Download as pdf or txt
Download as pdf or txt
You are on page 1of 47

Accepted Manuscript

Effects of Short-Term Random Noise Electrical Stimulation on


Dissociated Pyramidal Neurons from the Cerebral Cortex

Leonardo Remedios, Pedro Mabil, Jorge Flores-Hernandez,


Oswaldo Torres-Ramírez, Nayeli Huidobro, Gerardo Castro,
Lucia Cervantes, Jesus A. Tapia, Braniff De la Torre Valdovinos,
Elias Manjarrez

PII: S0306-4522(19)30055-7
DOI: https://doi.org/10.1016/j.neuroscience.2019.01.035
Reference: NSC 18861
To appear in: Neuroscience
Received date: 25 June 2018
Accepted date: 21 January 2019

Please cite this article as: L. Remedios, P. Mabil, J. Flores-Hernandez, et al., Effects of
Short-Term Random Noise Electrical Stimulation on Dissociated Pyramidal Neurons from
the Cerebral Cortex, Neuroscience, https://doi.org/10.1016/j.neuroscience.2019.01.035

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 1

Effects of short-term random noise electrical


stimulation on dissociated pyramidal neurons from
the cerebral cortex
Leonardo Remedios3, Pedro Mabil1, Jorge Flores-Hernandez2, Oswaldo
Torres-Ramírez2, Nayeli Huidobro1, Gerardo Castro1, Lucia Cervantes3, Jesus
A Tapia4, Braniff De la Torre Valdovinos5, Elias Manjarrez1*
1
Laboratorio de Neurofisiología Integrativa

T
2
Laboratorio de Neuromodulación

IP
Instituto de Fisiología
3
Facultad de Cs. Físico-Matemáticas

CR
4
Escuela de Biología
Benemérita Universidad Autónoma de Puebla.
Puebla, Puebla, México.

US
5
Centro Universitario de Ciencias Exactas e Ingenierías, Universidad de
Guadalajara, Jalisco, México
AN
Running title: Effects of short-term noisy electrical stimulation on sodium currents
M

Number of words: Text 7923, abstract (248), introduction (629), discussion (828)
Number of text pages: 247
Number of figures: 9
ED

Corresponding Author:
Prof. Dr. Elías Manjarrez
PT

Instituto de Fisiología, Benemérita Universidad Autónoma de Puebla.


14 sur 6301, Col. San Manuel A.P. 406, C.P. 72570
Puebla, Pue., México
CE

Tels.: +5222-22-29-5500 Ext 7326, Fax: +5222-22-33-4511


eliasmanjarrez@gmail.com
AC

Acknowledgments:
The following grants supported this research: CONACyT Fronteras en la Ciencia #536
(E.M), Cátedra Moshinsky (E.M), F1-62610 (E.M) and CONACyT 229866 (E.M), and
VIEP-PIFI-FOMES-PROMEP-BUAP-Puebla (E.M), VIEP-BUAP Apoyos Especiales
(L.C.). We thank Lorena Arroyo for technical help to dissociate pyramidal neurons and
Manuel Alva for their participation in the adaptation of some figures. LR, PM and NH
contributed equally to the present article. We thank Robert Simpson for proof reading the
English manuscript
Key words: tRNS, electrical noise, somatosensory cortex, pyramidal neurons, stochastic
resonance.
Competing Financial Interests statement: The authors declare no competing financial
interest
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 2

Abstract

Transcranial random noise electrical stimulation (tRNS) of the human brain is a

non-invasive technique that can be employed to increase the excitability of the cerebral

cortex; however, the physiological mechanisms remain unclear. Here we report for the first

T
time the effects of short-term (250 ms) random noise electrical stimulation (RNS) on in-

IP
vitro acutely-isolated brain pyramidal neurons from the somatosensory and auditory

CR
cerebral cortex. We analyzed the correlation between the peak amplitude of the Na+ current

and its latency for different levels of RNS. We found three groups of neurons. The first

US
group exhibited a positive correlation, the second, a negative correlation, and the third
AN
group of neurons did not exhibit correlation. In the first group, both the peak amplitude of a

TTX-sensitive Na+ current and its inverse of latency followed similar inverted U-like
M

functions relative to the electrical RNS level. In this group, the RNS levels in which the
ED

maximal values of the inverted U-like functions occurred were the same. In the second

group, the maximal values of the inverted U-like functions occurred at different levels. In
PT

the third group, only the peak amplitude of the Na+ current exhibited a clear inverted U-like
CE

function, but the inverse of the latency versus the electrical RNS, did not exhibit a clear

inverted U-like function. A Hodgkin-Huxley neuron model reproduces our experimental


AC

results and shows that the observed behavior in the Na+ current could be due to the impact

of RNS on the kinetics of activation and inactivation of the Na+ channels.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 3

Introduction

Terney et al., 2008, found that the tRNS of the human brain induces strong

excitability increases lasting up to 60 min after stimulation. These authors suggested that

such phenomenon could be attributed to the repeated opening of Na+ channels (see also

Chaieb et al., 2015); however, there is a lack of knowledge of the underlying neuronal

T
mechanisms. The first logical step to understanding mechanisms at the cellular level is to

IP
analyze the impact of RNS on single neurons from the brain and in a time window in which

CR
the Na+ current occurs. In the literature, there is compelling evidence supporting this idea.

Bezrukov and Bodyanoy (1995, 1997), showed the existence of effects of electrical noise at

US
a sub-cellular level in artificial lipid bilayers. These authors added the polypeptide
AN
alamethicin to an artificial lipid bilayer to promote the formation of voltage-dependent

alamethicin ion channels. They found an inverted U-like shape profile of the current
M

magnitude throughout these channels as a function of the electrical noise level applied at
ED

the input. Gluckman and colleagues (1996) showed that noise electric fields (i.e., electrical

RNS) applied directly to hippocampal neurons in brain slices of the rat produced an
PT

increased extracellular electrical activity of the neurons. In such study, the signal and noise
CE

were delivered by the same electrodes (parallel nonpolarizing Ag-AgCl plates) and on the

whole brain slice. Subsequently, Stacey and Durand (2000) performed a more detailed
AC

experimental design in the same type of whole brain slice. They demonstrated that

multiunit recordings of CA1 neurons could increase their firing activity via synaptic inputs

in the hippocampus for a particular level of electrical noise. In a more recent study, Onorato

et al. (2016) performed the whole cell patch-clamp recordings in culture neurons from the

mouse dorsal root ganglia (DRG) and added noise (i.e., RNS) before or during depolarizing

current steps. They found that electrical RNS enhanced action potential firing in DRG
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 4

neurons and suggested that this increase occurred by the concurrent activation of voltage-

gated Na+ channels, via the stochastic resonance phenomenon. However, in this study, the

authors did not perform voltage-clamp experiments to elicit Na+ currents or the verification

that the Na+ currents were sensitive to TTX.

To our knowledge, there are no reports in the literature about the application of

T
electrical RNS on isolated cells from the brain. Possibly, because the main problem found

IP
to explore functional mechanisms at the cellular level, has been the difficulty of directly

CR
adjusting the noise level of neurons with continuous electrical noise. This issue has been a

challenging task since many years ago (Segundo et al., 1994; Faisal et al., 2008). The root

US
of the problem could be that the electrical stimulation in real biological neurons could
AN
produce ambiguous effects because it activates many types of ionic channels. Therefore, it

is necessary to introduce new experimental paradigms to activate ion channels with noise
M

selectively.
ED

The present study was aimed to examine the effects of short-term electrical RNS

(during 250 ms) on the Na+ currents of pyramidal neurons from the cerebral cortex and to
PT

explain the results with an H-H neuronal model. We employed the whole-cell technique to
CE

obtain voltage-clamp recordings of TTX sensitive Na+ currents in in-vitro acutely-isolated

brain pyramidal neurons of Wistar rats, with a blockage of K+ and Ca2+ channels. We
AC

examined the effects of short-term electrical noise applied to the voltage-clamp ramps on

the kinetics of the Na+ current. Our findings can be explained by a theoretical model

involving the kinetics of activation and inactivation of the Na+ channels. Our results will

help in the advancement and exploration of the physiological mechanisms of short-term

RNS. The results obtained from our study will be of interest to understand the mechanisms

of tRNS in the human brain (Antal and Hermann, 2016; Fertonani and Minussi, 2016).
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 5

Methods

Experiments were performed in 7 Wistar rats (mean weight 100-150 g). The animals

were kept in light and temperature controlled rooms (lights on at 6 a.m. and lights off at 6

p.m.) with free access to food and water. Experiments were performed following the

European Communities Council Directive of 24-November-1986 (86/609/EEC), the

T
guidelines contained in the “Norma-Oficial-Mexicana-NOM-062-ZOO-1999” and the

IP
National Institutes of Health Guide for the Care and Use of Laboratory Animals (85–23,

CR
revised in 1985). Wistar rats raised in the animal facility from Benemérita Universidad

Autónoma de Puebla (BUAP), México. The experimental protocols were approved by the

US
ethics committee (CICUAL-Proyecto-00489) from BUAP.
AN
Isolated pyramidal neurons

We obtained recordings of the Na+ current of 34 isolated pyramidal neurons from


M

the cerebral cortex. Pyramidal neurons from the primary somatosensory (SN, n=16) and
ED

auditory cortex (AN, n=18) were acutely isolated following methods previously described

(Bargas et al., 1994; Flores-Hernandez et al., 2002). The rats were anesthetized with
PT

halothane and decapitated. Their brains were dissected and placed in a cold solution of
CE

isethionate with low calcium (in mM): 140 sodium isethionate, 2 KCl, 2 MgCl2, 0.1 CaCl2,

23 glucose, 15 HEPES, pH = 7.4, 300-310 mOsm / L. Also were supplemented with (in
AC

mM): 1 pyruvic acid, 0.005 glutathione, 0.1 NG-nitro-L-arginine, 1 kynurenic acid, gassed

with O2. Subsequently, coronal slices of 350 m from the barrel somatosensory or auditory

cortex were obtained using a commercial vibratome (Campden). Slices were kept between

1 to 6 hours at room temperature (20-22 °C) in Earle's balanced salt solution (EBSS,

SIGMA), buffered with sodium bicarbonate (NaHCO3). They were supplemented with (in
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 6

mM): 1 pyruvic acid, 0.005 glutathione, 0.1 NG-nitro-L-arginine, one kynurenic acid;

gassed with 95 % O2 / 5 % CO2, pH = 7.4 adjusted with NaOH, 300-310 mOsm/l.

After at least 1 hour of incubation, every slice was placed in enzymatic digestion.

We employed a growth chamber with 40 ml of Hanks balanced salt solution (HBSS,

Sigma) buffered with HEPES bubbled O2 containing 0.375 mg/ml of Papain (Calbiochem)

T
at 35 °C for 15 minutes. It was supplemented with (in mM): 1 pyruvic acid, 0.005

IP
glutathione, 0.1 NG-nitro-L-arginine, 1 kynurenic acid. After the enzymatic digestion, the

CR
tissue was washed with cold solution isethionate and mechanically dissociated with several

Pasteur pipettes of different calibers (fire polished). With this procedure, we obtained a cell

US
suspension with the acutely isolated neurons. A sample of such cell suspension was placed
AN
in a Petri dish (35mm polystyrene; Nunclon Surface, NUNC) and observed with an

inverted microscope to identify live isolated pyramidal-cells. Figure 1A shows two


M

pyramidal neurons from the auditory cortex (labeled with AN) and the somatosensory
ED

cortex (labeled with SN). After 10 minutes of incubation the suspension was washed with a

solution of background (in mM): 140 NaCl, 23 glucose, 15 HEPES, 2 KCl, 2 MgCl2, 1
PT

CaCl2, and 1% phenol red, gassed with O2, pH = 7.4 adjusted with NaOH, 300-310
CE

mOsm/l. Finally, whole-cell voltage clamp recordings were obtained from the selected

neurons. We employed a voltage clamp amplifier Multiclamp 700A (Molecular Devices,


AC

Axon Instruments Foster City, CA), controlled by the pClamp-10 (Axon Instruments)

software, running on a PC with a Digidata 1440A (Molecular Devices, Axon Instruments).

The resistance of the electrodes in the bath was 4-8MΩ. After breaking the seal, we

selected cells when the access resistance (series resistance) was equal or less than 30 MΩ.

The internal solution for the whole cell recordings of Ca2+ and Na+ currents was (in mM):

175 N-methyl-D-glutamine (NMDG), 40 HEPES, 2 MgCl2, 10 acid ethylene glycol-bis (β-


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 7

aminoethyl ether) -N,N,N',N'-tetraacetic acid (EGTA), 12 phosphocreatine, 3 Na2ATP,

0.35 Na3GTP, 0.1 leupeptin, pH = 7.3 with H2SO4, 265- 270 mOsm / L. The external

solution was composed of (in mM): 127 NaCl, 20 CsCl, 5 BaCl2, 2 CaCl2, 12 glucose, 10-

HEPES, ph = 7.3 with NaOH, 300-305 mOsm / L.

Drugs were applied via the SF77B system (Instruments Warner Company) directly

T
to the surface of the recorded neuron. We performed the experiments in a solution free of

IP
TTX, but we blocked Ca+ and K+ channels using the application of 300 µM of cadmium, 20

CR
mM of Cesium and five mM of Barium. To verify that we recorded TTX-sensitive Na+

currents we applied 300 nM of TTX (Sigma Aldrich) at the end of the experiment.

US
Details of whole-cell patch clamp ramps on pyramidal cells
AN
The whole-cell recorded neurons were stimulated with six groups of 10 voltage-

clamp ramps of 100 ms duration, from -100 to +40 mV, with a holding potential of -80 mV
M

(Figure 1B). The pre-pulse duration from -80 to -100 mV was 125 ms. At the end of each
ED

ramp, the voltage returns to the holding potential. The voltage-clamp ramps have the

advantage of generating voltage-current relations directly. Therefore, they are suitable for
PT

the study of voltage-dependent activation latency.


CE

Details of the electrical noise applied to pyramidal cells of Wistar rats

Pyramidal neurons were stimulated with short-term electrical RNS, which consisted
AC

of random fluctuations around the voltage-clamp profiles during a “short-term” of 250 ms.

This RNS was applied throughout the same glass microelectrode. The experimental

paradigm of electrical RNS is illustrated in the right panel of Figure 1A. Note that the RNS

was continuously applied during the complete protocol of voltage-clamp ramps (Figure

1A). For clarity, the second set of traces in the right panel of Figure 1A show the voltage

noise that was summated to the protocol of voltage-clamp ramps showed above. The power
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 8

spectrum of this RNS was Brownian in the frequency range from 0 to 5000 Hz (Huidobro

et al., 2017; 2018).

Stimulation protocol on the pyramidal cells

The stimulation protocol consisted of a series of six voltage-clamp ramps for every

level of RNS (Figure 1A). We applied zero-noise of RNS (control, Z-RNS). We also

T
applied five other different levels of RNS. For illustrative purposes, in Figure 1A, we

IP
illustrate first the zero noise level Z RNS, second the noise level RNS1, third the noise

CR
level RNS2, fourth the noise level RNS3, fifth the noise level RNS4, and sixth the noise

level RNS5. Every level of RNS was continuously applied during all the patch clamp

US
protocol of such six voltage-clamp ramps. The six RNS levels were presented in a pseudo-
AN
randomized fashion. Furthermore, to avoid adaptation, rest intervals of 3 s were included

between the RNS levels.


M

Peak amplitude and latency of the Na+ current


ED

We computed the peak amplitude of the Na+ current for all RNS levels. Because the

different levels of RNS produced changes in the latency of the Na+ current, we also used
PT

this variable to analyze the effects of RNS on the recorded pyramidal cells. We measured
CE

such latency from the beginning of the ramp to the peak of the Na+ current as illustrated in

Figure 1B. This measure was used to quantify the inverse of the latency of the Na+ current
AC

peak.

Statistical analysis

To test for any statistical difference in the peak amplitude of the sodium current and

the inverse of latency of the Na+ current, we compared them in the following conditions: Z-

RNS vs. RNS1, Z-RNS vs. RNS2, Z-RNS vs. RNS3, Z-RNS vs. RNS4, Z-RNS vs. RNS5.

We used the non-parametric pairwise Signed-Rank Tests to examine the statistical


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 9

significance of the peak amplitude of Na+ current or the inverse of its latency. We

compared the abovementioned conditions of RNS, in all the 34 neurons, under the null

hypothesis that the differences in the means between such conditions were zero. Due to the

multiple comparisons, we used a corrected Bonferroni adjustment. We also performed an

analysis of linear regression, and we calculated the Pearson’s correlation and its

T
significance. Error bars indicate standard deviation. The comparisons were considered as

IP
significant if p < 0.05. Data and detailed methods are available from the authors upon

CR
request.

Results

Experimental results
US
AN
We analyzed the effects of RNS on the TTX-sensitive Na+ currents of 34 pyramidal

cells (18 from the primary auditory cerebral cortex and other 16 from the primary
M

somatosensory cortex). We verified that all the neurons exhibited a pyramidal shape
ED

irrespective of its size as illustrated in the pictures of Figure 1A. In particular, we analyzed

the effects of RNS on the peak amplitude and latency of Na+ currents recorded in these
PT

dissociated cells. We recorded the Na+ current from auditory and somatosensory pyramidal-
CE

neurons as illustrated in Figure 1A. The lower right panels of Figure 1A show the effects of

six levels of electrical RNS (Z-RNS, RNS1, RNS2, RNS3, RNS4, RNS5) on the peak
AC

amplitude of the Na+ current elicited by the voltage-clamp ramp as indicated (Z is for zero

noise). Figure 1C shows superimposed recordings of Na+ currents for these neurons during

the Z RNS (green trace), optimal (O) RNS (red trace) and high (H) RNS (blue trace). Note

how the Na+ current peak amplitude and its latency change by the addition of such levels of

RNS. Note that for these selected neurons, the application of an optimal level of noise

produces an increase in the Na+ current peak and a concomitant reduction in its latency (see
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 10

traces in red color in the lower panel of Figure 1C). However, we found that not always it

was the rule. The correlation between the peak amplitude of the Na+ current and its inverse

of latency for the different levels of RNS exhibited different profiles. Therefore, we

analyzed in detail such correlation. Surprisingly, this correlation analysis served as a

grouping criterion. We found that all the neurons could be classified into three different

T
groups. The first group, exhibited a positive correlation, the second, a negative correlation,

IP
and the third group of neurons did not exhibit correlation. In Figure 2, we illustrate

CR
examples of graphs for the Na+ peak amplitude versus its latency for 15 pyramidal neurons.

Each point corresponds to one level of RNS. We are also showing the correlation

US
coefficients r for these 15 pyramidal cells. All the 34 neurons were classified into these
AN
three groups (see Table 1 for the Pearson’s correlation analysis). The sign of the Pearson’s

correlation coefficient (r) and the value of its significance, between the “peak amplitude of
M

the Na+ current” and the “inverse of latency,” were used as the grouping criteria. Group 1
ED

(+ sign and/or p<0.05, termed positive correlation, i.e., +r), Group 2 (- sign and/or p<0.05,

termed negative correlation, i.e., -r), Group 3 (+ or – sign but p>0.2, termed no correlation).
PT

Table 1 shows that there was a significant positive Pearson’s correlation (+r,
CE

p<0.05, *) between the “peak amplitude of the Na+ current” and the “inverse of latency” for

six neurons (AN-13, AN-14, SN-12, SN-15, SN-16, and SN-24). Moreover, for ten neurons
AC

(AN-07, AN-09, AN-15, AN-18, AN-24, SN-09, SN-11, SN-13, SN-14, and SN-23), there

was a significant negative Pearson’s correlation (-r, p<0.05, *) between the “peak

amplitude of the Na+ current” and the “inverse of latency.” Table 1 also shows ten neurons

with a non-significant Pearson’s correlation (-r or +r, p>0.2, NS) between the “peak

amplitude of the Na+ current” and the “inverse of latency” (AN-03, AN-08, AN-12, AN-16,

AN-17, AN-22, AN-23, SN-05, SN-20, and SN-22).


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 11

Following the grouping criteria illustrated in Figure 2, we grouped the graphs of the

peak amplitude of the Na+ current and its inverse of latency versus the electrical RNS level

for all the pyramidal neurons. Figure 3 shows examples of grouped graphs (three neurons

per group) for such peak amplitude and its inverse of latency versus the electrical RNS

level. All the 34 neurons were successfully grouped as in the examples of Figure 3. In

T
Figure 5, we show pooling data for all these neurons separated into the three groups

IP
(detailed explanation afterward).

CR
In the first group (positive correlation), both the peak amplitude of the Na+ current

and its inverse of latency followed similar profiles of inverted U-like functions relative to

US
the electrical RNS level (see Figure 3A). In this group, the maximal values of these
AN
inverted U-like functions occurred at the same RNS levels (see red circles of the

experimental data in Figure 3A). In the second group (negative correlation), both the peak
M

amplitude of a TTX-sensitive Na+ current and its inverse of latency followed different
ED

profiles of inverted U-like functions relative to the electrical RNS level (see Figure 3B). In

this group, the maximal values of these inverted U-like functions occurred at different RNS
PT

levels (see red circles of the experimental data in Figure 3B). In the third group (no
CE

correlation), only the peak amplitude of the Na+ current exhibited a clear inverted U-like

function, but the inverse of latency versus the electrical RNS did not exhibit a clear inverted
AC

U-like function (see red circles of the experimental data in Figure 3C). In this group, the

maximal values of the functions occurred at different RNS levels (see red circles of the

experimental data in Figure 3C).

To understand in more detail the changes in amplitude and latency, we show in

Figure 4 a zoom of the Na+ current peaks for representative pyramidal neurons (PN) of the

three groups, recorded in the three conditions, zero, optimal and high electrical RNS. Note
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 12

that the peak amplitude for the ON condition (red traces) is larger than the peak amplitude

associated with the ZN (green traces) or HN (blue traces) conditions. Figure 4A illustrates

Na+ current recordings for the first group (positive correlation). It is clear that both the

amplitude and latency are positively correlated. Both the amplitude and the inverse of

latency are increased for the optimal level of RNS (red traces). Note that the latency of red

T
traces is lower than the latency for the control ZN condition (green traces). This indicates

IP
that there is an optimal level of RNS for which the latency of the Na+ current is reduced

CR
(i.e., its inverse is increased). However, Figure 4B (for the second group, negative

correlation) and Figure 4C (for the third group, no correlation), show that the latency of

US
red traces (optimal response in amplitude) is not as short as the latency in Figure 4A.
AN
Figures 5A, 5B and 5C show pooled data for the amplitude and inverse of latency

of the Na+ current, respectively, for all the auditory and somatosensory pyramidal neurons
M

for the three groups (positive correlation, negative correlation, and no correlation). Note
ED

that the peak amplitude of the Na+ current and the inverse of latency follow an inverted U-

like shape as a function of the RNS level for all the pyramidal neurons from the first and
PT

second groups. However, for the third group (no correlation), only the peak amplitude of
CE

the Na+ current exhibits the inverted U-like shape as a function of the RNS level.

We can compare Figures 5A, 5B and 5C. Note that in Figure 5A (positive
AC

correlation), the maximal values for the peak amplitude and inverse of latency occur at the

same level of RNS (see red circles). In contrast, in Figure 5B (negative correlation) and

Figure 5C (no correlation), the maximal values for the peak amplitude and inverse of

latency occurs at different levels of RNS (see red circles). Furthermore, there are other clear

differences among the three groups. First, the group with a positive correlation (Figure 5A)

exhibits a similar shape in the inverted U-like functions for both the amplitude and the
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 13

inverse of latency of the Na+ current versus the level of RNS. Second, the group with

negative correlation (Figure 5B) exhibits different shapes in the inverted U-like functions

for both the amplitude and the inverse of latency versus the RNS level. Third, the group of

no correlation (Figure 5C) does not exhibit an inverted U-like function in the inverse of

latency versus the RNS level.

T
In the data of Figure 5, the non-parametric pairwise Signed-Rank Tests also

IP
uncovered significant differences between “zero RNS” versus “optimal RNS” and “optimal

CR
RNS” versus “high RNS” for the amplitude of the Na+ current from all the groups, but no

for the inverse of latency from the third group. The p values are indicated in each graph.

The model
US
AN
We modeled the effect of noise over Na+ current under different voltage-clamp

conditions following the function V(t):


M
ED

−80 𝑖𝑓 0 ≤ 𝑡 ≤ 125
PT

−100 𝑖𝑓 125 < 𝑡 ≤ 250


𝑉(𝑡) = { (1)
1.4𝑡 − 600 𝑖𝑓 250 < 𝑡 ≤ 350
−80 𝑖𝑓 350 < 𝑡 ≤ 500
CE
AC

The electrical noise was modeled with the function:

𝑓(𝑥𝑚𝑒𝑎𝑛 , 𝑡) = 𝑎𝑖 ∗ (𝑠𝑖𝑛(𝜋 ∗ 𝑡)) 𝑖𝑓 2𝑖 ≤ 𝑡 ≤ 2𝑖+1 (2)

Where 𝑎𝑖 is a random number (𝑖 ∈ {0,1, … ,250}) following a normal distribution with a

mean equal to 𝑥𝑚𝑒𝑎𝑛 and variance 𝜎.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 14

Therefore, the voltage function is related to the ramp stimulus V(t) and noise 𝑓(𝑥𝑚𝑒𝑎𝑛 , 𝑡):

𝑉𝑠𝑡𝑖𝑚 = 𝑉(𝑡) + 𝑓(𝑥𝑚𝑒𝑎𝑛 , 𝑡) (3)

T
The differential equations for the activation and inactivation of the Na+ current are:

IP
CR
𝑑𝑚
= 𝛼𝑚 (𝑉𝑠𝑡𝑖𝑚 )(1 − 𝑚) − 𝛽𝑚 (𝑉𝑠𝑡𝑖𝑚 )𝑚 (4)
𝑑𝑡

US
and

𝑑ℎ
AN
= 𝛼ℎ (𝑉𝑠𝑡𝑖𝑚 )(1 − ℎ) − 𝛽ℎ (𝑉𝑠𝑡𝑖𝑚 )ℎ (5)
𝑑𝑡
M
ED

where the functions αm,  m, h, and  h were modified from the Hodgkin-Huxley model,

including the effect of noise with the functions 𝑉𝑠𝑡𝑖𝑚 , p(𝑥𝑚𝑒𝑎𝑛 ) and q(𝑥𝑚𝑒𝑎𝑛 ):
PT
CE

𝜂𝑚
𝛼𝑚 (𝑉𝑠𝑡𝑖𝑚 ) = exp [0.5 (𝑉 − 𝑝(𝑥𝑚𝑒𝑎𝑛 ))] (6)
𝑣𝑇 𝑠𝑡𝑖𝑚
AC

𝜂𝑚
𝛽𝑚 (𝑉𝑠𝑡𝑖𝑚 ) = exp [−0.5 (𝑉 − 𝑝(𝑥𝑚𝑒𝑎𝑛 ))] (7)
𝑣𝑇 𝑠𝑡𝑖𝑚

𝜂ℎ
𝛼ℎ (𝑉𝑠𝑡𝑖𝑚 ) = exp [0.5 (𝑉 − 𝑞(𝑥𝑚𝑒𝑎𝑛 ))] (8)
𝑣𝑇 𝑠𝑡𝑖𝑚

𝜂ℎ
𝛽ℎ (𝑉𝑠𝑡𝑖𝑚 ) = exp [−0.5 (𝑉 − 𝑞(𝑥𝑚𝑒𝑎𝑛 ))] (9)
𝑣𝑇 𝑠𝑡𝑖𝑚
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 15

𝛼𝑚 𝛼
𝑚∞ (𝑉) = ⁄(𝛼 + 𝛽 ) and ℎ∞ (𝑉) = ℎ⁄(𝛼 + 𝛽 ) are sigmoidal functions
𝑚 𝑚 ℎ ℎ

representing the steady state for the activation and inactivation variables experimentally

obtained (see equations 6-9). Where 𝜂𝑚 and 𝜂ℎ are the slopes of these sigmoidal functions

and 𝑣𝑇 is the Boltzmann potential.

Here (𝑝(𝑥𝑚𝑒𝑎𝑛 ), 0.5) and (𝑞(𝑥𝑚𝑒𝑎𝑛 ), 0.5) are the inflection points of 𝑚∞ (𝑉) and

T
ℎ∞ (𝑉) respectively. In our model, we assume that the noise modifies the kinetics of the Na

IP
channels throughout an impact on these inflection points. Here we represent this

CR
modification of the inflection points horizontally scanning the voltages that determine the

US
inflection points as a function of noise.

Here 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) are polynomial functions depending of the modeled
AN
group of neurons.
M

For the groups 1 and 2:


ED

𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2 (10)


PT

and

𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2 (11)


CE
AC

Here we assumed b2>0 and c2>0 for group 1 (i.e., positive correlation, Pearson’s correlation

r>0). Whereas b2<0 and c2<0 for group 2 (negative correlation, Pearson’s correlation r<0).

For group 3 (no correlation) the polynomials were of fourth order:

𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑏3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑏4 𝑥𝑚𝑒𝑎𝑛 4 (12)


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 16

and

𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑐3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑐4 𝑥𝑚𝑒𝑎𝑛 4 (13)

Therefore, we describe the Na+ current as follows:

T
𝐼𝑁𝑎 = 𝑔𝑁𝑎 𝑅(𝑥𝑚𝑒𝑎𝑛 ) 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 )) ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))(𝑉𝑠𝑡𝑖𝑚 − 𝑉𝑁𝑎 ) (14)

IP
CR
Where 𝑔𝑁𝑎 is the maximal Na+ conductance, 𝑅(𝑥𝑚𝑒𝑎𝑛 ) is the maximal percentage

of recruited channels as a function of noise, 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 )) and ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))

US
are the activation and inactivation variables for the Na+ current as a function of noise, and
AN
𝑉𝑁𝑎 is the reversal potential for the Na+ current.

For the groups 1 and 2 the function 𝑅(𝑥𝑚𝑒𝑎𝑛 ) is represented by:


M
ED

𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2 (15)


PT

Here we assumed a2>0 for the groups 1 and 2, moreover, for group 3, the function
CE

𝑅(𝑥𝑚𝑒𝑎𝑛 ) is a polynomial of fourth order:


AC

𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑎3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑎4 𝑥𝑚𝑒𝑎𝑛 4 (16)

These assumptions allowed us to predict that the impact of noise on the Na+

channels differentially amplifies the Na+ currents, depending on the activation and

inactivation gates and the number of recruited channels.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 17

First, we reproduced the waveform of the Na+ currents, including the characteristics

for the latency and amplitude. To guarantee appropriate modeling for the Na+ current, we

chose parameter values with a precision larger than 99 %. It means that the error was

measured as the maximum of the differences between the experimental data and its

corresponding simulated data: max │dexp – dsim│≤ c. In this form, when the maximal

T
amplitude of the current was c = 1 pA, then the error was less than 0.02 % (i.e., with a

IP
precision of 99.98 %). Moreover, when the inverse of latency was c = 1*e-4 (ms)-1, then the

CR
estimated error was less than 0.06 %, and the precision was 99.4 %. Table 1 also shows

these errors for all the modeled pyramidal neurons.

US
Second, to reproduce the inverted-U profile elicited by the input noise we employed
AN
the functions 𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) where 𝑥𝑚𝑒𝑎𝑛 is the mean value of the

noise amplitude. Note that in equation (14), we incorporated the electrical noise in the
M

functions R, m and h. In this form, we guarantee that the added electrical noise is producing
ED

an impact on the kinetics of the Na+ channels.

Furthermore, we observed that modifying 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) we can modify


PT

the inverse of latency of the Na+ current in the model due to a change in the activation and
CE

inactivation parameters of the Na+ channels. Modifying 𝑅(𝑥𝑚𝑒𝑎𝑛 ) in the model we can

modify the peak amplitude of the Na+ current due to a change in the maximal percentage of
AC

recruited channels. These observations were useful to explain the mechanisms and grouping

criteria for our experimental results.

In our experiments, we found that the variables “peak amplitude” and “inverse of

latency” of the Na+ current versus the RNS level for each neuron can be classified in three

groups based on the correlation between both variables. Therefore, we examined the class
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 18

of simplified functions for 𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) that can reproduce this

grouping in our experimental findings. We assumed three possible correlations between the

“peak amplitude” versus the “inverse of latency” of the modeled Na+ current: positive

correlation, negative correlation and no correlation. We also assumed that the best

𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) polynomials must be parabolic functions of 𝑥𝑚𝑒𝑎𝑛 as

T
illustrated in Figure 7.

IP
Moreover, for the case of “no correlation”, we assumed that the best 𝑅(𝑥𝑚𝑒𝑎𝑛 ),

CR
𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) polynomials must be functions of fourth order of 𝑥𝑚𝑒𝑎𝑛 .

US
We successfully reproduced the experimental results with our model. For example,

in Figure 6 we show that a specific selection of parameters 𝑎𝑖 and 𝑏𝑖 in the polynomial


AN
functions (10) to (13) can reproduce the effects of short term electrical noise on the

amplitude and latency of the Na+ current. We assumed that the polynomial coefficients in
M

the functions are related to the intrinsic neuronal properties, and that they are different for
ED

each neuron. Table 2 shows polynomial functions that reproduce our experimental results.

With the polynomials of 𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) showed in equations


PT

(10) to (13), we also successfully modeled the three types of correlations (see Figures
CE

7DHL) between the “peak amplitude” versus the “inverse of latency” of the modeled Na+

current: positive correlation, negative correlation and no correlation. The red points in
AC

Figure 7 show the results of 5 calculations obtained with the model. Note that in the model

we are going from the hypothetic profiles of the 𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 )

polynomials (Figures 7ABC, 7EFG and 7IJK) to the predicted grouping of correlations

between the peak amplitude and inverse of latency of the Na+ current (Figures 7DHL). The

rationale of this procedure (see arrow in Figure 7) is that we can perform the inverse
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 19

procedure; i.e., going from the experimental data to the predicted values for the 𝑅(𝑥𝑚𝑒𝑎𝑛 ),

𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) polynomials that could explain plausible physiological

mechanisms (see arrow in Figure 8). Figures 8AEI illustrate experimental data using our

grouping criterion. We found that our model was capable to predict the expected profiles of

the 𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) polynomials (see Figures 8BCD, 8FGH and

T
8JKL).

IP
Finally, we also successfully reproduced the experimental results for the inverted U-

CR
like functions depicted in Figure 3. The gray traces in Figure 3 represent the results

US
obtained from the model, and the black lines, correspond to the experimental data. Note the

close resemblance between the experimental and theoretical superimposed curves for the
AN
three groups of neurons. We found that an appropriated selection, of the polynomials

𝑅(𝑥𝑚𝑒𝑎𝑛 ), 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) and parameters 𝑎𝑖 , 𝑏𝑖 and 𝑐𝑖 , allows a good


M

reproducibility of the experimental data. Note that for the groups “positive correlation” and
ED

“negative correlation”, both the experimental and theoretical results show that the Na+

current amplitude is maximal for an intermediate level of input RNS. However, for the
PT

group of “no correlation” the inverse of latency of the Na+ current does not exhibit such
CE

phenomenon. Table 2 shows the appropriate values for the selected 𝑎𝑖 , 𝑏𝑖 and 𝑐𝑖 parameters

for which the modeling results exhibit a close resemblance with the experimental curves.
AC

To illustrate that the Na+ current recorded in our experiments was sensitive to TTX

we are showing in Figure 9 how the application of 300 nM of TTX abolished this current

during the application of the different levels of RNS.

Discussion
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 20

We found that for all the recorded pyramidal neurons the amplitude of a TTX-

sensitive Na+ inward current followed an inverted U-like shape as a function of the short-

term electrical RNS level. Moreover, we observed that the electrical RNS produced a

modulation in the inverse of the latency of the Na+ current: 1) following an inverted U-like

shape as a function of the electrical RNS for neurons of the group 1 or 2, or 2) following an

T
arbitrary profile as a function of the electrical RNS for neurons from the group 3.

IP
Interestingly, the effects of RNS on the peak amplitude and inverse of latency of the Na+

CR
current were successfully explained by our H-H-like model and separated into three groups

(Figure 3). These results were predicted with our model, showing that the electrical effect

US
of RNS on the Na+ current (the inverted U-like shape) is not an electrical artifact. We found
AN
that the RNS has an impact on the activation and inactivation parameters of the Na+

channels. It means that the RNS exerts a modulation on the activation or inactivation gates
M

of the Na+ channels. It reinforces the hypothesis that the electrical RNS could induce
ED

facilitation of the TTX-sensitive Na+ current via an excitability increase at an intermediate

level of noise.
PT

The studies by Bezrukov and Bodyanoy (1995, 1997) suggested the existence of
CE

stochastic resonance at a sub-cellular level in artificial lipid bilayers and opened the

question whether this phenomenon could occur in isolated cells from the brain. Our results
AC

respond to such a question. Our results are also consistent with previous studies in humans

made by Chaieb et al. (2015), who showed that transcranial random noise stimulation-

induced plasticity is NMDA-receptor independent but sodium-channel blocker and

benzodiazepines sensitive. Our findings are also consistent with previous studies in which

the electrical tRNS increased brain excitability, enhanced perception and learning in

humans (Terney et al., 2008; Van der Groen et al., 2016; Van Koningsbruggen et al., 2016;
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 21

Romanska et al., 2015; Laczo et al., 2014; Van Doren et al., 2014; Pasqualotto, 2016;

Popescu et al., 2016; Penton et al., 2017; Looi et al., 2017; Rufener et al., 2017;

Mammarella et al., 2017; Yang and Banissy, 2017; Antal et al., 2017).

We applied RNS instead of other stimulation patterns, first, because we wanted to

explain the results obtained by Terney et al. (2008), and second, because there is

T
compelling evidence that the tRNS stimulation exceeds the beneficial influences of the

IP
transcranial Direct Current Stimulation (tDCS) (Fertonani et al., 2011, Jausovec and Pahor,

CR
2017).

Our model shows that the modulation of the inverse of latency and peak amplitude

US
of the Na+ inward current by RNS could occur by a noisy disturbance in the kinetics of
AN
activation and inactivation of Na+ channels, and not only by the impact of RNS in the

transmembrane potential. The underlying reason for the postulated differences in activation
M

and inactivation variable/functions is that the noise produces an impact on the kinetics of
ED

the Na+ current depending on possible intrinsic properties. We modeled such properties of

the Na+ channels by parameters in the polynomials. This means that the RNS acts on the
PT

activation and inactivation gates of the three groups of cells in a different form. For
CE

example, in neurons from group 1, the RNS produces a parabolic decrease in the

parameters 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞(𝑥𝑚𝑒𝑎𝑛 ) associated with the activation and inactivation gates (see
AC

Figure 7 and equations in the model section). However, for the group 2, the RNS produces

a parabolic increase in the parameters 𝑝(𝑥𝑚𝑒𝑎𝑛 ) and 𝑞 (𝑥𝑚𝑒𝑎𝑛 ) associated with these gates

(see Figure 7 and equations in the model section). In the model, we used a2>0, b2>0 and

c2>0 for group 1 and a2>0, b2<0 and c2<0 for group 2 (see formulas (10), (11) and (15) in

the model section). For group 3, we used polynomials of fourth order (see formulas (12),

(13) and (16) in the model section). As illustrated in Figure 7, our model was capable to
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 22

predict three groups in the experimental relationship between the amplitude and inverse of

latency of the Na+ currents (see Figures 7DHL). Such finding was useful to obtain

meaningful conclusions. On the other hand, our model shows that a change in the activation

and inactivation variables produces a significant change in the amplitude and latency of the

Na+ current. This type of procedure, in which the “activation” or “inactivation” variables

T
were altered, has been employed in other modeling studies (Borg-Graham, 1999). In

IP
particular, Traub et al., (1991, 1994) and Poirazi et al., (2003) developed models of

CR
pyramidal neurons from the hippocampus, in which they changed the typical values of the

power in the activation and inactivation variables of the Na+ current.

US
Our study could motivate the implementation of other more realistic biophysical
AN
models using neuronal networks, which could include the dynamics of ion channel

populations, the traditional mechanisms like the ball and chain model (Armstrong and
M

Bezanilla, 1977), or the time-delayed self-feedback in pyramidal neurons called autapse


ED

(Yilmaz et al., 2016), among other assumptions.

We conclude that an appropriate intermediate-level of RNS can increase the peak


PT

amplitude of the TTX-sensitive Na+ current and its inverse of latency throughout its
CE

“parabolic action” on the activation or inactivation gates of the Na+ channel. The term

“parabolic action” means that the RNS can exert its effects on both gates following a
AC

dynamics like a parabolic function. In other words, there is an intermediate level of RNS

that enhances the activation or inactivation processes occurring in the Na+ channels of the

pyramidal neurons. Our model provides compelling evidence for differences in the impact

of noise on three groups of pyramidal cells. It is important to mention that this is a first

modeling study that could be improved including other variables associated with the
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 23

intrinsic properties of the Na+ channels. Some of these properties could include the channel

phosphorylation, accessory subunits of Na+ channels, membrane fluidity, among others.

Our findings shed light on the possible physiological process related to the

successful amplification of Na+ currents during the application of short-term RNS. These

findings could be necessary for the design of new stimulation protocols with short-term

T
tRNS in humans.

IP
References

CR
Antal A, Herrmann CS (2016) Transcranial Alternating Current and Random Noise

Stimulation: Possible Mechanisms. Neural Plast 2016.

US
Antal A, Alekseichuk I, Bikson M, Brockmöller J, Brunoni AR, Chen R, Cohen LG,
AN
Dowthwaite G, et al. (2017) Low intensity transcranial electric stimulation: Safety, ethical,

legal regulatory and application guidelines. Clin Neurophysiol 128:1774-1809.


M

Armstrong CM, Bezanilla F (1977) Inactivation of the sodium channel. II. Gating current
ED

experiments. J of Gen Physiol 70: 567–590.

Bargas J, Howe A, Eberwine, J, Cao Y, Surmeier DJ (1994) Cellular and molecular


PT

characterization of Ca2+ currents in acutely isolated, adult rat neostriatal neurons. J


CE

Neurosci 14:6667–6686.

Bezrukov SM, Vodyanoy I (1995) Noise-induced enhancement of signal transduction


AC

across voltage-dependent ion channels. Nature 378:362-364.

Bezrukov SM, Vodyanoy I (1997) Signal transduction across alamethicin ion channels in

the presence of noise. Biophys J 73:2456-2464.

Borg-Graham LJ (1999) Interpretations of data and mechanisms for hippocampal pyramidal

cell models. In: Models of Cortical Circuits, pp. 19-138. Springer Boston MA.
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 24

Chaieb L, Antal A, Paulus W (2015) Transcranial random noise stimulation-induced

plasticity is NMDA-receptor independent but sodium-channel blocker and benzodiazepines

sensitive. Front in Neurosci 9:125.

Faisal AA, Selen LPJ, Wolpert DM (2008) Noise in the Nervous System. Nature Rev

Neurosci 9:292–303.

T
Fertonani A, Miniussi C (2011) Random noise stimulation improves neuroplasticity in

IP
perceptual learning. J Neurosci. 31: 15416-15423.

CR
Fertonani A, Miniussi C (2016) Transcranial Electrical Stimulation: What We Know and

Do Not Know About Mechanisms. Neuroscientist 23:109-123.

US
Flores-Hernandez J, Cepeda C, Hernández-Echeagaray E, Calvert CR, Jokel ES, Fienberg
AN
AA, Greengard P, Levine MS (2002) Dopamine enhancement of NMDA currents in

dissociated medium–sized striatal neurons: role of D1 receptors and DARPP–32. J


M

Neurophysiol 88:3010-3020.
ED

Gluckman BJ, Netoff TI, Neel EJ, Ditto WL, Spano ML, Schiff SJ (1996) Stochastic

Resonance in a Neuronal Network from Mammalian Brain. Phys Rev Lett 77:4098-4101.
PT

Huidobro N, Mendez-Fernandez A, Mendez-Balbuena I, Gutierrez R, Kristeva R,


CE

Manjarrez E (2017) Brownian optogenetic-noise-photostimulation on the brain amplifies

somatosensory-evoked field potentials. Front Neuroci 11:464.


AC

Huidobro N, De la Torre-Valdovinos B, Mendez A, Treviño M, Arias-Carrion O, Chavez

F, Gutierrez R, Manjarrez, E (2018) Optogenetic noise-photostimulation on the brain

increases somatosensory spike firing responses. Neurosci Lett 664:51-57.

Jausovec N, Pahor A (2017) Changing brain activity, increasing intelligence: Transcranial

electrical and magnetic stimulation. In Increasing intelligence. Chapter 4, Pages 175-236.

Eds. Jausovec N, Pahor A. Academic Press, Elsevier.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 25

Laczo B, Antal A, Rothkegel H, Paulus W (2014) Increasing human leg motor cortex

excitability by transcranial high frequency random noise stimulation. Restor Neurol

Neurosci 32:403-410.

Looi CY, Lim J, Sella F, Lolliot S, Duta M, Avramenko AA, Kadosh RC (2017)

Transcranial random noise stimulation and cognitive training to improve learning and

T
cognition of the atypically developing brain: A pilot study. Sci Rep 7:4633.

IP
Mammarella N, Di Domenico A, Palumbo R, Fairfield B (2017) Self-generation and

CR
positivity effects following transcranial random noise stimulation in medial prefrontal

cortex: A reality monitoring task in older adults. Cortex 91:186-196.

US
Onorato I, D'Alessandro G, Di Castro MA, Renzi M, Dobrowolny G, Musarò, A, Salvetti
AN
M, Limatola C et al. (2016) Noise Enhances Action Potential Generation in Mouse Sensory

Neurons via Stochastic Resonance. PLoS One 11:e0160950.


M

Pasqualotto A (2016) Transcranial random noise stimulation benefits arithmetic skills.


ED

Neurobiol Learn Mem 133:7-12.

Penton T, Dixon L, Evans LJ, Banissy MJ (2017) Emotion perception improvement


PT

following high frequency transcranial random noise stimulation of the inferior frontal
CE

cortex. Sci Rep 7:11278.

Poirazi P, Brannon T, Mel BW (2003) Arithmetic of subthreshold synaptic summation in a


AC

model CA1 pyramidal cell. Neuron, 37:977-987.

Popescu T, Krause B, Terhune DB, Twose O, Page T, Humphreys G, Kadosh RC (2016)

Transcranial random noise stimulation mitigates increased difficulty in an arithmetic

learning task. Neuropsychol 81:255-264.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 26

Romanska A, Rezlescu C, Susilo T, Duchaine B, Banissy MJ (2015) High-Frequency

Transcranial Random Noise Stimulation Enhances Perception of Facial Identity. Cereb

Cortex 25:4334-4340.

Rufener KS, Ruhnau P, Heinze, HJ, Zaehle T (2017) Transcranial Random Noise

Stimulation (tRNS) Shapes the Processing of Rapidly Changing Auditory Information.

T
Front Cell Neurosci 11:162.

IP
Segundo JP, Vibert JF, Pakdaman K, Stiber M, Diez Martinez O (1994) Noise and the

CR
neurosciences: a long history, a recent revival and some theory (eds K.H. Pribram)

Origins: Brain and Self- Organization. pp. 299–333. Elbaum, Hillsdale, NJ.

US
Stacey WC, Durand DM SR (2000) Stochastic resonance improves signal detection in
AN
hippocampal CA1 neurons. J Neurophysiol 83:1394-1402.

Terney D, Chaieb L, Moliadze V, Antal A, Paulus W (2008) Increasing human brain


M

excitability by transcranial high-frequency random noise stimulation. J Neurosci 28:14147-


ED

1455.

Traub RD, Wong RK, Miles R, Michelson H (1991) A model of a CA3 hippocampal
PT

pyramidal neuron incorporating voltage-clamp data on intrinsic conductances. J


CE

Neurophysiol 66:635-650.

Traub RD, Jefferys JG, Miles R, Whittington MA, Tóth K (1994) A branching dendritic
AC

model of a rodent CA3 pyramidal neuron. J Physiol 481:79-95.

Van der Groen O, Wenderoth N (2016) Transcranial Random Noise Stimulation of Visual

Cortex: Stochastic Resonance Enhances Central Mechanisms of Perception. J Neurosci

36:5289-98.

Van Doren J, Langguth B, Schecklmann M (2014) Electroencephalographic effects of

transcranial random noise stimulation in the auditory cortex. Brain Stimul 7:807-812.
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 27

Van Koningsbruggen MG, Ficarella SC, Battelli L, Hickey C (2016) Transcranial random-

noise stimulation of visual cortex potentiates value-driven attentional capture. Soc Cogn

Affect Neurosci 11:1481-1488.

Yang T, Banissy MJ (2017) Enhancing anger perception in older adults by stimulating

inferior frontal cortex with high frequency transcranial random noise stimulation.

T
Neuropsychol 102:163-169.

IP
Yilmaz E, Ozer M, Baysal V, Perc M (2016) Autapse-induced multiple coherence

CR
resonance in single neurons and neuronal networks. Sci Rep 6:30914.

Figure legends
US
AN
Figure 1. The method to analyze the effects of electrical RNS on the peak amplitude of

Na+ currents elicited by a voltage-clamp-ramp protocol in dissociated cortical neurons


M

of Wistar rats. A. Left panel, pictures of two pyramidal cells from the auditory and
ED

somatosensory cortex. Right panel, voltage-clamp ramps and the associated Na+ currents

for these cells in conditions of zero RNS and five different levels of RNS as indicated
PT

above. Note that there is an increase in the peak amplitude of the Na+ current for
CE

intermediate intensities of RNS (red recordings). B. Details of the voltage-clamp ramp and

how the peak amplitude and latency of the Na+ current were measured. C. Superimposed
AC

traces of three voltage clamp ramps and Na+ currents for three levels of noise: zero RNS

(green), optimal RNS (red) and high RNS (blue).

Figure 2. Typical graphs of the Na+ current peak-amplitude versus its inverse of

latency grouped according to a correlation-criterion. The amplitude and latency of the

Na+ current were measured as indicated in Figure 1B for all the six levels of RNS. A.

Graphs obtained from five pyramidal neurons that exhibited a positive correlation between
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 28

Na+ current peak-amplitude versus its inverse of latency. B. The same as A, but for other

five pyramidal neurons exhibiting a negative correlation. C. The same as B but for other

five pyramidal neurons not exhibiting correlation. Every gray point represents the

measurements made for a level of RNS. The green line was obtained with linear regression.

Figure 3. Graphs of the peak amplitude and inverse of latency of the Na+ current

T
versus the level of RNS, for nine pyramidal cells (black traces) and our mathematical

IP
model (gray traces). A. Graphs obtained from three pyramidal cells of a group termed

CR
“Positive correlation.” For these cells, the same level of electrical RNS produced a maximal

increase in both the peak amplitude of the Na+ current and its inverse of latency. Note that

US
the red circles are associated with the same level of RNS. B. Graphs obtained from the
AN
second group termed “Negative correlation,” of other three pyramidal cells, in which the

maximal increase in the peak amplitude of the Na+ current and its inverse of latency were
M

produced by different levels of electrical RNS. Note that the red circles are not related to
ED

the same level of RNS. C. Graphs obtained from the third group termed “No correlation” of

another three pyramidal cells, in which the maximal increase in the peak amplitude of the
PT

Na+ current and its inverse of latency were produced by different levels of electrical RNS
CE

and the graphs of the inverse of latency did not exhibit the inverted U-like shape. Error bars

indicate standard error (SE). Note the great resemblance among the experimental (black
AC

traces) and theoretical (gray traces) data.

Figure 4. Superimposed recordings of Na+ currents for some pyramidal cells for three

levels of noise, zero (Z RNS, green), optimal (O RNS, red) and high (H RNS, blue)

RNS. A. Recordings of Na+ currents from three pyramidal cells in the first group termed

“Positive Correlation.” B. The same as A but for other three neurons in the second group
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 29

“Negative Correlation.” C. The same as B but for other three neurons in the third group

termed “No Correlation.”

Figure 5. Pooled data obtained from the analysis illustrated in Figure 3 for all the

pyramidal neurons. A. Na+ current amplitude and its inverse of latency versus the input

electrical RNS for pyramidal cells from the group “Positive Correlation.” B and C. The

T
same as A, but for pyramidal cells from the groups “Negative Correlation” and “No

IP
Correlation” as indicated. Error bars indicate standard deviation (SD).

CR
Figure 6. Theoretical and experimental Na+ currents. The green traces show the

modeled Na+ currents obtained from equation (14) and the appropriate variables R(𝑥𝑚𝑒𝑎𝑛 ),

US
p(𝑥𝑚𝑒𝑎𝑛 ) and q(𝑥𝑚𝑒𝑎𝑛 ) illustrated in Table 2. The gray traces are the Na+ currents obtained
AN
from the auditory pyramidal neuron SN-12.

Figure 7. Graphs of the functions R(𝑥𝑚𝑒𝑎𝑛 ), p(𝑥𝑚𝑒𝑎𝑛 ) and q(𝑥𝑚𝑒𝑎𝑛 ) versus the level of
M

RNS obtained from our model. These functions are included in the activation and
ED

inactivation parameters m and h. Note that in these functions, 𝑥𝑚𝑒𝑎𝑛 is the mean input

electrical-RNS. A, B and C. Parabolic graphs obtained from the functions illustrated in


PT

formulas (10), (11) and (15) when a2>0, b2>0 and c2>0. D. Predicted positive correlation
CE

(group 1) between the peak amplitude of the theoretical Na+ current and its inverse of

latency. The arrow indicates that the assumptions in A, B and C generate the graph in D. E,
AC

F and G. The same as A, B and C but when a2>0, b2<0 and c2<0. H. Predicted negative

correlation (group 2) between the peak amplitude of the theoretical Na+ current and its

inverse of latency. The arrow indicates that the assumptions in E, F and G generate the

graph in H. I, J and K. Parabolic graphs obtained from the functions illustrated in formulas

(12), (13) and (16) in which the polynomials are of fourth order. L. Predicted no-correlation
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 30

between the peak amplitude of the theoretical Na+ current and its inverse of latency. The

arrow indicates that the assumptions in I, J and K generate the graph in L. These theoretical

assumptions allowed us to obtain a robust grouping criterion for our experimental data.

Figure 8. Graphs of the functions R(𝒙𝒎𝒆𝒂𝒏 ), p(𝑥𝑚𝑒𝑎𝑛 ) and q(𝑥𝑚𝑒𝑎𝑛 ) versus the level of

RNS obtained from our experimental data illustrated in graphs A, E and I. The arrow

T
indicates that the graphs for R(𝑥𝑚𝑒𝑎𝑛 ), p(𝑥𝑚𝑒𝑎𝑛 ) and q(𝑥𝑚𝑒𝑎𝑛 ) versus the input electrical-

IP
RNS were obtained from the experimental data in graphs A, E and I. Note that these graphs

CR
are consistent with the graphs showed in Figure 7.

Figure 9. Effects of the application of 300 nM of TTX on the Na+ currents recorded in our

US
experiments. This figure illustrates the results from one pyramidal neuron. The two upper
AN
panels show the voltage ramps and the input RNS. The two lower panels show the response

of the pyramidal neuron in control conditions and after the application of 300 nM of TTX
M

in the bath.
ED

Tables
PT

Table 1. Parameters of the electrical properties of the pyramidal recorded neurons and the
CE

linear regression for the experimental data. The last two columns show the maximum

relative errors for the amplitude and inverse of latency of the modeled Na+ current.
AC

Table 2. Parameters employed in the modeling of the Na+ current.


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 31

T
IP
CR
US
AN
M
ED
PT

Figure 1
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 32

T
IP
CR
US
AN
M
ED
PT
CE
AC

Figure 2
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 33

T
IP
CR
US
AN
M
ED
PT
CE
AC

Figure 3
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 34

T
IP
CR
US
AN
M
ED
PT
CE

Figure 4
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 35

T
IP
CR
US
AN
M
ED

Figure 5
PT
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 36

T
IP
CR
US
AN
M

Figure 6
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 37

T
IP
CR
US
AN
M
ED
PT
CE
AC

Figure 7
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 38

T
IP
CR
US
AN
M
ED
PT
CE
AC

Figure 8
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 39

T
IP
CR
US
AN
M
ED

Figure 9
PT
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 40

Table 1

Linear regression
Electrical equation Pearson´s correlation Maximum relative error
properties
y=y0+a*x
for the for the inverse of
Parameters Parameters Parameters
amplitude latency
Significance

T
Cm Ra y0(pA*S) a(pA*S) z r (* p<0.05) (%)
Neurons (pF) (M) (NS p>0.2)

IP
E.9 AN-11 12 12 -0.025 179.41 0.5374 0.491 0.161 0.4 1.8
Positive correlation

E.10 AN-13 14 18 -2.74 263.11 0.7633 0.643 0.035* 0.5 1.3


E.10 AN-14 10 18 2.08 33.15 0.8404 0.686 0.046* 0.81 1.4

CR
E.10 SN-12 16 15.9 2.57 80.94 0.9868 0.756 0.04* 0.8 1.2
E.11 SN-15 11 13 0.54 149.37 1.8745 0.954 0.002* 0.5 0.7
E.11 SN-16 15 15 -1.64 163.41 1.5698 0.917 0.005* 0.9 2
E.11 SN-17 9 22 -3.21 150.54 0.646 0.569 0.078 2.4 1.1
E.12 SN-19 11 22 -2.42 190.47 0.6994 0.604 0.1 1 1.3

US
E.12 SN-24 10 15 -6.05 329.68 1.2044 0.835 0.019* 1.5 1.7
E.8 AN-07 13 20 6.75 -143.86 -1.132 -0.812 0.025* 1.52 2.01
E.9 AN-09 13 23 4.51 -83.88 -1.188 -0.83 0.021* 1.8 1.4
E.9 AN-10 16 14 5.72 -92.09 -0.459 -0.43 0.197 1.3 2.6
AN
E.11 AN-15 13 14 14.07 -485.99 -0.905 -0.719 0.041* 1.2 0.76
E.11 AN-18 16 24 11.5 -220.36 -0.984 -0.755 0.04* 0.85 2.3
E.11 AN-19 15 17 9.29 -118.42 -0.637 -0.563 0.1 0.52 1.6
E.11 AN-20 8 12 3.69 -90.14 -0.506 -0.467 0.175 2.1 1.2
Negative Correlation

E.12 AN-24 10 19 5.54 -65.66 -0.989 -0.757 0.041* 1 2.3


M

E.9 SN-09 14 23 7.11 -144.26 -0.787 -0.657 0.049* 1.2 0.9


E.9 SN-10 12 10 11.15 -289.39 -0.480 -0.447 0.18 1.8 1
E.9 SN-11 10 16 14.57 -344.09 -2.442 -0.985 0.0001* 0.97 1.88
E.10 SN-13 11 14 20.49 -619.87 -1.270 -0.854 0.015* 1.2 1.6
ED

E.10 SN-14 7 30 3.82 -56.62 -1.345 -0.873 0.012* 1.2 1.9


E.12 SN-21 9 11 5.53 -108.22 -0.523 -0.48 0.167 1.2 1.6
E.12 SN-23 12 20 11.05 -213.33 -0.873 -0.703 0.04* 1.09 1.9
E.4 AN-03 11 25 2.85 -64.87 -0.098 -0.098 0.426NS 0.54 2.2
PT

E.9 AN-08 16.9 18 2.44 16.48 0.2888 0.281 0.294 NS 1.3 1.3
E.10 AN-12 11 20 2.27 9.64 0.1634 0.162 0.38 NS 0.9 0.59
No Correlation

E.11 AN-16 19 26 5.74 -82.25 -0.243 -0.239 0.324 NS 1 1.5


E.11 AN-17 11 20 2.01 10.29 0.2007 0.198 0.353 NS 1.7 1
CE

E.12 AN-22 8 15 1.06 19.46 0.2163 0.213 0.343 NS 1.5 1.2


E.12 AN-23 22 29 3.77 5.75 0.023 0.023 0.483 NS 0.9 2.1
E.5 SN-05 5 20 1.61 3.41 0.049 0.049 0.463 NS 1.5 1.5
E.12 SN-20 8 10 3.14 34.69 0.3826 0.365 0.238 NS 1 1.9
1.9
AC

E.12
SN-22 11 18 2.57 55.75 0.3128 0.303 0.279 NS 0.7
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 41

Table 2

Modeling equations

𝐼𝑁𝑎 = 𝑔𝑁𝑎 𝑅(𝑥𝑚𝑒𝑎𝑛 ) 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 ))ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))(𝑉𝑠𝑡𝑖𝑚 − 𝑉𝑁𝑎 )

T
𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2
𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2

IP
𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2

CR
Parameters
Neurons a0 a1 a2 b0 b1 b2 c0 c1 c2
AN-11 0.933 0.651 -1.86 -50.03 -42.184 79.911 -48.13 -42.184 75.911
AN-13 0.7935 3.0236 -12.321 -82.23 -42.387 82.235 -51.06 -42.387 82.235
Positive correlation

AN-14 0.96 0.287 0.778 -61.71 -28.004 36.986 -58.56 -28.004 36.986

US
SN-12 0.968 0.355 -0.812 -52.29 -27.484 55.913 -50.4 -27.47 55.912
SN-15 0.891 0.501 -0.598 -46.87 -54.946 89.981 -44.77 -54.954 89.992
SN-16 0.99 -0.04 -0.165 -58.86 -5.833 19.831 -56.56 -5.833 19.831
SN-17 0.788 1.257 -4.33 -54.37 -26.285 66.142 -52.17 -26.285 66.142
AN
SN-19 0.823 1.684 -4.3733 -51.98 -28.486 55.432 -50.98 -28.486 55.432
SN-24 0.727 0.949 -2.043 -37.75 -18.701 34.24 -34.95 18.701 34.24
AN-07 0.461 1.27 -2.519 -59.59 5.533 -29.97 -57.49 5.533 -29.974
AN-09 0.656 2.728 -10.717 -65.39 46.183 -211.69 -61.92 35.323 -164.26
AN-10 0.905 -0.022 -0.614 -56.23 14.248 -32.385 -54.43 14.248 -32.385
M

AN-15 0.777 0.891 -2.05 -42.43 2.591 -3.391 -40.15 2.1619 -2.319
AN-18 0.788 1.808 -4.799 -68.1 5.767 -18.014 -64.73 7.2133 -23.595
Negative correlation

AN-19 0.959 0.678 -3.478 -65.31 -3.233 -9.09 -62.74 -0.642 -15.851
ED

AN-20 0.887 -0.633 -0.32 -40.34 -7.3131 -31.127 -36.74 -7.316 -31.125
AN-24 0.965 0.208 -0.862 -59.03 13.542 -38.903 -56.83 13.481 -38.819
SN-09 0.817 0.877 -2.144 -52.56 41.26 -217.13 -50.16 41.26 -217.13
SN-10 0.953 -1.456 -0.13 -51.37 -7.147 -79.23 -48.12 -7.147 -79.23
SN-11 0.643 2.152 -4.863 -57.59 38.11 -88.726 -55.39 38.111 -88.726
PT

SN-13 0.741 2.463 -5.556 -53.16 14.578 -32.564 -50.21 14.578 -32.563
SN-14 0.93 0.629 -1.3177 -53.54 26.27 -57.469 -51.29 26.281 -57.496
SN-21 0.641 1.981 -12.873 -57.99 9.5242 -133.52 -56.01 9.5242 -133.52
SN-23 0.909 1.007 -5.82 -59.26 5.339 -20.032 -56.73 5.269 -19.927
CE

𝐼𝑁𝑎 = 𝑔𝑁𝑎 𝑅(𝑥𝑚𝑒𝑎𝑛 ) 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 ))ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))(𝑉𝑠𝑡𝑖𝑚 − 𝑉𝑁𝑎 )


𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑎3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑎4 𝑥𝑚𝑒𝑎𝑛 4
𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑏3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑏4 𝑥𝑚𝑒𝑎𝑛 4
𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑐3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑐4 𝑥𝑚𝑒𝑎𝑛 4
AC

Parameters
a0 a1 a2 a3 a4 b0 b1 b2 b3 b4 c0 c1 c2 c3 c4
AN-03 0.7599 -14.14 262.47 1263.71 1668.12 -54.99 -69.75 757.68 -2847.7 3321.1 -52.89 -69.75 757.68 -2847.7 3321.1
AN-08 0.955 -0.64 -8.06 34.014 -42.962 -54.49 -60.028 786.95 -3591.1 4818.3 -51.82 -60 786.95 -3591.1 4818.3
AN-12 0.912 2.81 -39.46 167.98 -210.188 -42.24 39.08 -763.64 3920.1 -5576.2 -40.19 39.065 -763.37 3918.9 -5574.7
No correlation

AN-16 0.867 1.153 2.633 -32.208 46.716 -60.15 34.69 -381.25 1451.3 -1685.1 -56.75 34.73 -381.59 1452.6 -1686.6
AN-17 0.904 2.87 -30.95 120.63 -146.7 -48 179.44 -2367.2 9344.5 -11192 -45.95 179.44 -2367.2 9344.5 -11192
AN-22 0.901 2.45 22.53 78.11 -87.25 -56.33 -45.182 689.6 -3265.3 4403.2 -53.58 -45.19 690.08 -3269.2 4412.4
AN-23 0.95 1.49 -18.82 65.6 -72.77 -54.69 6.14 151.67 -1620.6 2841.6 -52.63 9.33 101.89 -1406.2 2574.7
SN-05 0.89 2.65 -7.43 261.26 349.32 -56.21 -47.45 634.1 -2786.3 3967.2 -53.8 -46.91 790.32 -3198.4 4980.3
SN-20 0.867 0.626 -13.14 83.53 -129.87 -49.91 -207.53 2569 -10993 -13729 -47.21 -207.53 2569 -10993 -13729
SN-22 0.58 0.002 30.82 -137.17 160.89 -63.97 77.08 -1393.4 6869.1 -8999.6 -60.17 55.25 -1086.1 5275.9 -6787.5
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 42

Linear regression
Electrical equation
Pearson´s correlation Maximum relative error
properties
y=y0+a*x

for the for the inverse of


Parameters Parameters Parameters
amplitude latency

Significance

y0(pA*S) a(pA*S) z r (* p<0.05) (%)


Cm Ra
(NS p>0.2)

T
Neurons (pF) (M)

IP
E.9 AN-11 12 12 -0.025 179.41 0.5374 0.491 0.161 0.4 1.8

E.10 AN-13 14 18 -2.74 263.11 0.7633 0.643 0.035* 0.5 1.3

CR
E.10 AN-14 10 18 2.08 33.15 0.8404 0.686 0.046* 0.81 1.4

E.10 SN-12 16 15.9 2.57 80.94 0.9868 0.756 0.04* 0.8 1.2

US
E.11 SN-15 11 13 0.54 149.37 1.8745 0.954 0.002* 0.5 0.7
Positive correlation

E.11 SN-16 15 15 -1.64 163.41 1.5698 0.917 0.005* 0.9 2


AN
E.11 SN-17 9 22 -3.21 150.54 0.646 0.569 0.078 2.4 1.1

E.12 SN-19 11 22 -2.42 190.47 0.6994 0.604 0.1 1 1.3


M

E.12 SN-24 10 15 -6.05 329.68 1.2044 0.835 0.019* 1.5 1.7

E.8 AN-07 13 20 6.75 -143.86 -1.132 -0.812 0.025* 1.52 2.01


ED

E.9 AN-09 13 23 4.51 -83.88 -1.188 -0.83 0.021* 1.8 1.4

E.9 AN-10 16 14 5.72 -92.09 -0.459 -0.43 0.197 1.3 2.6


PT

E.11 AN-15 13 14 14.07 -485.99 -0.905 -0.719 0.041* 1.2 0.76

E.11 AN-18 16 24 11.5 -220.36 -0.984 -0.755 0.04* 0.85 2.3


CE

E.11 AN-19 15 17 9.29 -118.42 -0.637 -0.563 0.1 0.52 1.6

E.11 AN-20 8 12 3.69 -90.14 -0.506 -0.467 0.175 2.1 1.2


AC
Negative Correlation

E.12 AN-24 10 19 5.54 -65.66 -0.989 -0.757 0.041* 1 2.3

E.9 SN-09 14 23 7.11 -144.26 -0.787 -0.657 0.049* 1.2 0.9

E.9 SN-10 12 10 11.15 -289.39 -0.480 -0.447 0.18 1.8 1

E.9 SN-11 10 16 14.57 -344.09 -2.442 -0.985 0.0001* 0.97 1.88

E.10 SN-13 11 14 20.49 -619.87 -1.270 -0.854 0.015* 1.2 1.6


No Correlation

E.10 SN-14 7 30 3.82 -56.62 -1.345 -0.873 0.012* 1.2 1.9

E.12 SN-21 9 11 5.53 -108.22 -0.523 -0.48 0.167 1.2 1.6


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 43

E.12 SN-23 12 20 11.05 -213.33 -0.873 -0.703 0.04* 1.09 1.9

E.4 AN-03 11 25 2.85 -64.87 -0.098 -0.098 0.426NS 0.54 2.2

E.9 AN-08 16.9 18 2.44 16.48 0.2888 0.281 0.294 NS 1.3 1.3

E.10 AN-12 11 20 2.27 9.64 0.1634 0.162 0.38 NS 0.9 0.59

E.11 AN-16 19 26 5.74 -82.25 -0.243 -0.239 0.324 NS 1 1.5

E.11 AN-17 11 20 2.01 10.29 0.2007 0.198 0.353 NS 1.7 1

E.12 AN-22 8 15 1.06 19.46 0.2163 0.213 0.343 NS 1.5 1.2

T
0.483 NS

IP
E.12 AN-23 22 29 3.77 5.75 0.023 0.023 0.9 2.1

E.5 SN-05 5 20 1.61 3.41 0.049 0.049 0.463 NS 1.5 1.5

CR
E.12 SN-20 8 10 3.14 34.69 0.3826 0.365 0.238 NS 1 1.9

1.9

US
E.12
SN-22 11 18 2.57 55.75 AN 0.3128 0.303 0.279 NS 0.7
M
ED
PT

Table 1
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 44

Modeling equations

𝐼𝑁𝑎 = 𝑔𝑁𝑎 𝑅(𝑥𝑚𝑒𝑎𝑛 ) 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 ))ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))(𝑉𝑠𝑡𝑖𝑚 − 𝑉𝑁𝑎 )

𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2

𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2

𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2

T
Parameters

IP
Neurons a0 a1 a2 b0 b1 b2 c0 c1 c2

AN-11 0.933 0.651 -1.86 -50.03 -42.184 79.911 -48.13 -42.184 75.911

CR
AN-13 0.7935 3.0236 -12.321 -82.23 -42.387 82.235 -51.06 -42.387 82.235

AN-14 0.96 0.287 0.778 -61.71 -28.004 36.986 -58.56 -28.004 36.986

US
Positive correlation

SN-12 0.968 0.355 -0.812 -52.29 -27.484 55.913 -50.4 -27.47 55.912

SN-15 0.891 0.501 -0.598 -46.87 -54.946 89.981 -44.77 -54.954 89.992

SN-16 0.99 -0.04 -0.165 -58.86 -5.833 19.831 -56.56 -5.833 19.831
AN
SN-17 0.788 1.257 -4.33 -54.37 -26.285 66.142 -52.17 -26.285 66.142

SN-19 0.823 1.684 -4.3733 -51.98 -28.486 55.432 -50.98 -28.486 55.432
M

SN-24 0.727 0.949 -2.043 -37.75 -18.701 34.24 -34.95 18.701 34.24

AN-07 0.461 1.27 -2.519 -59.59 5.533 -29.97 -57.49 5.533 -29.974
ED

AN-09 0.656 2.728 -10.717 -65.39 46.183 -211.69 -61.92 35.323 -164.26

AN-10 0.905 -0.022 -0.614 -56.23 14.248 -32.385 -54.43 14.248 -32.385
PT

AN-15 0.777 0.891 -2.05 -42.43 2.591 -3.391 -40.15 2.1619 -2.319

AN-18 0.788 1.808 -4.799 -68.1 5.767 -18.014 -64.73 7.2133 -23.595

AN-19 0.959 0.678 -3.478 -65.31 -3.233 -9.09 -62.74 -0.642 -15.851
CE
Negative correlation

AN-20 0.887 -0.633 -0.32 -40.34 -7.3131 -31.127 -36.74 -7.316 -31.125

AN-24 0.965 0.208 -0.862 -59.03 13.542 -38.903 -56.83 13.481 -38.819
AC

SN-09 0.817 0.877 -2.144 -52.56 41.26 -217.13 -50.16 41.26 -217.13

SN-10 0.953 -1.456 -0.13 -51.37 -7.147 -79.23 -48.12 -7.147 -79.23

SN-11 0.643 2.152 -4.863 -57.59 38.11 -88.726 -55.39 38.111 -88.726

SN-13 0.741 2.463 -5.556 -53.16 14.578 -32.564 -50.21 14.578 -32.563

SN-14 0.93 0.629 -1.3177 -53.54 26.27 -57.469 -51.29 26.281 -57.496

SN-21 0.641 1.981 -12.873 -57.99 9.5242 -133.52 -56.01 9.5242 -133.52

SN-23 0.909 1.007 -5.82 -59.26 5.339 -20.032 -56.73 5.269 -19.927

𝐼𝑁𝑎 = 𝑔𝑁𝑎 𝑅(𝑥𝑚𝑒𝑎𝑛 ) 𝑚(𝑉𝑠𝑡𝑖𝑚 , 𝑝(𝑥𝑚𝑒𝑎𝑛 ))ℎ(𝑉𝑠𝑡𝑖𝑚 , 𝑞(𝑥𝑚𝑒𝑎𝑛 ))(𝑉𝑠𝑡𝑖𝑚 − 𝑉𝑁𝑎 )

𝑅(𝑥𝑚𝑒𝑎𝑛 ) = 𝑎0 + 𝑎1 𝑥𝑚𝑒𝑎𝑛 + 𝑎2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑎3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑎4 𝑥𝑚𝑒𝑎𝑛 4


ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 45

𝑝(𝑥𝑚𝑒𝑎𝑛 ) = 𝑏0 + 𝑏1 𝑥𝑚𝑒𝑎𝑛 + 𝑏2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑏3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑏4 𝑥𝑚𝑒𝑎𝑛 4

𝑞(𝑥𝑚𝑒𝑎𝑛 ) = 𝑐0 + 𝑐1 𝑥𝑚𝑒𝑎𝑛 + 𝑐2 𝑥𝑚𝑒𝑎𝑛 2 + 𝑐3 𝑥𝑚𝑒𝑎𝑛 3 + 𝑐4 𝑥𝑚𝑒𝑎𝑛 4

Parameters

a0 a1 a2 a3 a4 b0 b1 b2 b3 b4 c0 c1 c2 c3 c4

AN-03 0.7599 -14.14 262.47 1263.71 1668.12 -54.99 -69.75 757.68 -2847.7 3321.1 -52.89 -69.75 757.68 -2847.7 3321.1

AN-08 0.955 -0.64 -8.06 34.014 -42.962 -54.49 -60.028 786.95 -3591.1 4818.3 -51.82 -60 786.95 -3591.1 4818.3

AN-12 0.912 2.81 -39.46 167.98 -210.188 -42.24 39.08 -763.64 3920.1 -5576.2 -40.19 39.065 -763.37 3918.9 -5574.7

T
AN-16 0.867 1.153 2.633 -32.208 46.716 -60.15 34.69 -381.25 1451.3 -1685.1 -56.75 34.73 -381.59 1452.6 -1686.6
No correlation

IP
AN-17 0.904 2.87 -30.95 120.63 -146.7 -48 179.44 -2367.2 9344.5 -11192 -45.95 179.44 -2367.2 9344.5 -11192

AN-22 0.901 2.45 22.53 78.11 -87.25 -56.33 -45.182 689.6 -3265.3 4403.2 -53.58 -45.19 690.08 -3269.2 4412.4

CR
AN-23 0.95 1.49 -18.82 65.6 -72.77 -54.69 6.14 151.67 -1620.6 2841.6 -52.63 9.33 101.89 -1406.2 2574.7

SN-05 0.89 2.65 -7.43 261.26 349.32 -56.21 -47.45 634.1 -2786.3 3967.2 -53.8 -46.91 790.32 -3198.4 4980.3

SN-20 0.867 0.626 -13.14 83.53 -129.87 -49.91 -207.53 2569 -10993 -13729 -47.21 -207.53 2569 -10993 -13729

US
SN-22 0.58 0.002 30.82 -137.17 160.89 -63.97
AN 77.08 -1393.4 6869.1 -8999.6 -60.17 55.25 -1086.1 5275.9 -6787.5
M
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
January 5th 2019 Revised Ms. NSC-18-1034-R1 46

 Random noise electrical stimulation (electrical RNS) amplifies Na+ currents


 Electrical RNS modulates the latency of Na+ currents
 A modeling study shows that electrical RNS alters the kinetics of Na+ channels

T
IP
CR
US
AN
M
ED
PT
CE
AC

You might also like