Learning in a simple biological system: a pilot study of classical conditioning of human macrophages in vitro
© Nilsonne et al; licensee BioMed Central Ltd. 2011
Received: 27 April 2011
Accepted: 18 November 2011
Published: 18 November 2011
Recent advances in cell biology and gene regulation suggest mechanisms whereby associative learning could be performed by single cells. Therefore, we explored a model of classical conditioning in human macrophages in vitro. In macrophage cultures, bacterial lipopolysaccharide (LPS; unconditioned stimulus) was paired once with streptomycin (conditioned stimulus). Secretion of interleukin-6 (IL-6) was used as response measure. At evocation, conditioning was not observed. Levels of IL-6 were higher only in those cultures that had been exposed to LPS in the learning phase (p's < .05), regardless whether they received the conditioned stimulus or not at evocation.
However, habituation was evident, with a 62% loss of the IL-6 response after three LPS presentations (p < .001). If further experiments confirm that simple learning can occur in immune cells, this may have bearings not only on immune regulation, but also on the brain response to molecular signals detected in the periphery. Importantly, whether capacities for simple learning in single cells extend beyond habituation, and how this would be demonstrated, remain open questions.
Associative learning has been demonstrated in organisms with relatively simple nervous systems, such as snails  and cephalopods . In principle, however, associative learning does not necessarily presuppose a nervous system. The essential prerequisites - sensory mechanisms capable of detecting a variety of stimuli, and an information processing system that can assume several stable states encoding past exposures and directing differential responses - could well be found in simpler biological systems such as a single cell. Early studies indicated that the unicellular protozoon Tetrahymena could be conditioned to a light avoidance behavior [3–5], although another study  failed to replicate these findings. Evidence has later been presented suggesting that the unicellular ciliate Paramecium can learn to associate light and electric stimulations . Other studies have shown that membrane changes in a single photoreceptor in the Hermissenda, produced by pairing light with current, was sufficient to cause a learned suppression of phototactic behavior . More recently, predictive behavior has been shown in slime molds and populations of E. coli, which are able to anticipate periodic environmental perturbations [9, 10].
Advances in the understanding of signal transduction and gene regulation now provide a theoretical framework for information processing in the cell, which suggests that mechanisms for learning are likely to exist even in single cells. Epigenetic regulatory mechanisms can function in analogy to a simple type of memory, encoding past experiences of a single cell as DNA methylation patterns, histone modifications, or by action on transcription factors. Such mechanisms can underpin adaptive response patterns that depend on prior exposure to various stimuli [11, 12]. Another possible level of regulation is through changes in faster-acting transcription-independent cell signaling networks. Such networks can assume several semi-stable states, functionally encoding past exposures and leading to adaptive responses on a rapid time-scale .
In the case of the immune system, a large body of research has demonstrated successful conditioning in vivo in several species . Simpler (non-associative) aspects of learning have been demonstrated also in immune cells in vitro. Cultured human macrophages show habituation to LPS , an effect mediated in part by histone modification [16, 17]. However, no reports of behavioral conditioning of leukocytes in vitro have to our knowledge been published before.
Thus, several observations raise the question whether isolated immune cells could be capable of more complex aspects of learning, such as classical conditioning, as part of their functional repertoire. Therefore, we here describe and discuss a tentative model for ex vivo immune conditioning. We tested a classical conditioning paradigm on human macrophages in vitro. The canonical activating stimulus LPS was used as unconditioned stimulus (UCS). As conditioned stimulus (CS), we used streptomycin, a substance well-known to be non-toxic and which the macrophages could potentially recognize through interaction with the acetylcholine (ACh) receptor [18, 19]. As conditioned (CR) and unconditioned (UCR) responses, we measured levels of IL-6 in the cell culture medium. In addition, in a subsample of cultures, we tested whether repeated exposures to LPS would cause habituation of the IL-6 response.
Materials and methods
Macrophage isolation and culture
Monocytes were isolated from human buffy coats using Lymphoprep (Axis-Shield PoC AS, Oslo, Norway) density gradient separation. Buffy coats were obtained from blood donors who had reported that they had not recently undergone antibiotic treatment. Cell suspensions were cultured at 1.8 × 107 cells/ml in 6-well Primaria culture plates (Becton Dickinson, San José, California, USA). After 2 h, non-adherent cells were washed away with phosphate-buffered saline (PBS) and 1 ml of a solution containing supernatants produced by allogeneic stimulation of peripheral blood mononuclear cells was added to promote macrophage differentiation . After 24 h incubation, non-adherent cells were again washed away with PBS and medium was added. Macrophages were fully differentiated after seven days. Cells were cultured at 37°C in a humidified CO2 incubator in Iscove's Modified Dulbecco's Medium, IMDM (Sigma-Aldrich Company Ltd, Irvine, Ayrshire, UK.) with 10% human AB-serum, without antibiotics.
Analysis of variance (ANOVA) was performed with condition as independent variable and IL-6 level after evocation as dependent variable in the first analysis, and IL-6 levels across pairings in the second analysis; the second analysis being a repeated measures ANOVA. Data were log transformed before analyses. For repeated measures the Greenhouse-Geisser epsilon correction was used. T-tests were used to analyse which conditions that differed significantly from each other (one tailed t-test were used to test whether condition 1 had higher levels than the other ones and two-tailed tests were used between the other conditions). The analyses were performed with Stata 9.2 (Statacorp, Texas, US).
The ANOVA showed a significant effect of condition (F (6, 41) = 13, p < .0001). T-tests between respective conditions showed that group 1, 2, 5, and 6, that is, all groups that had been exposed to LPS in the learning phase, responded with IL-6 secretion in the evocation phase regardless whether they received CS or not at evocation (Figure 1, all p's < .05). Cells in group 1, which received streptomycin (CS) and LPS (UCS) in the learning phase and streptomycin (CS) as evoking stimulus, did not show higher IL-6 concentrations compared to group 2. Thus, streptomycin was ineffective as a conditioned stimulus to evoke a conditioned release of IL-6 from macrophages in this experimental setting.
We investigated a classical conditioning paradigm in human macrophages using streptomycin as CS and LPS as UCS. Conditioning was not shown. Rather, the groups that showed an IL-6 response at evocation were those that had been exposed to LPS in the learning phase, irrespective of presence of the CS in either the learning or the evocation phase. The reason for this is not clear, but the effect could possibly be due to sustained activation. Markedly lower responses to UCS were observed after repeated exposures to LPS (and streptomycin) in a subsample of cultures. This indicates that habituation has occurred, and agrees well with previous studies in human macrophage cultures , experimental animals  and humans [22, 23]. Because habituation enables the organism to save energy by down-scaling the response to known stimuli, or to protect the organism from e.g. inflammatory damage, this mechanism could be advantageous also in immune cells.
One of several limitations of this pilot experiment is the scant knowledge of what might constitute a suitable stimulus repertoire to test simple learning in immune cells. While LPS is a canonical activating stimulus for macrophages, it is less obvious what sort of substances, if any, could constitute conditioned stimuli without interfering with the biology of the unconditioned response. We argued that streptomycin fulfills fundamental criteria for a CS, as it is non-cytotoxic and not known to induce specific functional responses in macrophages, while being able to bind to ACh receptors on the cell surface and thus being detected by the cell . Binding of streptomycin to ACh receptors has been shown in the context of neuromuscular junctions, where streptomycin has inhibitory activity [18, 19]. Streptomycin is not considered to invoke specific responses in macrophages, and is frequently used in culture of macrophages to prevent bacterial infection.
Whether conditioning may be achieved in macrophages or other immune cells using a different stimulus set, or different CS/UCS duration and intensity, remain open questions. Another area of obscurity concerns the appropriate number of pairings in the acquisition phase to obtain a conditioned response. Along with a strong tradition of effective aversive taste conditioning  and in vivo immune conditioning , and in line with some prior work on unicellular organisms , we used only one pairing between the CS and the UCS. Another limitation is that habituation was tested in a group of cells that were simultaneously exposed to streptomycin. However, our results do not indicate that streptomycin in itself affected IL-6 secretion, nor that it interacted synergistically or antagonistically with the IL-6 response to LPS after one pairing.
When classical conditioning of the immune system is investigated in mammals, it is presumed that a conditioned stimulus such as a taste is sensed by the CNS via neural afferents, while the neural or immunological changes induced by the US is sensed via neural or humoral afferent pathways . In simple models such as the present using single cells, albeit co-cultured with peers, corresponding sensory signaling tasks need to be carried out within the unicellular system. The surprising flexibility observed in simple systems through e.g. epigenetic mechanisms, and evidence for conditioning in simple organisms  or even in photoreceptors , suggests that highly competent cells in the immune system could potentially show associative learning capacities even in the absence of a nervous system. Hypothetically, if mechanisms of simple learning exist in non-neural cells, it is possible that a hierarchically organized regulation of behavioral responses to actual or anticipated stimuli is performed by the nervous system in conjunction with local circuits that also include cells of the immune system. Thus, the abilities attributed to the immune system to serve sensory functions [27, 28] may include aspects of plasticity, so that facilitation and inhibition of signaling from the micro (e.g. detection of danger associated molecular patterns) to the macro level occurs in a way analogous with how neural responses are modulated by e.g. serotonin release in micro circuits after learning in behavioral models . The implications of understanding learning at the cellular, immune or epigenetic level may therefore have bearings on how we understand learning on a more general level, as the brain response to molecular signals detected in the periphery, and therefore also on regulation of human behavior.
Future studies might explore the possibility to investigate simple learning in immune cells ex vivo by varying e.g. the conditioned stimulus, the number of pairings and by testing other leukocytes, e.g. NK cells which display immunological memory  and which can be cultured without proliferation. If associative learning capacities, as suggested from recent findings in cell biology and gene regulation, could be demonstrated in leukocytes in vitro, the old analogy of the immune system as a mobile brain  would appear even more warranted.
The authors are grateful to Anders Hjerpe and Katalin Dobra for generously providing laboratory facilities, and to Martin Ingvar for support. We also thank Antonio Valentin and Karl Johan Rydgård for early methodological discussions, Martin Hammar and Agnes Ohlsson Gotby for comments on an earlier version of the manuscript, and Lotta Tammik, Klas Strååt and Cecilia Söderberg-Nauclér for valuable methodological advice, and for generously sharing supernatants from allogeneically stimulated monocyte derived macrophages. Furthermore, we thank Mélanie Thessén Hedreul, Amennai Beyeen and Roham Parsa for support and methodological advice.
- Walters ET, Carew TJ, Kandel ER: Classical conditioning in Aplysia californica. Proc Natl Acad Sci USA. 1979, 76: 6675-6679. 10.1073/pnas.76.12.6675.PubMed CentralView ArticlePubMedGoogle Scholar
- Glanzman DL: Octopus conditioning: a multi-armed approach to the LTP--learning question. Curr Biol. 2008, 18: R527-530. 10.1016/j.cub.2008.04.046.View ArticlePubMedGoogle Scholar
- Bergström SR: Induced avoidance behaviour in the protozoa Tetrahymena. Scand J Psychol. 1968, 9: 215-219. 10.1111/j.1467-9450.1968.tb00536.x.View ArticleGoogle Scholar
- Bergström SR: Acquisition of an avoidance reaction to light in the Protozoa Tetrahymena. Scand J Psychol. 1968, 9: 220-224. 10.1111/j.1467-9450.1968.tb00537.x.View ArticleGoogle Scholar
- Bergström SR: Amount of induced avoidance behaviour to light in the protozoa Tetrahymena as a function of time after training and cell fission. Scand J Psychol. 1969, 10: 16-20. 10.1111/j.1467-9450.1969.tb00002.x.View ArticlePubMedGoogle Scholar
- Applewhite PB, Gardner F, Foley D, Clendenin M: Failure to condition Tetrahymena. Scand J Psychol. 1971, 12: 65-67. 10.1111/j.1467-9450.1971.tb00605.x.View ArticleGoogle Scholar
- Armus HL, Montgomery AR, Gurney RL: Discrimination learning and extinction in paramecia (P. caudatum). Psychol Rep. 2006, 98: 705-711. 10.2466/pr0.98.3.705-711.View ArticlePubMedGoogle Scholar
- Farley J, Richards WG, Ling LJ, Liman E, Alkon DL: Membrane changes in a single photoreceptor cause associative learning in Hermissenda. Science. 1983, 221: 1201-1203. 10.1126/science.6612335.View ArticlePubMedGoogle Scholar
- Saigusa T, Tero A, Nakagaki T, Kuramoto Y: Amoebae anticipate periodic events. Phys Rev Lett. 2008, 100: 018101View ArticlePubMedGoogle Scholar
- Tagkopoulos I, Liu YC, Tavazoie S: Predictive behavior within microbial genetic networks. Science. 2008, 320: 1313-1317. 10.1126/science.1154456.PubMed CentralView ArticlePubMedGoogle Scholar
- Ginsburg S, Jablonka E: Epigenetic learning in non-neural organisms. J Biosci. 2009, 34: 633-646. 10.1007/s12038-009-0081-8.View ArticlePubMedGoogle Scholar
- Fernando CT, Liekens AM, Bingle LE, Beck C, Lenser T, Stekel DJ, Rowe JE: Molecular circuits for associative learning in single-celled organisms. J R Soc Interface. 2009, 6: 463-469.PubMed CentralView ArticlePubMedGoogle Scholar
- Guantes R, Poyatos JF: Multistable decision switches for flexible control of epigenetic differentiation. PLoS Comput Biol. 2008, 4: e1000235-10.1371/journal.pcbi.1000235.PubMed CentralView ArticlePubMedGoogle Scholar
- Schedlowski M, Pacheco-Lopez G: The learned immune response: Pavlov and beyond. Brain Behav Immun. 2010, 24: 176-185. 10.1016/j.bbi.2009.08.007.View ArticlePubMedGoogle Scholar
- Ziegler-Heitbrock HW, Blumenstein M, Kafferlein E, Kieper D, Petersmann I, Endres S, Flegel WA, Northoff H, Riethmuller G, Haas JG: In vitro desensitization to lipopolysaccharide suppresses tumour necrosis factor, interleukin-1 and interleukin-6 gene expression in a similar fashion. Immunology. 1992, 75: 264-268.PubMed CentralPubMedGoogle Scholar
- Foster SL, Hargreaves DC, Medzhitov R: Gene-specific control of inflammation by TLR-induced chromatin modifications. Nature. 2007, 447: 972-978.PubMedGoogle Scholar
- Gantner BN, Singh H: Immunology: Short-term memory. Nature. 2007, 447: 916-917. 10.1038/447916a.View ArticlePubMedGoogle Scholar
- Jindal MN, Deshpande VR: Neuromuscular blockade by streptomycin and dihydrostreptomycin. Br J Pharmacol Chemother. 1960, 15: 506-509.PubMed CentralView ArticlePubMedGoogle Scholar
- Dretchen KL, Gergis SD, Sokoll MD, Long JP: Effect of various antibiotics on neuromuscular transmission. Eur J Pharmacol. 1972, 18: 201-203. 10.1016/0014-2999(72)90243-9.View ArticlePubMedGoogle Scholar
- Söderberg-Nauclér C, Streblow DN, Fish KN, Allan-Yorke J, Smith PP, Nelson JA: Reactivation of Latent Human Cytomegalovirus in CD14+ Monocytes Is Differentiation Dependent. J Virol. 2001, 75: 7543-7554. 10.1128/JVI.75.16.7543-7554.2001.PubMed CentralView ArticlePubMedGoogle Scholar
- Randow F, Syrbe U, Meisel C, Krausch D, Zuckermann H, Platzer C, Volk HD: Mechanism of endotoxin desensitization: involvement of interleukin 10 and transforming growth factor beta. J Exp Med. 1995, 181: 1887-1892. 10.1084/jem.181.5.1887.View ArticlePubMedGoogle Scholar
- Hållström L, Berghall E, Frostell C, Sollevi A, Soop AL: Nitric oxide inhalation and glucocorticoids as combined treatment in human experimental endotoxemia. Crit Care Med. 2008, 36: 3043-3047. 10.1097/CCM.0b013e318186f5b2.View ArticlePubMedGoogle Scholar
- Biswas SK, Lopez-Collazo E: Endotoxin tolerance: new mechanisms, molecules and clinical significance. Trends in Immunology. 2009, 30: 475-487. 10.1016/j.it.2009.07.009.View ArticlePubMedGoogle Scholar
- Gwilt CR, Donnelly LE, Rogers DF: The non-neuronal cholinergic system in the airways: an unappreciated regulatory role in pulmonary inflammation?. Pharmacol Ther. 2007, 115: 208-222. 10.1016/j.pharmthera.2007.05.007.View ArticlePubMedGoogle Scholar
- Garcia J, Lasiter PS, Bermudez-Rattoni F, Deems DA: A general theory of aversion learning. Ann N Y Acad Sci. 1985, 443: 8-21. 10.1111/j.1749-6632.1985.tb27060.x.View ArticlePubMedGoogle Scholar
- Ader B, Cohen L: Conditioning and immunity. Psychoneuroimmunology. Edited by: Ader R, Felten DL, Cohen L. 2001, San Diego, California, USA: Academic Press, 3-34. thirdGoogle Scholar
- Blalock JE: The immune system as the sixth sense. J Intern Med. 2005, 257: 126-138. 10.1111/j.1365-2796.2004.01441.x.View ArticlePubMedGoogle Scholar
- Tracey KJ: The inflammatory reflex. Nature. 2002, 420: 853-859. 10.1038/nature01321.View ArticlePubMedGoogle Scholar
- Brunelli M, Castellucci V, Kandel ER: Synaptic facilitation and behavioral sensitization in Aplysia: possible role of serotonin and cyclic AMP. Science. 1976, 194: 1178-1181. 10.1126/science.186870.View ArticlePubMedGoogle Scholar
- Ugolini S, Vivier E: Immunology: Natural killer cells remember. Nature. 2009, 457: 544-545. 10.1038/457544a.View ArticlePubMedGoogle Scholar
- Blalock JE, Smith EM: A complete regulatory loop between the immune and neuroendocrine systems. Fed Proc. 1985, 44: 108-111.PubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.