If you only needed to have the < and > encoded again, setting the flag JSON_HEX_TAG while re-encoding the data, would do.

If you want something that leaves all such unicode escape sequences in place … then you need to replace each \uXXXX with \\u0075XXXX first, then manipulate and re-encode your data – and then replace \u0075 with just u again at the end:

$json = preg_replace('#\\\\u([0-9A-F]{4})#', '\\\\\u0075$1', $json);
$dec = json_decode($json, true);
$enc = preg_replace('#\\\\u0075#', 'u', json_encode($dec));

If you do it like this, then you of course got to be aware, that your data now contains literal \uXXXX sequences. Meaning, if you wanted to replace < in one of the object’s string values now, you can’t search for <, but would need to search for \u003C.