Around 120 years ago, Ivan Pavlov unintentionally sparked a new field of psychology research. He did so by noting that his dogs had learned to associate the sound of the bell that he rang before feeding them with the food itself, such that they would salivate upon hearing the bell even when there was no food present. This form of learning—now known as associative learning—has since been demonstrated in species from honeybees to humans.

For the brain to associate two events, such as the sound of a bell and the delivery of food, it must encode the first event and keep that information available or ‘on-line’ until the occurrence of the second event, at which point the two can be linked together. This process takes place in part of the brain called the prefrontal cortex, but the mechanism by which it occurs is largely unclear.

Now, Iwashita has obtained new insights into the molecular basis of associative learning by studying how the activity of the prefrontal cortex is affected by the activity of a second region of the brain. This second region, called the ventral tegmental area, is part of the brain's reward circuit: it becomes active whenever an animal experiences a desirable event, such as receiving food, and supplies a neurotransmitter called dopamine to its target areas, which include the prefrontal cortex.

Electrodes were used to mimic the changes in brain activity that occur when a mouse learns to associate a visual stimulus with a reward: this involved repeatedly activating the visual cortex in a conscious mouse, followed by activation of the ventral tegmental area. Short-lived increases in calcium levels were seen in the prefrontal cortex, raising the possibility that these ‘calcium transients’ are the signal that enables the brain to link two events. Moreover, blocking proteins called dopamine D1 receptors in the prefrontal cortex reduced the calcium transients, which is consistent with existing evidence that dopamine from the ventral tegmental area is required for associative learning.

Intriguingly, the calcium transients lasted for roughly 30 s, which is also the maximum length of time by which a stimulus and a reward can be separated and still be associated. Given that the calcium transients could not be detected in anesthetized mice, a full understanding of the mechanisms underlying associative learning may require studies of the conscious brain.

The prefrontal cortex (PFC) plays an important role in adaptive behavior such as associative learning (Duncan, 2001). Dopaminergic input from the ventral tegmental area (VTA) is crucial for PFC function (Schultz, 2007). In primates, sensory cues, which are used in associative learning tasks, create specific temporal activity patterns in PFC neurons and a neural representation of the sensory cue (Jacob et al., 2013). However, how dopamine contributes to form neural representations at the neuronal network level is largely unknown. One potential mechanism is that dopaminergic (DA) neurons target dopamine over a number of inhibitory and excitatory neurons via their widespread axonal arborizations (Matsuda et al., 2009). Thus, investigating modification of neuronal activity at the population level will reveal the role of dopamine signaling in the formation of neural representation. Therefore, utilizing in vivo two-photon Ca²⁺ imaging in awake mice, this study investigated how neural representation in PFC neurons is developed under regulation by dopamine, in response to visual sensory input.

PFC neuronal activity was recorded through a cranial window at the secondary motor cortex (M2). The M2 is categorized as the dorsomedial PFC in rodents in some publications, based on both its anatomical features, including thalamocortical and cortical–basal ganglia connections, and its functional role in motor decision (Sul et al., 2011; Uylings et al., 2003; Hoover and Vertes, 2007). The M2 receives inputs from both the VTA and the secondary visual cortex lateral area (V2L) (Figure 1A,B). Microelectrodes were implanted in these two brain areas to supply electrical stimulation (Figure 1C,D).

Figure 1

Download asset Open asset

To determine the single input response, the PFC neuronal response to VTA stimulation was first measured by recording Ca²⁺ transients. Most DA neurons are spontaneously active with firing patterns that range from regular tonic firing (1–5 Hz) to phasic firing (40–50 Hz) (Robinson et al., 2004; Grace and Bunney, 1984a; Grace and Bunney, 1984b). In addition, salient events such as experiencing a novel environment or receiving a reward-associated signal, evoke phasic firing in the VTA (Schultz, 2007; Horvitz, 2000). The PFC activity evoked by VTA stimulation in these physiological ranges was therefore examined.

When low frequency VTA microstimulation was used (1 Hz–10 Hz, tonic range), Ca²⁺ transients in the PFC neurons were evoked and decayed within 5 s. This steep increase and short decay is comparable with the reported response to sensory input observed in cortical neurons of anesthetized animals (Ohki et al., 2005). In contrast, high frequency VTA stimulation (40–50 Hz, phasic range) evoked substantially elongated Ca²⁺ transients that lasted 20–30 s and then returned to the original baseline (Figure 2). In addition, these Ca²⁺ transients of the PFC neurons reached a peak relatively slowly, 6–7 s after stimulation. Furthermore, high frequency VTA stimulation robustly evoked Ca²⁺ transients in all tested animals, whereas low frequency VTA stimulation did not as only about half of the animals tested showed detectable Ca²⁺ transients in the PFC. Importantly, these long-lasting Ca²⁺ transients were only detected in awake and not in anesthetized mice (Figure 1E,F), possibly because of the potential inhibition of voltage-gated calcium channels by isoflurane (Herring et al., 2009; Study, 1994). This result highlights the advantage of a system using awake animals to elucidate dopamine regulation of neuronal responses in the PFC.

Figure 2 with 1 supplement see all

Download asset Open asset

To test whether long-lasing Ca²⁺ transients can be evoked by other neural inputs, the same microstimulation protocol used in the VTA (1–50 Hz) was applied to a region of the visual cortex, the V2L. Both stimulation frequencies only evoked short Ca²⁺ transients in PFC neurons (Figure 2—figure supplement 1). Input from the VTA, especially through high frequency, phasic stimulation can therefore induce a specific signal required to evoke long-lasting Ca²⁺ transients in PFC neurons. Phasic stimulation of the VTA is known to facilitate DA release from its terminals (Tsai et al., 2009; Lavin et al., 2005), therefore, it was hypothesized that DA receptors are involved in induction of the long-lasting Ca²⁺ transients. DA receptors are members of the G-protein coupled receptor family and are composed of two groups: D1 and D2, which have opposing effects on intracellular signaling (Seamans and Yang, 2004; Soltani et al., 2013). Selective DA receptor antagonists for each family were administered by intraperitoneal injection (i.p.). Treatment with the D1 antagonist SCH23390 (1 mg/kg) reduced the long-lasting Ca²⁺ transients by 50% and the decay time constant by 60%. In contrast, no detectable attenuation was observed in mice treated with the D2 antagonist eticlopride (0.5 mg/kg) (Figure 3A,B). In addition, the short Ca²⁺ response evoked by a 5 Hz VTA stimulation was not affected by the D1 or D2 antagonist (Figure 3—figure supplement 1A,B). However, this short Ca²⁺ response was reduced by NMDA and AMPA antagonists (Figure 3—figure supplement 1C,D). In the VTA, up to 65% of the neurons are dopaminergic and the others are GABAergic or glutamatergic (Nair-Roberts et al., 2008; Gorelova et al., 2012), and some DA neurons co-release glutamate (Hnasko et al., 2010). However, glutamatergic terminals are dominant in the mesocortical pathway (Gorelova et al., 2012). Inhibition of the short Ca²⁺ response by glutamate receptor antagonists suggests that glutamatergic neurons contribute to the 5 Hz responses. Initial experiments implicated the D1 receptors for the long-lasting Ca²⁺ transient. To exclude a potential role of glutamatergic neurons in initiating the long calcium transients, a cocktail of NMDA and AMPA receptor antagonists was used. However, the long-lasting Ca²⁺ transients were not affected by NMDA and AMPA antagonists (Figure 3C,D). This result further suggests that the long-sustained increase in intracellular Ca²⁺ concentration is not mediated by glutamatergic local recurrent networks. Over all, pharmacological experiments clearly suggest D1 receptors are mainly involved in the long-lasting Ca²⁺ response in PFC neurons.

Figure 3 with 1 supplement see all

Download asset Open asset

To resolve the potential role of dopamine-induced, long-lasting Ca²⁺ responses for PFC circuitry modification, the PFC neuronal response to sensory input from the V2L after combined repetitive stimulation of the V2L and VTA was investigated. An experimental paradigm composed of a pre-conditioning phase, conditioning phase, and test phase was used (Figure 4A). In the pre-conditioning phase, V2L stimulations (20 Hz, five pulses) were given three times to acquire the base response of PFC neurons to V2L inputs. In the conditioning phase, the VTA and V2L were simultaneously stimulated every minute for 30 min. After this conditioning phase, repetitive V2L stimulation (20 Hz, five pulses) was applied at three time points: right after, 1 hr after, and 2 hr after completion of the conditioning phase (test phase, Figure 4A), to determine whether or not the responses of the PFC neurons were modified.

Figure 4 with 2 supplements see all

Download asset Open asset

For the conditioning phase, the effect of timing on V2L and VTA stimulation was also tested by using two different time intervals, Timing1 (T1) and Timing2 (T2) (Figure 4A). In T1, V2L stimulation was applied half a second before VTA phasic stimulation. T1 simultaneously simulates a visual experience and VTA activation. For T2, V2L stimulation was applied 45 s after VTA phasic stimulation, which is 15 s before the next VTA stimulation, when no Ca²⁺ transients were observed (Figure 4A). Therefore, T2 is used as a control, at a point when visual and VTA inputs are temporally separated. As additional controls, the V2L and the VTA were stimulated separately during the conditioning phase.

Two-way repeated measures ANOVA with temporal change (before, right after, 1 hr after, and 2 hr after conditioning) and conditioning paradigm (V2L only, VTA only, Timing1, and Timing2) as factors, revealed that population averages of dF/F values of Ca²⁺ transients signal showed significant differences in temporal change (F(3, 84) = 48.486, p < 0.0001) and the conditioning paradigm × temporal change interaction (F(9,84) = 3.173, p = 0.0024) (Figure 4B). T1 conditioning significantly increased the dF/F value more than other conditioning paradigms, indicating that T1 conditioning has the largest modification effect on the PFC response to a sensory input (one-way ANOVA: F(3,28) = 6.504, p = 0.0018; post hoc Ryan's test: p < 0.01; Figure 4C). This increase in value is due to an increased number of neurons showing high dF/F values (Figure 4—figure supplement 1A). In addition, there was no significant correlation between the Ca²⁺ level in response to VTA stimulation (at conditioning) and the increased Ca²⁺ level in response to V2L stimulation ‘2 hr after’ T1 conditioning (test phase) (Figure 4—figure supplement 2), suggesting that the increased neuronal response with V2L stimulation is independent of how well the neuron responds to the DA input. This relatively unexpected conclusion may be explained by the fact that the network dynamics of the PFC microcircuit are composed of inhibitory and excitatory neurons, both of which express DA receptors.

Finally, the neuronal population dynamics were investigated through analysis of the pattern similarity (cosine similarity, see the ‘Materials and methods’ section) of neuronal activity in the PFC across three repetitions of V2L stimuli, conducted at four time points: ‘before’, ‘right after’, ‘1 hr after’, and ‘2 hr after’ after the conditioning phase. Two-way repeated measures ANOVA on the temporal changes of the mean cosine similarities in each conditioning paradigm revealed significant differences in the temporal factor (F(3, 84) = 7.676, p = 0.0001) and the time × condition paradigm interaction (F(9, 84) = 2.483, p = 0.0145; Figure 4D). Post hoc tests also showed that the cosine similarity in T1 was significantly increased in ‘2 hr after’ compared with ‘before’ (Ryan's test, p < 0.01; Figure 4D). This increase in cosine similarity in T1 conditioning is not simply due to an increased number of high dF/F neurons (Figure 4—figure supplement 1A), because ‘1 hr after’ in T1 conditioning had a significantly increased cosine similarity (Figure 4D) but its population probability distribution was not significantly different from that of VTA only (2 hr after) or T2 (2 hr after) (two-sample Kolmogorov–Smirnov test, p = 0.5128 and p = 0.1418, respectively) (Figure 4—figure supplement 1A). This suggests that the increase in cosine similarity in T1 conditioning was not due to an increased number of high dF/F neurons. In addition, the increased cosine similarity was not due to increased reliability of Ca²⁺ transients occurrence in response to three repetitive V2L stimulations, because the reliability calculated by Cronbach's alpha did not change between ‘before’ and ‘2 hr after’ in any of the conditioning groups, including T1 (Figure 4—figure supplement 1B). These results indicate that only T1 conditioning improved the pattern similarity in PFC neurons, and the increase in pattern similarity could be the result of circuit (network) modification, and not simply the result of increased reliability of Ca²⁺ transients occurrence or an increased number of high responding dF/F neurons.

This study revealed that only coincident visual sensory input with long-lasting Ca²⁺ transients increased neuronal activity and induced robust repeatable neuronal responses, at the population level, to the same sensory input. By increasing both dF/F and pattern similarity, DA input may enhance PFC activity and establish cortical representation (Xue et al., 2010). In fact, it is known that DA release plays a major role in dynamic cortical remodeling in the auditory cortex (Bao et al., 2001), suggesting that the phenomenon observed here in the T1 conditioning paradigm may represent the neurophysiological basis for dynamic neural events. Besides network modification in PFC, it has also been demonstrated that PFC neurons show long-lasting Ca²⁺ transients that depend on the D1 receptor pathway. This newly identified physiological phenomenon might have an important function in Ca²⁺ regulation of neuronal processes (Berridge, 1998; Brenowitz et al., 2006; Roussel et al., 2006; Bardo et al., 2006). For example, association learning is only successful if the cue signal is less than 20–30 s before the reward signal. Long-lasting Ca²⁺ transients may be tightly associated with this time window in association learning. Taken together, the long-lasting Ca²⁺ transients reported here could be a key phenomenon required to explain the dynamics of the dopaminergic neural network and its role in PFC cognitive functions.

1. 1. Bao S
   2. Chan VT
   3. Merzenich MM

   (2001) Cortical remodelling induced by activity of ventral tegmental dopamine neurons

   Nature 412:79–83.

   https://doi.org/10.1038/35083586

   • Google Scholar
2. 1. Bardo S
   2. Cavazzini MG
   3. Emptage N

   (2006) The role of the endoplasmic reticulum Ca2+ store in the plasticity of central neurons

   Trends in Pharmacological Sciences 27:78–84.

   https://doi.org/10.1016/j.tips.2005.12.008

   • Google Scholar
3. 1. Berridge MJ

   (1998) Neuronal calcium signaling

   Neuron 21:13–26.

   https://doi.org/10.1016/S0896-6273(00)80510-3

   • Google Scholar
4. 1. Brenowitz SD
   2. Best AR
   3. Regehr WG

   (2006) Sustained elevation of dendritic calcium evokes widespread endocannabinoid release and suppression of synapses onto cerebellar Purkinje cells

   The Journal of Neuroscience 26:6841–6850.

   https://doi.org/10.1523/JNEUROSCI.1280-06.2006

   • Google Scholar
5. 1. Duncan J

   (2001) An adaptive coding model of neural function in prefrontal cortex

   Nature Reviews Neuroscience 2:820–829.

   https://doi.org/10.1038/35097575

   • Google Scholar
6. 1. Gorelova N
   2. Mulholland PJ
   3. Chandler LJ
   4. Seamans JK

   (2012) The glutamatergic component of the mesocortical pathway emanating from different subregions of the ventral midbrain

   Cerebral Cortex 22:327–336.

   https://doi.org/10.1093/cercor/bhr107

   • Google Scholar
7. 1. Grace AA
   2. Bunney BS

   (1984a)

   The control of firing pattern in nigral dopamine neurons: burst firing

   The Journal of Neuroscience 4:2877–2890.

   • Google Scholar
8. 1. Grace AA
   2. Bunney BS

   (1984b)

   The control of firing pattern in nigral dopamine neurons: single spike firing

   The Journal of Neuroscience 4:2866–2876.

   • Google Scholar
9. 1. Herring BE
   2. Xie Z
   3. Marks J
   4. Fox AP

   (2009) Isoflurane inhibits the neurotransmitter release machinery

   Journal of Neurophysiology 102:1265–1273.

   https://doi.org/10.1152/jn.00252.2009

   • Google Scholar
10. 1. Hnasko TS
    2. Chuhma N
    3. Zhang H
    4. Goh GY
    5. Sulzer D
    6. Palmiter RD
    7. Rayport S
    8. Edwards RH

    (2010) Vesicular glutamate transport promotes dopamine storage and glutamate corelease in vivo

    Neuron 65:643–656.

    https://doi.org/10.1016/j.neuron.2010.02.012

    • Google Scholar
11. 1. Hoover WB
    2. Vertes RP

    (2007) Anatomical analysis of afferent projections to the medial prefrontal cortex in the rat

    Brain Structure & Function 212:149–179.

    https://doi.org/10.1007/s00429-007-0150-4

    • Google Scholar
12. 1. Horvitz JC

    (2000) Mesolimbocortical and nigrostriatal dopamine responses to salient non-reward events

    Neuroscience 96:651–656.

    https://doi.org/10.1016/S0306-4522(00)00019-1

    • Google Scholar
13. 1. Jacob SN
    2. Ott T
    3. Nieder A

    (2013) Dopamine regulates two classes of primate prefrontal neurons that represent sensory signals

    The Journal of Neuroscience 33:13724–13734.

    https://doi.org/10.1523/JNEUROSCI.0210-13.2013

    • Google Scholar
14. 1. Lavin A
    2. Nogueira L
    3. Lapish CC
    4. Wightman RM
    5. Phillips PEM
    6. Seamans JK

    (2005) Mesocortical dopamine neurons operate in distinct temporal domains using multimodal signaling

    The Journal of Neuroscience 25:5013–5023.

    https://doi.org/10.1523/JNEUROSCI.0557-05.2005

    • Google Scholar
15. 1. Matsuda W
    2. Furuta T
    3. Nakamura KC
    4. Hioki H
    5. Fujiyama F
    6. Arai R
    7. Kaneko T

    (2009) Single nigrostriatal dopaminergic neurons form widely spread and highly dense axonal arborizations in the neostriatum

    The Journal of Neuroscience 29:444–453.

    https://doi.org/10.1523/JNEUROSCI.4029-08.2009

    • Google Scholar
16. 1. Nair-Roberts RG
    2. Chatelain-Badie SD
    3. Benson E
    4. White-Cooper H
    5. Bolam JP
    6. Ungless MA

    (2008) Stereological estimates of dopaminergic, GABAergic and glutamatergic neurons in the ventral tegmental area, substantia nigra and retrorubral field in the rat

    Neuroscience 152:1024–1031.

    https://doi.org/10.1016/j.neuroscience.2008.01.046

    • Google Scholar
17. 1. Nimmerjahn A
    2. Kirchhoff F
    3. Kerr JN
    4. Helmchen F

    (2004) Sulforhodamine 101 as a specific marker of astroglia in the neocortex in vivo

    Nature Methods 1:31–37.

    https://doi.org/10.1038/nmeth706

    • Google Scholar
18. 1. Ohki K
    2. Chung S
    3. Ch'ng YH
    4. Kara P
    5. Reid RC

    (2005) Functional imaging with cellular resolution reveals precise micro-architecture in visual cortex

    Nature 433:597–603.

    https://doi.org/10.1038/nature03274

    • Google Scholar
19. 1. Robinson S
    2. Smith DM
    3. Mizumori SJ
    4. Palmiter RD

    (2004) Firing properties of dopamine neurons in freely moving dopamine-deficient mice: effects of dopamine receptor activation and anesthesia

    Proceedings of the National Academy of Sciences of USA 101:13329–13334.

    https://doi.org/10.1073/pnas.0405084101

    • Google Scholar
20. 1. Roussel C
    2. Erneux T
    3. Schiffmann SN
    4. Gall D

    (2006) Modulation of neuronal excitability by intracellular calcium buffering: from spiking to bursting

    Cell Calcium 39:455–466.

    https://doi.org/10.1016/j.ceca.2006.01.004

    • Google Scholar
21. 1. Schultz W

    (2007) Multiple dopamine functions at different time courses

    Annual Review of Neuroscience 30:259–288.

    https://doi.org/10.1146/annurev.neuro.28.061604.135722

    • Google Scholar
22. 1. Seamans JK
    2. Yang CR

    (2004) The principal features and mechanisms of dopamine modulation in the prefrontal cortex

    Progress in Neurobiology 74:1–58.

    https://doi.org/10.1016/j.pneurobio.2004.05.006

    • Google Scholar
23. 1. Soltani A
    2. Noudoost B
    3. Moore T

    (2013) Dissociable dopaminergic control of saccadic target selection and its implications for reward modulation

    Proceedings of the National Academy of Sciences of USA 110:3579–3584.

    https://doi.org/10.1073/pnas.1221236110

    • Google Scholar
24. 1. Stosiek C
    2. Garaschuk O
    3. Holthoff K
    4. Konnerth A

    (2003) In vivo two-photon calcium imaging of neuronal networks

    Proceedings of the National Academy of Sciences of USA 100:7319–7324.

    https://doi.org/10.1073/pnas.1232232100

    • Google Scholar
25. 1. Study RE

    (1994) Isoflurane inhibits multiple voltage-gated calcium currents in hippocampal pyramidal neurons

    Anesthesiology 81:104–116.

    https://doi.org/10.1097/00000542-199407000-00016

    • Google Scholar
26. 1. Sul JH
    2. Jo S
    3. Lee D
    4. Jung MW

    (2011) Role of rodent secondary motor cortex in value-based action selection

    Nature Neuroscience 14:1202–1208.

    https://doi.org/10.1038/nn.2881

    • Google Scholar
27. 1. Thévenaz P
    2. Ruttimann UE
    3. Unser M

    (1998) A pyramid approach to subpixel registration based on intensity

    IEEE Transactions on Image Processing 7:27–41.

    https://doi.org/10.1109/83.650848

    • Google Scholar
28. 1. Tsai HC
    2. Zhang F
    3. Adamantidis A
    4. Stuber GD
    5. Bonci A
    6. de Lecea L
    7. Deisseroth K

    (2009) Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning

    Science 324:1080–1084.

    https://doi.org/10.1126/science.1168878

    • Google Scholar
29. 1. Uylings HB
    2. Groenewegen HJ
    3. Kolb B

    (2003) Do rats have a prefrontal cortex?

    Behavioural Brain Research 146:3–17.

    https://doi.org/10.1016/j.bbr.2003.09.028

    • Google Scholar
30. 1. Xue G
    2. Dong Q
    3. Chen C
    4. Lu Z
    5. Mumford JA
    6. Poldrack RA

    (2010) Greater neural pattern similarity across repetitions is associated with better memory

    Science 330:97–101.

    https://doi.org/10.1126/science.1193125

    • Google Scholar

Author details

1. Motoko Iwashita

   National Institute of Mental Health, National Institutes of Health, Bethesda, United States

   Contribution

   MI, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   moko0927@gmail.com

   Competing interests

   The author declares that no competing interests exist.

• 2,333

  views

• 162

  downloads

• 6

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.