Some lumbar sympathetic neurons develop a glutamatergic phenotype after peripheral axotomy with a note on VGLUT₂-positive perineuronal baskets

Abstract

Glutamate is the main excitatory neurotransmitter in the nervous system, including in primary afferent neurons. However, to date a glutamatergic phenotype of autonomic neurons has not been described. Therefore, we explored the expression of vesicular glutamate transporters (VGLUTs) type 1, 2 and 3 in lumbar sympathetic chain (LSC) and major pelvic ganglion (MPG) of naïve BALB/C mice, as well as after pelvic nerve axotomy (PNA), using immunohistochemistry and in situ hybridization. Colocalization with activating transcription factor-3 (ATF-3), tyrosine hydroxylase (TH), vesicular acetylcholine transporter (VAChT) and calcitonin generelated peptide was also examined. Sham-PNA, sciatic nerve axotomy (SNA) or naïve mice were included. In naïve mice, VGLUT₂-like immunoreactivity (LI) was only detected in fibers and varicosities in LSC and MPG; no ATF-3-immunoreactive (IR) neurons were visible. In contrast, PNA induced upregulation of VGLUT₂ protein and transcript, as well as of ATF-3-LI in subpopulations of LSC neurons. Interestingly, VGLUT₂-IR LSC neurons coexpressed ATF-3, and often lacked the noradrenergic marker TH. SNA only increased VGLUT₂ protein and transcript in scattered LSC neurons. Neither PNA nor SNA upregulated VGLUT₂ in MPG neurons. We also found perineuronal baskets immunoreactive either for VGLUT₂ or the acetylcholinergic marker VAChT in non-PNA MPGs, usually around TH-IR neurons. VGLUT₁-LI was restricted to some varicosities in MPGs, was absent in LSCs, and remained largely unaffected by PNA or SNA. This was confirmed by the lack of expression of VGLUT₁ or VGLUT₃ mRNAs in LSCs, even after PNA or SNA. Taken together, axotomy of visceral and non-visceral nerves results in a glutamatergic phenotype of some LSC neurons. In addition, we show previously non-described MPG perineuronal glutamatergic baskets.

Introduction

Vesicular glutamate transporters (VGLUTs) incorporate the excitatory neurotransmitter glutamate into synaptic vesicles, and three subtypes have been identified to date, VGLUT₁, VGLUT₂ and VGLUT₃ (Kaneko and Fujiyama 2002; Fremeau, Jr. et al., 2004; Liguz-Lecznar and Skangiel-Kramska 2007). VGLUTs are tightly related to the glutamatergic neuronal phenotype and have become standard markers for the study of neurons using glutamate as neurotransmitter (see Brumovsky et al., 2007). Their expression has been analyzed in detail in rodent dorsal root ganglion (DRG) sensory neurons innervating non-visceral structures like skin (Oliveira et al., 2003; Hwang et al., 2004; Landry et al., 2004; Morris et al., 2005; Brumovsky et al., 2007; Seal et al., 2009), where VGLUT₂ appears to be the most abundantly expressed transporter (Morris et al., 2005; Brumovsky et al., 2007). We also have detected expression of VGLUT₂, and to a lesser extent VGLUT₁, in DRG neurons retrogradely traced from the mouse colorectum (unpublished data), in agreement with the presence of VGLUT₁ and VGLUT₂ protein in nerve fibers innervating guinea pig (Olsson et al., 2004) and mouse (Spencer et al., 2008) rectum.

Visceral organs like the colorectum are innervated in rodents by: 1) peripheral projections of thoracolumbar and lumbosacral DRG neurons (see Robinson and Gebhart 2008); 2) postganglionic projections of sympathetic neurons contained in the lumbar sympathetic chain (LSC); 3) sympathetic and parasympathetic neurons present in the ‘mixed’ major pelvic ganglion (MPG) (Furness 2006; Keast 2006), and 4) enteric neurons, residing within the gut walls and creating an ‘intrinsic’ sensory and motor neuronal network (Furness et al., 2004). Peripheral afferent sensory, postganglionic sympathetic and preganglionic parasympathetic efferent axons innervating these organs travel together in the pelvic (PN) and lumbar splanchnic (LSN) nerves. The effects of injuries to these nerves have not been thoroughly explored.

The pathophysiology of non-visceral pain induced by peripheral nerve lesion has been intensely studied over the last few decades. Several models are utilized in rodents and include complete axotomy (Wall et al., 1979), different types of compression of the sciatic nerve (Bennett and Xie 1988; Brumovsky et al., 2004) as well as section of nerve branches (see Jaggi et al., 2011). These procedures are well known to induce a complex array of up- and downregulation of a variety of molecules in their respective cell somata in DRG (e.g. sodium and potassium channels, neuropeptides, neurotrophic factors and their receptors, and inflammation-associated receptors (Xiao et al., 2002; Costigan et al., 2002; Navarro et al., 2007)). Recently, changes in the expression of VGLUTs in DRG neurons of rat and mouse have also been described (Hughes et al., 2004; Brumovsky et al., 2007).

Sympathetic neurons are classically identified by their expression of the noradrenergic marker tyrosine hydroxylase (TH) (Apostolova and Dechant 2009), but are also capable of synthesizing a diversity of neurotransmitters, including acetylcholine and neuropeptides (Elfvin et al., 1993; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007). Interestingly, the possibility of a glutamatergic phenotype for autonomic neurons has not previously been reported. Sympathetic neurons are also affected by axotomy of their postganglionic projections (see (Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007)). Thus, downregulation of the neuropeptide tyrosine (NPY) or TH (Cheah and Geffen 1973; Sun and Zigmond 1996; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007), and upregulation of galanin (Lindh et al., 1993; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007), vasoactive intestinal peptide (VIP) and substance P (SP) (Rao et al., 1993; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007) have been shown in superior cervical ganglion neurons after axotomy of their projections in rodents. Similar effects on some of the neurotransmitters mentioned above have also been shown in cat LSC (Lindh et al., 1993). Many of these neurochemical changes could relate to neuropathic pain mechanisms after nerve injury by sympathetic-sensory coupling (Shinder et al., 1999; Habler et al., 2000; Jänig 2003; Gibbs et al., 2008). Thus, identification of key sensory and autonomic molecules participating in the generation, maintenance or attenuation of pain remains essential for the development of novel analgesic/regenerative drugs.

To date a glutamatergic phenotype of autonomic neurons has not been studied. Here we use VGLUT1-3 as selective markers of a glutamatergic phenotype for analysis of mouse LSC and MPG, before and after pelvic nerve axotomy (PNA). The widely utilized sciatic nerve axotomy (SNA) model was also included. Colocalization analysis using antibodies raised against TH, the vesicular acetylcholine transporter (VAChT) and calcitonin gene-related peptide (CGRP) were also performed. Finally, we evaluated the expression of transcripts for VGLUT₁, VGLUT₂ and VGLUT₃ in the LSC and MPG, to compare with the immunohistochemical data, and to compensate for the lack of a reliable VGLUT₃ antibody.

Experimental Procedures

Animals

Male BALB/c mice (Taconic, Germantown, NJ or Jackson Laboratories, Bar Harbor, Maine, USA; 7–8 weeks old) were used. Experiments adhered to the United States Public Health Service policies regarding the care and use of animals in research, followed the Uniform Requirements for manuscripts submitted to Biomedical journals and were approved by the Institutional Animal Care Use Committee (University of Pittsburgh).

Pelvic or sciatic nerve axotomy

(Fig. 1). Axotomy of the pelvic nerve (PN) in mouse is complicated by the proximity of the bifurcation of the aorta into the iliac arteries and thus difficult to execute without damaging other tissues. Thus, an alternative approach was taken to transect major contributing nerves, which we refer to for convenience throughout this report as PNA (PN axotomy). In twelve isoflurane-anesthetized mice (Hospira Inc, Lake Forest, IL USA) injected with buprenorphine (0.1 mg/kg; Bedford Labs, Bedford, OH, USA) for postoperative analgesia, the ventral branches of the left 6^th lumbar and 1^st sacral spinal nerves, both major contributors to the PN, were exposed by aseptic laparotomy and dissection (S2 also contributes some fibers, but their percentage is very small and could be considered less relevant than L6 or S1). In eight mice, we tightly ligated and distally axotomized these two spinal nerves contributing to the PN (i.e., PNA); in the remaining four mice the same spinal nerves were exposed as described, but left intact (sham group). After suturing of the abdominal wall and topical application of 1% dibucaine ointment (1%; Perrigo, Allegan, MI, USA), mice recovered in a warm environment.

Figure 1.

Schematic diagram showing the general anatomical distribution of the lumbar sympathetic chain (LSC), the ventral branches of the spinal nerves L4-L6 and S1, the major pelvic ganglion (MPG), and the pelvic (PN), lumbar splanchnic (LSN) and sciatic (SN) nerves in the male mouse. Their relation with the aorta and stemming branches (celiac artery (CA), superior mesenteric artery (SMA), renal artery (RA), inferior mesenteric artery (IMA), iliac artery (IA)) is emphasized. Not illustrated for clarity is the spine, spinal cord or dorsal root ganglia, from which the axons of the spinal nerves arise. The nerves were exposed by blunt dissection of the overlying lumbar muscles (not shown); the location where transection of the pelvic nerve tributaries was executed is shown with an asterisk. The MPG is located between the colorectum and the urinary bladder, with branches from the pelvic nerve traveling through it (for clarity, other branches entering or exiting the MPG are not illustrated).

Additional mice were subjected to tight ligation and axotomy of the sciatic nerve (SNA; n=8) or sham surgery (n=4) as previously described (see Brumovsky et al., 2007) for comparison with the effects of PNA on LSC and MPG neurons.

Immunohistochemistry

Seven (PNA) or ten (SNA) days after lesion, mice were deeply anaesthetized using sodium pentobarbital (60 mg/kg, i.p.; Ovation Pharmaceuticals, Deerfield, IL) and perfused via the ascending aorta with 20 ml of Tyrode’s buffer (37°C), followed by 20 ml of a mixture of 4% paraformaldehyde and 0.2% picric acid dissolved in 0.16 M phosphate buffer (pH 6.9) (37°C) and 50 ml of the same mixture at 4°C, the latter for approximately 5–7 min (Zamboni and De Martino 1967). Four naïve mice were also included. The ipsilateral and contralateral LSCs and MPGs were dissected out and post fixed for 90 min at 4°C in the same fixative and finally immersed in 10–20% sucrose in phosphate-buffered saline (PBS) (pH 7.4) containing 0.01% sodium azide and 0.02% bacitracin (both from Sigma, St. Louis, MO) (4°C) for 48 h. After embedding in Tissue-Tek O.C.T. compound (Sakura, Torrence, CA) and deep freezing, all tissues were sectioned in a cryostat (Leica, Heidelberg, Germany) at 12 µm (LSC) or 20 µm thicknesses (MPG).

Tissue sections were incubated in a series of primary and secondary antibodies (Table 1), according to the tyramide signal amplification technique (Adams 1992) for single-staining immunohistochemistry, and following a previously described protocol (see (Brumovsky et al., 2007)). The VGLUT₁ and VGLUT₂ antibodies were raised against the C-terminal sequence of mouse VGLUT₁ [531–560 amino acid residues, GeneBank accession number U07609 (Kawamura et al., 2006; Miura et al., 2006)] or mouse VGLUT₂ [549–582 amino acid residues, GeneBank accession number BC038375 (Miyazaki et al., 2003)] and were expressed as glutathione-S transferase fusion proteins for immunization. Antibody specificity was tested by immunoblot of mouse brain extracts, with detection of a single band at 60kDa for both VGLUT₁ and VGLUT₂ (Miyazaki et al., 2003; Kawamura et al., 2006; Miura et al., 2006), and also by immunostaining patterns in the adult mouse brain, which were identical to previous reports by immunohistochemistry and in situ hybridization.

Table 1.

Primary and secondary antibodies/fluorophores used for the identification of VGLUT₁, VGLUT₂, ATF-3, TH, VAChT and CGRP. Single staining was achieved using the TSA Plus technique (T). In some cases, more than one primary antibody was used for colocalization experiments (C) and thus processed with different combinations of TSA Plus and indirect immunofluorescence (I) techniques.

PRIMARY ANTIBODIES
Target antigen and host species	Dilution	Reference or source
Rabbit anti-VGLUT1	1:4,000 (T)	Kawamura et al., 2006
Guinea Pig anti-VGLUT2	1:8,000 (T)	Miyazaki et al., 2003; Brumovsky et al., 2007
Rabbit anti-ATF-3	1:8,000 (T)/ 1:400(C; I)	sc-188; Santa Cruz
Rabbit anti-TH	1:4,000 (T) / 1:400 (C; I)	AB152; Millipore
Goat anti-VAChT	1:4,000(C; T)	AB1578; Millipore
Rabbit anti-CGRP	1:60,000 (C; T)	C8198; Sigma
SECONDARY ANTIBODIES / FLUOROPHORES
HRP-donkey anti-rabbit	1:200	711-035-152; Jackson Immunoresearch
HRP-donkey anti-guinea pig	1:200	706-035-148; Jackson Immunoresearch
HRP-donkey anti-goat	1:200	805-035-180; Jackson Immunoresearch
TRITC-donkey anti-rabbit	1:400	711-025-152; Jackson Immunoresearch
Cyanine 5-donkey anti-rabbit	1:600	711-175-152; Jackson Immunoresearch
Fluorescein tyramide	1:200	NEL741; Perkin Elmer
Tetramethylrhodamine tyramide	1:600	NEL742; Perkin Elmer

Some sections underwent additional primary and secondary antibody incubation steps (Table 1) for double- or triple-staining, following the protocol of indirect immunofluorescence by Coons and collaborators (see Coons 1958) or additional TSA protocols (see Brumovsky et al., 2007). LSC sections were also incubated with DAPI, for the identification of the nuclear profiles of all cells present per section (Kubista et al., 1987).

Specificity of the VGLUT₂ antibody has been previously tested (Kawamura et al., 2006; Brumovsky et al., 2007). The rabbit VGLUT₁ antiserum was studied after preadsorption with the C-terminal 531–560 amino acid peptide (10⁻⁵ and 10⁻⁶M), this combination giving no visible fluorescent signal (data not shown). ATF-3, TH, VAChT and CGRP antibodies are commercially available and repeatedly validated. Additional control experiments included the comparison of single- and double-stained sections; and the omission of the primary antibody in a few sections to test for non-specific staining by the secondary antibodies.

Riboprobe in situ hybridization

Male, 7 week old, BALB/c mice (PNA=4; sham-PNA=2; SNA=4; sham-SNA=2; naïve=2) were used for in situ hybridization analysis. Mice were sedated using CO₂ before rapid decapitation. The LSC and the MPG were quickly removed, embedded in O.C.T. (Tissue-Tek, Sakura, Torrance, CA) and frozen over dry ice. Sections of LSCs and MPGs (12 µm and 20 µm, respectively) were cut in a cryostat, thaw-mounted onto Superfrost Plus (Fisher Scientific, Waltham, MA) glass slides, and stored at −20°C until hybridization. Adjacent sections through the different tissues were processed for in situ hybridization localization of VGLUT₁, VGLUT₂ and VGLUT₃ mRNAs, using ³⁵S-labeled cRNA probes as described previously (Seroogy and Herman 1997; Numan et al., 2005; Dickerson et al., 2009).

Sense and antisense probes complementary to the coding region of mouse VGLUTs were generously provided by Drs. Akiya Watakabe, National Institute for Basic Biology, Okazaki, Japan [VGLUT₁, and VGLUT₂ (Nakamura et al., 2007)] and Jeffery Erickson, Louisiana State University, Baton Rouge, LA [VGLUT₃ (Schäfer et al., 2002)]. The sense and antisense cRNA probes were prepared by in vitro transcription using appropriate linearized DNA constructs in the presence of the proper RNA polymerase (T3, T7, or SP6) and ³⁵S-UTP (PerkinElmer, Chelton, CT).

Hybridized sections were dipped in NTB2 nuclear track emulsion (Kodak, City, State; 1:1 in dH₂O), air-dried, and exposed in sealed slide boxes at 4°C for 3 and 7 days for brain and brainstem or LSC and MPG sections, respectively. The emulsion was developed in D19 (Kodak) and fixed with Rapidfix (Kodak). The slides were counterstained with cresyl violet (Sigma) and coverslipped with DPX mounting solution (Fluka, Saint Louis, MO).

As controls for specificity, some sections were pretreated with ribonuclease A (0.05 mg/mL) for 30 min at 45°C before hybridization with the ³⁵S-labeled cRNA probes. Some sections were also hybridized with sense-strand ³⁵S-labeled riboprobes against each VGLUT. Finally, several tissue sections were incubated in hybridization cocktail that lacked the radioactive probe as a chemography control. No specific labeling was observed under any of these control conditions.

Microscopy and image processing

Sections processed for immunofluorescence were cover-slipped using 2.5% DABCO in glycerol (Sigma) and examined using a Nikon Eclipse E600 fluorescence microscope (Nikon, Tokyo, Japan) provided with appropriate objectives, filters and a Retiga 2000 R Fast CCD camera (Q-Imaging, Surrey, BC, Canada), using IPLab software (Scanalytics Inc., Vancouver, BC, Canada). For colocalization analysis, a Fluoview FV 1000 confocal laser scanning biological microscope equipped with 10× (0.45 N.A.), 20× (0.75 N.A.) and 60× oil (1.40 N.A.) objectives was used (Olympus, Tokyo, Japan). The FITC labelling was excited using a 547-514 nm argon multi-line laser. For the detection of TRITC and TMR, the 543 nm HeNe laser was used.

Sections processed for riboprobe in situ hybridization were examined using a Nikon Eclipse E800 fluorescence microscope provided with appropriate objectives and filters for bright- and dark-field visualization, and a DN100 Nikon Digital Net Camera.

Resolution, brightness and contrast of the images were optimized using the Adobe Photoshop CS3 software (Adobe Systems Inc., San Jose, CA).

Quantification and statistical analysis

Every fifth LSC section was used to quantify the number of VGLUT₁, VGLUT₂ or VGLUT₃ mRNA positive neuron profiles (NPs), as well as the total number of cresyl violet-stained NPs, giving a typical nuclear staining. Only nucleated NPs were counted. In total, 5 sections per LSC were used for quantification. An average of 752 NPs per mouse was counted in this study.

Data are presented as mean ± SEM and were statistically analyzed using the Student t test.

Results

Pelvic nerve axotomy induces upregulation of VGLUT₂ protein and mRNA, but not VGLUT₁ or VGLUT₃, in LSC NPs

In LSCs from naïve (Fig. 2A), as well as in contralateral sham-PNA, PNA or SNA mice (data not shown), VGLUT₁-like immunoreactivity (LI) was virtually absent, as it also was after PNA (Fig. 2B). Some VGLUT₂-immunoreactive (IR) fibers and varicosities, but no neuronal cell bodies, were detected in LSCs from naïve mice (Fig. 2C) and in contralateral LSCs from mice subjected to sham-PNA (Fig. 3A), PNA or SNA (data not shown). In contrast, PNA induced upregulation of VGLUT₂-LI in a considerable number of LSC neurons (Fig. 3C). This was accompanied by a dramatic upregulation of ATF-3-LI in LSC neurons (Fig. 3D), as compared to its absence in contralateral LSCs from sham-PNA (Fig. 3B), naïve or contralateral LSCs from PNA and SNA mice (data not shown). Sham-PNA (Fig. 3E–H) induced at most a moderate, variable upregulation of VGLUT₂ in LSC NPs-LI (Fig. 3E, G) and VGLUT₂ transcript (Fig. 4C, D), with parallel increases in the number of ATF-3-IR NPs (Fig. 3F, H).

Figure 2.

Confocal immunofluorescence photomicrographs of sections of the LSC from naïve mice (A, C) or PNA mice (B), incubated with VGLUT₁ (A, B) or VGLUT₂ (C) antisera. Arrowheads and arrow in (C) show VGLUT₂-IR varicosities and nerve fibers, respectively in naïve LSCs. Scale bar: 50 µm.

Figure 3.

Confocal immunofluorescence photomicrographs of sections of the contralateral (A, B) or ipsilateral (C–H) LSC of mice 7 days after sham-PNA (A,B; E–H) or PNA (C, D), incubated with VGLUT₂ (A, C, E, G) or ATF-3 (B, D, F, H) antisera (each corresponding pair of VGLUT₂ or ATF-3 micrographs shows sections separated by 36 µm). Arrows and arrowheads in (A, B) show VGLUT₂-IR fibers and varicosities, respectively. Black double arrowheads and double arrowheads in (C, D, E–H) show VGLUT₂- and ATF-3-IR LSC NPs. Scale bar: 50 µm (H=A–G).

Figure 4.

Dark-(A, C, E, G, I, J, K) and bright-field (B, D, F, H) photomicrographs of LSC (A–H; I) or the PAG (J, K), in naïve (A, B, J, K), ipsilateral sham-PNA (C, D), ipsilateral PNA (E, F, I) or ipsilateral SNA (G, H), after hybridization with antisense (A–H; J) and sense (I, K) riboprobes for VGLUT₂ mRNA (B, D, F and H show high power magnification images of boxes in A, C, E and G, respectively). (Arrowheads in B show NPs lacking VGLUT₂ mRNA. Double arrowheads in D, F, H show VGLUT₂ mRNA-expressing NPs. Asterisk shows location of the aqueduct of Sylvius. Scale bars: 50 µm (H=B, D, F); 100 µm (I=A, C, E, G); 500 µm (J, K).

Quantification of the number of NPs upregulating VGLUTs in the LSC was done on riboprobe in situ hybridization-processed slides (Table 2). We found that VGLUT₂ mRNA was also absent in naïve LSC (Table 2; Fig. 4A, B), and only detected in occasional NPs in the ipsilateral side of sham-PNA mice (Table 2; Fig. 4C, D). In contrast, PNA induced a marked upregulation of VGLUT₂-mRNA-positive LSC NPs (Table 2; Fig. 4E, F), whereas SNA resulted in a more modest upregulation than after PNA (Table 2; Fig. 4G, H). Overall, the grain density of upregulated VGLUT₂ riboprobe labeling after PNA (Fig. 4E, F) or SNA (Fig. 4G, H) appeared to vary from medium to strong, and clearly differentiated from the surrounding background grain density. However, a few NPs showing a lesser grain density related to VGLUT₂ mRNA were also detected in the ipsilateral LSC of sham-PNA mice (Fig. 4C, D), as well as in some NPs after PNA (Fig. 4E, F) or SNA (Fig. 4G, H). Certifying the sensitivity and specificity of the VGLUT₂ riboprobe, its application to brainstem sections showed established and abundant labeling in regions such as the periaqueductal gray (Fig. 4J) (Lein et al., 2007; Ng et al., 2009).

Table 2.

Percentage of LSC NPs expressing VGLUT₁, VGLUT₂ or VGLUT₃ mRNAs in naïve, sham-PNA, contralateral PNA (Cont-PNA), ipsilateral-PNA (Ipsi-PNA); sham SNA, Cont-SNA or Ipsi-SNA.

	Naїve (n=4)	Sham-PNA (n=2)	Cont-PNA (n=3)	Ipsi-PNA (n=4)	Sham-SNA (n=2)	Cont-SNA (n=1)	Ipsi-SNA (n=4)
VGLUT1	0	0	0	0	0	0	0
VGLUT2	0	0.2 ± 0.1 %	0.4 ± 0.4 %	7 ± 1.6 %**;†	0	0	3 ± 1 %*
VGLUT3	0	0	0	0	0	0	0

^*†P<0.05;

^**P<0.01.

^*compares naïve vs. Ipsi-SNA;

^**compares naive vs. Ipsi-PNA;

^†compares Cont- vs. Ipsi PNA.

As observed with immunohistochemistry, VGLUT₁- (Table 2; Fig. 5A) and VGLUT₃-(Table 2; Fig. 5B) mRNA-positive NPs are absent in naïve LSC. Likewise, sham-PNA (Fig. 5C, D), PNA (Fig. 5E, F) or SNA (Fig. 5G, H) failed to induce any change in the expression of either transcript. The riboprobes for VGLUT₁ or VGLUT₃ mRNAs did, however, efficiently label neurons in brain regions such as the hippocampus (Fig. 5I) and the dorsal raphe nucleus (Fig. 4K), respectively (cf. Lein et al., 2007; Ng et al., 2009).

Figure 5.

Dark-field photomicrographs of LSC (A–H), the hippocampus (HC) (I, J) or the dorsal raphe nucleus (DR) (K, L), in naïve (A, B, I–L), ipsilateral sham-PNA (C, D), ipsilateral PNA (E, F) or ipsilateral SNA (G, H), after hybridization with antisense (A–H; I, K) and sense (J, L) riboprobes for VGLUT₁ (A, C, E, G, I, J) or VGLUT₃ (B, D, F, H, K, L) mRNAs. Double arrow in (I) and double arrows in (K) show mRNA labeling in the CA3 region of the HC or the DR, respectively. Scale bars: 50 µm (H=A–G); 200 µm (K, L); 500 µm (I, J).

Hybridization of brain sections with appropriate control sense riboprobes resulted in absence of the normal labeling patterns previously published and shown in the present study (compare Fig. 5I, J for VGLUT₁; Fig. 5K, L for VGLUT₃; and Fig. 4E, I and J, K for VGLUT₂).

VGLUT₂ protein is selectively upregulated in axotomized LSC neurons

Colocalization analysis showed that virtually all VGLUT₂-IR neurons in LSCs ipsilateral to PNA (Fig. 6A–C), SNA (Fig. 6D–I) or sham-PNA (Fig. 6J–L), as well as rare NPs found contralateral to a sham-PNA (Fig. 6M–O), also upregulated ATF-3, a marker of mechanically injured neurons (Tsujino et al., 2000).

Figure 6.

Confocal immunofluorescence photomicrographs of sections of the ipsilateral LSC of mice 7 days after PNA (A–C), 10 days after SNA (D–I), ipsilateral (J–L) or contralateral to a sham-PNA (M–O), incubated with VGLUT₂ (A, D, G, J, M) and ATF-3 (B, E, H, K, N) antisera (C, F, I, L, O show merged images). (D–I; G–H show a magnified view of the box in F). Double arrowheads in (A, G, J, M, B, H, K, N, C, I, L, O) and arrows in (B, H, K) show VGLUT2/ATF-3 or ATF-3-only neurons, respectively. Scale bars: 20 µm (C=A, B; F=D, E; I=G, H; L=J, K; O=M, N).

Colocalization analysis with TH revealed that the great majority of the VGLUT₂-IR neurons in LSCs from PNA mice (Fig. 7A–I) appeared to lack the noradrenergic marker. However, a few neurons clearly coexpressing both VGLUT₂ and TH could be detected (Fig. 7D–I). TH-positive neurons were either strongly or weakly immunoreactive (Fig. 7B, E, H), but the few somata co-expressing VGLUT₂-IR most often belonged to the second group (Fig. 7G–I). In LSCs ipsilateral to a sham-PNA (Fig. 7J–L) or SNA (data not shown), the few neurons upregulating VGLUT₂ also lacked TH-LI, with the exception of occasional neurons positive for both markers (Fig. 7J–L). In LSCs contralateral to sham-PNA (Fig. 7M–R), PNA or SNA mice (data not shown), thick VGLUT₂-IR, TH-negative nerve bundles could be seen in close association with the LSC, apparently projecting smaller bundles towards the ganglion (Fig. 7M–O). Occasionally, clusters of perineuronal baskets formed by VGLUT₂-IR varicosities surrounded TH-IR LSC neurons in the contralateral LSC of a sham-PNA mouse (Fig. 7P–R).

Figure 7.

Confocal immunofluorescence photomicrographs of sections of the ipsilateral (A–L) or contralateral (M–R) LSCs of mice 7 days after PNA (A–I) or sham-PNA (J–R), incubated with VGLUT₂ (A, D, G, J, M, P) and TH (B, E, H, K, N, Q) antisera (C, F, I, L, O, R show merged images) (asterisks in E, H, K, N, Q indicate TH-negative LSC NPs). (A–I; boxes 1 and 2 in C are shown at higher magnification in D–F and G–I, respectively). Arrowheads in (D, G, J) and double arrowheads in (D, G, J, B, H, K, F, I, L) show TH-only or VGLUT₂/TH neurons, respectively. White arrows and black arrows show light or strong TH-IR-only NPs, respectively (M–R). Black asterisk in M shows thick, VGLUT₂-IR fiber bundles, from which smaller bundles appear to detach (double arrowheads in M). Black arrowheads show TH-IR fibers. Perineuronal VGLUT₂-IR baskets (arrowheads in P) are occasionally seen around TH-IR NPs (arrows in Q). Scale bars: 50 µm (C=A, B); 20 µm (I= D–H; L=J, K; O=M, N; R=P, Q).

VGLUT₂-IR perineuronal baskets, present in the mouse MPG, lack CGRP, VAChT and TH

In MPGs from contralateral (Figs. 8AF) or ipsilateral (Figs. 9E–L) sham-PNA, or contralateral PNA (Figs. 9A–D, M–P), VGLUT₂-LI was often detected in nerve fibers (Figs. 8A–C; 9A–D), as was as in perineuronal baskets built up of varicose nerve endings (Figs. 8A–F; 9E–P). VGLUT₁-LI was confined to a few varicosities sparsely distributed throughout the ganglion, as observed in contralateral PNA (Fig. 10A) or ipsilateral sham-PNA (Fig. 10C).

Figure 8.

Confocal immunofluorescence photomicrographs of sections of the contralateral MPGs of sham-PNA mice, co-incubated with VGLUT₂ (A, D) and CGRP (B, E) antisera (C, F, show merged images). Black double arrowheads in (A) and arrowheads in (A, D) show VGLUT₂-IR nerves or perineuronal baskets, respectively. Arrows in (B, E) and double arrowhead in (D–F) show CGRP fibers or VGLUT₂/CGRP-double stained nerve bundles. Scale bars: 20 µm (C=A, B; F=D, E).

Figure 9.

Confocal immunofluorescence photomicrographs of sections from contralateral (A–D; M–P) or ipsilateral (E–L) MPGs after PNA (A–D; M–P) or sham-PNA (E–L), incubated with VGLUT₂ (A, E, I, M), VAChT (B, F, J, N) and TH (C, G, K, O) antisera (D, H, L, P, show merged images). Arrowheads in (A, E, I) and white arrows in (B, F, J) show two separate populations of TH-IR neurons showing VGLUT₂- or VAChT-IR perineuronal baskets, respectively. Double arrowhead in (A) and black double arrowhead in (C) show thick VGLUT₂- or TH-IR nerve bundles, respectively. Black arrows in (M–P) show some TH-IR NPs innervated by both VGLUT₂- and VAChT-IR baskets. Scale bars: 50 µm (D=A–C); 10 µm (H=E–G; L=I–K; P=M–O).

Figure 10.

Confocal immunofluorescence photomicrographs of sections from contralateral (A, B) or ipsilateral (C–F) MPGs after PNA (A, B, E, F) or sham-PNA (c, D), incubated with VGLUT₁ (A, C, E) or VGLUT₂ (B, D, F) antisera. White arrowheads in (A, C) show VGLUT₁-IR varicosities (shown at higher magnification in inset (black arrowhead)). White or black double arrowheads show nerve fibers (B, D, F) or perineuronal baskets (B, D), respectively. Scale bars: 100 µm (A–F); 10 µm (inset in A).

Colocalization analysis showed that the VGLUT₂-IR baskets in the MPG virtually never colocalize with CGRP (Fig. 8A–F), although VGLUT₂- and CGRP-IRs intermingled in nerve bundles in this ganglion, possibly partly coexisting (Fig. 8D–F). VGLUT₂-IR perineuronal baskets were often detected around TH-IR MPG neurons (Fig. 9A–P), although some were occasionally found around neurons negative for the noradrenergic marker (data not shown). Colocalization analysis with VAChT (Fig. 9B, F, J) and TH (Figs. 9C, G, K) showed that the majority of the VGLUT₂-IR baskets did not coexpress either the cholinergic or noradrenergic marker, respectively. In fact, VGLUT₂- (Figs. 9A, E, I) and VAChT-LIs (Figs. 9B, F, J) were detected in different perineuronal baskets. However, in a few cases, VGLUT₂- (Figs. 9M) or VAChT-IR (Figs. 9N) varicosities innervating the same TH-IR MPG NP showed overlap (Figs. 9P).

Analysis of MPGs after its deafferentation showed that the occasional VGLUT₁-IR varicosities detected in contralateral PNA (Fig. 10A) or ipsilateral sham-PNA (Fig. 10C) disappear after PNA in the affected MPG (Fig. 10E). Accordingly, while abundant VGLUT₂-IR nerve fibers and several perineuronal baskets were observed in the contralateral PNA (Fig. 10B) or ipsilateral sham-PNA (Fig. 10D), the expression of VGLUT₂ was reduced to a few nerve fibers in the ipsilateral PNA (Fig. 10F). Neither VGLUT₁- or VGLUT₂-IR NPs were observed in MPGs after PNA (Fig. 10E, F) or SNA (data not shown). Riboprobe in situ hybridization of MPG sections in naïve mice failed to reveal signs of cellular expression of any of the studied VGLUT mRNAs (data not shown).

Discussion

In the present work, we show that axotomy of both visceral and non-visceral nerves induce the upregulation of VGLUT₂ protein and transcript in a subpopulation of LSC neurons. These observations suggest that some sympathetic neurons, upon direct injury of their postganglionic projections, can develop a glutamatergic phenotype. The presence of such a subpopulation of noradrenergic neurons in the LSC may seem surprising. However, coexpression of VGLUT₂ (Sulzer et al., 1998; Kawano et al., 2006; Yamaguchi et al., 2007) or glutamate (Liu et al., 1995; Sulzer et al., 1998) with TH, the latter a marker for dopaminergic neurons in this case, has been shown in rat (Sulzer et al., 1998; Stornetta et al., 2002a; Stornetta et al., 2002b; Kawano et al., 2006; Yamaguchi et al., 2007) and cat (Liu et al., 1995) brainstem, as well as in rat hypothalamus (Kawano et al., 2006). Interestingly, expression and upregulation of VGLUT₂ in supraoptic and paraventricular neurons synthesizing vasopressin was also described in rats subjected to water deprivation (Kawasaki et al., 2005).

Postganglionic axotomy of sympathetic nerves in cat LSC (Lindh et al., 1993) or rodent SCG (Zhang et al., 1994; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007) induces downregulation of neuropeptides such as CGRP (Lindh et al., 1993) and NPY, as well as the noradrenergic marker TH (Cheah and Geffen 1973; Sun and Zigmond 1996; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007), and upregulation of galanin (Lindh et al., 1993; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007), VIP, SP (Rao et al., 1993; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007) and the NPY Y₂-receptor (Lindh et al., 1993; Zhang et al., 1994; Zigmond and Sun 1997; Landry et al., 2000; Navarro et al., 2007). More recently, in a study in pigs, it was shown that axotomy of the nerves innervating the colon and containing sympathetic fibers projecting from the LSC, results in the upregulation of galanin and to some extent somatostatin, paralleled by the downregulation of TH (Skobowiat et al., 2011). In the present study in mouse, we observed that PNA, a lesion that targets both sensory afferent as well as autonomic efferent nerves, induces upregulation of VGLUT₂ mRNA in about 7% of all LSC neurons, an event paralleled by the upregulation of VGLUT₂ protein. This lesion-induced plasticity was selective in that expression of VGLUT₁ and VGLUT₃ mRNAs was not induced.

Whether the upregulation of VGLUT₂ reflects induction (i.e., de novo synthesis of transporter protein and transcript) or an increase from low levels not detected with our methods is not known. The physiological consequences of such an event remain unknown, but VGLUT₂ upregulation in sympathetic neurons could result in increased synthesis, vesicular uptake and synaptic release of glutamate, from cell bodies or axonal projections, potentially contributing to increased fast synaptic transmission and nociceptive mechanisms. In fact, glutamate could be released from the soma of LSC neurons. There is abundant evidence supporting the concept of somatic/dendritic release for neuropeptides, monoamines and glutamate (Duguid et al., 2007; Colgan et al., 2009; Xia et al., 2009). Moreover, VGLUT₂ immunoreactivity can be observed close to the plasma membrane in non-visceral DRG cell bodies, suggesting somatic glutamate release (Brumovsky et al., 2007).

Because sensory neurons in rodents express VGLUTs/glutamate (Oliveira et al., 2003; Hwang et al., 2004; Landry et al., 2004; Morris et al., 2005; Brumovsky et al., 2007; Seal et al., 2009), glutamate acting on peripherally expressed glutamatergic receptors (see below) most likely derives from afferent fibers. However, and as suggested by our present findings, the axonal projections of LSC neurons could be an additional source of glutamate, participating in a process of sympathetic-sensory neuron coupling (Shinder et al., 1999; Habler et al., 2000; Janig 2003; Gibbs et al., 2008). It has been demonstrated that activation of peripheral glutamate receptors by exogenous glutamate (Carlton et al., 1995; Jackson et al., 1995; Du et al., 2001), NMDA (Du et al., 2003) or kainate (Du et al., 2006) receptor agonists in primary afferents present in the glabrous skin of the rat hindpaw induces their depolarization and promotes pain-related behavior; a similar effect was also observed in hairy skin (see (Tian et al., 2005)). Nevertheless, Moalem and colleagues (Moalem et al., 2005) reported that a number of chemical mediators including glutamate, did not affect the excitability of unmyelinated sensory axons in normal or injured peripheral nerves of the rat. Unfortunately, the potential role of glutamate and other chemical mediators on the excitability of thick and thinly myelinated fibers was not assessed in that study.

Evidence for glutamatergic signaling in autonomic neurons has previously been presented. In fact, application of kainate to peripheral noradrenergic ganglia innervating the rat vas deferens was earlier shown to induce a time-dependent decrease of noradrenaline in this organ (Lara and Bastos-Ramos 1988). Moreover, NMDA receptor activation in isolated segments of the guinea pig colon potentiates the electrically stimulated noradrenaline overflow, a presynaptic effect probably exerted at sympathetic nerve terminals (Cosentino et al., 1995). In these studies, it was assumed that glutamate originated from sensory nerve terminals. However, our results on VGLUT₂ suggest that sympathetic neurons, at least after injury, may also contribute to peripherally released glutamate. Finally, supporting a role of glutamate in pain modulation, recent studies show that partial or total deletion of VGLUT₂ protein in primary afferent neurons markedly reduces sensory hypersensitivity induced by peripheral nerve lesions (Moechars et al., 2006; Leo et al., 2009; Liu et al., 2010; Lagerström et al., 2010). However, the expression of VGLUT₂ in the LSC of these mice, before and after injury, or the role of impairment of VGLUT₂ in the peripheral projections of DRG neurons were not assessed.

Expression of group I metabotropic (see (Shigemoto et al., 1992; Kammermeier and Ikeda 2002; Higashida et al., 2003)), as well as AMPA (Kiyama et al., 1993; Carlton et al., 1998; Coggeshall and Carlton 1999), NMDA (Carlton et al., 1998) (Coggeshall and Carlton 1999) and kainate (Coggeshall and Carlton 1999) ionotropic glutamate receptors in rodent superior cervical ganglion (SCG) (Shigemoto et al., 1992; Kiyama et al., 1993; Kammermeier and Ikeda 2002; Higashida et al., 2003) and LSC (Carlton et al., 1998; Coggeshall and Carlton 1999) neurons and their projections further supports a role of glutamate in the autonomic nervous system. Interestingly, Coggeshall and Carlton (Coggeshall and Carlton 1999) showed that the proportion of NMDA, AMPA and kainate receptors in the gray rami associated with the LSC in rat is considerably increased 48 hr after hindpaw inflammation. These glutamatergic receptors, under ‘normal’ circumstances presumably targeted by glutamate released from primary afferent nerve terminals could promote an augmented release of noradrenaline and other transmitters, potentially exacerbating the activation of sensory neurons in inflammatory states (Carlton et al., 1998; Coggeshall and Carlton 1999). Our results also suggest that autonomic neurons themselves could release glutamate, acting on primary afferent and/or autonomic nerve terminals, thus acting on hetero- or autoreceptors, or both. Also, enteric neurons in the gut or non-neuronal visceral tissues are putative targets of glutamate released by LSC nerve endings. It remains to be established whether inflammation of visceral organs or the hindpaw also induces VGLUT₂ upregulation in the LSC.

We also found that all NPs upregulating VGLUT₂-LI in the LSC coexpressed ATF-3, a transcription factor dramatically upregulated in lumbar DRG neurons in rodents after peripheral nerve injury (Tsujino et al., 2000). In addition, postganglionic axotomy of the superior cervical ganglion also upregulates ATF-3 in rat and mouse (Hyatt et al., 2007). Autonomic neurons, similar to sensory neurons, are characterized by their considerable capacity to survive and regenerate after peripheral nerve injury. Several molecules, including ATF-3 (see (Seijffers et al., 2007)), as well as classical ones like the nerve growth factor, have been proposed to be central to mechanisms of nerve regeneration (see (Keast 2006; Navarro et al., 2007)). Could the upregulation of VGLUT₂ and possibly also glutamate reflect a role in survival and regeneration of LSC neurons? (Balazs 2006). Interestingly, VGLUT₂- LI is present in neurons migrating from the olfactory placode toward the forebrain in the developing rat brain, gradually decreasing toward adulthood (Honma et al., 2004). Furthermore, an association between the expression of VGLUT₂ protein in mesencephalic dopaminergic axonal terminals and the formation of synaptic junctions by these terminals was recently reported, suggesting a role in axonal growth (Bérubé-Carriere et al., 2009). Classically, excessive activation of glutamatergic receptors has been linked to neurotoxicity (Mattson 2003). However, activation of different metabotropic glutamate receptors in the central nervous system has also been shown to be neuroprotective (Byrnes et al., 2009).

Unexpectedly, we noted in sham mice, or the contralateral side of PNA mice, the existence of perineuronal baskets in MPGs expressing VGLUT₂ and lacking both TH and VAChT. These baskets appeared sparsely distributed throughout the ganglion, often but not exclusively around TH-IR MPG neurons. VAChT-IR baskets were also present in the ganglion, but were virtually always different than those expressing VGLUT₂, suggesting that these baskets may target different subpopulations of MPG neurons. Interestingly, deafferentation of the MPG by PNA induced a noticeable reduction in the number of VGLUT₂-IR nerve fibers, and abolished most of the VGLUT₂-IR baskets ipsilateral to the lesion, suggesting that these two structures are contributed by the pelvic nerve. Perineuronal baskets in the MPG have been described in rat (Keast 1995) and mouse (Wanigasekara et al., 2003). These are typically cholinergic, and often coexpress other transmitters, including neuropeptides (de Groat 1987; Wanigasekara et al., 2003; Keast 2006). Evidence of a glutamatergic trait in MPG perineuronal baskets has not been reported previously, with the exception of an electron microscopy study in rat showing glutamate in axons and terminals in the MPG, also containing large dense core vesicles (Aïoun and Rampin 2006). However, and possibly because of the nature of the experimental approach (electron microscopy), the authors did not assess the presence of perineuronal baskets containing glutamate in the MPG. The present results support the notion of expression of glutamate in nerve fibers innervating the MPG and also forming perineuronal baskets. Finally, in their study, Aïoun and Rampin also reported that the distribution and appearance of glutamatergic terminals in the MPG was similar to those of retrogradely traced penile sensory nerve terminals in this ganglion, targeting a specific subset of MPG neurons (Aïoun and Rampin 2006). Whether this is the case in our present work in mouse requires further investigation.

Thus, the origin of the VGLUT₂-IR baskets remains to be elucidated. They could derive from preganglionic autonomic spinal neurons. However, we found that the VGLUT₂-IR baskets did not coexpress VAChT, one of the standard immunohistochemical markers for identification of sympathetic or parasympathetic preganglionic neurons (Keast 2006). Alternatively, the baskets could arise from sensory DRG neurons (Lundberg et al., 1980; Matthews and Cuello 1984; Keast 2006). We explored this by co-staining for VGLUT₂ and CGRP, the latter often used as a marker for peptidergic sensory nerves due to its abundance in DRG neurons, and found that they virtually never coexist. However, only half of the VGLUT₂-IR neurons in lumbar (Brumovsky et al., 2007), thoracic and sacral DRGs (manuscript in preparation) coexpress CGRP. Therefore, the VGLUT₂-IR MPG baskets described here could be derived from non-peptidergic sensory fibers. Finally, the VAChT-positive baskets could have a sensory origin, since cholinergic DRG neurons have been described in the mouse (Bernardini et al., 2004; Tata et al., 2004).

Although less likely, two additional sources of VGLUT₂-IR baskets could be LSC or myenteric neurons. LSC neurons project their fibers, traversing the MPG, on their way to the target organs (Costa and Funess 1973; Hulsebosch and Coggeshall 1982; Kuo et al., 1984). Even though these neurons only seem to upregulate VGLUT₂ after nerve injury, they may normally express and transport peripherally very low levels of the transporter, accumulating in the terminals and becoming immunohistochemically detectable only at this location. Concerning myenteric neurons in the colorectum, we have recently discovered a very small subpopulation of these neurons in the mouse capable of expressing VGLUT₂ mRNA (unpublished data). Viscerofugal neurons terminating in the MPG have been described in the rat (Luckensmeyer and Keast 1995; Luckensmeyer and Keast 1996; Luckensmeyer and Keast 1998). Similar viscerofugal neurons have been described in mouse gut (Miller and Szurszewski 2002), although it remains to be established if they express VGLUT₂.

In conclusion, the present results highlight the possibility that glutamate may be upregulated and have a role in sympathetic neurons after injury. It remains to be established if this role could relate to pain, survival or regeneration mechanisms, or all functions. Furthermore, the presence of perineuronal baskets expressing VGLUT₂ of potential sensory origin in the mouse MPG supports the hypothesis of glutamatergic sensory-autonomic interactions.