Objectives Previous research has suggested that different manual acupuncture (MA) manipulations may have different physiological effects. Recent studies have demonstrated that neural electrical signals are generated or changed when acupuncture is administered. In order to explore the effects of different MA manipulations on the neural system, an experiment was designed to record the discharges of wide dynamic range (WDR) neurons in the spinal dorsal horn evoked by MA at different frequencies (0.5, 1, 2 and 3 Hz) at ST36.
Methods Microelectrode extracellular recordings were used to record the discharges of WDR neurons evoked by different MA manipulations. Approximate firing rate and coefficient of variation of interspike interval (ISI) were used to extract the characteristic parameters of the neural electrical signals after spike sorting, and the neural coding of the evoked discharges by different MA manipulations was obtained.
Results Our results indicated that the neuronal firing rate and time sequences of ISI showed distinct clustering properties for different MA manipulations, which could distinguish them effectively.
Conclusions The combination of firing rate and ISI codes carries information about the acupuncture stimulus frequency. Different MA manipulations appear to change the neural coding of electrical signals in the spinal dorsal horn through WDR neurons.
Statistics from Altmetric.com
In recent decades, researchers have qualitatively explored the therapeutic effects and mechanisms of acupuncture based on adenosine, brain-derived neurotrophic factor, fibroblast growth factor 2 (FGF-2), neural acupuncture unit and endorphins.1–4 Manual acupuncture (MA) manipulation is regarded as one of the major components of acupuncture and is an important skill for traditional clinical practitioners. During acupuncture treatments, acupuncture needles are manipulated to elicit the characteristic de qi reaction, which some studies suggest is essential to acupuncture having a therapeutic effect.5–7
A previous study showed that distinct needling styles and techniques produced different needle motion and force patterns.8 A few studies have also shown that different MA manipulations produced different responses in the brain9–14 and one study suggested different therapeutic effects.15 However, manipulation is not easily explained or learnt by practitioners, and we are still not clear about the similarities and differences of the physiological mechanisms generated by different MA manipulations.
Classical acupuncture literature held that the regulatory effects of acupuncture could be influenced by various factors, such as needling intensity, needling frequency, needling duration and interval between stimuli.16 The ST36 point is one of the most frequently used acupuncture points in acupuncture treatment.17 Previous studies have demonstrated that neural electrical signals are generated or altered when the body is stimulated by acupuncture at ST36.18–22
Chaos theory is the study of non-linear dynamics, in which seemingly random events are actually predictable from simple deterministic equations. Although chaos is often thought to refer to randomness and lack of order, it is more accurate to think of it as an apparent randomness that results from complex systems and interactions among systems.23 Chaos theory is currently being applied to medical studies of epilepsy, specifically to the prediction of seemingly random seizures by observing initial conditions.24 Based on non-linear dynamics analysis, our previous studies have confirmed that spike discharges of both primary afferents in dorsal root ganglion and wide dynamic range (WDR) second-order neurons in dorsal horn evoked by different MA manipulations had distinguished chaotic features, which showed that different MA manipulations evoked various kinds of neural electrical signals; however, the seemingly random interspike interval (ISI) sequences of the evoked discharges were not random. It is thought they may in fact be chaotic.25–28
However, these findings have not provided an explicit physiological understanding of the underlying mechanisms of the conduction of acupuncture information. The neural system characterises information from external stimulations by spatiotemporal encoding, which has been widely studied using a spike sorting algorithm (in particular, visual analysis).22 ,29 ,30
Furthermore, computational neuroscience has played an important role in bridging the gap between the sophisticated function and the neuronal substrate. This notion should provide a new way to understand different MA manipulations based on neural information from the computational perspective.
Therefore, in the present work, approximate firing rates and coefficients of variation (CV) of ISI were used to extract neural information on spike discharges after spike sorting, and rate codings and temporal codings of spike discharges evoked by MA with different frequencies were obtained.
The experiments were conducted on adult Sprague–Dawley rats, weight 200–300 g, obtained locally from Japan SLC, Inc. Rats were housed in pairs on a 12:12 h light/dark cycle. Food and water were provided ad libitum.
All procedures were carried out in accordance with the USA Public Health Services Guide for care and use of laboratory animals and were approved by the Institutional Animals Care and Use Committee at Suzuka University of Medical Science. All efforts were made to minimise the number of animals used and the number of procedures per animal. Rats were anesthetised by injection of 20% Urethane (1.5 g/kg). The trachea was cannulated to provide unobstructed ventilation. Core body temperature was maintained at 37–38°C by a feedback-controlled heating pad. After shaving the hair over the midline of the animal's back, an incision was made followed by blunt dissection of the muscles and connective tissue to expose the vertebral column. The animal was secured within a stereotaxic frame to maintain stability (SR-5R, Narishige, Japan). A laminectomy was performed, vertebrae T13–L2 were located, and segments L3–L5 of the spinal cord were exposed and a small well was formed within the surrounding muscle. The spinal cord was continually bathed in a pool of warm (37°C) mineral oil.
Extracellular recordings of single dorsal horn neurons with receptive fields (RFs) located on ST36 were obtained by using glass micropipettes (A-M Systems Co., impedance 5–15 MΩ) filled with 0.5 mol/L sodium acetate. Recording electrodes were lowered into the spinal cord adjacent to the central vessel with a microdrive (MO-10, Narishige, Japan) while the skin surface near ST36 was tapped. Recordings were made from neurons whose amplitude could be easily discriminated. Electrophysiological activity was amplified (EX-1 Amplifier and 4001 headstage, DAGAN, USA), filtered (bandwidth: 300–5000), audio-monitored and recorded with data acquisition systems (Powerlab 2/26, AD instruments, Australia).
ST36 was located according to the positioning method of comparative anatomy. The position of ST36 was as follows: on the posterolateral knee, about 5 mm below the fibular head.
Functional classification of spinal neurons
Search stimuli consisted of mechanical stimulation (stroking the skin near ST36 and mild pinching with the experimenter's fingers). Each spinal neuron was characterised based on its response to graded intensities of mechanical stimulation applied to the RFs. Innocuous stimuli consisted of stroking the skin with a cotton swab. Noxious stimulation included mild pinching with the experimenter's fingers and with serrated forceps, but this latter stimulus was applied sparingly to avoid neuronal sensitisation. Neurons were classed functionally according to responses evoked by mechanical stimuli as (1) low threshold (LT) if they were excited maximally by innocuous stimulation, (2) WDR if they responded in a graded fashion to increasing intensity of stimulation and (3) high threshold (HT) if responses were evoked by noxious stimulation only.
Data acquisition and preprocessing
Extracellular recordings of WDR neurons (depth of 500–1000 µm, corresponding to lamina V–VI) were made with glass micropipettes. Based on clinical practice, an acupuncture needle was inserted into the skin and underlying muscles to a depth of 5 mm, and was lifted and thrusted manually approximately 0.5, 1, 2 or 3 times/s.
Each MA manipulation in lifting and thrusting type were applied for 120 s in duration in sequence (0.5 Hz→1 Hz→2 Hz→3 Hz). Manipulations were stopped at 5 min intervals between different patterns of MA to eliminate the effects of the previous MA. To maintain precision of neural encoding, the acupuncture needle was kept in the skin for the whole process of the experiment. Spike discharges of WDR neurons evoked by different MA manipulations were recorded in sequence.
Spikes were separated from the ‘biological’ noise by using the threshold method. The threshold for acceptance of a signal was set at 4σ (where σ was the SD of the quiescent signal). To diminish the interference of spikes, this estimation was based on the median value of the filtered signal.29 ,30
To provide an intuitive example of spike distribution, the left subplot in figure 1A was used for describing discharge activity of WDR neurons evoked by different MA manipulations, in which each line denoted a spike.
The recorded activity was viewed using Labchart V.6.0 (AD instruments, Australia), which extracted the timing of spike occurrence. These data were stored as ISIs. The ISI data were imported into Matlab V.8.0 for more detailed analysis.
The electrode may measure a different contribution from each of the different neurons around the microelectrode tip. Since the spike shape was unique and quite reproducible for each neuron, it could be used to distinguish spikes produced by different neurons; that is, to separate the activity produced by each neuron from background electrical noise. As shown in figure 1B, spike sorting algorithms used the shape(s) of waveforms collected with one or more electrodes to distinguish the activity of one or more neurons from background electrical noise.29
Approximate firing rates (rapprox)
Rate coding is a traditional coding scheme, in which information about the stimulus is contained in the firing rate of the neuron. Because the sequence of action potentials generated by a given stimulus varies from trial to trial, neuronal responses are typically treated statistically or probabilistically.
During rate coding, precise calculation of the firing rate is very important and different averaging procedures are used. In the present work, a time-window spike frequency analysis method was proposed to reflect the intensity changes of the discharges evoked by MA at different frequencies.
Figure 2 compares a number of ways of approximating firing rate from a spike sequence. Figure 2A shows a spike train from WDR neuron evoked by different MA manipulations at ST36; figure 2B shows a simple way of extracting an estimate of the firing rate from the spike train, which was to divide time into discrete bins of duration (Δt), count the number of spikes within each bin and divide by Δt=100 ms. The jagged curve in figure 2C shows the result of sliding a 100-ms wide window along the spike train. The firing rate approximated in this way could be expressed by sliding a rectangular window function along the spike train with Δt=100 ms. Figure 2D shows the approximate firing rate computed using a Gaussian window function with σw=0.08, which could generate a firing-rate estimate that was a smooth function of time.31 ,32
Temporal coding uses the internal time structure of spike trains to convey information of stimulus properties. The simplest temporal code is the ISI sequence, which is defined as the history of time intervals between consecutive spikes in chronological order.33
The ISI histogram and the joint-interval histogram were used to investigate the characteristics of temporal coding in spontaneous spike trains or spike trains generated by periodic stimulus.
The coefficient of variance (CV) of the ISI
A convenient and often used summary measure to quantify the firing variability of an ISI distribution is the CV, defined as the SD divided by the mean.34
Through reconstructing time series of the evoked discharges by the ISI method, variation analysis of ISI was used to analyse the ISI distributions for different acupuncture signals:
where is the mean ISI, σ0 is the SD of the ISI distribution and denotes averaging over k intervals. The mean firing rate is .
Statistical comparisons were carried out with SPSS using one-way analysis of variance (ANOVA) and post hoc tests. For all comparisons, p<0.05 was accepted as indicating significant differences.
General characteristics of WDR
Only WDR neurons without spontaneous activity were selected for study in this report. In all, 13 neurons with RFs located on ST36 were classified as WDR and were excited by innocuous and noxious mechanical stimuli. They responded in a graded fashion to increasing frequency of acupuncture stimulation.
Approximate firing rate of WDR neurons
As shown in figure 3, the approximate firing rate of WDR neurons increased with the increase of needling frequency.
The approximate firing rate for WDR neurons evoked by 0.5 Hz acupuncture stimuli was 0.355±0.130. The approximate firing rate of WDR neurons evoked by 1 Hz acupuncture stimuli was 54.1% greater than the response by 0.5 Hz acupuncture stimuli (p<0.05), whereas the mean response of WDR neurons to 2 and 3 Hz acupuncture stimuli increased 107.6% and 164.5% compared with the response by 0.5 Hz acupuncture stimuli (p<0.01).
ISI distribution of WDR neurons
Figure 4 illustrates the change of the ISI distribution for different MA manipulations. As shown in figure 4A, The peak of the ISI histogram occurred between 50 and 150 ms for all. The number of ISIs between 50 and 150 ms for 0.5 Hz acupuncture stimuli was 27, and the numbers for 1, 2 and 3 Hz acupuncture stimuli were 76, 125 and 146, respectively. This shows that an increase in neuronal activity followed almost immediately when the frequency of acupuncture stimulus was increased.
As shown in figure 4B, the firing pattern for 0.5 or 1 Hz acupuncture stimuli was characterised by a broad ISI histogram (2500 ms×2500 ms or 1500 ms×1500 ms), whereas the firing pattern for 2 or 3 Hz acupuncture stimuli was characterised by a narrow ISI histogram (800 ms×800 ms or 600 ms×600 ms).
CV of ISIs of WDR neurons
Investigation of the firing rate based on the spike counts is often accompanied by studying their variability. As shown in figure 5, with the increase of needling frequency, the CV of ISIs decreased.
The CV of ISIs for 0.5 Hz acupuncture stimuli was 0.931±0.108. The CV of ISIs for 1 Hz acupuncture stimuli was 19.8% less than the response to 0.5 Hz acupuncture stimuli (p<0.05), whereas the CV of ISIs for 2 and 3 Hz acupuncture stimuli decreased 37.6% and 34.8% compared with the response to 0.5 Hz acupuncture stimuli (p<0.01). However, when needling frequency exceeded 2 Hz, neuronal saturation phenomenon of the CV was observed.
There were three types of neuron in the upper lamina of the spinal dorsal horn: WDR, HT and LT.35 ,36 Previous research indicated that convergence of impulses originating from pain sites and acupuncture points occurs in the spinal dorsal horn and medial thalamus, where integration of two kinds of impulse takes place.16 A unique feature of WDR second-order neurons is that there is convergence of sensory information from the afferents from skin, muscle, viscera, tendons and joints.37 This convergence of sensory information opened the door to understanding how sensory stimulation with acupuncture could influence other visceral and somatic structures and function.38 ,39 Thus, WDR neurons without spontaneous activity were selected to explore the characteristics and features of neural discharges evoked by different MA manipulations in the present work.
The neural system characterises information in external stimulations by spatiotemporal encoding. Acupuncture, a form of external stimulus to the nervous system, is a somatic stimulation treatment that evokes various kinds of neural electrical signals due to the variation of the acupuncture manipulation and the non-linearity of the nervous system itself.25 ,28
Different MA manipulations induced the nervous system to generate or change neural electrical signals, which represent a spatiotemporal grouped sequence of input neuroinformation.26 ,40–43 Previous reports have demonstrated that acupuncture is an effective approach that regulated the human body through encoding external stimulation.22 ,44 However the relationship between the evoked discharges and MA manipulations is rarely investigated. We are still not clear on how the nervous system generates, codes and expresses neural electrical signals evoked by MA.
In the present work, rate coding and temporal coding of neural electrical signals evoked by different MA manipulations were studied based on firing rate and CV. Rate coding, first advocated by Adrian and Zotterman, held that the average number of spikes per time window carried information.45 In the present work, a time-window spike frequency analysis method was employed to calculate the number of spikes. Extracellular microelectrode recording results showed that WDR neurons responded progressively as needling frequency was increased.
Temporal coding provides a larger capacity for carrying information than pure ‘rate coding’ due to the abundant variability in the timing of spikes or ISI.46–48 Neurons responded to stimuli with temporally precise firing events and used precisely temporal spike patterns to encode information.49–51 Our results show that the CV of ISI decreased with the increase of needling frequency, which indicates that the firing variability increased with the increase of needling frequency. However, when needling frequency exceeded 2 Hz, neuronal saturation of the CV was observed, which had also been found in the spike discharges of primary afferents in dorsal root ganglion evoked by MA manipulations with different frequencies.22 ,52
The results above indicate that combining firing rate and ISI codes carried more information about the repetitive acupuncture stimulus. Neuronal firing rate and time sequences of ISI showed distinct clustering properties for different MA manipulations, which could distinguish different MA manipulations effectively. There are many hypotheses about the mechanism of MA manipulation on therapeutic effect, and an encoding method is one of the important factors. The regulatory effects of acupuncture could be impacted by various factors. Given that the acupuncture needle is a physical sensory stimulus, many important factors (such as the intensity, frequency, duration, interval between stimuli, acupuncture point selection and the physical condition of the subject) should directly influence the kind of receptors or peripheral nerve fibres activated.16 Increasing evidence has revealed that the types of afferent nerve fibres activated by acupuncture are diverse, depending upon the different manipulation methods of acupuncture and individual differences in acupuncture sensitisation.16 ,18
Most studies have proved that the action potential initiation dynamic may be an important mechanism of neuronal coding. The way that external stimuli (eg, acupuncture) change the action potential initiation dynamics mechanism of neurons has important biological physical meaning.53–55 Different MA manipulations might change the neural coding and result in different acupuncture efficacies by influencing the action potential initiation dynamics mechanism of neurons. Further investigation is needed to provide more evidence in support of this proposition.
Wide dynamic range (WDR) neurons in the dorsal horn relay information about acupuncture stimulation.
We analysed the response of WDR neurons to manual acupuncture stimulation at 0.5, 1, 2 and 3 Hz.
Information is coded by the WDR neurons in firing rate and interspike interval patterns.
We would like to thank the Institute of Traditional Chinese Medicine, Suzuka University of Medical Science.
Contributors IT and YG: made substantial contributions to conception and design; revised the article critically for important intellectual content. JW, C-XH and TZ: made contributions to acquisition, analysis and interpretation of data. TZ: drafted the article.
Funding This work was supported by the Natural Science Foundation of China (grant number 50537030, 81102642).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.