How do I filter columns in a dataframe by the keys in another dataframe

My first dataframe is

obj1 obj2 obj3
 01   02   03
 02   03   04

another dataframe is:

col1 col2 col3
obj1  M    N
obj2  M    T
obj3  M    N

I only want the columns in the first dataframe who are marked with

{'col2':M, 'col3':N}

The wanted result is

obj1 obj3
 01   03
 02   04

I can't seem to achieve this in a decent way.

One idea is convert dictionary to one row DataFrame and merge, because no parameter on it merging by intersection of all columns names (in helper DataFrame by keys of dictionary):

d = {'col2':'M', 'col3':'N'}
vals = pd.DataFrame([d]).merge(df2)['col1']

Or use DataFrame.query with dict comprehension:

vals = df2.query(' & '.join(['{}=={}'.format(i,repr(j)) 
                              for i, j in d.items()]))['col1']

And then filter by Index.isin and DataFrame.loc, : means all rows:

df = df1.loc[:, df1.columns.isin(vals)]
print (df)
  obj1 obj3
0   01   03
1   02   04