This is not a Client issue, but an IBM limit:
DSNTIAUL (and DSNTEP2/DSNTEP4/DSNTIAD) all only support SYSIN first 72 bytes.
Beyond that the data must be split to multiple lines with no data in column 72.
This is IBM limit. In CNTL datasets, "Normally" columns 72-80 are used for numbering.
As it turns out, the Subsetting column names end exactly on column 72. As a result, following the above requirements, there is an error being generated.