I find the solution eventually, Anyway, share with others, hope it is helpful in your routine analysis. Also, I found gridGraphics and grid is very useful for complicated figure preparation.

library(gridGraphics)
library(grid)

grab_grob <- function(){
  grid.echo()
  grid.grab()
}

data<-read.table("heatmap.input.matrix.data.txt")

library(gplots)
gl <- lapply(1:ncol(data), function(i){
  heatmap.2(data.matrix(data[[i]]))
  grab_grob()
})

grid.newpage()
library(gridExtra)
grid.arrange(grobs=gl, ncol=5, clip=TRUE)