Perception of sensory information by the brain requires highly ordered synaptic connectivity between peripheral sensory neurons and their targets in the central nervous system. Somatosensory information is processed in the spinal cord dorsal horn, which is responsible for modulating, integrating, and relaying to higher brain centers various types of sensory input using a complex circuit that is established during embryonic development (for review, see Gillespie and Walker,2001; Julius and Basbaum,2001). The molecular events that govern the generation of specific types of dorsal horn sensory neurons along the neural tube began to be unmasked only recently. Sensory neurons are born and start to migrate towards their final location along the dorso-ventral axis in a highly spatial and temporal order (Goulding et al.,2002; Caspary and Anderson,2003; Helms and Johnson,2003) due to the expression of multiple patterning genes that encode a set of homeodomain transcription factors in a combinatorial manner (Lee and Jessell,1999; Jessell,2000; Goulding et al.,2002; Müller et al.,2002; Helms et al.,2005). These genes are important to instruct neurons to fulfill their differentiation program (Chen et al.,2001; Müller et al.,2002; Qian et al.,2002, Zhou and Anderson,2002; Cheng et al.,2004; Ding et al.,2004).
In the mouse spinal cord, dorsal horn neurons arise from progenitors in the ventricular zone in two neurogenic waves between embryonic days 10 and 14 (E10–14). The first wave takes place between E10.5 and E11.5 and generates six subpopulations (dI1–6) of early-born neurons that will populate the deep dorsal horn. The second wave (E12–14.5) originates two late-born neuronal populations (dILA and dILB), which migrate dorsally and form the superficial dorsal horn (Gross et al.,2002; Müller et al.,2002). The early-born neurons can be subdivided into two major classes: Class A (dI1–3) neurons are born in the dorsal alar plate, depend on roof plate signals, and are Lbx1-independent (Liem et al.,1997; Lee et al.,1998,2000; Wine-Lee et al.,2004); Class B neurons (dI4–6) arise from the ventral alar plate, are independent of roof plate signals, and are Lbx1-dependent (Pierani et al.,2001; Gross et al.,2002; Matise, 2002; Müller et al.,2002; Cheng et al.,2004). dI3 and dI5 neurons express the glutamatergic fate determinant Tlx3 (Cheng et al.,2004). dI1–3 neurons are thought to be involved in proprioceptive processing (Bermingham et al.,2001; Gowan et al.,2001). As to late-born neurons, dILA neurons express Pax2, Lhx1/5, and Ptf1a, and dILB neurons are Tlx1/3-positive and Lmx1b-positive (Gross et al.,2002; Müller et al.,2002; Cheng et al.,2005). Glutamatergic neurons differentiate from the dILB population, while dILA neurons follow the GABAergic fate (Gross et al.,2002; Müller et al.,2002; Cheng et al.,2005; Glasgow et al.,2005).
Prrxl1 (also known as Drg11) is a paired-like homeodomain protein that is expressed both in early-born (dI3 and dI5) and glutamatergic late-born (dILB) neurons (Rebelo et al.,2007). Mice with deletions in Prrxl1 gene present defects in the spinal cord superficial dorsal horn similar to those observed in Tlx3/1 or Lmx1b knockout mice. This is suggestive that the three transcription factors take part in the genetic cascade involved in building up the superficial spinal cord neuronal circuitry (Chen et al.,2001; Gross et al.,2002; Müller et al.,2002; Qian et al.,2002). The specific role of these genes remains, however, to be characterized. In Prrxl1 knockout mice (Prrxl1−/−), abnormalities in DRG spinal projections, superficial dorsal horn structure and neurochemistry, and nociceptive responses have been identified (Chen et al.,2001). Recent data point to a role for Prrxl1 in early postnatal survival of normally differentiated small-size primary afferent neurons innervating various kinds of peripheral tissues, which would explain the nociceptive deficits observed in Prrxl1-null mutant mice (Rebelo et al.,2006).
This study investigates the role of Prrxl1 in the specification of nociceptor targets in the spinal superficial dorsal horn. Taking advantage of multiple-labeling combining immunohistochemical detection of Prrxl1 and the two functionally related transcription factors, Tlx3 and Lmx1b, which are likely to be upstream to Prrxl1 (Ding et al.,2004), we were able to define various subsets of early- and late-born neuronal populations. Moreover, in perinatal Prrxl1−/− mice, expression of Tlx3 and Lmx1b in the spinal cord superficial dorsal horn was deeply reduced, the preserved neurons equaling in number those that did not express Prrxl1 in the wild-type mice. Golgi studies confirmed the absence of a large amount of neurons in laminae I–III of Prrxl1−/− mice but not in the deep dorsal horn. Accordingly, the number of neurons c-fos activated after cutaneous and visceral noxious stimulation was reduced but only in the superficial dorsal horn, suggesting that the remaining neurons are not functionally affected by the loss of Prrxl1-dependent neurons.
Migration Pattern of Glutamatergic Neurons in the Developing Spinal Cord
Previous immunohistochemical studies showed that Prrxl1 is detected in mice from the time of neural tube closure (E10.5) up to P21, becoming restricted to the spinal cord superficial dorsal horn (Rebelo et al.,2007). Two functionally related transcription factors, Tlx3 and Lmx1b, have expression patterns similar to Prrxl1 and co-localize with VGLUT2, which was taken as indicative of their involvement in glutamatergic specification (Cheng et al.,2004). To better define the role of Prrxl1 in the differentiation of the glutamatergic neuronal population, a time course analysis of the expression of Prrxl1 and either Tlx3 or Lmx1b was performed using double-labeling immunohistochemistry (Fig. 1). At E10.5, Prrxl1 was exclusively present in the dI3 and dI5 early-born neuronal subpopulations, as confirmed by the absence of co-localization with Lim1/2, a marker of dI2, dI4, and dI6 interneurons (Fig. 1A, H). Prrxl1-positive neurons lied adjacent to but outside the ventricular zone, indicating that they are postmitotic neurons. Tlx3 marked dI3 and dI5 neurons (Fig. 1A), while Lmx1b marked dI5 neurons exclusively (Fig. 1H). Prrxl1 co-localized extensively both with Tlx3 (Fig. 1A) and Lmx1b (Fig. 1H). At E11.0, both dI3 and dI5 Prrxl1-expressing neurons started the migration towards the neural tube (Rebelo et al.,2007), but a few dI5 neurons appeared to form streams that extended to the ventral neural tube (Fig. 1B–F, I–M; small arrows). At E12.0 and E12.5, numerous newly formed (late-born) neurons were concentrated adjacent to the ventricular zone and along a stream in lateral migration (Fig. 1C,D,J,K; large arrows). Between E14.5 and E16.5, neurons assumed their final position in the superficial dorsal horn (Fig. 1E,F,L,M). At these ages, as well as at P7 (Fig. 1G,N), Prrxl1-positive neurons were mainly located in lamina II (Tables 1–3; see also Supp. Fig. S1, which is available online) and co-expressed extensively either Tlx3 (Fig. 1E–G; Table 1) or Lmx1b (Fig. 1L–N; Table 3).
Table 1. Prrxl1 and Tlx3 Co-Expression in Superficial Laminae of P7 Dorsal Horn Spinal Cord
Nuclei number per section
62 ± 20
58 ± 18
100 ± 30
224 ± 52
196 ± 44
270 ± 38
82 ± 22
42 ± 22
64 ± 26
368 ± 92
296 ± 84
434 ± 94
Table 3. Prrxl1 and Lmx1b Co-Expression in Superficial Laminae of P7 Dorsal Horn Spinal Cord
Nuclei number per section
Prrxl1 + Lmx1b
52 ± 24
50 ± 24
94 ± 30
230 ± 54
198 ± 58
278 ± 80
60 ± 24
56 ± 22
94 ± 34
342 ± 102
304 ± 104
466 ± 144
Altogether, these data unveil a highly dynamic synchrony in the migration of neurons expressing Prrxl1, Tlx3, and Lmx1b, and show a high degree of co-expression between the three transcription factors along development of glutamatergic neurons.
Co-expression of Prrxl1, Tlx3, and Lmx1b Defines Subsets of Glutamatergic Superficial Dorsal Horn Neurons
In order to ascertain whether co-expression of the various transcription factors identifies different categories among dorsal horn glutamatergic neurons, counting of early-born and late-born neurons was performed at E10.5 and P7, respectively. It should be noted that staining intensity varied considerably, suggesting differential levels of expression for each transcription factor. Nevertheless, all labeled neurons were considered for counting irrespective of their staining level.
Prrxl1-positive neurons were present in similar numbers in dI3 and dI5 regions (Fig. 2B,D). As expected from the results obtained by Müller et al. (2002) and Cheng et al. (2005), the entire dI5 Prrxl1-positive neuronal population co-expressed the dI4–dI6 marker, Lbx1 (Fig. 2A).
All Prrxl1-positive neurons present in both the dI3 and dI5 regions also expressed Tlx3 (Fig. 2B,B′,B″). However, a large subpopulation of Tlx3-positive neurons in both domains did not co-localize with Prrxl1 (Fig. 2B,B′,B″). dI3 glutamatergic neurons fell into two subpopulations: one expressing only Tlx3 (Prrxl1−/Tlx3+, 66%; Fig. 2E) and another expressing both Prrxl1 and Tlx3 (Prrxl1+/Tlx3+, 34%; Fig. 2E).
In the dI5 region, Tlx3-positive neurons also co-expressed Lmx1b extensively (Fig. 2C,C″). Since, in addition, some Prrxl1 neurons did not express Lmx1b (Fig. 2D,D″) and all Lmx1b-positive neurons co-expressed Tlx3 (Fig. 2C,C″), three distinct dI5 neuronal populations were identified: a prevalent one, expressing Tlx3 and Lmx1b (Prrxl1−/Tlx3+/Lmx1b+, 50%), and two other identical in number expressing either Prrxl1 and Tlx3 (Prrxl1+/Tlx3+/Lmx1b−, 23%) or Prrxl1, Tlx3, and Lmx1b (Prrxl1+/Tlx3+/Lmx1b+, 27%) (Fig. 2F).
It is known that by E14.5, neurogenesis in the dorsal spinal cord has already ceased (Nornes and Carry,1978). From this time point on, dorsal spinal cord neurons mature and start to populate the dorsal horn at their final position in the various laminae that constitute the spinal dorsal horn (Müller et al.,2002). Counting of late-born neurons was performed at P7 because this age has the advantage of presenting the dorsal spinal cord neatly laminated and expressing Prrxl1 at levels high enough to allow its detection (Rebelo et al.,2007).
Prrxl1-positive neurons spanned the entire superficial dorsal horn from laminae I to III (Fig. 3A,B,D; Tables 1 and 3) and co-localized extensively with Lbx1 in lamina III (Fig. 3A), being extremely scattered in the deep dorsal horn (lamina IV–V, 5 neurons on average; 1.8% of all Prrxl1-expressing neurons; Fig. 3A,B,D). They were more abundant in lamina II (Tables 1 and 3; chart in Fig. 3), where they amounted to 65% of all Prrxl1-immunoreactive neurons (Supp. Fig. S1). The distribution per lamina of Tlx3- and Lmx1b-positive neurons followed the same pattern (Tables 1–3; chart in Fig. 3; Supp. Fig. S1).
A large amount of Prrxl1-positive neurons co-expressed Tlx3 (81%, Table 1) or Lmx1b (89%, Table 3) (Fig. 3B, D). Tlx3 neurons (Fig. 3B) and Lmx1b neurons (Fig. 3D) co-expressed Prrxl1 in about 60 to 70% in all superficial dorsal horn laminae (Tables 1 and 3; chart in Fig. 3). Considering that all Tlx3-positive neurons also expressed Lmx1b (Fig. 3C; Table 2; Dai et al.,2008), four distinct neuronal populations could be identified (Fig. 3E). Neurons expressing Prrxl1 contributed altogether to 74% of the entire superficial dorsal horn glutamatergic neuronal population and belonged in three categories (Fig. 3E): neurons expressing Prrxl1, Tlx3, and Lmx1b (Prrxl1+/Tlx3+/Lmx1b+, 58%); neurons expressing both Prrxl1 and Lmx1b (Prrxl1+/Tlx3−/Lmx1b+, 7%); and neurons expressing only Prrxl1 (Prrxl1+/Tlx3−/Lmx1b−, 8%). The remaining 27% neurons were Prrxl1-negative and expressed both Tlx3 and Lmx1b (Prrxl1−/Tlx3+/Lmx1b+) (Fig. 3E). Notably, neurons co-localizing Tlx3 and Lmx1b with or without Prrxl1 amounted to 85% of all superficial dorsal horn glutamatergic neurons (Fig. 3E) and predominated in laminae I–II (Fig. 3C), whereas Lmx1b neurons that did not express Tlx3 were concentrated in lamina III (Fig. 3C). This lamina III population is likely to correspond to the Prrxl1+/Tlx3−/Lmx1b+ neurons since Prrxl1 neurons that did not co-express Tlx3 in lamina III (Fig. 3B) were much more abundant (40 in Table 1) than those that did not co-express Lmx1b (Prrxl1+/Tlx3−/Lmx1b−, 4 in Table 3; Supp. Fig. S2). Moreover, the Prrxl1+/Tlx3−/Lmx1b− subpopulation prevailed in lamina II since Prrxl1-positive neurons non-expressing either Tlx3 (Table 1) or Lmx1b (Table 3) were similar in number and concentrated in lamina II (84%; Supp. Fig. S2).
Table 2. Tlx3 and Lmx1b Co-Expression in Superficial Laminae of P7 Dorsal Horn Spinal Cord
Nuclei number per section
Tlx3 + Lmx1b
94 ± 30
94 ± 30
95 ± 30
264 ± 38
264 ± 38
275 ± 70
64 ± 26
64 ± 26
100 ± 34
422 ± 94
422 ± 94
470 ± 134
Defective Development of Superficial Dorsal Horn Glutamatergic Neurons in Prrxl1 Mutant Mice
In the spinal cord of Prrxl1−/− embryos, the differentiation and migration of superficial dorsal horn neurons expressing Tlx3 or Lmx1b occurred normally until E14.5, in accordance with the results of Chen et al. (2001). From E18.5 on, a progressive decrease in their number was, however, observed (Fig. 4), in accordance with previous findings showing apoptosis from E17.5 (Chen et al.,2001). In order to clarify whether the absence of Prrxl1 induces cell death only in neurons that normally express it or, alternatively, also affects the other Prrxl1-negative glutamatergic and the GABAergic subpopulations by following another differentiation pathway, the expression profile of the spinal dorsal horn of P7 Prrxl1−/− mice was compared with that of wild-type mice (Fig. 5). This developmental age was selected because Prrxl1 expression markedly decreases after this age (Rebelo et al.,2007), while the amount of Tlx3- and Lmx1b-expressing neurons is similar to that observed at P21.
In the P7 mutant mice, Tlx3- and Lmx1b-positive neurons were highly packed in a narrow strand at the dorsal horn surface (Fig. 5B,E). Notably, the number of Tlx3- or Lmx1b-positive neurons in the Prrxl1−/− mice was similar to those of neurons Prrxl1-negative but Tlx3- and Lmx1b-positive (Fig. 5F) in wild-type mice. These data indicate that, in the absence of Prrxl1, there are no additional neurons differentiating into the Prrxl1−/ Tlx3+/Lmx1b+ subpopulation (Supp. Fig. S3), which accounts for about 30% of the entire glutamatergic population of the superficial dorsal horn (Fig. 3E). Similarly, the number of Pax2-positive neurons was identical in wild-type and Prrxl1−/− mice (Fig. 5G–I), suggesting that the development of GABAergic neurons in the superficial dorsal horn is not influenced by Prrxl1.
Departing from the dramatic loss of superficial dorsal horn neurons in Prrxl1−/− mice, we investigated how such loss affected the anatomy of the spinal cord dorsal horn using the Golgi-Rio Hortega method. The dorsal funiculus and the dorsal horn of Prrxl1−/− mice were shorter dorso-ventrally as compared to the wild-type (Fig. 6A,B,E), as previously described (Chen et al.,2001). As in other mammals (Beal et al.,1981; Lima and Coimbra,1986), the most superficial laminae of the spinal dorsal horn of wild-type mice were characterized by the presence of small-size spiny neurons, which amounted to 352. In contrast, the population of the deep dorsal horn (390) included a large amount of neurons with large cell bodies and aspiny dendritic arbors (Fig. 6B–D). In the mutant mice, small neurons at the surface of the dorsal horn were scarce (56), although neurons belonging in the four neuronal types typically present in lamina I (Lima and Coimbra,1983; Galhardo and Lima, 1999; Galhardo et al.,2000) could be recognized (Fig. 6A,E). Deep dorsal horn neurons were similar in number (422) to the wild-type and were located near the dorsal surface (Fig. 6A,E). These data indicate that, in the absence of Prrxl1, superficial dorsal horn laminae are reduced to a narrow strand and deprived of most of their neurons, while the deep dorsal horn is not affected.
Defective Nociception Activation of the Spinal Cord Dorsal Horn in Prrxl1−/− Mice
Taking into account the marked fall in number of small size neurons in the superficial dorsal horn of Prrxl1−/− mice, we investigated to what extent the mutation affected noxious-induced neuronal activation at the spinal level. To address this issue, spinal expression of the Fos protein, a marker of neuronal activity (Hunt et al.,1987; Menétrey et al.,1989), was assessed by immunohistochemistry in wild-type and mutant P21 mice after somatic and visceral noxious stimulation. Counting of Fos-immunoreactive cells was referred to the superficial dorsal horn and the deep dorsal horn. Bearing in mind that the aberrant morphology of the spinal cord of Prrxl1−/− mice impaired its correct lamination and that the deep dorsal horn seemed not to be affected in Golgi impregnations, we considered as the deep dorsal horn in the mutant the area equal in ventro-dorsal extent from the central canal plane to that defined as the deep dorsal horn in wild-type mice.
No Fos-immunoreactivity was detected in the spinal cord of either wild-type or mutant animals not subject to noxious stimulation. In wild-type mice, neurons Fos-immunoreactive after thermal noxious stimulation were located in the medial third of the superficial dorsal horn (31 ± 9) and in the deep dorsal horn (27 ± 8) (Fig. 7A,C,D). In Prrxl1−/− mice, a notable reduction in the number of Fos-expressing neurons occurred in the superficial dorsal horn (5 ± 2), but not in the deep dorsal horn (19 ± 8) (Fig. 7B–D). In wild-type mice, after visceral stimulation, Fos-staining was observed in the superficial dorsal horn (38 ± 6), as well as in the neck of the dorsal horn and in lamina X (79 ± 10) (Fig. 7E,G,H). In mutant mice, the superficial dorsal horn was almost deprived of immunoreactive neurons (9 ± 6), whereas the deep dorsal horn presented a similar amount (58 ± 18) of similarly located Fos-positive neurons (Fig. 7F–H).
The fact that Fos expression in the Prrxl1−/− mice was dramatically reduced in the superficial dorsal horn, but not in the deep dorsal horn, confirms the marked impairment of superficial dorsal horn development in the mutant mice.
Differential Co-Expression of Prrxl1, Tlx3, and Lmx1b Underlies Cellular Heterogeneity Among Superficial Dorsal Horn Glutamatergic Neurons
Most of the ascending projection neurons and local circuit interneurons present in the dorsal horn of the spinal cord are excitatory and use glutamate as neurotransmitter (Azkue et al.,1998; Lu and Perl,2003; Santos et al.,2007). Here we show for the first time that the superficial dorsal horn glutamatergic population expressing Tlx3 and/or Lmx1b (Cheng at al.,2004), comprises various subsets of neurons as defined by the differential combinatorial expression with the transcription factor Prrxl1. Postnatally four subpopulations were identified (Prrxl1+/Tlx3−/Lmx1b−, Prrxl1+/Tlx3+/Lmx1b+, Prrxl1+/Tlx3−/Lmx1b+, and Prrxl1−/Tlx3+/Lmx1b+), while early-born neurons were distributed through two subpopulations in dI3 (Prrxl1+/Tlx3+ and Prrxl1−/Tlx3+) and three subpopulations in dI5 (Prrxl1+/Tlx3+/Lmx1b+, Prrxl1+/Tlx3+/Lmx1b− and Prrxl1−/Tlx3+/Lmx1b+). Curiously, while all newly formed glutamatergic neurons express the glutamatergic fate determinant gene, Tlx3 (Cheng et al.,2004), postnatally two of the four subpopulations described do not express Tlx3 and the great majority of glutamatergic neurons are Lmx1b-positive. This is, however, in accordance with recent findings from Xu and collaborators (2008) showing that some Tlx3-dependent neurons located in lamina III, in the same location as the Prrxl1+/Tlx3−/Lmx1b+ described here, do no longer express Tlx3 postnatally.
In Prrxl1−/− spinal cord, both the initial migration pattern and the differentiation of Lmx1b neurons do not seem to be affected until E14.5 (Chen et al.,2001). At E15.5, a decrease in Nissl staining compared to wild type mice is detected, whereas higher levels of apoptosis are observed at E17.5 (Chen et al.,2001). The present observation that, in this mutant, a large fraction of Tlx3-positive and Lmx1b-positive superficial dorsal horn neurons is absent from E18.5 on is in line with these previous findings on putative neuronal degeneration after E15.5 in the absence of Prrxl1 (Chen et al.,2001), and reveals the glutamatergic nature of this Prrxl1-dependent population. In this respect, it should be noted that, although lamina II neurons have for long been claimed to exert an inhibitory role (Willis and Coggeshall,1991), the occurrence of an excitatory, glutamatergic subpopulation amounting to 85% of lamina II neurons was recently demonstrated (Santos et al.,2007,2009), in agreement with developmental data showing that Tlx3-positive neurons populate this lamina (Cheng et al.,2004) and the fact that Pax2-positive neurons amount to about 20% of laminae I–III neurons (present results). Moreover, the present study clearly shows that in Prrxl1−/− mice, the superficial dorsal horn glutamatergic population that does not express Prrxl1(Prrxl1−/Tlx3+/Lmx1b+) is preserved in amounts similar to those occurring in the wild-type mice. The same occurred with the GABAergic (Pax-2-positive) population, further supporting that Prrxl1-dependent neurons degenerate instead of following another differentiation pathway.
Prrxl1 spinal expression is highly reduced in Tlx3 knockout mice around E16.5, and completely abolished in Tlx3/Tlx1 double knockouts (Qian et al.,2002). This suggests that the Prrxl1-expressing neurons present in the Tlx3 mutant mice may correspond to the Prrxl1+/Tlx3−/Lmx1b+ and Prrxl1+/Tlx3−/Lmx1b− subpopulations described here, which are therefore likely to also express Tlx1. On the other hand, in Lmx1b−/− mice at E15.5, Prrxl1 expression was reported to be totally absent (Ding et al.,2004). Here, however, we uncovered a minute proportion of both early-born and late-born Prrxl1-positive neurons that does not express Lmx1b. Taking into account that the anti-Prrxl1 antibody used in this study does not discriminate between Prrxl1 and its recently described spliced variant Prrxl1-b (Rebelo et al.,2009), it is possible that the Prrxl1+/Tlx3−/Lmx1b− neurons express the splice variant of Prrxl1b.
Spinal Nociceptive Processing Impairment in the Absence of Prrxl1
In the spinal cord dorsal horn, neurons relaying nociceptive input to supraspinal levels are distributed through lamina I and the deep dorsal horn, while lamina II is almost exclusively populated by local circuit, modulatory interneurons (Willis and Coggeshall,1991). Here we used the Golgi-Rio Hortega silver impregnation method to better evaluate to which extent the loss of Prrxl1-expressing neurons affected the neuronal population of the superficial and deep dorsal horn laminae. In Prrxl1−/− mice, a clear reduction (85%) in the number of superficial dorsal horn neurons was observed, whereas no changes were detected among neurons of the deep dorsal horn. The number of Golgi-impregnated superficial neurons was, however, far below what would be expected from the sum of glutamatergic Prrxl1-independent neurons and GABAergic Pax2-positive neurons at P7 (about 50% of the total dorsal horn neurons) (Fig. 5). The fact that the Golgi method impregnates a small fraction of the neurons present in a certain area may explain these figures, although the possibility that, due to reorganization of the dorsal horn, some of these neurons are now intermingled with deep dorsal horn neurons cannot be ruled out.
The marked decrease in the number of noxious-evoked Fos-positive neurons in the superficial dorsal horn of Prrxl1−/− mice was not accompanied by any change in Fos- activation in the deep dorsal horn. This points to preserved peripheral nociceptive innervation of the deep dorsal horn in the absence of Prrxl1. Previous studies have shown that, although in Prrxl1−/− mice marked postnatal death of small-size DRG neurons takes place with disruption of sensory innervation of various peripheral tissues, about 2/3 of the peptidergic and non-peptidergic DRG neurons survive (Rebelo et al.,2006). The present finding strongly suggests that these remaining primary afferent neurons are those that innervate the deep dorsal horn, and reinforces the hypothesis that postnatal death of DRG neurons in Prrxl1−/− mice might be due to them not finding their neuronal targets in the superficial dorsal horn.
On the other hand, preserved noxious-evoked activation of deep dorsal horn neurons in Prrxl1−/− mice suggests that the amputation of the spinal cord superficial modulatory circuitry does not significantly affect nociceptive processing in the deep dorsal horn. According to the molecular characterization of the superficial dorsal horn neuronal population carried out here, the local spinal cord pain control circuit of Prrxl1−/− mice must be deprived of an important excitatory component. The absence of lamina II excitatory interneurons should result in substantially reduced activation of deep dorsal horn projecting neurons by disrupting the balance between excitation and inhibition of nociceptive transmission, namely through local GABAergic neurons, whose numbers were not changed. The unexpected observation of unchanged c-fos induction levels in the deep dorsal horn raises the possibility that the spinal and supraspinal pain control system has adapted to this condition during development in order to reestablish the lost balance.
However, the apparently normal activation of nociceptive deep dorsal horn neurons in Prrxl1−/− mice (present data) coexists with a significant depression of nociceptive responses in a variety of acute pain behavioral tests (Chen et al.,2001). This poses interesting questions as to the relative role of lamina I and deep dorsal horn transmission neurons in pain processing and reveals that Prrxl1−/− mice may constitute a unique model to address this issue, provided that postnatal survival can be prolonged. Anatomofunctional studies of the nociceptive neuronal network in this model would give invaluable information on the role of deep dorsal horn projection neurons and substantia gelatinosa modulatory interneurons in pain processing.
Wild-type mice and Prrxl1−/− littermates were generated by heterozygote intercrosses and genotyped as previously described (Chen et al.,2001). Animals were bred and maintained at the IBMC animal facility under temperature- and light-controlled conditions. The embryonic day 0.5 (E0.5) was considered to be the midday of the vaginal plug. Experiments were carried out in accordance with the European Community Council Directive (86/609/EEC) and the ethical guidelines for pain investigation in animals (Zimmermann,1983).
Embryos were removed by caesarian surgery of pregnant females under sodium pentobarbital anaesthesia (50 mg/kg i.p.), fixed by immersion for 4 hr in freshly prepared 4% paraformaldehyde in 0.1M phosphate-buffered saline pH 7.4 (PBS), cryoprotected in 30% sucrose in PBS overnight, and sectioned transversally at the hind-paw level on a cryostat at 12 μm. Post-natal mice were perfused through the ascending aorta with 5 ml of PBS followed by 50 ml of 4% paraformaldehyde in PBS. Spinal cords were removed, immersed in the same fixative for 2 hr, and cryoprotected in 30% sucrose in PBS overnight. Lumbar coronal frozen sections were cut on a cryostat at 12 μm, except for tissue that underwent c-fos analysis or Golgi impregnation, in which case one in every four 40-μm-thick cryostat sections and serial 150-μm frozen sections, respectively, were cut from spinal segments L4 and L5.
Primary antibodies used were: rabbit anti-Prrxl1 (1:500; Rebelo et al.,2007), guinea pig anti-Tlx3, anti-Lmx1b, and anti-Lbx1 (1:1,000; kindly provided by C. Birchmeier and T. Müller), rabbit anti-Pax2 (1:1,000, Invitrogen), rabbit anti-Fos (1:10,000, Oncogene Science, Gaithersburg, MD), and mouse anti-Lim1/2 (1:20, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA). Secondary Alexa-conjugated antibodies (Invitrogen, Carlsbad, CA) were used to detect Prrxl1, Tlx3, Lmx1b, Lbx1, and Pax2, and biotinylated swine anti-rabbit antibody to detect the Fos protein (Dako, Carpinteria, CA). Vectastain ABC-HRP standard kit (Vector Labs, Burlingame, CA) was used to detect the Fos signal with the DAB chromogen. Sections were cleared in xylol and mounted in Eukitt®. Spinal cord counter-staining was accomplished using Topro (Invitrogen) and images were captured on a Zeiss (Z1, Thornwood, NY) fluorescence microscope equipped with an ApoTome system.
Spinal cord blocks from L4–L5 removed from wild-type (n = 2) and Prrxl1−/− (n = 2) P7 mice were double impregnated by the Golgi-Río Hortega method (Ramón y Cajal and de Castro,1933; Sotelo and Palay,1968). Briefly, blocks were immersed in the potassium dichromate solution for 48 hr (renewed after 24 hr), 0.75% silver nitrate overnight, the potassium dichromate solution for another 24 hr, and the silver nitrate solution overnight. Frozen serial 150-μm-thick transverse sections were obtained, and neuronal counting performed using five sections randomly taken from each animal.
Noxious Stimulation for c-fos Induction
Experiments were carried out in 22 P21 mice (12 wild-type and 10 Prrxl1−/−), which were further subdivided in 4 groups. The first group consisted of 4 wild-type and 4 Prrxl1−/− mice receiving cutaneous noxious thermal stimulation. Animals were anaesthetized with sodium pentobarbital and their right hindpaw was submersed for 10 sec in hot water (52°C) every minute, during 20 min. The second group, again 4 wild-type and 4 Prrxl1−/− mice, was used for visceral noxious stimulation. Anaesthetized mice were placed on a cotton pad, the bladder was exposed through a low midline abdominal incision, and a 25-gauge needle was inserted in the bladder dome. Bladders were then distended with saline (0.9%) at room temperature during 15 min at a constant pressure of 50 cm of H20. Free outflow through the urethra allowed fluids to be expelled preventing the occurrence of bladder over distension. A third group (n = 4) was used for controlling c-fos induction, and consisted of 2 wild-type and 2 Prrxl1−/− animals that were maintained for 2 hr under sodium pentobarbital anaesthesia (50 mg/kg, i.p.) without any further manipulation. These mice were assumed to indicate the level of c-fos induction in non-stimulated animals. The last group consisted of 2 wild-type mice that were only subjected to bladder surgery. Two hours after the onset of manipulation, all animals were transcardially perfused as described above.
Cell Countings/Data Analysis
To quantify Prrxl1-, Lmx1b-, and Tlx3-stained neurons, 15–20 sections from lumbar cord of wild-type (n = 3) and Prrxl1−/− mice (n = 3) were used. Cell counting was performed with the ImageJ open source software (Rasband, W., NIH, Bethesda, MD).
Fos-immunoreactive cells present in 10 sections randomly taken from each animal were counted. In order to prevent double counting, sections were separated by at least 120 μm. All Fos-immunoreactive neurons were counted irrespective of the staining intensity, and plotted on drawings of the sections with the aid of a camera lucida. The total number of Fos-positive cells was depicted for each section and the mean number of neurons/section in the superficial dorsal horn (SDH, laminae I–III) and deep dorsal horn (DDH, laminae IV–VI and X) was determined. The distance between a line passing through the central canal and the ventral border of lamina III in wild-type animals was used to define the border between the DDH and the SDH in Prrxl1−/− mice (based on the data from Golgi impregnations, which indicated that the deep dorsal horn was not affected in Prrxl1 knockout mice). Results are presented as the mean ± SEM; differences were compared by one-way ANOVA and statistical significance taken at P < 0.5.
Numbers of Golgi impregnated neurons are presented as the total number of neurons observed in the superficial and deep dorsal horn of two wild-type and two knockout mice.
Lamination of the spinal cord dorsal horn was based in Nissl-stained sections, as described previously (Rebelo et al.,2007).
The authors thank Qiufu Ma for a critical reading of the manuscript and Thomas Müller and Carmen Birchmeier for gifts of antibodies.