Corticospinal output and cortical excitation‐inhibition balance in distal hand muscle representations in nonprimary motor area

Abstract

Transcranial magnetic stimulation (TMS) of the superior frontal gyrus in the non‐primary motor area (NPMA) can evoke motor‐evoked potentials (MEPs) at 20 ms latency range in contralateral distal hand muscles similar to stimulation of M1 and indicating monosynaptic corticospinal tracts. We compared the intracortical inhibitory and excitatory balance in primary motor cortex (M1) and in NPMA by navigated single‐ and paired‐pulse TMS (ppTMS). We also evaluated the spatial stability of muscle representations in M1 and NPMA by remapping 11 healthy subjects one year after the initial mapping. Resting motor threshold (rMT) was higher in NPMA than in M1 as were the MEP amplitudes evoked by 120% rMT stimulation intensity of the local MT. Short‐interval intracortical inhibition (SICI) was significantly weaker in NPMA than in M1 at ISI of 2 ms and conditioning stimulus (CS) 80% rMT. Our findings suggest that the cortical hand representations in NPMA 1) are connected to lower motoneurons monosynaptically, 2) are less strictly organized, i.e. motoneuron population representing a discrete hand muscle is sparser and less dense than in M1 and 3) have the capacity to generate powerful, rapid muscle contraction if sufficient number of motoneurones are activated. In NPMA, local intracortical inhibitory and excitatory activity is mainly similar to that in M1. The lower SICI in NPMA at an ISI of 2 ms may reflect less strict topographic organization and readiness to reorganization of neural circuits during motor learning or after motor deficits. Hum Brain Mapp, 2010. © 2010 Wiley‐Liss, Inc.

INTRODUCTION

We have recently shown that motor‐evoked potentials (MEPs) can be evoked in contralateral distal hand muscles only 20 ms after transcranial magnetic stimulation (TMS) of the superior frontal gyrus in nonprimary motor areas (NPMA) [Teitti et al.,2008]. MEPs were elicited when stimuli were targeted to the sulcus between the medial and the superior frontal gyrus and the elicited electric field (E‐field) was directed from lateral to medial direction towards the superior frontal gyrus.

Motor areas in the frontal lobe are classically divided by physiological and cytoarchitectonic features to primary motor cortex (M1) and nonprimary motor areas (NPMA). M1 covers the posterior part of precentral gyrus [Fulton,1935]. This is the area where gigantopyramidal cells (Betz cells) exist (Brodmann area 4) [Brodmann,1909]. Electrical microstimulations of this area evoke simple muscle movements from topographically organized muscle representations [Penfield W.1954]. NPMAs anterior to M1 are also part of agranular cortex but lack the thick zone of Betz cells [Brodmann,1909]. These areas are taken to participate in higher order motor functions [Fulton,1935] and electrical stimulations of Brodmann area 6 produce complex movements [Penfield and Welch,1951; Penfield,1954]. More recent studies divide NPMAs based on noninvasive imaging methods. Anatomically and by connectivity profiles, NPMAs can first be divided into premotor cortex (PMA) and supplementary motor area (SMA) [Orgogozo and Larsen,1979; Roland et al.,1980]. PMA is furthermore divided into dorsal and ventral premotor areas (PMd, PMv) and SMA into pre‐SMA and SMA‐proper [Fink et al.,1997; Hikosaka et al.,1996; Johansen‐Berg et al.,2004; Lee et al.,1999; Picard and Strick2001; Pochon et al.,2001; Toni et al.,1999].

Neurons activated by navigated TMS in superior frontal gyrus reside most likely in the PMd [Picard and Strick,2001]. This area corresponds best to Brodmann area 6 [Geyer,2004]. Posterior parts of PMd, anterior to hand representation in M1 in precental gyrus, are active during movement preparation or execution [Boussaoud,2001; Pochon et al.,2001; Toni et al.,1999] while more frontal parts of PMd are active during motor imaginary and cognitively demanding motor tasks [Gerardin et al.,2000; Grafton et al.,1998; Picard and Strick2001; Toni et al.,1999]. In our previous study, MEPs were evoked from both of these areas. In some subjects occasional MEPs were evoked near the interhemispheric fissure, corresponding most likely to SMA, or more frontal areas of superior frontal gyrus, which may correspond to posterior part of Brodmann area 8 [Teitti et al.,2008].

Earlier studies in healthy human subjects [Fink et al.,1997; Johansen‐Berg et al.,2004; Kim et al.,2004; Partanen et al.,2000; Uozumi et al.,2004] and in patients with brain lesions [Fridman et al.,2004; Ikeda et al.,1992; Krainik et al.,2001; Laplane et al.,1977; Uematsu et al.,1992] have suggested that NPMAs may contain direct corticospinal tracts in addition to the well‐known role in the selection and motor control of complex movements via cortico‐cortical networks. In non‐human primates, it has been well established that, the NPMAs in the frontal lobe contain multiple fore‐ and hindlimb muscles representation areas with monosynaptic connections to lower motor neurones [Boudrias et al.,2010a; Boudrias et al.,2010b; Dum and Strick,1991; Dum and Strick,2002]. Medial wall of each hemisphere, in particular superior frontal gyrus, is one of the areas outside M1 with the highest number of corticospinally projecting neurones [Dum and Strick,1991]. Although the functional significance of these corticospinal motoneurons outside M1 is still unknown, observed increase in activation of NPMAs during paretic limb movements in stroke patients strongly suggests that these areas have a role in rehabilitation from motor deficit [Carey et al.,2006; Fridman et al.,2004; Johansen‐Berg et al.,2002; Ward et al.,2006; Weiller et al.,1992].

The suppression of excitability in an area surrounding an activated neural population in the sensory systems is one of the best‐known physiological mechanisms to focus neuronal activity and to control neuronal firing, providing spatiotemporal discrimination of sensory inputs [Blakemore et al.,1970]. Several studies support the presence of surround inhibition in the human motor system [Beck et al.,2008; Hallett,2003; Hallett,2004; Sohn and Hallett,2004a,b; Ziemann et al.,1996c], although the exact physiological mechanism remain elusive. One possible candidate which reflects the surround inhibition at the motor network level is the short‐interval intracortical inhibition (SICI) [Beck et al.,2008].

Paired‐pulse TMS (ppTMS) provides a tool to investigate both intracortical inhibitory and excitatory (intracortical facilitation; ICF) balance in M1 [Claus et al.,1992; Di Lazzaro et al.,1998; Kujirai et al.,1993; Nakamura et al.,1997; Valls‐Sole et al.,1992; Ziemann et al.,1996a,b,c]. In M1, ppTMS with a subthreshold conditioning stimulus (CS) which is followed by a suprathreshold test stimulus activates most effectively inhibitory interneurons and induces a decrease in MEP amplitude at interstimulus intervals (ISIs) shorter than 6 ms (SICI). At ISIs over 7 ms, ppTMS elicits an increase in MEP amplitude resulting from the activity of excitatory interneurons (ICF) [Claus et al.,1992; Kujirai et al.,1993; Nakamura et al.,1997; Valls‐Sole et al.,1992; Ziemann et al.,1996a,b,c].

SICI and ICF are modulated by intracortical neurotransmitter systems, mainly by the balance of inhibitory, GABAergic and excitatory, most probably glutamatergic, interneuron activity [Hanajima et al.,1998; Schwenkreis et al.,1999; Ziemann et al.,1996a,1998]. It has been proposed that changes in the balance of the intracortical neurotransmitter system enable cortical reorganization after stroke [Butefisch et al.,2003; Fridman et al.,2004; Marshall et al.,2000; Seitz et al.,1998; Weiller et al.,1992], after peripheral nerve lesion [Brasil‐Neto et al.,1992b; Donoghue et al.,1990; Franchi and Veronesi2006], and during motor learning [Karni et al.,1995; Karni et al.,1998; Kleim et al.,2004; Muellbacher et al.,2001; Nudo et al.,1996; Pascual‐Leone et al.,1995; Plautz et al.,2000]. Plasticity, as the reorganization is called, is postulated to result from the unmasking of existing horizontal connections, i.e. activation of physiologically inactive interneurons by disinhibition [Sanes and Donoghue,2000], in addition to other forms of plasticity [Dancause et al.,2005; Hess and Donoghue,1994; Kleim et al.,2004; Rosenkranz et al.,2007; Ziemann et al.,2004].

Motor output and cortical inhibition‐excitation balance in NPMA are previously unknown and accumulating evidence points to NPMA contributing to recovery from motor deficits in man. Hence, our aim was to characterize the basic functional properties of the muscle representations and output properties of NPMA in healthy subjects by ppTMS.

MATERIALS AND METHODS

Eleven healthy subjects were studied (7 female, 4 male; age range 21–31 years, mean 25.3 ± 2.9 years). They had also participated in our previous study one year earlier [Teitti et al.,2008]. Ten of the subjects were right‐handed and one was left‐handed, according to the Waterloo Handedness Questionnaire‐revised and reduced form with 20 items [Elias et al.,1998]. Nature of experimental procedures was explained to subjects and all the subjects gave their written informed consent. The study was performed in compliance with the Declaration of Helsinki and was approved by the local ethics committee.

Navigated Brain Stimulation and EMG

Navigated brain stimulation, NBS (eXimia NBS, Nexstim Ltd., Helsinki, Finland), was used for mapping the M1 and the NPMA in the nondominant hemisphere. NBS combines neuronavigation with transcranial magnetic stimulation (TMS). It is guided by individual MR images and a computational electric field display which allow stimulation of discrete cortical areas and enable the investigation of spatially nearby but functionally separate motor representations [Hannula et al.,2005; Julkunen et al.,2009; Ruohonen,2003; Säisänen et al.,2008]. The most probable stimulation area is located on the cortex where the induced electric field is strongest [Ravazzani et al.,1996; Ruohonen,2003; Thielscher and Kammer,2002]. The computational electric field is calculated in the spherical model [Sarvas,1987; Tarkiainen et al.,2003], matched to the individual MRIs. The computed electric field does not account for details of geometry or material conductivity differences. It does however account for the stimulation intensity, coil parameters (shape of copper wiring inside the coil, 3D position, and orientation of the coil), and the head and brain anatomy [Hannula et al.,2005]. Three‐dimensional subject‐specific T1‐weighted MR images (TR 1980 ms, TE 3.93 ms, FOV 256 mm, slice thickness 1.0 mm) (Siemens Avanto 1.5 T, Erlangen, Germany) were used for the navigation. A three‐dimensional brain surface was reconstructed to a depth of 25 mm from the scalp to visualize the mapping surface at the region of the anatomical hand knob. At this depth, anatomical structures such as the sulci and gyri are easily identified.

Single and paired monophasic TMS pulses were delivered with a Magstim BiStim 200² stimulator (Magstim Company Ltd., Whitland, Wales, UK) via a 70‐mm figure‐of‐eight coil. Only the nondominant hemisphere was chosen for examination because we found no significant differences in frontal representations between the hemispheres in our previous study [Teitti et al.,2008].

During the NBS study, muscle activation was monitored on‐line and recorded continuously by electromyography (EMG) (ME 6000, Mega Electronics Ltd., Kuopio, Finland). Disposable surface electrodes (circular, diameter 9 mm, Ag‐AgCl) were positioned bilaterally on the opponens pollicis (OP) muscles (referred to the 1st metacarpal bone in the metacarpophalangeal joint). The EMG signals were sampled at a rate of 1000 Hz, filtered (8–500 Hz), amplified and stored for off‐line analyses. In these analyses, only MEPs without preceding muscle activity were included in the analysis, and amplitudes and latencies were measured for each response.

At first, the cortical representation area of the OP was mapped from the M1 with stimulations along the posterior part of the precentral gyrus (distance between stimulation points ∼2 mm, interstimulus interval ≥5 s, direction of the induced electric field perpendicular to the central sulcus). Stimulations were focused on the surroundings of a previously determined optimal representation area [Teitti et al.,2008]. The stimulus location on the M1 eliciting repeatedly the OP response with the highest peak‐to‐peak amplitude was taken as the optimal M1 representation location. The optimal direction for the induced electric field in this target point was determined by rotating the coil (±45°, ±90°, ±135°, and 180°). The coil orientation which produced the highest MEP was chosen as the final location in which rMT was measured (see Fig. 1). The first estimate of rMT was determined with 10 steps using a threshold‐hunting paradigm [Awiszus2003]. Thereafter, rMT was defined as the minimum stimulus intensity (percent of maximal stimulator output) and the corresponding electric field value (V/m) that produced 5 MEPs out of 10 stimuli (peak‐to‐peak amplitudes ≥50 μV) [Chen et al.,2008; Rossini et al.,1994]. Subsequently, 10 single‐pulse MEPs were recorded at an intensity of 120% rMT to establish baseline for the MEP amplitude.

Figure 1.

Optimal representation areas of OP muscle in M1 and NPMA in 10 subjects (visual data of one subject is lacking because of technical reason). Cylinders show the coil locations, orange dots the sites of the maximal electric fields on the cortex, and arrows the direction of the current. Cylinders filled with black color mark the optimal OP representations in this study and yellow cylinders the mapping results of our previous study. Two‐colored cylinders mean that optimal representation location were the same in both studies.

NPMAs anterior to M1 were mapped with an intensity of approximately 55% of the maximal stimulator output. The mapping intensity was set to be ∼15% greater than the previously defined MT value in each subject [Teitti et al.,2008]. Mapping was focused on the superior frontal gyrus and the sulcus between the superior and medial frontal gyrus (direction of induced electric field perpendicular to the sulci) from where MEPs could previously be evoked in the contralateral hand muscles without concomitant stimulation of M1 [Teitti et al.,2008]. The optimal location of OP muscle representation and rMT was detected as in the M1. The corresponding electric field value in M1 was defined to ensure that it remained below the measured M1 rMT electric field value, to exclude the simultaneous stimulation of M1 while mapping NPMA. Subsequently, 10 single‐pulse MEPs induced at an intensity of 120% rMT were recorded.

Paired‐pulses were delivered to both M1 and NPMA, first to M1. In the first ppTMS experiment, CS was set at 80% and the test stimulus at 120% of local rMT. Pulses were delivered at ISIs 2, 3, 10, and 15 ms. For each ISI, 10 trials were recorded with an inter‐trial interval of 5–10 s. The order of trials with different ISIs was randomized. In the second experiment, CS intensities of 30, 50, 70, and 90% of local rMT were used at a constant ISI of 2 ms. Ten trials were performed for each CS intensity in randomized order, as in the first experiment.

Analyses

Single‐pulse and ppMEPs which could be distinguished repeatedly from the baseline EMG were considered as responses even if the peak‐to‐peak amplitude were smaller than 50 μV. MEPs smaller than 50 μV were taken into account to avoid emphasis of higher responses in the averages especially in pp‐examinations with inhibitory ISIs. In the case of very small uncertain responses (typically <20 μV), the result was considered to be a no‐response and MEP amplitude was marked as 0 μV in the average value. In the analyses, the highest and lowest MEPs (of 10 trials) were excluded to ensure that average MEP amplitudes were not affected by any outliers. Thereby, eight responses/trial were considered for average MEP amplitudes. Mean ppMEP amplitudes were normalized to the amplitude of the single‐pulse MEP (120% rMT) in M1 and in NPMA, respectively.

The distances between two stimulation locations were measured and defined as the Euclidean distances between coil locations and between the locations of maximal electric fields on the cortex at a depth of 25 mm from the scalp. Data of 10 subjects were used in the analyses of optimal OP locations distances. The TMS log data of one subject were not available because of a technical failure.

Statistical Analyses

Single‐ and paired‐pulse MEP amplitudes (absolute and normalized) were compared between M1 and NPMA by using the Wilcoxon signed ranks test. MEP latencies for the single‐pulse trials as well as rMTs for the stimulation sites were also compared using this test. On the basis of Kolmogorov‐Smirnov test, the data was not normally distributed (P = 0.104–0.998), and therefore nonparametric test was used. The values of rMTs, amplitudes and latencies are presented as mean ± S.D. All the statistical tests were performed with SPSS 16 (SPSS Inc., Chicago, IL).

RESULTS

Mapping of the M1 showed the typical representation of the OP muscle on the anatomical hand knob [Ro et al.,1999; Rossini et al.,1994; Yousry et al.,1997] (see Fig. 1). MEPs could be elicited from contralateral distal hand muscles by targeted stimulation of the superior frontal gyrus as in previous study of the same subjects one year earlier [Teitti et al.,2008] (see Fig. 1). In M1, the distance between optimal OP representation sites mapped in these two separate studies, was 2.0 ± 2.8 mm when the centers of maximal electric fields were compared and in NPMA, the distance was 9.7 ± 5.6 mm.

The average distance between the optimal stimulation targets in M1 and NPMA was 32.9 ± 4.2 mm when defined as the distance between the locations of maximal electric fields on the cortex and 30.1 ± 6.5 mm when defined as the distance between the coil locations.

The average rMT for OP in M1 was 37 ± 5%, the single‐pulse MEP (120% rMT) amplitude was 924 ± 791 μV, and latency 22.4 ± 1.3 ms (Table I). The rMT for OP in NPMA was 46 ± 6%, the single‐pulse MEP (120% rMT) amplitude was 1201 ± 750 μV, and latency 21.8 ± 1.4 ms (Table I). rMTs in NPMA were higher than in M1 (P < 0.01), as were the single‐pulse MEP amplitudes evoked by stimulation intensities relative to local MTs (P < 0.01). Moreover, the latencies of MEPs evoked by stimulation of NPMA were shorter (P < 0.01) than those evoked by stimulation of M1.

Table I.

Individual motor threshold (MT) values (per cents of maximum stimulator intensity), single‐pulse MEP amplitudes (mean ± SD), single‐pulse MEP latencies (mean ± SD)

	M1			NPMA
Subject	MT (%)	MEP amplitude (μV)	MEP latency (ms)	MT (%)	MEP amplitude (μV)	MEP latency (ms)
1	39	2405 ± 1230	20.8 ± 1.2	53	2597 ± 995	19.7 ± 1.2
2	48	806 ± 387	21.0 ± 1.0	59	1120 ± 572	20.6 ± 0.7
3	36	1226 ± 749	24.0 ± 0.6	46	1094 ± 583	23.6 ± 0.5
4	37	237 ± 214	21.9 ± 0.9	42	466 ± 380	21.9 ± 1.6
5	31	1909 ± 556	21.0 ± 0.6	37	1840 ± 1243	19.9 ± 1.4
6	38	452 ± 252	23.9 ± 0.7	46	645 ± 455	23.7 ± 1.1
7	40	132 ± 39	23.5 ± 0.8	48	326 ± 166	23.0 ± 0.6
8	35	369 ± 225	23.0 ± 1.0	50	1178 ± 429	21.5 ± 0.7
9	37	1819 ± 1100	22.1 ± 0.9	47	2345 ± 1766	22.4 ± 0.8
10	40	496 ± 294	21.1 ± 0.6	42	759 ± 560	20.9 ± 1.0
11	29	310 ± 264	23.9 ± 0.9	40	835 ± 198	22.3 ± 1.9
Mean	37	924	22.4	46	1201	21.8
SD	5	791	1.3	6	750	1.4

Group means and SDs (based on individual mean values) are also presented.

ppMEPs at ISIs of 2 and 3 ms, with CS set at 80% of rMT, showed a clear SICI both in M1 and in NPMA. At an ISI of 2 ms, the normalized ppMEP amplitude was more inhibited in M1 than in NPMA (35.3 ± 31.3% and 47.1 ± 34.8%, respectively; P < 0.05) (Figs. 2 and 3). At an ISI of 3 ms, the normalized amplitudes were 22.9 ± 15.3% in M1 and 34.1 ± 25.4% in NPMA (P = 0.33). ISIs of 10 and 15 ms showed facilitation in both areas with no significant differences. At an ISI of 10 ms, normalized ppMEP amplitudes were 140.1 ± 58.0% in M1 and 134.6 ± 126.0% in NPMA (P = 0.23). At an ISI of 15 ms, normalized ppMEP amplitudes were 141.7 ± 59.7% in M1 and 109.7 ± 52.0% in NPMA (P = 0.18) (Fig. 3, Tables II and III).

Figure 2.

Paired‐pulse MEP amplitudes were significantly more inhibited in M1 than in NPMA at an ISI of 2 ms. This figure illustrates average single‐pulse and ppMEP amplitudes and latencies in one subject at an ISI of 2 ms.

Figure 3.

Mean normalized ppMEP amplitudes at different ISIs. Asterisks indicate significant differences (P < 0.05) in the ppMEP amplitude between M1 and NPMA. At an ISI of 2 ms, the ppMEP amplitude was significantly more inhibited in M1 than in NPMA.

Table II.

Stimulation of M1: Individual absolute ppMEP amplitudes (μV) (mean ± SD) and relative ppMEP amplitudes (%) in parenthesis

Subject	CS 80% MT, ISI (ms) variable				ISI 2 ms, CS (%) variable
	2 ms	3 ms	10 ms	15 ms	30%MT	50%MT	70%MT	90%MT
1	278 ± 301 (11.6)	588 ± 510 (24.5)	2626 ± 643 (109.2)	2954 ± 869 (122.9)	1196 ± 700 (49.7)	1285 ± 1284 (53.4)	278 ± 314 (11.6)	201 ± 217 (8.4)
2	47 ± 42 (5.9)	108 ± 53 (13.3)	1597 ± 617 (198.3)	1789 ± 541 (222.1)	833 ± 629 (103.5)	540 ± 408 (67.0)	442 ± 384 (54.8)	102 ± 82 (12.6)
3	571 ± 494 (46.6)	106 ± 50 (8.6)	2492 ± 677 (203.2)	1000 ± 525 (81.6)	630 ± 375 (51.4)	515 ± 427 (42.0)	342 ± 205 (27.9)	223 ± 97 (18.2)
4	26 ± 17 (10.8)	63 ± 23 (26.7)	127 ± 52 (53.5)	146 ± 56 (61.7)	97 ± 84 (40.8)	94 ± 60 (39.5)	32 ± 11 (13.7)	16 ± 12 (6.6)
5	241 ± 75 (12.6)	386 ± 160 (20.2)	1450 ± 412 (76.0)	1127 ± 500 (59.0)	626 ± 783 (32.8)	872 ± 731 (45.7)	192 ± 93 (10.0)	140 ± 43 (7.3)
6	460 ± 465 (101.8)	293 ± 190 (64.9)	481 ± 228 (106.4)	1058 ± 287 (233.9)	333 ± 152 (73.6)	270 ± 180 (59.8)	131 ± 56 (29.0)	326 ± 186 (72.1)
7	75 ± 33 (57.2)	23 ± 16 (17.8)	126 ± 30 (95.7)	195 ± 116 (148.0)	725 ± 208 (551.1)	387 ± 319 (294.2)	59 ± 22 (45.0)	49 ± 57 (37.0)
8	100 ± 28 (27.0)	75 ± 29 (20.3)	643 ± 296 (174.3)	604 ± 136 (163.6)	122 ± 67 (33.0)	135 ± 61 (36.5)	33 ± 18 (9.0)	20 ± 13 (5.5)
9	907 ± 750 (49.8)	511 ± 490 (28.1)	2313 ± 1239 (127.1)	3441 ± 1072 (189.2)	3056 ± 1268 (168.0)	4216 ± 1847 (231.7)	926 ± 1159 (50.9)	2595 ± 949 (142.6)
10	6 ± 9 (1.3)	88 ± 50 (17.7)	1158 ± 589 (233.6)	771 ± 523 (155.6)	180 ± 108 (36.2)	228 ± 104 (46.1)	43 ± 22 (8.8)	40 ± 21 (8.1)
11	198 ± 142 (63.7)	30 ± 32 (9.6)	508 ± 143 (163.9)	375 ± 153 (120.9)	389 ± 267 (125.3)	527 ± 155 (169.8)	459 ± 173 (148.0)	250 ± 122 (80.5)
mean	264 (35.3)	206 (22.9)	1229 (140.1)	1224 (141.7)	744 (115.0)	824 (98.7)	267 (37.1)	360 (36.3)
SD	267	204	936	1088	837	1177	271	748

The interstimulus interval (ISI) and the conditioning stimulus (CS) intensity are shown in header line. The test stimulus intensity was always 120% MT.

Table III.

Stimulation of NPMA: Individual absolute ppMEP amplitudes (μV) (mean ± SD) and relative ppMEP amplitudes (%) in parenthesis

Subject	CS 80% MT, ISI (ms) variable				ISI 2 ms, CS (%) variable
	2 ms	3 ms	10 ms	15 ms	30%MT	50%MT	70%MT	90%MT
1	1793 ± 582 (69.0)	667 ± 454 (25.7)	1755 ± 725 (67.6)	3078 ± 1007(118.5)	2487 ± 1285 (95.7)	896 ± 884 (34.5)	826 ± 752 (31.8)	1259 ± 563 (48.5)
2	906 ± 559 (80.9)	522 ± 330 (46.6)	1103 ± 332 (98.4)	1328 ± 396 (118.5)	1221 ± 147 (109.0)	1272 ± 632 (113.5)	795 ± 386 (70.9)	1656 ± 276 (147.8)
3	1189 ± 597 (108.7)	100 ± 43 (9.1)	2188 ± 573 (200.1)	2076 ± 873 (189.9)	1092 ± 713 (99.9)	585 ± 348 (53.5)	605 ± 201 (55.3)	759 ± 412 (69.4)
4	151 ± 91 (32.3)	72 ± 31 (15.5)	407 ± 264 (87.2)	577 ± 206 (123.8)	416 ± 253 (89.2)	72 ± 33 (15.5)	23 ± 11 (5.0)	251 ± 186 (53.7)
5	453 ± 244 (24.6)	290 ± 141 (15.8)	502 ± 598 (27.3)	718 ± 595 (39.0)	418 ± 191 (22.7)	176 ± 155 (9.6)	129 ± 56 (7.0)	259 ± 157 (14.1)
6	575 ± 309 (89.1)	359 ± 149 (55.6)	878 ± 560 (136.1)	981 ± 609 (152.0)	453 ± 280 (70.2)	250 ± 137 (38.7)	186 ± 138 (28.9)	521 ± 229 (80.8)
7	54 ± 35 (16.7)	175 ± 83 (53.8)	1443 ± 535 (442.8)	624 ± 205 (191.6)	766 ± 356 (235.1)	696 ± 436 (213.6)	78 ± 44 (24.0)	558 ± 397 (171.2)
8	37 ± 20 (3.1)	35 ± 18 (3.0)	342 ± 214 (29.0)	411 ± 361 (34.9)	326 ± 162 (27.7)	932 ± 749 (79.1)	205 ± 127 (17.4)	170 ± 162 (14.5)
9	1061 ± 489 (45.3)	1381 ± 973 (58.9)	2198 ± 1071(93.7)	1475 ± 1395 (62.9)	1411 ± 1309 (60.2)	1871 ± 1382 (79.8)	1102 ± 983 (47.0)	358 ± 389 (15.2)
10	92 ± 132 (12.1)	598 ± 645 (78.8)	2009 ± 537 (264.7)	818 ± 817 (107.8)	490 ± 510 (64.5)	714 ± 260 (94.1)	169 ± 147 (22.3)	248 ± 225 (32.6)
11	305 ± 175 (36.6)	104 ± 25 (12.4)	284 ± 161 (34.0)	570 ± 213 (68.2)	339 ± 176 (40.6)	743 ± 395 (88.9)	251 ± 144 (30.1)	336 ± 120 (40.2)
Mean	601 (47.1)	391 (34.1)	1192 (134.6)	1151 (109.7)	856 (83.2)	746 (74.6)	397 (30.9)	580 (62.5)
SD	571	396	761	808	663	515	367	475

The interstimulus interval (ISI) and the conditioning stimulus (CS) intensity are shown in header line. The test stimulus intensity was always 120% MT.

In M1, CS intensities of 70% rMT and 90% rMT created a significant and almost equally intense inhibition at a constant ISI of 2 ms (P < 0.05). The normalized ppMEP amplitude was 37.1 ± 40.7% with a CS intensity of 70% rMT and 36.3 ± 44.4% with a CS intensity of 90% rMT. At lower CS intensities, the inhibitory effect was weaker or even absent. With a CS intensity of 30% rMT, the normalized ppMEP amplitude was 115.0 ± 151.2%, and CS set at 50% of rMT 98.7 ± 90.4% (Fig. 4, Table II). In NPMA significant (P < 0.05) inhibition was observed with CS intensities of 70% and 90% rMT (30.9 ± 20.0% and 62.5 ± 52.9%), respectively. Contrary to M1, also lower CS intensities showed significant inhibitory effect in NPMA (P < 0.05) but there was not significant difference in inhibitory effect when the effect in NPMA and in M1 was compared. At a CS intensity of 30% rMT, normalized ppMEP amplitude was 83.2 ± 58.1%, and at a CS intensity of 50% rMT 74.6 ± 57.0% (Fig. 4, Table III).

Figure 4.

Mean normalized ppMEP amplitudes at different CS intensities (ISI = 2 ms). In M1, ppMEP amplitudes are significantly inhibited with CS intensities 70 and 90% of rMT. In NPMA, ppMEP amplitudes are inhibited with all CS intensities, most with 70 and 90% of rMT. There were not statistical differences in inhibition when results of M1 and NPMA were compared.

DISCUSSION

Short‐latency MEPs were elicited in contralateral distal hand muscles in all subjects after TMS of the superior frontal gyrus. Stimuli evoking responses in OP muscle were found in the same cortical areas as in our previous study of the same subjects one year earlier [Teitti et al.,2008]. These areas of superior frontal gyrus correspond most likely to PMd and to Brodmann area 6 [Geyer2004; Picard and Strick2001]. In some subjects occasional responses were elicited quite medially, corresponding probably to SMA, or even more frontal areas, which may be part of Brodmann area 8 [Geyer2004]. As in our previous study, rMT of OP at NPMA was a slightly higher than the corresponding rMT at M1. Average MEP latencies were significantly shorter after stimulations of NPMA. Since the simultaneous stimulation of M1 during stimulation targeted to NPMA was explicitly excluded (the individually evaluated activating electric field in M1 stayed clearly below the electric field required to reach rMT in M1), the short latencies of MEPs from NPMA strongly suggest that motoneuron populations with direct monosynaptic connections to the spinal cord exist both in M1 and in NPMA.

The higher rMT in NPMA probably reflects functionally poorly organized, less dense corticospinal projections, as described previously in studies comparing MTs of different muscles [Chen et al.,1998]. In nonhuman primates, the MT of digit representations in PMA and SMA is two to three times higher than the MT in M1 [Boudrias et al.,2006; Dum and Strick,2002; Dum and Strick,2005; Luppino et al.,1991], and corticospinal projections from SMA to lower cervical segments of spinal cord are sparser and less dense than projections originating from M1 [Dum and Strick,1996]. Another explanation for higher rMT is a lower membrane excitability of motor neurones in NPMA (despite the lower SICI in this area). On the basis of earlier observations, the local motoneuron membrane excitability is not directly modulated by surrounding intracortical excitability [Ziemann et al.,1996b].

Optimal OP representations had more anatomical variation in NPMA than in M1 when mapping results of this study and our previous study were compared. As already suggested by the higher rMT, this may result from less dense and functionally less strictly organized muscle representations in NPMA when compared with representations in M1. This hypothesis would be in line with our earlier observation showing that stimulation of superior frontal gyrus elicited more often simultaneous activation of proximal and distal upper limb muscles from the same stimulation site than did the stimuli targeted to the optimal OP representation in M1 [Teitti et al.,2008].

Single‐pulse MEP amplitudes were higher from NPMA than from M1. Since the rMT was also higher in NPMA, the stimulator output corresponding to 120% of NPMA rMT could theoretically generate an electric field exceeding the rMT value on M1. However, this is a very unlikely since: (1) the optimal direction of current eliciting MEPs in NPMA was from lateral to medial direction contrasting that in M1, where optimal current flow was from posterior to anterior direction and perpendicular to the central sulcus, in line with previous studies [Brasil‐Neto et al.,1992a; Mills et al.,1992]; (2) the optimal site for stimulation of NPMA was over 30 mm distant from the optimal stimulation site in M1; (3) the MEP latencies from NPMA were shorter than those elicited by stimulation of M1. Polysynaptic activation of M1 motor neurones via cortico‐cortical projections from NPMA to M1 should produce prolonged MEP latencies [Tokuno and Nambu2000].

Higher single‐pulse MEP amplitudes indicate most probably that simultaneous activation of upper motoneurons in NPMA may generate powerful and well synchronized corticospinal output. Our observation is conflicting with electrical microstimulation studies of NPMA and M1 in nonhuman primates [Boudrias et al.,2006; Boudrias et al.,2010a; Boudrias et al.,2010b], where muscle responses are significantly larger when M1 is stimulated. This difference between electrical cortical stimulation studies and TMS may derive from the fact that the size of stimulated cortical area is orders of magnitude larger when cortical activation elicited by TMS is observed. In line with this hypothesis, lower single‐pulse MEP amplitudes from M1 may result from incomplete saturation of M1 motoneurons although anatomical density is high. Functionally, MT is lower when upper motoneurons density is higher [Chen et al.,1998] and, subsequently, stimulated cortical area remains smaller. Thus, it cannot be concluded that corticospinal output from NPMA is stronger than output from M1 although MEP amplitudes are higher. In NPMA, more extensive saturation of less dense motoneuron population may have been achieved by the higher stimulation intensity.

Shorter latencies of MEPs evoked from NPMA are likely related to higher MEP amplitudes. Suprathreshold stimuli elicit larger MEPs with shorter latencies and longer duration than weaker stimuli [Caramia et al.,1989]. We may speculate that the stimulation of OP representation in M1 with the same absolute intensity as representation in NPMA would evoke MEPs with equal latencies when compared to stimulations of NPMA as was observed in our previous study [Teitti et al.,2008]. Another explanation for shorter MEP latencies could be direct axonal stimulation. Direct axonal stimulation is more common if the TMS induced current flows latero‐medially, as in our stimulations of NPMA, than to postero‐anterior direction [Werhahn et al.,1994]. Although direct axonal stimulation could explain shorter latencies in single‐pulse MEPs it is implausible explanation because in paired‐pulse measurements the significant inhibition of test‐pulses was observed also in NPMA. SICI should be absent if the test‐pulse activates axons directly [Kujirai et al.,1993].

NPMA exhibited a similar pattern of inhibition (SICI) with short ISIs (2–3 ms) and facilitation (ICF) with longer ISIs (10–15 ms), as described previously for M1 [Chen et al.,1998; Di Lazzaro et al.,1998; Kujirai et al.,1993; Nakamura et al.,1997; Valls‐Sole et al.,1992; Ziemann et al.,1996c]. The effect of different CS intensities showed the same typical U‐shaped curve (SICI curve) in NPMA as in M1, with most prominent inhibition in the mid‐range of the stimulus intensities tested, and less inhibition at higher or lower intensities [Chen et al.,1998; Schafer et al.,1997]. Even though the basic pattern was similar, the inhibition elicited by 2 ms ISI and CS at 80% of rMT was significantly smaller in NPMA than in M1. It is unlikely that differences in current direction in M1 and in NPMA would explain the decreased SICI in NPMA [Ziemann et al.,1996c], since the tangential fibers which mediate the effect of CS do not have a uniform organizational pattern in cortex. We suggest that the difference in SICI may indicate either diminished activation of inhibitory circuits or relatively enhanced activation of excitatory circuits. The stronger activation of excitatory circuits inducing higher ppMEPs at short ISIs has been reported by Ridding et al., [1995], providing evidence that the overall changes in ppMEP amplitudes are always net‐effects of both excitatory and inhibitory circuits.

The similarity of the inhibitory effect in M1 and NPMA at an ISI of 3 ms, and the disparity at an ISI of 2 ms, may reflect the function of an isolated circuit, which would then be mostly responsible for the short‐interval ICI in NPMA. Even with lower SICI at an ISI of 2 ms in NPMA, we did not observe stronger facilitation at longer ISIs, which might have been expected considering the findings of Hanajima et al., [1998] showing that ICF is modulated by the preceding GABAergic inhibition. The inhibitory effects at short ISIs may be divided into at least two conditioning intervals producing maximal inhibition, suggesting the activity of two separate intracortical neuronal circuits. In earlier studies, the maximal inhibition was observed at 1 ms and 2.5 ms ISIs [Fisher et al.,2002; Roshan et al.,2003], and different I‐waves were inhibited at an ISI of 1 ms and of 3‐5 ms [Hanajima et al.,2003]. The latest studies suggest that SICI at 2 ms ISI represents the most pronounced SICI, and SICI at slightly longer ISIs (>2 ms) is contaminated by simultaneous short‐interval cortical facilitation (SICF) that has been described at discrete ISI of 1.5, 2.5–3.1, and 4.5 ms [Peurala et al.,2008]. It is thus highly unlikely that the dissociation of 2 ms and 3 ms SICI would reflect the refractoriness of activated neuronal population, and we speculate that the dissociation of 2 and 3 ms inhibitory effects is either due to two separate local circuits, or significantly different net‐effect of the circuits mediating SICI and SICF in M1 and in NPMA.

The less prominent SICI in NPMA at an ISI of 2 ms (CS 80% of rMT) may be related to poorer topographic organization. Hypothetically, it may reflect higher intracortical excitability and readiness to provide strong excitatory output despite strong inhibitory input from the surroundings. This capacity may facilitate the mobilization of these motor representational resources during post‐lesional changes in the motor network [Fridman et al.,2004; Johansen‐Berg et al.,2002; Rossini et al.,2003] or while practising a new motor skill [Karni et al.,1995; Karni et al.,1998; Kim et al.,2004; Kleim et al.,2004; Muellbacher et al.,2001; Pascual‐Leone et al.,1995; Rosenkranz et al.,2007; Tyc et al.,2005].

CONCLUSIONS

Our findings suggest that distal hand muscle representations in superior frontal gyrus are connected monosynaptically to lower motoneurons, and are less dense and less strictly organised as those in M1. Simultaneous extraneous, and probably intrinsic, activation of these sparser and less dense corticospinal projections which reside in NPMA may activate lower motoneurons quite effectively and produce powerful muscle responses. Local intracortical inhibitory and excitatory activity is quite similar to that in M1. The lower SICI in NPMA at an ISI of 2 ms is taken to reflect less strict topographic organization and cortical readiness to react to powerful input from other parts of sensorimotor network, which may be relevant to the reorganization of neural circuits in recovery from motor deficits or during motor learning.