One solution is to adjust globally the x-axis so that it fits the data range:
par(xaxs="i")
heatmap.2(z, trace='none', dendrogram='row', Colv=F, scale='row',
hclust=hclustfunc, distfun=distfunc, col=greenred(256), symbreak=T,
margins=c(10,20), keysize=0.5, labRow=data$Gene.symbol,
lwid=c(1,0.05,1), lhei=c(0.03,1), lmat=rbind(c(5,0,4),c(3,1,2)),
RowSideColors=as.character(clusters))