Altered Loyalties of Neuronal Markers in Cultured Slices of Resected Human Brain Tissue

Abstract

Organotypic cultures from normal neocortical tissue obtained at epilepsy surgery show a severe injury response. This response involves both neuronal degeneration and the proliferation of reactive cells. A salient feature of the reactive cells is the co‐expression of microglial and astrocytic markers. Surprisingly, the reactive cells also began to express neuronal markers Tubulin βIII and MAP2 adding to the confusion about their origin. Concomitant with their appearance in reactive cells MAP2 and Tubulin βIII expression disappeared from neurons. While NeuN expression decreased significantly, it did not entirely disappear from many neurons. Moreover, it was not observed in reactive cells, showing that NeuN is a reliable marker of neurons.

Introduction

To explore physiological properties of the human brain, organotypic slices provide a model system that differs in several aspects from the more extensively studied experimental animal models. Acute experiments with resected human brain tissue have revealed intrinsic differences between the electrophysiological characteristics of human neurons and those of mice 24, 27. Moreover, human astrocytes are larger, possess more elaborate arborizations and have electrical properties that differ from their rodent and non‐human primate equivalents 17, 18. Longer lasting experiments with human brain slices kept in culture conditions 28 would allow cellular properties to be compared between control subjects and patients with a neurological disease such as Alzheimer 32. However, it should be kept in mind that properties of the cells in the slices may change due to tissue processing and in vitro conditions. There are two sources of human brain tissue: one concerns tissue donated by deceased human subjects or patients [post‐mortem tissue 28, 29] and the other consists of tissue removed during an operation [resected tissue 31]. Recently, we showed that slice cultures of resected human brain tissue exhibit a very strong injury response 31. This injury response was characterized by the emergence of many reactive cells, which could be divided into two major types: IVR cells (in vitro reactive cells) and IVFL cells (in vitro foam‐like cells). Usually, IVR cells were large with long processes and were associated with blood vessels. They expressed reactive cell markers (Nestin and Vimentin) as well as astrocytic markers [glial fibrillary acidic protein (GFAP) and calcium binding protein S100β] and microglia marker [human leukocyte antigen (HLA)]. The IVFL cells had an amoeboid (round) morphology and no, or short, processes. An association with blood vessels was less clear and IVFL cells expressed a reactive marker (Vimentin), an astrocytic marker (S100β) and microglial markers (HLA, calcium binding protein Iba1 and CD68). Moreover, the non‐specific cytoskeletal protein Tubulin βI that was mostly found in microglia in noncultured slices, labeled both IVR and IVFL cells in cultured slices 31. Further, oligodendrocyte precursor marker (Olig2) was found in smaller IVR cells. We established that several of the aforementioned markers were expressed in the same cell. Concomitant with the emergence of reactive cells neurons degenerated. Here, we specifically examine the expression patterns of neuronal markers in relation with the in vitro injury response. We found that MAP2 and Tubulin βIII appeared in many IVR cells, whereas NeuN remained loyal to its neuronal lineage. Moreover, MAP2 and Tubulin βIII co‐localized with each other, and with Nestin, Vimentin, GFAP, S100β and Tubulin βI. Tubulin βIII was also found together with HLA and Olig2, but not with Iba1.

Materials and Methods

Brain samples

For this study, specimens of normal temporal cortex from 45 patients with medication‐refractory temporal lobe epilepsy were used (Supporting Information Table S1). These tissue specimens comprise a subpopulation of a larger collection, which overlaps with the subpopulation used for a previous report 31. For ethical reasons, the size of the specimens is limited, which in turn restricted the number of features (ie, gene expression, protein expression and immunocytochemical detection of various antigens and combinations thereof) of the tissue that we could assess. The tissue had to be removed to reach the epileptic focus and was not needed for diagnostic purposes. Prior to resection, this tissue had not been probed by invasive recording techniques such as depth electrodes or subdural grids and there was no evidence of previous mechanical damage 8. Adjacent tissue samples from many of the epilepsy patients described in this study have also been used for acute electrophysiological recordings 24, 27. Permission for the use of human brain tissue for in vitro research was granted by the Ethical Medical Committee of the VU University Medical Center (VUmc) in Amsterdam, where the operations took place. Brain tissue specimens were obtained with written informed consent and patient information was treated in accordance with the Declaration of Helsinki.

Culture procedure

For the processing of brain tissue and the culture procedure, we refer to previous reports 28, 29, 31, 32. Essentially, the tissue was cut into rod‐like pieces and chopped using a McIlwain tissue chopper, in such a way that 300‐μm thick slices comprising all cortical layers and a part of the white matter were obtained 29. Reference slices were fixed immediately for immunocytochemical staining or snap frozen for quantitative PCR analysis and Western blotting. The rest was distributed systematically over 24 wells plates, each well containing 1 mL of culture medium and harvested at later time points. Such slices were maintained in vitro for periods up to day‐in‐vitro (DIV) 37 and occasionally longer. Three times a week 200 μL of medium was replaced.

Immunohistochemical procedures

For fixation, slices were incubated overnight in 4% formaldehyde in phosphate buffered saline (PBS). Primary antibody data are summarized in Supporting Information Table S2. For diamino‐benzidine (DAB) staining, HRP‐conjugated antibodies (1:100, DAKO Cytomation, Heverlee, Belgium) or biotinylated antibodies (1:400, Vector Labs, Burlingame, CA, USA) were used. Biotinylated second antibodies were followed by incubation with the Vectastain ABC kit (1:800, Elite PK 6100 Standard, Vector Labs). In DAB protocols, the slices were pretreated with 3% hydrogen peroxide in 10% methanol to destroy endogenous peroxidase activity. Secondary antibodies conjugated with Cy‐2 (1:100) or Cy‐3 (1:400, Jackson ImmunoResearch, West Grove, PA, USA), or with Alexa‐488 or Alexa‐594 (1:800, Life Technologies) were used for fluorescent detection. Sudan black (Brunswig Chemie, Amsterdam, Netherlands) was used to reduce auto‐fluorescence. In some fluorescent procedures, ethidium‐bromide homodimer (2 μL/mL, Life Technologies) was added during the last antibody incubation to visualize the nuclei. For staining with PCNA or Olig2, slices were pretreated with 10 mM citrate buffer (pH 6.0, 90°C) containing 0.05% Tween. Fluorescence images were recorded using a confocal laser scanning microscope (Zeiss 410 or Leica TSC SP5 II), DAB stained images were obtained using a Zeiss Axioplan 2. To check for possible background staining present in fluorescent images, negative control experiments were performed. In these controls, the first antibody was omitted and the Alexa‐conjugated second antibody was incubated according to the normal procedure. The slices were then washed and imaged. The signal in negative controls comes from antibodies trapped in the tissue in a non‐specific way and from incompletely blocked auto‐fluorescence (Supporting Information Figure S1). This background staining is somewhat nondescript but occasionally reveals dim features of the environment in which the always much brighter positively stained structures are embedded.

Rationale behind the chosen antibodies

Here, we briefly recapitulate the rationale for the antibodies used and refer to Verwer et al 31 for pertinent literature. Nestin and Vimentin were used as markers for astroglia precursors and reactive cells. In noncultured slices GFAP and calcium binding protein (S100β) stain astrocytes, while HLA and ionized calcium binding adaptor (Iba1) visualize microglia. Human transcription factor Olig2 is a marker of oligodendrocyte and astrocyte precursors. Neuronal nucleic protein (NeuN) and microtubule‐associated protein (MAP2) were used to visualize neurons. Tubulin βIII is often used as a pan‐neuronal marker 10, 12, but several papers have emphasized the specificity of the Tubulin βIII clone TuJ1 for neurons 6, 23. We have used four different Tubulin βIII antibodies (ie, three TuJ1 clones and one clone SDL.3D10). It may be noted that during mid‐gestational stages TuJ1 is co‐localized with GFAP and Nestin in human brain 4. For double staining protocols in combination with rabbit antibodies, the TuJ1 antibody from Covance was used, whereas the TuJ1 antibody from Sigma was used in combination with mouse antibodies (Supporting Information Table S2). We have reported before that Tubulin βI (TubβI) often stains microglial cells, and much less frequently neuronal somata, apical dendrites and perivascular cells in noncultured slices 31.

PAGE electrophoresis

For protein extraction, we used at least 12 tissue slices per condition (DIV 0 or DIV 28). The tissue was homogenized in 500‐μL suspension buffer (100 mM NaCl, 10 mM Tris‐HCl, 1 mM EDTA, 0.5% Triton X‐100 (pH 8.0) to which the following protease inhibitors (Sigma) were added: chymostatin (f.c. 20 μM), leupeptin (f.c. 10 μg/mL), antipain (f.c. 20 μM), pepstatin (f.c. 1 μM), PMSF (f.c. 100 μg/mL). We used the BCA Protein Kit Assay (Thermo Scientific, Breda, Netherlands) to determine the protein concentration. Per lane, a total of 10 μg of homogenate was loaded on an 8% polyacrylamide gel. Images of gels were obtained with an Odyssey infrared system (LI‐COR, Lincoln, NE, USA). The BenchMark Pre‐Stained Protein Ladder (Life Technologies) was used as molecular weight marker. β‐Actin was used as reference for total protein content loaded.

Reverse transcriptase quantitative polymerase chain reaction (RT‐qPCR)

RT‐qPCR was used to determine RNA expression levels of Tubulin subtype βI (TUBB1) and Tubulin subtype βIII (TUBB3). Total RNA was isolated using Trizol (Invitrogen), followed by DNAse I (Amplification Grade, Life Technologies) treatment and first strand cDNA synthesis using Superscript II (Life Technologies) reverse transcriptase. For each RNA isolation, 24 slices per time point (DIV 0, 10 or 28) were used. All steps were performed according to the manufacturer's recommendations. Primer information for TUBB1 and TUBB3 is summarized in Supporting Information Table S3. Amplification was detected with ABI 7300 real time PCR systems, using SYBR green master mix (Life Technologies). PCR reactions included controls omitting reverse transcriptase or template. We estimated crossing point (Cq) values, reaction efficiencies and the fluorescence threshold using version 12 of LinRegPCR shareware 22, which were then used to calculate corresponding RNA quantities. Eukaryotic Elongation factor 1 alpha (EF1α) and Glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) were used for normalization of the data. The primer data of EF1α and GAPDH have been detailed previously 31. These reference genes were selected based on expression stability among a group of putative reference genes 26, and on the proportion of explained variability regarding all target genes.

Statistical analysis

For statistical analysis and preparation of graphs, we used TIBCO Spotfire S+ software (version 8.2.0; TIBCO, Seattle, WA, USA). Differences in ¹⁰log‐transformed gene expression data were analyzed with paired t‐tests. The presented P‐values pertain to two‐sided alternative hypotheses. The Benjamini–Hochberg false discovery/coverage criterion (FDR/FCR) 1 was used to correct P‐values or confidence intervals for multiple testing. It may be noted that the P‐values presented here and those in a previous article 31 were adjusted together and are, therefore, all simultaneously valid.

Results

Co‐localization of neuronal markers

At DIV 0, the neuronal marker TuJ1 showed a relatively strong apparent background staining (see below) with vague outlines of apical dendrites and fibers, and less frequently neuronal somata (Figure 1A), which was confirmed by partial co‐localization with NeuN (Figure 1B,C). During the reactive process, neurons shrunk and lost much of their NeuN expression. At the same time also TuJ1 labeling of neurons disappeared, whereas it appeared in some cells with a reactive morphology (Figure 1D,E). A similar pattern was found with double staining of MAP2 and NeuN (Figure 1F,G). Thus, there was a partial overlap of TuJ1 and MAP2 in both non‐cultured and cultured tissue slices (Figure 1H,I). Western blots showed that protein expression of TuJ1, Tubulin βIII (clone SDL.3D10) and the βI subtype of Tubulin was reduced at DIV 28 (Supporting Information Figure S2), but less dramatically than NeuN and MAP2 31. Conversely, TUBB3 and TUBB1 gene expression was not noticeably affected by the injury process (Supporting Information Figures S3, Supporting Information Table S4).

Figure 1.

Co‐localization of neuronal marker NeuN, neuronal precursor cell marker Tubulin βIII (clone TuJ1) and adult neuronal marker MAP2 in resected human brain tissue. A. DAB staining of TuJ1 in a slice at DIV 0 showed a relatively high background with sometimes vague apical dendrites, fibers and pyramidal somata (arrows). B,C. Co‐localization of TuJ1 with NeuN at DIV 0 in neuronal cell bodies (asterisks) among single stained NeuN⁺ somata (arrow heads). D. At DIV 28, some shrunken neurons might still be vaguely TuJ1⁺ with a NeuN⁺ nucleus (asterisks), while others are only NeuN⁺ (arrow head). E. In other slices at DIV 28, TuJ1⁺ profiles of reactive cells (arrow) may be found among single stained NeuN⁺ neurons (arrow heads). F. In non cultured slices, many NeuN⁺ neurons exhibit variable co‐expression of MAP2 (asterisks). Some neurons seem to be devoid of MAP2 in the cell body (arrow heads). G. After some time in vitro, reactive cells may express MAP2 (arrow), while in neurons MAP2 was below the detection level (arrow heads). H. At DIV 0, MAP2⁺/TuJ1^‐‐ (arrow heads) and double labeled neurons (asterisks) and fibers (arrows) could be observed. I. At DIV 28, reactive cells were present that contained both TuJ1 and MAP2 (asterisk). d0: DIV 0. d28: DIV 28. g: green. r: red. Rabbit TuJ1 (Sigma) was used in A–E. Mouse TuJ1 (Covance) was used in H, I. Scale bars: A = 100 μm; C–F, H, I = 50 μm; G = 25 μm; B = 20 μm.

In cultured slices NeuN remained associated with adult (degenerating) neurons

NeuN was associated with pyramidal neurons and did not show any co‐localization with non‐neuronal markers in non‐neuronal cells at DIV 0 (Figure 2). However, occasionally, Nestin stained neurons were observed at DIV 0 (Supporting Information Figure S4), which could also co‐express NeuN (Figure 2A) and MAP2 (Figure 3A). Other non‐neuronal markers were not observed in neurons at DIV 0 (Figure 2). The separated localization of NeuN and non‐neuronal markers persisted when the slices were maintained in vitro and reactive cells appeared (Figure 2).

Figure 2.

The adult neuronal marker NeuN was limited to neurons both in cultured and non‐cultured tissue. A. Arrow heads indicate NeuN⁺ neurons at DIV 0. Nestin is usually present in blood vessels (not shown) and more or less prominent background. However, infrequently Nestin⁺/NeuN⁺ neurons were also observed (asterisks). B. At DIV 28, neurons are NeuN positive (arrow heads) and reactive cells show only Nestin (asterisks). C. At Div 0, Vimentin labeled blood vessels and NeuN neurons (arrow heads). D. At DIV 28, both reactive cells (asterisks) and resident astrocytes (arrows) could be labeled by Vimentin. Shrunken neurons might still be NeuN positive (arrow head). E. NeuN⁺ neurons (arrow heads) intermingled with S100β⁺ cells (arrows) at DIV 0. F. S100β⁺ reactive cells (asterisks) among shrunken NeuN⁺ neurons (arrow heads) at DIV 26. G. Double labeling of GFAP and NeuN at DIV 0. Arrows point at astrocytic cell bodies and arrow heads indicate neurons. H. Mildly hypertrophic astrocytes (arrows) with neurons (arrow heads) at DIV 28. It is unclear whether these astrocytes were already hypertrophic before the in vitro conditions. I. Iba1⁺ microglia (arrows) surrounding NeuN⁺ neurons (arrow heads) at DIV 0. J. Iba1⁺ reactive foam‐like cells (asterisks) with degenerated NeuN⁺ neurons (arrow heads) at DIV 26. K. Olig2⁺ cells (arrows) were often found in close association with neurons (arrow heads) at DIV 0. L. Shrunken neurons (arrow heads) and an inflated Olig2 cell (asterisk) at DIV 26. It is unclear whether the Olig2 cell was really reactive. bv: blood vessel. d0: DIV 0. d28: DIV 28. g: green. r: red. Scale bars: E–J = 50 μm; D, K, L = 25 μm; A–C = 20 μm.

Figure 3.

Combinations of MAP2 with non‐neuronal markers. A. In noncultured slices MAP2 (arrow heads) labeled a varying proportion of neuronal somata and apical dendrites. Infrequently, Nestin⁺/MAP2⁺ neurons (asterisks) were observed. B. At DIV 28, some reactive cells contained both Nestin and MAP2 (asterisks), while others were only Nestin positive (arrows). C. At DIV 0, Vimentin stained mainly blood vessels and MAP2 stained neurons (arrow head). D. Reactive cells that contained both MAP2 and Vimentin (asterisk) at DIV 28. E. MAP2⁺ neurons (arrow heads) with neighboring GFAP⁺ astrocytes (arrows) at DIV 0. F. At DIV 28, reactive cells could be GFAP⁺/MAP2⁺ (asterisks), or alternatively, mainly MAP2⁺ (arrow heads) or mainly GFAP⁺ (arrows). G. At DIV 0, S100β labeled astrocytes (arrows) and MAP2 labeled neurons (arrow heads). H. At DIV 28, single S100β⁺ (arrows) and S100β⁺/MAP2⁺ (asterisk) reactive cells were observed. I. MAP2⁺ neurons (arrow heads) and HLA⁺ microglia (arrows) present at DIV 0. J. At DIV 29, reactive cells were either MAP2⁺ (arrow heads) or HLA⁺ (arrows), suggesting that they constituted different subpopulations. K. Iba1⁺ microglia cells surrounding MAP2⁺ neurons at DIV 0. L. Iba1 (arrows) had a preference for foam‐like reactive cells, whereas MAP2 (arrow head) was not expressed by these cells. Consequently, they were not present in the same cells. M. TubβI⁺ microglia cells (arrow) and MAP2⁺ neurons at DIV 0. N. A reactive cell with a MAP2⁺ soma and MAP2⁺/TubβI⁺ processes. d0: DIV 0. d26: DIV 26. d28: DIV 28. g: green. r: red. Scale bars: C, F, H, I–K, N = 50 μm; A, B, E, G, L = 25 μm; D = 20 μm; M = 10 μm.

Reactive cells expressed MAP2 and Tubulin βIII

At DIV 0, MAP2 could be co‐localized with Nestin (Figure 3A), but not with Vimentin (Figure 3C), GFAP (Figure 3E), S100β (Figure 3G), HLA (Figure 3I), Iba1 (Figure 3K) or TubI (Figure 3M). Apparently, double labeled MAP2⁺ neurons (asterisk) in Figure 3I are due to HLA⁺ net‐like structures engulfing the neurons. After some time in vitro, reactive cells appeared and MAP2 was found to co‐localize in these reactive cells with Nestin (Figure 3B), Vimentin (Figure 3D), GFAP (Figure 3F), S100β (Figure 3H) and TubI (Figure 3N), but not with HLA (Figure 3J) and Iba1 (Figure 3L). We did not find co‐localization of TuJ1 with Nestin (Figure 4A) in noncultured slices. Further, at DIV 0, TuJ1 did not show co‐localization with Vimentin (Figure 4C), GFAP (Figure 4E), S100β (Figure 4G), HLA (Figure 4I), Iba1 (Figure 4K), TubI (Figure 4M) or Olig2 (Figure 4O). In reactive cells, TuJ1 was co‐localized with Nestin (Figure 4B), Vimentin (Figure 4D), GFAP (Figure 4F), S100β (Figure 4H), HLA (Figure 4J), TubßI (Figure 4N) and Olig2 (Figure 4P), but not with Iba1 (Figure 4L). It may be noted that the amount of TuJ1⁺ reactive cells in slices could vary considerably (Supporting Information Figure 5). MAP2⁺ reactive cells were usually less abundant and more vaguely stained than TuJ1 (not shown). Clone SDL.3D10 of Tubulin βIII can be found in reactive cells to the same extent as TuJ1 (not shown). During the time in vitro, the expression of MAP2 and Tubulin βIII by neurons was lost, whereas NeuN was merely reduced (Figure 1). In the white matter of some slices, small cells were observed whose morphology differed markedly from IVR and IVFL cells and that partly displayed co‐localization of TuJ1 with GFAP and S100β (Figure 5). The simple morphology might indicate that such cells constituted a population of precursor cells instead of reactive cells. In slices stained for TuJ1 and MAP2, but also for Nestin, bright puntate structures often stand out against the amorphous background. Comparison of Figures 1A,C–E,H, 2A and 4A,N with Supporting Information Figure S1 illustrates that the puntate structures are not part of the background. In noncultured slices, the punctate staining was generally more extensive than in cultured slices, which is in agreement with the decreased protein expression (Supporting Information Figure S2) and the loss of neuronal staining. Therefore, the punctate staining probably pertained to randomly oriented processes and fibers (Supporting Information Figure S5B).

Figure 4.

Combinations of TuJ1 with non‐neuronal markers. A. At DIV 0, both TuJ1⁺ and Nestin⁺ neurons were quite rare and usually only vaguely stained. A TuJ1⁺ neuron (arrow head) near two Nestin⁺ neurons (arrows). B. At DIV 28, reactive cells could be both TuJ1 and Nestin positive (asterisks). The arrow points to predominantly Nestin expressing cells (B). C. Two TuJ1⁺ neurons close to a Vimentin stained blood vessel at DIV 0. D. Vimentin⁺/TuJ1⁺ reactive cells at DIV 28. E. TuJ1⁺ neurons (arrow heads) and GFAP⁺ astrocytes (arrows) at DIV 0. F. A reactive cell double labeled for GFAP and TuJ1 (asterisk) at DIV 26. G. S100β⁺ astrocytes (arrows) and TuJ1⁺ neurons (arrow heads) at DIV 0. H. Reactive cells expressed variable amounts of S100β and TuJ1 (asterisks) at DIV 26. I. A TuJ1⁺ neuron (arrow head) with HLA⁺ microglia cells (arrows) at DIV 0. J. Reactive cells labeled for TuJ1 (arrow head), for HLA (arrow) and for both (asterisks) at DIV 28. K. TuJ1⁺ neurons (arrow heads) among Iba1⁺ microglia cells (arrows) at DIV 0. L. TuJ1⁺ reactive cells along blood vessels (arrow heads) and a Iba1⁺ foam‐like reactive cell (arrow) at DIV 26. M. At DIV 0, TubβI often stained microglia (arrows). Here, two TubβI⁺ microglial cells are shown with a TuJ1 stained neuron (arrow head). N. At DIV 28, TubβI could label both foam‐like reactive cells (arrows) and the other type of reactive cells (asterisks; here in combination with TuJ1). O. Double staining of Olig2 and TuJ1 at DIV 0. No TuJ1⁺ neurons were present in this cluster of Olig2 cells (arrows). P. At DIV 26, Olig2 and TuJ1 could both be present in the cytoplasm of small reactive cells (asterisks). However, also single TuJ1⁺ cells (arrow heads) and cells with TuJ1⁺ cytoplasm and an Olig2⁺ nucleus (arrows) were present. d0: DIV 0. d28: DIV 28. g: green. r: red. Mouse TuJ1 (Covance) was used in A–H, K, L, O, P. Rabbit TuJ1 (Sigma) was used in I, J, M, N. Scale bars: B, D‐L, N–P = 50 μm; A, M = 25 μm; C = 10 μm.

Figure 5.

Some small reactive cells in the white matter have a simple morphology. Double staining for GFAP (A) or S100β (B) with TuJ1 might suggest that these cells are neuronal precursors. Arrows indicate single labeled GFAP⁺ (A) or S100β⁺ (B) cells, arrow heads point to single TuJ1⁺ cells and asterisks mark GFAP⁺/TuJ1⁺ (A) or S100β⁺/TuJ1⁺ (B) cells. d26: DIV 26; g = green; r = red. Scale bars: A, B = 50 μm.

Discussion

Resected normal brain tissue is healthy, very valuable material to study human CNS cells. In principle, these samples could be instrumental in uncovering unknown features of human brain cells. However, the healthy status of the constituting cells also raises a problem: the ensuing injury response. This response develops over the first week after resection and, probably, has no consequences for acute electrophysiological studies. It might be argued that the culture conditions do not truly mimic the effects of injury in situ, which may involve hypoxia. Indeed, the oxygen level to which the tissue is exposed in vitro is even significantly higher than the normal physiological condition in situ, let alone putative hypoxic conditions 3. Still, the normoxic (20% oxygen) condition in vitro is substantially lower than the 95% oxygen applied in electrophysiological experiments. Nevertheless, with careful experimentation, resected tissue could well serve to study the role of hypoxia in brain injury responses. In this report, we describe what happens to three markers of neuronal cells and various markers of resident cells in resected tissue as a result of the in vitro conditions. First, we confirmed that in noncultured resected brain slices astrocytic markers (GFAP, S100β) are confined to cells with an astrocytic morphology, Vimentin is associated to a variable degree with blood vessels and present in some astrocytes while microglia markers (HLA, Iba1) are present in cells with a typical microglia morphology 31. Likewise, neuronal markers (NeuN, MAP2) were observed in neuronal structures. Occasionally, we found some Nestin stained pyramidal neurons with a normal morphology in noncultured neocortical slices. Co‐localization of Nestin with MAP2 and NeuN confirmed these cells to be neurons. Nestin positive cells with a pyramidal morphology can be present in vivo in neocortex regions damaged by intracranial electrodes8. Unfortunately, these authors could not confirm the neuronal nature of the cells using NeuN or neurofilament staining8. Faint Nestin⁺ neocortical neurons have also been observed at the outer boundary of stroke lesions15. Therefore, it seems likely that the occasional Nestin stained pyramidal neurons in noncultured resected slices were a consequence of a minute in vivo injury. The TuJ1 antibody is considered by some investigators as a specific marker for new‐born/immature neurons 10, while others used it to identify neurons in adult rats6. It has been asserted that TuJ1 labels only certain types of mature neurons in human brain and that co‐expression of TuJ1 with GFAP or Nestin in beyond late gestation would indicate maturational delay or mixed lineage23. In some human, slices at DIV 0, TuJ1 antibodies appeared to vaguely stain apical dendrites and cell bodies of pyramidal shaped neurons, and (although rarely) a few axonal fibers. Tubulin βIII (clone SDL.3D10) antibody showed the same staining pattern as the TuJ1 antibodies. At DIV 0, TubβI was predominantly expressed by microglia and perivascular cells and occasionally by neuronal structures31. With respect to the different cellular antecedents of Tubulin subtypes βI and βIII at DIV 0, it is remarkable that their gene and protein expression reacted in a similar way to the in vitro conditions.

Reactive cells appeared during the first week in vitro at varying rates31. These cells expressed reactivity markers and combinations of astrocytic and microglial markers. In this study, we observed that reactive cells expressed neuronal markers MAP2 and TuJ1, which were co‐localized with Nestin, Vimentin, GFAP and S100β. Besides, TuJ1 was also found in combination with HLA in IVR cells, but not with Iba1, which was exclusively localized to IVFL cells. Of course, the fact that MAP2 was not observed in association with HLA and Iba1 does not mean that such combinations are excluded. Interestingly, some IVR cells co‐expressed TubβI with MAP2 or TuJ1, whereas TubβI⁺ IVFL cells did not show such co‐localization. Thus, IVR cells may express Vimentin, Nestin, GFAP, S100β, MAP2, TuJ1, TubβI and HLA, whereas IVFL cells tend to have a smaller repertoire: Vimentin, S100β, TubβI, HLA, Iba1 and CD68. The cell types from which the reactive cells originated are presently unknown, but it seems likely that resident microglia and astrocytes are major contributors. The detection of neuronal markers further complicates resolving the nature of the reactive cells.

We did not observe any co‐localization of NeuN with markers of reactive cells, astrocyte markers or microglia markers in cultured slices. Moreover, NeuN was not observed in cells with a reactive morphology. However, as not all neurons are NeuN positive23, 25, 30 and NeuN expression could also be actively down‐regulated during differentiation to the reactive phenotype, we cannot conclude that reactive cells do not derive from neurons. Such a, possibly temporary, down‐regulation is not incompatible with the role of the proposed NeuN antibody target, Rbfox3, which seems to function as a splicing regulator in neural cell‐specific alternative splicing 11, 12. Interestingly, hybrid cells involved in a so‐called neuron‐to‐astrocyte transition have been reported 14. Here, early neurons or neuronal precursors have acquired the capacity to express glial markers, due to the culture conditions. It may be significant that in the cultures of Laywell et al 14, no hybrid cells co‐expressing NeuN and glial or reactive markers were detected. Therefore, we conclude that NeuN is a faithful marker of neurons that express it.

The co‐expression of TuJ1 and MAP2 with glial markers in reactive cells also demands an explanation. The first and most trivial explanation is that reactive cells, whatever their origin may be, have the capacity to express unexpected combinations of proteins to adapt to altered circumstances. For instance, it has been reported before that reactive astrocytes may express neuron‐specific enolase (NSE) and MAP2 7, 21. Moreover, astrocytes in the cortical plate of midgestational human fetal brain show co‐expression of Tubulin βIII with GFAP and Nestin 4. Thus, depending on the circumstances expression of MAP2 and Tubulin, βIII is not restricted to neurons. Alternatively, it could be that some reactive cells are attempting to generate new neurons. It has been suggested that injury expands the repertoire of precursor cells to become neurogenic 10 and the combined expression of different markers might be a sign of multipotency. Continuous neurogenesis in the adult human brain has been established in the subventricular zone along the lateral ventricles 13 and the subgranular zone of the hippocampal dentate gyrus 5. It has been proposed that small capillaries in the subgranular zone of the adult rodent hippocampus provide an instructive niche for neurogenesis to occur 19. While the reactive cells in the human slices do in fact predominantly emerge along blood vessels, it must also be conceded that resected tissue concerns neocortex and that the blood vessels involved usually are larger than such capillaries. But rare residual neuronal precursors may also persist in the parenchyma of the temporal cortex of epilepsy patients 20. Indeed, it has been shown that resident glial or precursor cells in the neocortex of adult rodents can replace injured neurons in vivo after a targeted lesion 2, 16. In these latter studies, a possible association of such precursors or glial cells with the vasculature has not been investigated. The observations by Magavi et al 16 may be the result of the exquisite experimental strategy they used, because others have reported that astrocytes under in vitro conditions require highly careful reprogramming to transform into functional neurons 9. Alternatively, astrocytes or precursor cells in vivo may dispose of an innate reprogramming strategy, comparable to that used by Heinrich et al 9, that allows them to attain the replacement observed in vivo by Magavi et al 16. In the white matter of some slices, we observed clusters of small cells with a simple morphology some of which expressed both GFAP and TuJ1 or S100β and TuJ1. These cells might constitute a population of precursors or mixed lineage cells23. With respect to this possibility, it would be of interest to know the proportion of reactive cells expressing both astrocytic and neuronal markers. Of course, complementary questions involving quantitative estimates of cells expressing both neuronal and microglial markers or even combinations of neuronal, astrocytic and microglial markers would be equally important. A systematic investigation into this matter would require a dedicated collection of tissue. Lacking such a collection, we could only focus on revealing representative examples of co‐localization of neuronal and non‐neuronal markers in reactive cells and, in case of absence, search for separately labeled neighboring cells. We have the impression that on average co‐expression of neuronal and asrtocytic markers is common, but highly variable.

Our work clearly shows that the reactive cells in our slice preparations and culture conditions cannot accomplish the transition to bona fide functional neurons within roughly 4 weeks without extra stimuli. However, exciting opportunities for the development of therapeutic strategies would ensue if reprogramming using viral vector mediated expression of transcription factors 9 could induce the true conversion of some reactive cells into neurons. Although less exciting, strategies that can prevent or ameliorate the degeneration of neurons and the emergence of reactive cells are also crucial and can be studied using our model system.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher's web‐site:

Figure S1. Negative control stainings. Negative controls were obtained by omitting the primary, but normally incubating with a secondary fluorescent antibody. A. Slice incubated with Donkey‐anti‐Mouse antibody conjugated to Alexa‐488. B. Slice incubated with Donkey‐anti‐Rabbit antibody conjugated to Alexa‐594. These controls show vaguely fluorescent non‐descript structures and autofluorescence that may be observed in thick human slices. Note that the positive signal in the other figures in this report is much brighter than the signal displayed here. Scale bars: A, B = 50 μm.

Figure S2. Western blot analysis of Tubulin subtypes βI and βIII. Tubulin βIII was detected using several different antibodies. It appeared that protein expression of both subtypes decreased during the in vitro conditions. White signal in the bands indicates over‐exposure during imaging. The ß‐Actin bands serve as indicators of the amount of protein loaded. We refer to Supporting Information Table S1 for patient data corresponding to the E numbers. Benchmark markers were used for approximate size indication.

Figure S3. Gene expression levels of Tubulin subtypes βI (TUBB1) and βIII (TUBB3). The expression levels of both subtypes at DIV 10 and DIV 28 were not significantly different from DIV 0. Mean values of the respective differences and 95% confidence intervals are shown. The confidence intervals were adjusted for multiple testing using the Benjamini–Yekutieli procedure (1) together with the gene expression levels reported in the previous article (31). The scale of the x‐axis is comparable to that of Figure 2 in reference (31).

Figure S4. Nestin⁺ neurons in noncultured neocortex slices. Infrequently, neurons were found to be labeled by Nestin (arrows). d0 = DIV 0. bv = blood vessel. Scale bar: 30 μm.

Figure S5. Illustration of the variability in the number of TuJ1⁺ reactive cells. TuJ1 was visualized using the DAB procedure. A. An overview of a slice from patient E8 at DIV 25. Here, most TuJ1⁺ reactive cells were present in a small area, while the rest of the slice only contained a few isolated cells. B. In a slice from patient E9, the TuJ1⁺ reactive cells were much more numerous and evenly distributed. See Supporting Information Table S1 for patient data. The number of TuJ1⁺ reactive cells in slices from other patients ranged from zero to roughly the amount shown in (B). It should be noted that the thickness of the slices is not uniform due to compression of the tissue during chopping and to uneven flattening of the slices during culture. The background is darker at thicker parts of the slices. However, zooming in on panel (B) shows that the punctate staining clearly pertains to fibers (arrows) and processes (arrow heads). Scale bars: A, B = 500 μm.

Table S1. Clinico‐pathological data of resected brain tissue from patients used in this study.

Table S2. List of applied primary antibodies.

Table S3. Primer data for Homo sapiens genes used in this study.

Table S4. Changes in gene expression during the in vitro conditions.