In the human brain, many differentially expressed genes were observed from 2 to 6 days and from 9 to 20 days after stroke, with the majority being upregulated. The number of deregulated genes declined during 26 to 37 days after stroke, indicating that dynamic changes in gene expression occur during the first days to few weeks in the human postischaemic brain. In the rat brain, few differences were observed at 1 hour, while the number of differentially expressed genes steadily increased with time after MCAO, with a peak after 3 days, supporting the concept of active mechanisms initiated during the acute phase after experimental stroke and lasting for several days. The number of upregulated genes gradually increased, peaking at 3 days, while downregulated genes were detected 24 h after MCAO and increased dramatically until the final measured time-point at 21 days (Figure 1).
The limitations of post-mortem brain samples in cDNA microarray analysis concern the small sample size and potential low quality and the genetic heterogeneity and diversity in terms of age, sex and previous medical history within a group of patients [28, 29]. We found that analysis of postischaemic gene expression using a cDNA microarray can allow identification of known and novel transcriptional events, molecular participants and signalling mechanisms in cerebral ischaemia as previously suggested, but can also detect differences in gene expression between distinct organisms.
The present gene expression profile study is the first large-scale microarray report showing altered expression of several genes following human stroke. These included genes participating in transcription, apoptosis, inflammation and neuroprotection. Many genes/proteins previously shown to be deregulated following stroke were reported in our study too e.g. IL-10 [30, 31], PDGF , STAT3 [33, 34], MAPK1/2 . To test whether our microarray analysis could predict novel candidate genes involved in the cerebral response to ischaemia with possible functional importance and significance in stroke-induced neuronal damage, we measured protein expression and cellular localisation for three induced genes, INI1, PAK1 and MMP11. They were chosen because they showed at least 2-fold mRNA induction and there was no prior published evidence implicating them in human cerebral ischaemia.
PAK1 is a downstream Rac effector and a major cyclin-dependent kinase 5 (Cdk5) substrate and target that co-localizes with p35/Cdk5 at neuronal peripheries. P35/Cdk5 causes PAK1 hyperphosphorylation, which results in PAK1 down-regulation and is likely to have an impact on the dynamics of the reorganization of the actin cytoskeleton in neurons during dendrite development . Based on this evidence, these authors proposed the existence of a neuron-specific signalling complex involving Cdk5/p35-PAK1 that inhibits PAK1 activity in neurons. We have recently provided evidence for a potential role of Cdk5/p35 in the response to ischaemic injury as we showed association of Cdk5 with nuclear damage, by demonstrating co-expression of Cdk5 in TUNEL-positive neurons following human stroke and in propidium iodide-positive human foetal neurons following OGD . Here, we have reported for the first time an upregulation in PAK1 protein levels in human and rat brain samples following MCAO and in HFN following oxygen-glucose deprivation. Although in the animal model PAK1 protein levels returned to normal 3 days following stroke, some patients showed elevated levels for PAK1 at later time-points too. In both human and the animal model, neurons were the predominant type of cells stained positive for PAK1.
MMP11 or stromelysin-3 (ST3), first isolated as a breast cancer-associated protease, is not expressed in the majority of normal adult organs but is expressed during a number of pathological processes, including wound healing and atherosclerotic lesions [38, 39]. Although other metalloproteinases have been studied extensively following stroke [40, 41], there is no report of the expression of MMP11 following stroke in vivo or in vitro. Here we report an increase in protein levels of MMP11 following stroke in both human and rat brain, although the increase seen in man remained elevated much longer. Although MMP11 shares many similarities with other MMPs, it also differs in that it exhibits anti-apoptotic properties, a first-known activity for a MMP . Moreover, although it is expressed in many processes involving tissue remodelling, cell migration and cell death, the pathways through which it participates in pathogenesis remain unclear, largely due to the lack of information on its substrates in vivo .
INI1 is a tumour suppressor gene, thought to exert its tumour suppressor function by mediating cell cycle arrest . It was initially identified as a human homolog of yeast transcriptional activator SNF5 that binds to the HIV-1 integrase and stimulates its DNA-joining activity . Brains of AIDS patients had been shown to manifest neuronal injury and apoptotic-like cell death raising the question about the way HIV-1 resulted in neuronal damage, since neurons themselves are very rarely infected by the virus . Adler et al.  also reported an association of the human SNF5/INI1 protein with growth arrest and DNA damage-inducible protein 34 (GADD34) that mediates growth arrest and apoptosis in response to stress signals [48, 49]. Our study is the first to suggest a potential role for INI1 in pathways activated after stroke with a possible role in brain injury. However, in the animal model study, INI1 levels remained unchanged following stroke. The reason for this discrepancy warrants further studies.
Many experimental trials of stroke therapies have failed to translate to human clinical trials and one possible way to improve the success rate can be through comparative genomics. As it has been recently commented, it is very surprising that the exciting developments observed in basic and clinical stroke research over the past two decades have occurred in parallel, with too little direct translation between bench and bedside . Here, we have provided substantial evidence that, although the available animal models of MCAO may well be suitable to study the pathophysiological changes following the occlusion of a cerebral vessel, they may not entirely reflect the pathophysiological process through which stroke evolves in humans. The species difference is one of the main reasons accounting for the lack of success of bench to bedside translation in the stroke area. Limitations of our study include the fact that early acute phase changes in gene expression may have been missed since genes induced and returning to normal during the first 48 hours post-ischaemia in man could not have been detected. Moreover, since we analyzed pooled RNA samples, small changes in gene expression occurring in a minority of the samples may have been missed. However, there was only a small overlap of our results with prior studies in experimental stroke involving brain tissue, and the successful identification of novel ischaemia-related genes reported here suggests that performing a further study using whole genome microarrays would be valuable.