Merging advanced technologies with classical ...

G Model

ARTICLE IN PRESS

NSR 3813 1–11

Neuroscience Research xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Neuroscience Research journal homepage: www.elsevier.com/locate/neures

Review article

1

Merging advanced technologies with classical methods to uncover dendritic spine dynamics: A hot spot of synaptic plasticity

2

3

4 5 6 7

Q1

Panchanan Maiti a,∗ , Jayeeta Manna b , Michael P. McDonald a,c a b c

Department of Neurology, University of Tennessee Health Science Center, Memphis, TN 38163, USA Department of Physiology, University of Tennessee Health Science Center, Memphis, TN 38163, USA Department of Anatomy & Neurobiology, University of Tennessee Health Science Center, Memphis, TN 38163, USA

8

9 24

a r t i c l e

i n f o

a b s t r a c t

10 11 12 13 14 15

Article history: Received 26 September 2014 Received in revised form 17 February 2015 Accepted 19 February 2015 Available online xxx

16

23

Keywords: Dendritic spine Synaptic plasticity Golgi stain Fluorescent labeling Protein engineering Super resolved optical microscopes

25

Contents

17 18 19 20 21 22

26 27 28

1. 2. 3.

29 30 31 32 33 34

4.

The structure of dendritic spines determines synaptic efficacy, a plastic process that mediates information processing in the vertebrate nervous system. Aberrant spine morphology, including alterations in shape, size, and number, are common in different brain diseases. Because of this, accurate and unbiased characterization of dendritic spine structure is vital to our ability to explore and understand their involvement in neuronal development, synaptic plasticity, and synaptic failure in neurological diseases. Investigators have attempted to elucidate the precise structure and function of dendritic spines for more than a hundred years, but their fundamental role in synaptic plasticity and neurological diseases remains elusive. Limitations and ambiguities in imaging techniques have exacerbated the challenges of acquiring accurate information about spines and spine features. However, recent advancements in molecular biology, protein engineering, immuno-labeling techniques, and the use of super-resolution nano-microscopy along with powerful image analysis software have provided a better understanding of dendritic spine architecture. Here we describe the pros and cons of the classical staining techniques used to study spine morphology, and the alteration of dendritic spines in various neuropathological conditions. Finally, we highlight recent advances in super-resolved nanoscale microscopy, and their potentials and pitfalls when used to explore dendritic spine dynamics. © 2015 Published by Elsevier Ireland Ltd.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Why is it important to study dendritic spines? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classical techniques used to explore dendritic spine morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Golgi stain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Golgi-Cox staining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Modified Golgi-Cox/rapid Golgi stains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescent labeling of neurons and dendritic spines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Use of lipophilic dye . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Transfection methodology and protein engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

00 00 00 00 00 00 00 00 00

Abbreviations: 3-D, three-dimensional; EM, electron microscopy; CLSM, confocal laser scanning microscope; CCD, charged couple device; TPM, two-photon microscope; MPLSM, multi-photon laser scanning microscopy; sptPALM, single particle tracking photo activated localization microscopy; uPAINT, universal point accumulation imaging in nanoscale topography; STED, stimulated emission depletion; STORM, stochastic optical reconstruction microscopy; DH-QPM, digital holographic quantitative phase microscopy; AD, Alzheimer’s disease; PD, Parkinson’s disease; HD, Huntington’s diseases; GFP, green fluorescent protein; PSD, postsynaptic density; shRNA, short hair pin ribonucleic acid; YFP, yellow fluorescent protein; TEM, transmission electron microscope; ER, endoplasmic reticulum; sER, smooth ER; GABA, gamma amino butyric acid; IP3, inositol 1,4,5-triphosphate; STORM, stochastic optical reconstruction microscopy; PALM, photo activated localization microscopy; FPALM, fluorescence photo-activation localization microscopy; PAINT, point accumulation for imaging in nanoscale topography; SIM, structured illumination microscopy. ∗ Corresponding author at: Department of Neurology, University of Tennessee Health Science Center, 855 Monroe Avenue, Johnson Building, Memphis, TN 38163, USA. Tel.: +1 9012462649. E-mail addresses: [email protected] (P. Maiti), [email protected] (J. Manna), [email protected] (M.P. McDonald). http://dx.doi.org/10.1016/j.neures.2015.02.007 0168-0102/© 2015 Published by Elsevier Ireland Ltd.

Please cite this article in press as: Maiti, P., et al., Merging advanced technologies with classical methods to uncover dendritic spine dynamics: A hot spot of synaptic plasticity. Neurosci. Res. (2015), http://dx.doi.org/10.1016/j.neures.2015.02.007

G Model NSR 3813 1–11

P. Maiti et al. / Neuroscience Research xxx (2015) xxx–xxx

2

5.

6.

35

ARTICLE IN PRESS

Advanced microscopic tools to elucidate dendritic spine architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Ultra structure of dendritic spine by electron microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. High-resolution optical imaging to elucidate dendritic spine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1. Confocal laser scanning microscope (CLSM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2. Two/multi photon microscopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3. Stimulated emission depletion (STED) microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4. Super resolved single fluorphore microscopes (STORM, PALM, FPALM, PAINT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.5. Fiber-optic endomicroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Author’s contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction

Almost a hundred billion neurons and an estimated hundred 37 trillion (1014 ) synapses make the human brain the most complex 38Q3 structure known (Williams and Herrup, 1988; Nimchinsky et al., 39 2004). These neurons are involved in maintenance of basic brain 40 functions as well as learning, memory, and higher-order thought 41 processes. Maintaining healthy synaptic structure is therefore crit42 ical to the preservation of normal brain functions. Importantly, 43 all higher-order neuronal communications are mediated by den44 dritic spines – specialized knob-like structures protruding from 45 dendritic shafts (Hering and Sheng, 2001; Nimchinsky et al., 2002; 46 Bourne and Harris, 2007). They are considered specialized, semi47 autonomous postsynaptic compartments on which most excitatory 48 synapses (over 95% in vertebrate brain) impinge (Hering and Sheng, 49 2001; Nimchinsky et al., 2002; Bourne and Harris, 2007). The den50 dritic spines are of multiple shapes and sizes, with diverse functions 51 depending on type and activity of the neurons (Jones and Powell, 52 1969; Harris et al., 1992; Hering and Sheng, 2001; Nimchinsky et al., 53 2002; Bourne and Harris, 2007). Most importantly, spine morphol54 ogy determines the strength and stability of the synapse, and can 55 be significantly altered in neurodevelopmental and neurodegen56 erative diseases (Fiala et al., 2002; van Spronsen and Hoogenraad, 57 2010). Experimental evidence suggests that abnormal spine mor58 phology is a principal cause of synaptic dysfunction in a number 59 of neurological and neuropsychiatric disorders (Fiala et al., 2002; 60 Bredesen et al., 2006; Rubinsztein, 2006). However, in order to 61 understand the role of dendritic spines in synaptic plasticity and 62 disease, it is first vital to characterize them accurately – not only 63 their numbers, but also their three-dimensional (3D) structure 64 (Calabrese et al., 2006; Kasai et al., 2010). Because of diffrac65 tion limitations and lack of spatial resolution in light microscopy, 66 the dynamics and nanoscale structure of spine necks and distri67 butions of spine proteins remained unexplored for many years. 68 Recently, new technologies have facilitated significant advances in 69 our understanding of the basic structure and function of dendritic 70 spines. Notably, the development of super-resolution fluorescence 71 microscopes has enabled capture of nanoscale-level spine struc72 tures in living neurons non-invasively. However, a great deal still 73 needs to be done, particularly with respect to dendritic spine mor74 phology and its role in impaired synaptic plasticity in neurological 75 disorders. In this review we will discuss how classical methods 76 and novel approaches can be used in a complimentary fashion to 77 discover the detailed structures and functions of dendritic spines. 36Q2

is under stress or in an injury or disease state because structural abnormalities in dendritic spines are thought to underlie symptomatology in many neuropathological states (Fiala et al., 2002). Changes associated with impaired cognitive function include loss or decrease in spine number or density, reduced size, increased number of immature spines and varicosities, distortion of spine shape, and enhanced ectopic spine formation (Table 1; Fiala et al., 2002; Penzes et al., 2011). For example, a reduction in spine size ad reduced dendritic length has been reported in the striatum of schizophrenics and the motor cortex of infants with Down’s syndrome (Roberts et al., 1996; Marin-Padilla, 1972). In contrast to reductions in spine density, increased spine density has been observed during brain development in such conditions as phenylketonuria, fragile-X syndrome, and exposure to an enriched environment (Table 1; Huttenlocher and Dabholkar, 1997; Lacey, Q4 1985; Irwin et al., 2001; Globus et al., 1973; Berman et al., 1996). Formation of varicosities is another morphological change common following brain injury, which might be due to abnormal organization of microtubules or actin polymerization. Swelling of dendritic trunks can also produce varicosities, as observed in progressive neurodegenerative disorders such as Pick’s disease, frontal lobe dementia, and motor neuron disease (Sotrel et al., 1993; Hogan et al., 1987; Ferrer et al., 1990, 1991). Ectopic spines can be observed during early development in olivopontocerebellar atrophy, fetal alcohol syndrome, Menkes disease, epilepsy, and in cats with GM2 gangliosidosis (Ferrer et al., 1988; Mohamed et al., 1987; Hinton et al., 1991; Walkley et al., 1990). In addition to alterations in spine size and number, ultrastructural intra-spine changes, including cytoplasmic densification, hypertrophy of organelles or spine volume, and formation of aberrant synapse-like connections have all been linked with

Table 1 Dendritic spine pathology in neurological diseases. Spine pathology

Occurrences

Decreased spine density

Deafferentation, agenesis, mental retardation, malnutrition, poisoning, alcohol abuse, epilepsy, spongiform encephalopathies, Alzheimer’s disease, and others Some types of deafferentation, environmental enrichment, Fragile-X syndrome, sudden infant death syndrome, stimulatory drug use Sensory deprivation, schizophrenia, Down syndrome Deafferentation, agenesis, mental retardation, malnutrition, poisoning, alcohol abuse, epilepsy, spongiform encephalopathies Acute excitotoxicity, traumatic injury and edema, epilepsy, hypoxia/ischemia Olivopontocerebellar atrophy, Menkes disease, metabolic storage diseases

Increased spine density

Reduction in spine size Distortion of spine shape

78

79 80 81

2. Why is it important to study dendritic spines? There are several reasons to study the structure, function, genesis, and loss of dendritic spines. In addition to developmental changes, it is important to explore spine dynamics when the brain

00 00 00 00 00 00 00 00 00 00 00 00 00

Varicosity formation Ectopic spines

Reprinted with permission from Fiala et al. (2002).


82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112


ARTICLE IN PRESS P. Maiti et al. / Neuroscience Research xxx (2015) xxx–xxx

113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142

adverse neurodevelopmental conditions (Fiala et al., 2002). Distorted spines, including absent or enlarged heads and long, tortuous filopodia with multiple fusiform swellings are associated with mental retardation (Fiala et al., 1998). Aberrant spines may also result from alterations in postsynaptic density (PSD) protein, polyribosomes, endosomes, or the spine cytoskeleton following neurotoxic insult. For example, the PSD is thickened by anoxia or ischemia, and a higher density of polyribosomes has been observed after deafferentation (Martone et al., 2000; Steward, 1983). Similarly, elaborated or atrophied spine apparatus, hydropic swelling, and vacuolization of the endoplasmic ER of dendrites and spines have been observed in human edematous brain (Castejon et al., 1995). A complete absence of smooth endoplasmic reticulum (sER) in spines has been observed in Purkinje cells of an ataxic mutant rat expressing an endogenous in-frame deletion in the Myo5a gene (Dekker-Ohno et al., 1996; Futaki et al., 2000). Other spine abnormalities include formation of giant spines, which is observed following deafferentation. Dendritic spine anomalies are typically investigated in animal models, but aberrant spine density or morphology has also been observed post-mortem in patients with many brain diseases including mental retardation, Down syndrome, fragile X syndrome, epilepsy, and several neurodegenerative diseases. It is clear from the heterogeneity of spine abnormalities that nano-scale alterations in morphology are found in a wide variety of neurological abnormalities. These spine abnormalities have been explored using different classical staining techniques as well as advanced imaging tools, which are discussed in detail below. The goal of this anatomical approach is to document these pathological changes and learn how they affect synaptic efficacy and symptom expression.

144

3. Classical techniques used to explore dendritic spine morphology

145

3.1. Golgi stain

143

146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174

Golgi impregnation, also called ‘black reaction’, is a powerful classical histochemical technique. It has long been the gold standard with respect to neuronal and dendritic morphology (Mazzarello, 1999) (Fig. 1A–D). It is appropriate not only for visualization and investigation of neuronal anatomy in experimental animals, but also for autopsied human brain tissue (Millhouse, 1969; Glaser and Van der Loos, 1981; Spacek, 1989; Gibb and Kolb, 1998; Zhang et al., 2003). The Italian scientist Camillo Golgi introduced this eponymous stain and used it to sketch dendritic morphology (Mazzarello, 1999). Prof. Golgi was not able to successfully elucidate dendritic spines using this stain. Spines were first successfully visualized, sketched, and documented as genuine structures of neurons by neuro-anatomist Santiago Ramón y Cajal using the Golgi stain in 1888 (Garcia-Lopez et al., 2007; Buell, 1982; Shepherd, 1996; Hering and Sheng, 2001; Nimchinsky et al., 2002; Dhawale and Bhalla, 2008). Cajal also used the Golgi stain to precisely differentiate the now well-known categories of spines, including filopodia, thin, stubby, mushroom, and cup shaped, from different brain areas and in different species (Garcia-Lopez et al., 2007). The success of the Golgi stain is based on its tendency to stain a few neurons, seemingly at random, leaving the surrounding neurons unstained and thus providing strong contrast with the stained neurons (Fig. 1). An analysis of the ingredients used in the Golgi stain suggests that the potassium dichromate impregnates the neurons and glia and hardens them, whereas the silver nitrate turns the neurons black (Ranjan and Mallick, 2010, 2012; Mancuso et al., 2013). It is not known why some neurons are stained and others are not; in fact the staining mechanism remains enigmatic today, well over hundred years after its introduction. In spite of

3

its ability to reveal clear dendritic spine morphology from unfixed animal or human brain tissue, the Golgi stain has some major drawbacks when it comes to reliable characterization of dendritic spines. First, there is a bias or inconsistency in impregnation of the neurons, which can reduce staining specificity and affect reproducibility (Globus and Scheibel, 1966; Pasternak and Woolsey, 1975; Zhang et al., 2003). Second, it cannot stain formalin-fixed tissue properly even when immersed for more than a month in the stain (Marin-Padilla, 1990). Third, the staining success rate is unsatisfactory for many applications; in the best cases less than 5% of neurons are stained (Spacek, 1989). Fourth, because it is a random and unpredictable staining method with minimal background, it is often difficult to pinpoint a neuron’s precise location in the brain (Pilati et al., 2008). Finally, during the impregnation process in Golgi staining the granules of silver or mercury chromate are deposited on the surface of the neuronal cell body. They also precipitate as small, dense granules inside the nerve cells, but are unable to stain the nucleus or mitochondria reliably (Fairen, 2005). 3.2. Golgi-Cox staining By using the Golgi method we can achieve a clear picture of an entire neuron with minimal background along with relatively well-stained dendritic spines, but it has certain limitations as mentioned above. To overcome these limitations, including the low rate of impregnation, a slight modification of Golgi’s original stain was introduced called the Golgi-Cox stain (Cajal, 1988). The brain tissues are immersed in solutions containing potassium dichromate and mercuric chloride and then treated with sodium carbonate or ammonia instead of silver nitrate (Landas and Phillips, 1982; Gibb and Kolb, 1998) (Table 2). These modifications enhance the movements of the metallic ions in the staining solutions and increase the influx of those ions within the neurons, which increases the likelihood of staining (Ranjan and Mallick, 2010, 2012). This technique can be used for both less-myelinated and younger neurons as well as more heavily myelinated and older neurons. The Golgi-Cox method is also suitable for freshly or lightly fixed tissues (Rosoklija et al., 2003), as well as brain tissue that has been frozen for a long time (Melendez-Ferro et al., 2009). There are a number of distinct advantages of this method over the traditional Golgi method. First, the Golgi-Cox stain labels neurons and dendritic spines better. Second, the dendritic spine morphology and dendritic arborization is clearer than with the classical Golgi stain (Buell, 1982; Castano et al., 1995). Third, the probability of neuronal staining is increased (Scheibel and Tomiyasu, 1978). Despite its advantages, the Golgi-Cox method is not perfect. The assay is time-consuming and provides inconsistent and non-uniform staining, which is difficult to reproduce (Rutledge et al., 1969; Glaser and Van der Loos, 1981; Zhang et al., 2003). The Golgi-Cox stain is also not ideal for formalin-fixed tissue that has been stored for periods longer than a year (Rosoklija et al., 2003; Melendez-Ferro et al., 2009) (Table 2). 3.3. Modified Golgi-Cox/rapid Golgi stains A number of modifications have been made to the Golgi and Golgi-Cox methods to address the issues of long duration, low impregnation rate, inconsistent crystallization, and precipitation of chromogens in the sections (Pilati et al., 2008; Ranjan and Mallick, 2010). Both chemical and physical modifications have been made. Chemical modifications include replacing mercury chloride in the Golgi-Cox with osmium tetroxide to significantly reduce staining time (Williams et al., 1978; Marin-Padilla, 1990), substi- Q5 tution of aldehydes for the traditional osmium-based techniques (Armstrong and Parker, 1986), (iv) and including constituents such as sodium carbonate or ammonium solution in the staining solution to facilitate staining of glia (Ibrahim et al., 1968; Pessacq,


175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

193

194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223

224

225 226 227 228 229 230 231 232 233 234 235 236

G Model

ARTICLE IN PRESS

NSR 3813 1–11


4

Fig. 1. Fresh rat brain tissues were stained with classical Golgi and Golgi-Cox stain. Golgi stained CA1 (A) and CA3 (B) hippocampal neurons and their dendritic spines (C and D) respectively. E–M: neurons were stained by Golgi-Cox method. E, F, and G: CA1, CA3 and cortical pyramidal neurons; H, I, J: individual pyramidal neurons from CA1, CA3 of hippocampus and cortex; K, L, M: dendritic spines from CA1, CA3 of hippocampus and cortex respectively. The reproducibility of Golgi-Cox stain is more and dendritic spines were better stained by this method compared to classical Golgi stain.

Table 2 Comparison of different Golgi stains. Stain

Principal ingredients

Advantages

Limitations

Golgi

Potassium dichromate (2%) Silver nitrate (2%)

Provides a clear picture of the entire neuron with minimal background; Relatively stably stained spines; Inexpensive

Inconsistent specificity, reproducibility, and success rate; Time consuming; Unable to stain fixed tissue; Cannot be used with living neurons; Insufficient resolution in the z-axis; Lack of 3-D structure or ultrastructure

Golgi-Cox

Potassium dichromate (5%) Mercury chloride (5%) Potassium chromate (5%) Sodium carbonate (5%)/ammonia solution (5%)

Provides clear dendritic spine and arborization structures; Suitable for fresh or lightly fixed tissue; Increased probability of staining; Stains lightly myelinated tissue as well as heavily myelinated; Inexpensive

Inconsistent specificity and reproducibility; Time consuming; Cannot be used with living neurons; Insufficient resolution in the z-axis; Lack of 3-D structure or ultrastructure

Rapid or modified Golgi

Potassium dichromate (5%) Osmium tetroxide (OT)/aldehyde Sodium carbonate (5%)/ammonia solution (5%)

Thick sections (90–350 ␮m) can be stained; Stains both neurons and glia; Time efficient (48 h); Less chance of damaging cellular structures

Cannot be used with living neurons; Insufficient resolution in the z-axis; Lack of 3-D structure or ultrastructure




237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265

266

267 268 269 270 271 272 273 274 275 276 277 278 279

280

281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297

1970) (Table 2). Physical modifications include variations in pH (Gonzalez-Burgos et al., 1992; Angulo et al., 1994), the use of vacuum to change the temperature (Friedland et al., 2006) (Berbel, 1986; Angulo et al., 1994), using microwave energy (Armstrong and Parker, 1986; Zhang et al., 2003), coating the brain blocks with egg yolk (Zhang et al., 2003), and varying the temperature of the incubating medium (Armstrong and Parker, 1986; Berbel, 1986; Angulo et al., 1994). To increase impregnation rate, improve quality, and decrease reaction time Ranjan et al. performed a simple and inexpensive modification of Golgi-Cox stain, namely incubating the tissue at 37 ◦ C (Ranjan and Mallick, 2010, 2012). They sectioned an entire rat brain into small blocks and incubated them at 37 ◦ C for 1 day, and observed improvements in impregnation rate and staining consistency (Ranjan and Mallick, 2010, 2012). They claimed this these rapid, inexpensive and simple modification provided more reproducibility and more uniform staining, with very good resolution of neuronal soma dendrites, and spines (Ranjan and Mallick, 2010, 2012). Similarly, Pilati et al. shortened impregnation time by using an aldehyde-containing solution in place of mercury chloride (Pilati et al., 2008). Rosoklija et al. also improved the quality of impregnation rate of formalin fixed human brain tissues using the Golgi-Kopsch technique (Rosoklija et al., 2003). Overall, these variations reduced duration of staining time and improved quality, but the staining consistency and reproducibility remained unsatisfactory. Moreover, none of these modified techniques is satisfactory in terms of achieving detailed pictures of dendritic spines or being able to discern their numbers or morphology in different brain tissues. 4. Fluorescent labeling of neurons and dendritic spines The classical staining methods described above can be use to investigate the anatomy of neurons in vivo but are not appropriate for visualization of cultured living neurons. Several methods have been developed to overcome the limitations of Golgi stains for spine study, including the use of various commercially available tracer dyes, fluorochrome-labeled antibodies, and genetically encoded fluorescent proteins such as green fluorescent protein (GFP) or yellow fluorescent protein (YFP; Staffend and Meisel, 2011; Malinow et al., 2010). Among these methods, viral-transfected fluorescent protein engineering, immuno-labeling techniques, and transgenic animal engineering have been helpful in elucidating the detailed structure and dynamics of dendritic spines of different brain diseases (Sala and Segal, 2014). 4.1. Use of lipophilic dye Because of its excellent specificity, GFP is one of the most wellaccepted neuronal labeling techniques. However, a similar or even higher level of neuronal labeling can be achieved using lipophilic tracer dyes DiI and its derivatives, with lower cost and a shorter time frame (Cheng et al., 2014; Papa et al., 1995; Chen et al., 2011; Maiti et al., 2011). Dyes such as DiO, DiI, and DiD are considered one of the most effective method for labeling and detecting spines in cultured neurons (Hosokawa et al., 1995; Papa et al., 1995; Chen et al., 2011; Maiti et al., 2011; Attar et al., 2012). Endocytotic dyes such as FM1-43 (Dhawale and Bhalla, 2008) and bio-enzymebased markers are also frequently used for neuronal labeling (Ryan, 2001; Couch et al., 2010; Anderson et al., 2011). There are several approaches used to incorporate these dyes into lipid microcapsules, including biolistic delivery system and gene guns. These techniques non-specifically label the neurons in a slice or a chunk of brain tissue or dissociated cultured neurons. Microinjection techniques are also frequently used to label individual neurons in dissociated

5

cultures (Papa et al., 1995; Chen et al., 2011; Maiti et al., 2011; Attar et al., 2012). In this technique a glass micro-pipette loaded with a fluorescent dye dissolved in fish oil is controlled by an automated micromanipulator to touch the dye to the surface of the neuron (Fig. 2B–E). The lipid droplets containing fluorescent dye bind to the neuronal membrane after its application due to their lipophilic nature. The labeled neurons are left to disperse the dye throughout the entire neuron for 6–12 h, after which a fluorescent signal can be visualized by confocal or ordinary fluorescent microscopy. A number of research groups have been using this simple technique to label neurons and study synaptotoxicity in vivo and in vitro (Papa et al., 1995; Chen et al., 2011; Maiti et al., 2011; Attar et al., 2012). For example, in a recent study examining dendritic spine loss induced by amyloid beta protein (the misfolded protein that accumulates into plaques in Alzheimer’s disease), we used this technique to label 3-week-old cultured hippocampal neurons and capture clear images of individual neurons and dendritic spines (Fig. 2B–J) (Maiti et al., 2011; Attar et al., 2012). An advantage of this technique is its selectivity. Researchers can label the neurons they wish, e.g. they can select individual healthy neurons by observing them under microscope and dye them accordingly. Only a small population of neurons are labeled by lipophilic dyes, helping us to reduce image background, and conventional bright field microscopes are sufficient to capture the spine images. Most importantly, all kinds of spine morphology can be visualized in fine detail using this technique with high-resolution confocal microscopy. Thus lipophilic dyes are useful for quantification and morphological analysis of spines, including the study of abnormal spines from diseased brains. Moreover, using this technique spines can be studied from different neuronal populations, and qualitative as well as quantitative analyses can be compared with other traditional methods (Cheng et al., 2014). Due to the above-mentioned advantages, DiI fluorescence labeling has become very popular. However, major obstacles to more widespread use are its optimization and accurate quantification of spines (Cheng et al., 2014). In addition, after a period of time dye becomes diffuse, which makes it difficult to study spine morphology from the distal part of the original application site (Rasia-Filho et al., 2010). 4.2. Transfection methodology and protein engineering The use of lipophilic fluorescent dyes to label neurons is a laborious and time-consuming process. In addition, only small population of neurons can be labeled using these dyes (Papa et al., 1995; Chen et al., 2011). Transfection methodologies may be a way to address some of these issues (Feng et al., 2000; Gong et al., 2003; Livet et al., 2007). Transfection can be used in vivo as well as in cultured brain slices or dissociated neurons. Because of the high specificity achieved recently with enhancement and refinement of methods, the transfection of neurons with fluorescent markers is now one of the most widely used methods to study neuronal architecture (Colello et al., 2012; Table 3). In the last few years scientists have been able to quantify spine numbers, spine head diameter, and other post-synaptic marker proteins using fluorescence immuno-labeling techniques (Cheng et al., 2014). These techniques are relatively simple. A plasmid is used to carry the fluorescent protein (e.g. GFP, YFP, etc.). In the presence of transfecting reagent, the plasmid can be taken up by cultured neurons or the brain area into which the plasmid was injected. The plasmid has a structure homologous to that of a promoter gene of interest and can be incorporated into the protein-generating cellular machinery to produce large amounts of the transfected molecules. Ultimately, the fluorescent protein is expressed along with expression of the target protein (Matus et al., 2007). Both neurons and spines can be visualized with fluorescent microscopes using this labeling method. Recently, scientists have used plasmid-encoded soluble enhanced


298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335

336

337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361

G Model

ARTICLE IN PRESS

NSR 3813 1–11


6

Fig. 2. Fluorescent labeling of neurons by DiI. The DiI was dissolved in fish oil, loaded in glass micropipette and applied to 3 weeks cultured hippocampal neurons using the micromanipulator (Eppendorf Cell Technology). After 12–16 h of dye application, the images were taken by confocal laser scanning microscope. A: Chemical structure of DiI; B–E: steps of DiI labeling; F: Image just after applying DiI; G: neuronal images after 16 h of DiI labeling; H: Morphology of single hippocampal neuron after 16 h of DiI application; I: Images of dendritic branches; J: image of dendritic spines. Table 3 Comparative advantages and disadvantages of fluorescence dyes, transfection, and fluorescent protein engineering.

362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386

Spine pathology

Advantages

Limitations

Lipophilic dyes

Simple, easy to label; Requires less time; Can be used with in vivo cells; Lower background; Better resolution than bright field

Labels a small population of neurons; Images may be blurred; Cannot be used with live neurons; Micromanipulators and gene guns are required and costly; Labeling method is laborious; Dye becomes diffuse with time

Transfection methodologies; fluorescent protein engineering

Can be used in vivo as well as cultured sections and dissociated neurons; Provides a clear fluorescent signal; Shows dendritic architecture including spines; Can label individual neurons with multi-colored dyes

Transfection rate is low (10–20%) Transfected neurons are vulnerable to death; Transfection suppresses synaptic transmission; Not specific to neurons; Expensive; Time consuming

GFP (eGFP), membrane-tagged eGFP, and mRFP-ruby tagged Life act, a small actin-binding peptide, to label neurons (Mancuso et al., 2013). Similarly, neuron-specific promoters (e.g. CaMKII and Syn) can be used to distinguish neurons from glia. The major disadvantage of transfection techniques is the small population of neurons (10–20%) that are transfected, and that the transfected neurons are more vulnerable to death. In addition, synaptic transmission may be hampered or suppressed in transfected neurons, possibly leading to alterations of spine morphology. Therefore, depending on one’s needs, labeling neurons by transfection methods may not be ideal to study dendritic spine structure. Due to the low success rate and neuronal vulnerability to transfection reagents, some scientists prefer transgenic strategies or viral expression for labeling living neurons with fluorescent proteins. A number of viral vectors have been used to express fluorescent proteins in neurons for imaging, including recombinant adeno-associated virus (rAAV), lenti-virus, and rabies virus (Chamberlin et al., 1998). GFP-tagged molecules are routinely used to label neurons and synaptic proteins of interest. For example, GFP-tagged PSD-95 can be used to mark the post-synaptic side of a synapse (Ehrlich and Malinow, 2004; Racz and Weinberg, 2013). Similarly, the Thy-1 promoter can be incorporated into a vector with a YFP tag, making it possible to distinguish dendrites and spines of individual neurons from those of their unlabeled neuronal and non-neuronal neighbors (Feng et al., 2000). The brainbow

technique is another method of labeling individual neurons, using multiple fluorescence dyes that are expressed in random recombinations of Cre-based transgenes to produce a different color in each neuron (Gong et al., 2003; Livet et al., 2007). However, these methods are far from perfect. In a recent experiment we made bilateral intra-striatal stereotactic injections of an rAAV-mediated shRNA tagged with eGFP. After 3 weeks we observed widespread eGFP-fluorescence in the entire brain, from olfactory bulbs to the cerebellum. Unfortunately, the transfected signals were uneven, with some neurons showing strong signals and others weak. Additionally, it was difficult to capture clear spine images from the transfected neurons, especially those showing a faded or diffuse eGFP signal. The exact reason for this phenomenon is not immediately evident. However, for our purposes we feel that the use of a lipophilic dye would have been a better choice to label individual neurons to detect dendritic spine architecture and spine morphol- Q6 ogy (Fig. 3). 5. Advanced microscopic tools to elucidate dendritic spine architecture Light microscopes do not provide sufficient resolution for the study of dendritic spines. The discovery of transmission electron microscopy (TEM) allowed us to see the ultrastructure of dendritic spines for the first time (Harris and Stevens, 1989; Papa et al., 1995).


387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403

404 405

406 407 408 409



7

Fig. 3. Recombinant adeno-associated virus (rAAV) mediated labeling of neurons. A and B: rAAV mediated transfected mouse brain eGFP signals from cortex (A) and CA1 region of hippocampus (B); C: dendritic spines from cortical pyramidal neuron.

However, TEM cannot elucidate the activity-dependent spine dynamics in living neurons. To resolve this issue scientists developed a number of advanced microscopic tools, including confocal 412 laser scanning microscopy (CLSM) and two-photon microscopy 413 (TPM)/multi-photon laser scanning microscopy (MPLSM). Unlike 414 light microscopes, these techniques overcome the diffraction bar415 rier of light and provide spatial resolution from microscale to 416 nanoscale. Digital recording of spine images, including semi417 automated software-guided tracing systems (e.g. NeuroZoom, 418 Neurolucida), were later developed to characterize the fine 3-D 419 structure of dendritic spines (Glaser and Glaser, 1990; Jacobson 420 et al., 1997; Rodriguez et al., 2008; Sala and Segal, 2014). Recently, 421 fully automated computational products have been developed 422 for use in CLSM and TPM/MPLSM (Rodriguez et al., 2008; Sala 423 and Segal, 2014) to reconstruct the 3-D structure of spines auto424 matically. These tools are much faster and more accurate than 425 the existing semi-automated technology. Historically, the super426 resolution structure of the entire spine, including head, neck, 427 and base, were still difficult to image clearly using CLSM and 428 TPM/MPLSM (Rodriguez et al., 2008; Sala and Segal, 2014). To 429 achieve more precise spine geometry with high-resolution images, 430 scientists developed new algorithm-based software packages that 431 collect terabyte-sized image stacks and can reconstruct the entire 432 spine accurately with the full 3-D structure (Wearne et al., 2005; 433 Rodriguez et al., 2006, 2008). Unfortunately, there is still an issue 434 of image resolution, and this is a serious limitation to using CLSM 435 and TPM/MPLSM to study deeper brain tissues (Rodriguez et al., 436 2008). Additionally, these advanced methods cannot be used to 437 capture spine images from freely moving animals (Barretto and 438 Schnitzer, 2012; Gu et al., 2014). Recently, several super-resolution 439 microscopes, including nano-scopes, have been developed to cap440 ture highly resolved images of dendritic spines from living neurons. 441 The include single-particle tracking photo-activated localization 442 microscopy (sptPALM; Manley et al., 2008), universal point accu443 mulation imaging in nanoscale topography (uPAINT; Giannone 444 et al., 2010), stimulated emission depletion (STED) microscopy 445 Q7 (Willig et al., 2006; Hein et al., 2008), stochastic optical reconstruc446 tion microscopy (STORM; Heilemann, 2010), digital holographic 447 quantitative phase microscopy (DH-QPM; Marquet et al., 2013), 448 and fiber-optic endo-microscopes (Murari et al., 2011; Barretto and 449 Schnitzer, 2012; Gu et al., 2014). They produce very clear images 450 of living neurons and reveal unprecedented details of dendritic 451 spine structure. These tools can also be used to study the dis452 tributions and dynamics of receptors and signaling molecules in 453 dendritic spine (Murari et al., 2011; Ige et al., 2000; Gu et al., 2014). 454 Below we discuss in greater detail the various advanced micro455 scopes, their applications, advantages, and limitations in exploring 456 457 dendritic spine architecture. 410 411

458

459 460

5.1. Ultra structure of dendritic spine by electron microscopy After the discovery of TEM dendritic spines were imaged and studied from several aspects, and it is now widely accepted

Fig. 4. Comparative morphology of dendritic spines by confocal laser scanning microscopy and by TEM. A: different types of dendritic spines imaged by confocal laser scanning microscopy. B: ultra-structure of spine by TEM (total magnification: Q10 50 × 103 times). TEM can depict ultra-structure of spine which is the major limitation in light or fluorescent microscopes. Red arrow: presynaptic vesicles, green arrow: postsynaptic density protein; star: spine apparatus. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

that they are specialized and distinct compartments bearing neurotransmitter receptors and other sub-organelles involved in excitatory synaptic transmission (Castejon and Valero, 1980; Hering and Sheng, 2001; Fiala et al., 2002; Sala and Segal, 2014). Indeed, the inability of light microscopes to elucidate the ultrastructure of dendritic spines was only fully realized with the advent of TEM and scanning electron microscopy (SEM). Before EM it was not possible to study the volumetric changes in the dendritic spine head in spiny neurons during neurotransmission (Harris and Stevens, 1988). Due to its high resolving power, the fine intracellular structure inside the spine was firsts identified by TEM. This led to the observation of polyribosomes and the suggestion that dendritic spines are capable of local protein synthesis (Bodian, 1965). The presence of sub-organelles inside the spine, including PSD, sER, and the spine apparatus, and cytoskeletal proteins such as F-actin, coated vesicles, etc., were first identified using TEM (Papa et al., 1995; Hering and Sheng, 2001; Nimchinsky et al., 2002) (Fig. 4). TEM also provided the first clear information about the dendritic spine neck. This new information confirmed that many spines necks are beyond the limit of what conventional light microscopy is capable of visualizing in detail. Immuno-labeling techniques used with TEM have been used to reveal the distribution of neurotransmitter receptors on the surface of dendritic spines (Ottersen et al., 1988; Ige et al., 2000; Nimchinsky et al., 2002). For example, Ottersen et al., used immuno-gold to identify glutamate, GABA, and inositol 1,4,5triphosphate (IP3) receptors, and scaffolding proteins such as PSD95, shank, and homer in the post-synaptic membrane of cerebellar Purkinje neurons (Ottersen et al., 1988; Mateos et al., 1998; Ige et al., 2000; Sen and Gleason, 2006). EM can also be used to quantify the number of post-synaptic terminals, which in turn can be used to estimate the number of dendritic spines. One application of EM is to study spine organelles and the measurement of PSD area in individual spines. Similar to TEM, SEM can also be used to study the ultrastructure of dendritic spine (Castejon and Valero, 1980). EM provides a spatial resolution up to the nanometer level,


461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496

G Model NSR 3813 1–11 8 497 498 499 500 501 502 503 504

505 506

507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560


but requires fixed tissue. This restricts our ability to understand the dynamic events of a living neuron. Thus TEM is unable to provide a clear picture of activity-dependent changes in a single spine. Some disadvantages of using EM are that labeling of multiple dendritic spine proteins is difficult, and 3D reconstruction of single spines is laborious and time-consuming. In addition, it is a costly method that requires well-trained and dedicated researchers to view and capture the right images. 5.2. High-resolution optical imaging to elucidate dendritic spine structure 5.2.1. Confocal laser scanning microscope (CLSM) Using light or fluorescent wide-field microscopy to study dendritic spines is not ideal because the images produced are blurred, with insufficient resolution and inferior quality. Similarly, the use of EM to study spine ultra-structure has major limitations as mentioned above. The application of the pin-hole concept in CLSM to achieve high-resolution spine images resolved this issue to a certain degree (Moser et al., 1994; Papa et al., 1995). Importantly, spine images can be captured by CLSM from fluorescently labeled fixed tissues, living neurons from acute slices, or dissociated cultured neurons, using appropriate excitation/emission filters. Unlike classical techniques, CLSM can provide images from living neurons, thus revealing information about spine dynamics (Moser et al., 1994; Papa et al., 1995). CLSM is routinely used to study the structure–function relationships of spines or to evaluate the importance of dendritic spines in synaptic plasticity (Knott and Holtmaat, 2008). Using the appropriate software it can also be used to reconstruct image stacks to show high-resolution 3-D structure, thus providing information about hidden spines in the z-direction (Castano et al., 1995). The development of new fluorescent dyes, immuno-labeling techniques, protein engineering tools, and genetically engineered animals allowed for major advances in the study of the functional aspects of dendritic spines using CLSM (Mancuso et al., 2013; Sala and Segal, 2014). More recently the introduction of calcium-sensitive fluorescent dyes (e.g. fura-2, indo-1, fluo-3, fluo4, and Calcium Green-1) have helped us monitor spine dynamics in acute slices or dissociated cultures using CLSM. All these experimental approaches have significantly advanced our understanding of the role of dendritic spines in synaptic remodeling (Hosokawa et al., 1995; Korkotian and Segal, 2001; Eilers et al., 1995). CLSM also provides better information about individual spine dynamics, including their z-plane resolution, which is almost impossible by light microscopy (Chan-Palay et al., 1974; Feldman and Dowd, 1975; Harris and Stevens, 1988). Some limitations of CLSM are that the laser used cannot penetrate very deep into the tissue (