If you have the co-ordinates of your gene e.g. your genes are in .bed format then fetch their upstream sequences with a simple awk command:

If 4th column is strand in your file

awk '{ if ($4=="+") print $1,$2-5000,$2; else print $1,$3,$3+5000}' OFS="\t" genes.bed > upstream.bed

Now either you can look for ChIP-seq data in you cell line of interest and see if the peaks are there in your region of interest by a simple

intersectBed -a peaks.bed -b upstream.bed -f 0.5

or

if you want to make predictions then you can fetch the sequences from genome

fastaFromBed is powerful tool to do so

And then just align your sequences and run meme motif scan, once you get the motifs, compare the matrix with the JASPAR or TRANSFAC ones.

hth