuroendocrine hormones
is critical to the direction of the CR to the taste cue paired with CY. However, Siegel et al. (1987) posited an
alternative hypothesis, indicating that the observation that most conditioning studies of regulatory
responses use stimuli other than taste cues suggest that the immunosuppression results in evoking a
taste–CY association may be unique to the use of taste cues. Other cues, inadvertently present in the
association regimen may control CRs different than those CRs elicited by taste cues (e.g., cue to conse-
quence specificity, Garcia and Koelling, 1967b).
Consciousness: An interesting finding is that association and recall phases can occur in anesthetized
subjects. Hsueh et al. (1992) reported that mice associate CS (camphor odor) and US (poly I:C) under
anesthesia. Second, if conditioned consciously, they could recall under anesthesia, even when CS/US
interstimulus interval was separated by 1–2 days. Since this unusual long interval differed from conscious
learning where the organism seeks information using a logical perceptual relation among events, the
authors reasoned that CS/US learning must be taken place unconsciously by different rules.
Immune history: Different immune histories among subjects may result in a divergent response to the
same immune stimulus. This yields a different immune reaction signaled to the CNS that may become
Conditioned immune responses on animal disease models
Disease
model
Conditioned
response
Unconditioned
stimulus
Conditioned
stimulus Species References
Allergy Delayed‐typehypersensitivity
response ↓
Cyclophosphamide SAC Mouse Roudebush and
Bryant (1991)
Delayed‐typehypersensitivity
response ↑
Cyclophosphamide SAC Mouse Bovbjerg et al.
(1987)
Tumor
model
Survival time of
tumor‐bearinganimals ↑
Poly I:C Camphor odor Mouse Ghanta et al.
(1987, 1988)
Tumor growth ↓ Allogenic DBA/
2 spleen cells
Camphor odor Mouse Ghanta et al.
(1990)
Myeloma growth ↓
cytotoxic
T‐lymphocyte
response to
YC8 tumor ↑
Poly I:C Camphor odor Mouse Ghanta et al.
(1995)
Increased and
decreased
development of
DMBA‐inducedtumors ↑↓
Cyclophosphamide
cimedine (histamine
type II‐receptorantagonist)
SAC Mouse Gorczynski et al.
(1985)
Autoimmune
disease
Proteinuria and
mortality ↓
Cyclophosphamide SAC Mouse Ader and Cohen
(1982)
Grafting Cytotoxic
T‐lymphocyte
precursor cells ↑
C57BL/6 lymphoid
cells inoculated i.p.
Sham skin
grafts
Mouse Gorczynski et al.
(1982)
Heart graft survival ↑ Cyclosporine A SAC Rat Grochowicz
et al. (1991)
Arthritis Arthritic
inflammation
SAC‐vanilladrink
Rat Klosterhalfen and
Klosterhalfen
(1983)
Asthma Histamine release ↑ Bovine serum
albumin (with prior
sensitization)
Methylsulfide
and
triethylamine
odor
Guinea
pig
Russell et al.
(1984); Dark
et al. (1987);
Peeke et al.
(1987)
Histamine ralease ↑ Ovalbumin (with
prior sensitization)
Dimethylsulfide
odor
Guinea
pig
Irie et al. (2001,
2002, 2004)
134 6 Neuro-immune associative learning
and it has been postulated that the insular cortex may integrate gustatory and visceral stimuli (Sewards and
Sewards, 2001). More recently, using the neuronal activity marker c‐Fos, it was possible to confirm the
preponderant role of the insular cortex in conditioned increase of antibody production (taste–ovalbumin)
(Chen et al., 2004). In particular, reexposure to the CS yielded significant increase in c‐Fos expression in all
insular areas (granular, dysgranular, and agranular) 120 min postrecall, confirming previous observations
employing an excitotoxic lesioning approach (Ramırez‐Amaya and Bermudez‐Rattoni, 1999).Regarding other forebrain structures, the amygdala seems to play an important role during the
formation of aversive ingestive associations (Reilly and Bornovalova, 2005), and is also relevant for
US, the CS was a usually preferred dark compartment of a box. After association phase, animals avoided
this compartment and spent more time in the bright (usually aversive) compartment of the box. The
expression of c‐Fos expression was tracked after airway ovalbumin challenge and it was found to be
enhanced in the hypothalamic paraventricular nucleus and the central nucleus of the amygdala in
sensitized mice. Therefore, it is likely that the immediate hypersensitivity reaction in allergic asthma,
which is characterized by IgE‐mediated mast‐cell deregulation with concomitant histamine release, plays
a major role in the reported brain activity (Costa‐Pinto et al., 2005). In addition, these structures are
commonly linked to emotional behavioral patterns that are also involved in conditioned taste aversion
(Bermudez‐Rattoni, 2004).Regarding neural mechanisms behind the odor–poly I:C conditioning paradigm, it has been reported
that IFN‐b is the major input to the CNS during association phase. It can replace poly I:C as US to acquire a
conditioned increase in NK‐cell activity; interestingly, this is not the case for IFN‐a (Solvason et al., 1988).
This association can be achieved by intravenous (i.v.) administration of IFN‐b at a dose of 10,000 IU, but
not at 1,000 IU. Further, if IFN‐b was directly delivered into the CNS via the cisterna magna, a smaller dose
(100 IU) sufficed to be associated with the CS. In turn, the CR was blocked by injecting anti‐IFN‐b(100 neutralizing units) into the cisterna magna 24 h before association (Solvason et al., 1993). The afferent
pathway for association yielding conditioned increased NK‐cell activity seems to be a different one to that of
the conditioned fever in the same associative learning conditioning protocol (Rogers et al., 1992). The i.p.
injection of sodium carbonate blocked CS–US association for enhancement of NK‐cell activity, but leftconditioned fever response unaffected. Conversely, indomethacin treatment which is known to prevent
prostaglandin synthesis blocked the conditioned fever response but not the conditioned NK‐cell response.Using a pharmacological approach, the neurochemical features of the conditioned effect enhancing NK‐cellactivity in rodents (odor–poly I:C association) have been described in detail by one group. Administering
lidocaine centrally blocked both association and recall of the CR (Rogers et al., 1994a) and the CNS sensory
processes of the CS, but not of the US. Similarly, peripheral and central treatment with monosodium
glutamate (Ghanta et al., 1994) or sodium carbonate (Rogers et al., 1994b) apparently blocks association,
but not CS perception. Moreover, the hypothalamic arcuate nucleus is necessary for association, but not for
recall of the CR (Ghanta et al., 1994). Recall in this paradigm seems to be mediated by central opioid
pathways; peripheral injection of the opioid receptor antagonist quaternary naltrexone (not penetrating the
CNS) does not affect recall of the CR. Naltrexone given before association did not prevent the conditioned
effect (Solvason et al., 1989), indicating independency of central opioid pathways.
Central catecholamines seem to be essential, and glutamate – but not GABA – is also required at recall
stage (Hsueh et al., 1999; Kuo et al., 2001). In particular, reserpine treatment before recall, which
unspecifically depletes central and peripheral catecholamine contents, blocked the CR (Hiramoto et al.,
1990). More recently, it has been shown that a‐ and b‐adrenoceptor antagonists or dopamine (DA)‐1 and
DA‐2 receptor antagonists given shortly before recall also blocked the CR (Hsueh et al., 1999).
Furthermore, it has been demonstrated that cholinergic, as well as serotonergic, central systems are
required in triggering conditioned NK‐cell response (Hsueh et al., 2002). At association, acetylcholine is
believed to act through nicotinic, M2‐, and M3‐muscarinic receptors, whereas at recall M1‐, M2‐ and M3‐muscarinic receptors have been identified to be crucially involved. In both association and recall phase,
serotonin acts through 5‐HT1 and 5‐HT2 receptors to affect the CR.
Neurotransmitter contents in certain brain areas during recall stage have been analyzed, revealing a
significantly higher norepinephrine content in the cerebellum and dopamine content in the striatum and
hippocampus in conditioned animals compared with controls (Hsueh et al., 1999). Interestingly, glutamate
contents at recall in this paradigm in the same brain areas did not differ between groups (Kuo et al., 2001).
In an identical conditioning protocol (camphor odor, poly I:C) with enhanced neutrophil activity as CR
instead of NK‐cell activity (Chao et al., 2005) measured levels of tyrosine hydroxylase in several brain areas
24 h after recall to localize action sites of catecholamines. Conditioned animals displayed significantly more
tyrosine hydroxylase expressing neurons in the hypothalamus, cortex, and locus coeruleus compared with
control animals. But this is unlikely to be related to a neural memory process since a 24 h memory trace is
In addition to classical neurotransmitters, cytokines have been demonstrated to play an important role
within the CNS, modulating neuronal and glial function in nonpathological settings such as learning and
memory processes (Balschun et al., 2004; Dantzer, 2004; Tonelli et al., 2005). Specifically, proinflammatory
cytokines, such as IL‐1, IL‐6, and TNF‐a, have been shown to modulate spatial learning tasks, as well as
long‐term potentiation phenomena (Gibertini, 1996; Schneider et al., 1998; Fiore et al., 2000; Banks et al.,
2001; Matsumoto et al., 2001, 2002; Rachal Pugh et al., 2001; Lynch, 2002). In this sense, it can be assumed
that cytokines may be a significant factor in the associative processes occurring during behavioral condi-
tioning of immune functions. Apart from these neuromodulatory properties, proinflammatory cytokines
seem to play an important part in the afferent pathway between the immune system and the CNS
(Besedovsky and del Rey, 1996; Turnbull and Rivier, 1999; Dantzer, 2004). Therefore, it can be hypothesized
that central cytokines may act as mediators in the brain during an ‘‘immune‐sensing’’ phase in the
association phase. These hypotheses are supported by observations that (1) receptors for these proinflam-
matory cytokines are expressed in the CNS (Szelenyi, 2001; Sredni‐Kenigsbuch, 2002), (2) peripheral
immune changes affect central cytokine production and cytokine receptor expression in the brain (Pitossi
et al., 1997; Del Rey et al., 2000), and (3) cytokines can act as unconditioned stimuli to induce conditioned
taste aversion/avoidance (Tazi et al., 1988; Dyck et al., 1990; Janz et al., 1991; Hiramoto et al., 1993).
The underlying mechanisms when using SEB as US are completely unknown. However, systemic IL‐2administration has been found to modify central monoamine activity (Lacosta et al., 2000). Similarly,
striatum catecholamine concentrations followed a dose–response curve in reaction to increased peripheral
SEB immunization (Pacheco‐Lopez et al., 2004). Therefore, association may involve T‐cell‐derived cyto-
kines like IL‐2 signaling to the CNS.
5.2 Peripheral Mediation of the Conditioned Effects on Immunity
The available data suggest that the effects of conditioning could be mediated by a preferential effect on
T cells. Conditioned suppression of lymphoproliferative responses in rats and mice, for example, has been
observed in response to T‐cell mitogens but not (or less reliable) in response to B‐cell mitogens (Neveu
et al., 1986; Kusnecov et al., 1988; Lysle et al., 1990, 1991). Immune adoptive transfer experiments also
suggest that conditioning may be mediated by T‐cell changes (Gorczynski, 1987a). Splenocytes from
conditioned or experimentally naıve animals were transferred into irradiated naıve or conditioned animals
that were or were not subsequently reexposed to the CS. The observed increases or decreases in the
antibody‐forming cell response to SRBC depended on the donor cells and the conditioning treatment
experienced by the recipient. The separate transfer of enriched Tand B cells into naıve or conditioned animals
suggested that conditioning effects were attributable to the adoptively transferred T cells (Gorczynski, 1991).
However, the specificity of whether conditioning can modulate the antibody response to different types of
antigens has not been resolved.
With respect to adrenocortical influences, it is reasonable to hypothesize that conditioned alterations in
immunologic reactivity could be mediated by conditioned neuroendocrine changes. In experiments in
which humoral immune responses were assessed in conditioned animals, the recall of a taste–LiCl
association did not affect antibody responses to SRBC, in contrast to the immunosuppression that was
observed after evoking the taste–CY engram (Ader et al., 1979). This finding indicates that an immuno-
suppressive drug was needed to cause immunosuppression during association phase, thus supporting a
taste–immunosuppression engram interpretation.
In follow‐up experiments, circulating levels of adrenocortical steroids were artificially elevated by
exogenous administration of corticosterone at the time of antigen injection to mimic the stress response
occurring in animals reexposed to the taste paired with sickness (Ader et al., 1979). Elevated glucocorticoid
levels did not significantly lower antibody titers. These data would appear, a priori, to exclude the possibility
of a stress‐mediated phenomenon in the genesis of conditioned immunosuppression.
However, experimental data, in which the delayed‐type hypersensitivity (DTH) response was used as an
index of T‐cell function, support the hypothesis that changes in the immune function of animals subjected
to conditioned taste aversion might be better viewed as a secondary consequence of the psychological
conflict affecting thirsty animals that are exposed to a taste solution previously associated with sickness
(Kelley and Dantzer, 1988). According to this hypothesis, the immunosuppressive status after evocation
phase is a consequence of the conflict of the strong motivation to drink in fluid‐deprived animals against
the aversive memories associated to the CS. Accordingly, it should be possible to induce immunosuppres-
sion in conditioned animals even when no immunosuppressive drug is used during association phase (as
US). After a taste–LiCl association, conditioned animals showed a strong conditioned taste aversion, but
also an immunosuppressive status in the T‐cell function only when a forced choice test was implemented,
but not in a two‐bottle preference procedure which eliminates the psychological conflict of thirst versus
aversive memories (Kelley et al., 1984, 1985). In contrast to this stress hypothesis, a serum corticosterone
time–course study performed to examine the possible involvement of glucocorticoids in conditioned
immunosuppression of the DTH response has been published (Roudebush and Bryant, 1991). Animals
were sacrificed 30, 60, 90, 120 min and 24 h after evoking a taste–CY engram. No significant differences in
serum corticosterone levels were detected between nonconditioned controls and the conditioned groups at
any time point. Supporting this line of thinking, several other research groups have consistently reported that
the suppression of splenic T‐cell proliferation was independent of the stressor‐induced increase in adrenocor-
tical activity (Mormede et al., 1988; Lysle et al., 1990; Exton et al., 1998).
Importantly, both conditioned and stressor‐induced alterations in immune and nonspecific defense
responses have been attributed to the action of central and peripheral catecholamines. For instance, it has
been reported that both chlorpromazine and amitriptyline, both centrally acting, abolish the immunosup-
pressive status elicited by recalling a taste–CYengram (Gorczynski and Holmes, 1989). Furthermore, it has
also been reported that the b‐adrenergic antagonist, propranolol, blocked the immunosuppressive effects of
a conditioning stress paradigm (Lysle et al., 1992). Supporting the involvement of peripheral catechola-
mines, nadolol (a/b‐adrenergic antagonist that does not cross the blood–brain barrier) blocked the electric
shock‐induced suppression of splenic, but not peripheral, blood lymphocyte proliferation following
mitogenic stimulation ex vivo (Cunnick et al., 1990). In this regard, we have previously revealed that the
immunosuppressed status after evoking a taste–CsA engram is not related to the activation of the HPA axis
and is merely mediated by the neural innervation of the spleen, via noradrenaline–b‐adrenoceptors‐dependent mechanisms (Exton et al., 1999, 2002; Xie et al., 2002). It has been demonstrated that splenocyte
reactivity is modulated in part by tonic inhibition from the splenic nerve (Okamoto et al., 1996) and
sympathetic splenic innervation seems to be under the central control of the ventromedial hypothalamus
(Katafuchi et al., 1993, 1994). While electrical stimulation of the ventromedial hypothalamus has been
found to arouse sympathetic activity (Saito et al., 1989), the lateral hypothalamus seemed to do the
opposite (Bernardis and Bellinger, 1993). Thus, the lateral hypothalamus may exert immunoenhancing
properties (Wrona and Trojniar, 2003) in part by antagonizing the ventromedial hypothalamus, and thereby
reducing sympathetic tone inhibition. Importantly, such hypothalamic regulation of sympathetic activity
seems to be modulated by the insular cortex (Allen et al., 1991; Cechetto and Chen, 1992; Oppenheimer
et al., 1992; Butcher and Cechetto, 1998).
Concerning the conditioned enhancement of NK‐cell activity, several mechanisms have already been
elucidated. On the neural efferent pathway, sympathetic innervation of the spleen seems not to be responsi-
ble for mediating the conditioned response. Despite peripheral sympathectomy using 6‐hydroxydopamine
(6‐OHDA) between association and recall, the CR still occurred (Hiramoto et al., 1990). However, splenic
denervation should give final evidence. Interestingly, applying the same conditioning protocol, but measur-
ing neutrophil activity as UR and CR, respectively, peripheral sympathectomy before recall completely
abrogated the CR (Chao et al., 2005). On the humoral efferent pathway, elevated plasma adrenocorticotropic
hormone (ACTH) levels and splenic IFN‐a expression were measured in conditioned animals at recall time
(Hsueh et al., 1994b); no effect was found for b‐endorphin concentrations. Peripheral administration of the
synthetic glucocorticoid dexamethasone blocked recall, but not association of the CR, assumingly by
negative feedback inhibition of HPA axis activity (Hsueh et al., 1994a). Dexamethasone treatment before
recall resulted in elevated neutrophil activity in conditioned animals compared with controls, indicating that
this specific CR is independent of HPA activity (Chao et al., 2005).
The DTH response refers to an overreaction produced by the immune system given a presensitized
immune state of the host. A conditioned decrease (Roudebush and Bryant, 1991) and increase (Bovbjerg
et al., 1987) of this response has been observed by pairing saccharin with CY. Similarly, Exton et al. (2000)
reported a conditioned suppression of the contact hypersensitivity reaction by pairing saccharin with CsA.
Conditioning protocols attested a significant impact on tumor‐related models. For instance, tumor
growth was demonstrated to be enhanced as well as delayed by applying different US (CY and cimedine, a
histamine type II receptor antagonist) (Gorczynski et al., 1985). Moreover, the survival time of tumor‐bearing mice has been prolonged in conditioned animals (Ghanta et al., 1988).
Evoking a taste–CY engram, Ader and Cohen (1982) demonstrated a conditioned retardation in
proteinuria and mortality in New Zealand hybrid mice which are prone to develop an autoimmune disease
resembling human lupus erythematosus. Most interestingly, in a follow‐up study they could show that
the lupus‐prone MRL‐lpr/lpr mice displayed a weaker CTA compared with congenic controls. The
authors interpreted that the lupus‐prone animals seek to regain homeostasis by consuming the tasting
solution to achieve the therapeutic immunosuppressive status elicited by evoking the taste–CY engram
(Grota et al., 1987).
Another example of conditioning effects on immunity is documented by grafting experiments. Gorczynski
et al. (1982) performed skin grafting as CS in combination with i.p. injected lymphoid cells of another
mouse strain as US. Reexposing the conditioned animals to the sham‐grafting procedure, they showed
an increase in cytotoxic T‐lymphocyte precursor cells specific for alloantigens on the grafted tissue.
In addition, Grochowicz et al. (1991) reported a conditioned prolongation of the survival time of heart
allografts in rats evoking a saccharin–CsA engram. In follow‐up experiments, Exton and colleagues
(1999) demonstrated a significant prolongation of the survival of transplanted hearts, including long‐term survival (>100 days) of transplants in 20% of the animals that were conditioned and additionally
subtherapeutically treated with CsA. In addition, employing taste–CsA association Klosterhalfen and
Klosterhalfen (1983, 1990) extended the conditioned immunosuppression to an arthritis model. Before
induction of adjuvant arthritis, rats were dosed with cyclophosphamide (US) after presenting a distinctive
saccharin/vanilla solution (CS). Conditioned rats showed no external signs of a proliferation of inflamma-
tion, whereas approximately half of the animals in the control groups developed small lesions.
A series of studies dealt with conditioning of asthma‐like symptoms, anaphylactic shock (Noelpp
and Noelpp‐Eschenhagen, 1951c; Djuric et al., 1988; Palermo‐Neto and Guimaraes, 2000), or histamine
release, respectively (Dark et al., 1987; Irie et al., 2001, 2002, 2004) that have been already reviewed
(> Sections 1.3, 1.4, and 4.2).
6.2 Human Studies
This section summarizes the few, but promising research reports of conditioned effects modulating immune
responses in human subjects, indicating potential therapeutic outcomes of this behavioral approach.
A Japanese study from the early 1960s reported a conditioned dermatitis response in adult male subjects
elicited by evoking a specific association (CS: blue solution topically applied also contains 2% raw extract
Rhus vernicifera: US) (Ikemi et al., 1962). A case report of asthmatic patients suffering from skin sensitivities
to house‐dust extract and grass pollen shows a conditioned effect. The subjects were exposed to these
allergens by inhalation (Dekker et al., 1957). After a series of conditioning trials, they experienced allergic
attacks after inhalation of the neutral solvent used to deliver the allergens. This work showed not only fast
conditioning of the asthmatic attack (CR), but also tenacious retention, that is, lack of extinction. Similarly,
presenting a novel‐tasting and novel‐appearing drink as CS with house‐dust mite allergen as US to patients
with allergic rhinitis yielded an associative learning (Gauci et al., 1994). In addition, pairing an olfactory cue
as CS with nasal challenge with seasonal grass allergens to investigate conditioning of the immediate
hypersensitivity reaction in hay fever (seasonal allergic rhinitis) produced conditioned increase of histamine
release and decrease nasal airflow at recall (Barrett et al., 2000). Furthermore, there was evidence of
extinction and a follow‐up experiment showed a more pronounced conditioned effect after three associa-
tion trials compared with one, indicating a reinforcement effect. An elevated mast‐cell tryptase in mucosa
was observed when an intranasal saline application was given simultaneously with the CS at recall. Another
type of allergic reaction, the DTH response, was tested in healthy volunteers who received five monthly
tuberculin skin tests (Smith and McDaniel, 1983). In this conditioning protocol, both tuberculin (US) and
saline were injected; the latter was taken from a green vial (CS�) and the former was drawn from a red vial
(CSþ). On the test day, the color labeling of the substances was reversed. Although the saline injections did
not induce a skin reaction (erythema and induration), the severity of the symptoms was significantly
blunted in all the subjects tested when the tuberculin was drawn from the green vial (i.e., conditioned
compensatory effect), where subjects expected their reactions to be negative. However, a similar protocol
using various allergens (e.g., mite dust, fur) taken from colored vials did not result in conditioned
modulation of skin reactions in the subjects tested (Booth et al., 1995).
Another series of experiments investigated the role of conditioning in the context of cancer treatment
and chemotherapy (Bovbjerg, 2003). Some chemotherapeutic agents such as CY have immunosuppressive
effects. For the patient, often the treatment visits to the hospital become aversive and act as CS–US
association trials pairing whereby a variety of features such as white coat, distinct smell when entering
the building, the clinic itself, its smell, the clinician’s voice, and so on may act as distinctive salient stimuli
(CS) that are contingently paired with the chemotherapy (US) (e.g., CY). Immune function was assessed in
cancer patients in the hospital before chemotherapy and compared with assessments conducted at home.
Proliferative responses to T‐cell mitogens were lower for cells isolated from blood samples taken in the
hospital (i.e., after recall) than for home samples (Bovbjerg et al., 1990). These results were replicated in
ovarian cancer patients (Lekander et al., 1995) and pediatric cancer patients receiving chemotherapy (Stock-
horst et al., 2000). In addition, chemotherapy patients often develop conditioned/anticipatory nausea (Andry-
kowski, 1988; Bovbjerg et al., 1990; Morrow et al., 1991; Matteson et al., 2002), anxiety (Jacobsen et al., 1993;
DiLorenzo et al., 1995), and fatigue (Bovbjerg et al., 2005) responses to reminders of chemotherapy. To
resolve such severe consequence of chemotherapy treatment, one may use some of the learning principles
that have been shown to apply to neuroimmune associative learning (> Section 4), for example, latent
inhibition to prevent such undesirable conditioning effects. Several visits to the clinic where the chemo-
therapy will take place should be recommended before starting the chemotherapy. Another possible
behavioral prophylactic therapy may be accomplished before chemotherapy, a specific novel stimulus
that could be administered under the physician’s control, for example, a novel and distinctive tasting/
flavored/colored/sparkling solution. Such artificial stimulus will be categorized and associated to negative
chemotherapy hedonic values; however, it may prevent the development of possible associations to other
high‐protein/caloric food input (Scalera, 2002).
Regarding conditioning of cellular immune parameters, one group assessed the conditionability of
augmentation of NK‐cell numbers and their lytic activity in healthy subjects. Although a conditioned
response was evoked after pairing a given taste with subcutaneous administered adrenaline (Buske‐Kirschbaum et al., 1992, 1994), these effects could not be replicated (Kirschbaum et al., 1992). In multiple
sclerosis patients, four monthly CY infusions (US) were paired with an anise‐flavored syrup (CS) (Giang
et al., 1996). Long‐term treatment with CY decreases blood leukocyte numbers, which often leads to
leukopenia. After a long period such as 6 months of administering the placebo infusion paired with the
drink, 8 out of 10 patients showed a conditioned reduction in peripheral leukocytes numbers. In addition,
by pairing subcutaneously interferon‐g injections (US) with a distinctively flavored drink (CS), it was
possible to induce an elevation of neopterin and quinolinic acid serum levels after evoking such an
association in healthy volunteers (Longo et al., 1999). However, it has been hypothesized that more than
a single associative learning trial pairing a distinctive taste (CS) with interferon‐b injections (US) is